[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth created this revision.
paulkirth added reviewers: phosek, leonardchan, jakehehrlich, mcgrathr.
Herald added subscribers: cfe-commits, kristof.beyls, javed.absar, mgorny.
Herald added a project: clang.

**Overview**:
This patch contains a prototype of the basic functionality of clang-misexpect 
in the PGO pipeline.  clang-misexpect is a proposed clang-tool that can report 
potentially incorrect usage of __builtin_expect() by comparing the developer's 
annotation against a collected PGO profile.  A more detailed proposal and 
discussion appears on the CFE-dev mailing list 
(http://lists.llvm.org/pipermail/cfe-dev/2019-July/062971.html)

This patch adds the basic checking mechanisms to the compiler in the CodeGen 
library as a set of warnings usable when compiling with PGO. Once the logic in 
this patch is verified, it can be used to as the basis for a standalone tool.

This patch adds a new flag -fmisexpect to clang, and adds additional checks for 
branches and switch statements when branch weights are assigned in 
clang/lib/CodeGen/CodeGenFunction.cpp & clang/lib/CodeGen/CGStmt.cpp.

The bulk of the checking logic is implemented in 
clang/lib/CodeGen/MisExpect.cpp.  It has some minor changes to some of the 
logic in CGStmt.cpp to properly map the profiling counters to their concrete 
values in the case statements for switches.

The patch also provides a number of  lit based tests for various usage of 
__builtin_expect() for branches and switch statements.

**Details**:

The strategy for MisExpect checks is straightforward: when __builtin_expect() 
is used, then compare the relevant profile counter against a threshold value, 
and emit a warning if we've found a mismatch (i.e. the profile counter is 
outside the acceptable range).

For conditional statements this is simple.  We can determine whether the 
profile count agrees with the annotation via direct comparison w/ the 
appropriate hot/cold thresholds.

For switch statements the situation is slightly more complex due to the way 
profiling counters are assigned.

Each case statement in the switch body will get its own counter, except when 
the cases are empty fall-throughs, in which case they will share a counter.

So, in the case where:

  switch(x){
  case 1:
  case 2:
break;
  case default:
break;
  };

Here, values 1 & 2 will share a single counter, and all other cases will share 
a counter for the default case.

However, in the following example:

  switch(x){
  case 1:
   x = x+1;
  case 2:
break;
  case default:
break;
  };

In the example above, values 1 & 2 will each have their own profiling counter 
instead of sharing one, even though case 1 falls through to case 2.  The 
default case stays the same with a shared counter for every value not present 
in the switch body.

To be align with developer expectations we treat these empty fall-throughs 
conservatively for the purpose of generating warnings, i.e. if the expected 
value results in the hottest branch being executed we will not emit the 
warning, even though the strict semantics of __builtin_expect() are slightly 
different. We support this by maintaining a mapping from the concrete value in 
the switch arm back to its relevant profile counter.

In the case where the profile counters are not shared, we always do a direct 
check using the relevant profile counter.

Right now MisExpect only ensures that the hottest arm of the switch statement 
is the one the developer chosen, without reference to any particular threshold. 
It is arguable that we should have an additional check against a threshold, 
similar to what is done for conditional statements.

**Heuristics**: 
For branch 'hotness' I've used the existing thresholds for determining 
FFA_Hot/FFA_Cold function annotations. This is a dubious choice for the 
thresholds, but are a suitable heuristic for the prototype. Feedback on a more 
appropriate set of threshold values is welcome and desirable.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D65300

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/CodeGenFunction.h
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch.c

Index: clang/test/Profile/misexpect-switch.c
==

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 211831.
paulkirth added a comment.

Refactors some debug code to be more centralized and cleans up some comments.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/CodeGenFunction.h
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch.c

Index: clang/test/Profile/misexpect-switch.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch.c
@@ -0,0 +1,68 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 1;
+  switch (__builtin_expect(condition, 0)) { //expected-warning {{Current PGO counters disagree with the use of __builtin_expect().}}
+  case 0:
+val += sum(arry, arry_size);
+break;
+  case 1:
+  case 2:
+  case 3:
+  case 4:
+val += random_sample(arry, arry_size);
+break;
+  default:
+__builtin_unreachable();
+  } // end switch
+}   // end inner_loop
+  } // end outer_loop
+
+  return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-default.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-default.c
@@ -0,0 +1,68 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 5;
+  switch (__builtin_expect(condition, 6)) { //expected-warning {{Current PGO counters disagree with the use of __builtin_expect().}}
+  case 0:
+val += sum(arry, arry_size);
+break;
+  case 1:
+  case 2:
+  case 3:
+  case 4:
+val += random_sample(arry, arry_size);
+break;
+  default:
+__builtin_unreachable();
+  } // end switch
+}   // end inner_loop
+  } // end outer_loop
+
+  return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-no-warning-without-flag.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-no-warning-without-flag.c
@@ -0,0 +1,51 @@
+// Test that misexpect detects mis-annotated branches
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-branch.proftext -o %t.profdata

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

In D65300#1601733 , @xbolva00 wrote:

> Nice work!


Glad to hear you like it.

> It would be nice if we also have “-fsuggest-expect” so we could fix perf 
> issues thanks to PGO counters even for non-PGO builds. What do you think?

Supporting suggestions is something we're planning to do in the future. Non-PGO 
builds are even the main motivation. I wanted to get this logic correct first, 
and make sure that we're handling all the odd edge cases before making new 
suggestions. I feel like our approach is really straightforward, but some 
things in clang are spread out in surprising ways or have interactions I've 
been surprised to find. That said, if finding problematic usage is done 
correctly, then reversing the logic to make a suggestions about useful 
annotations should be pretty easy.

I still need to give some thought to how to express the right balance between 
execution count and frequency when suggesting new annotations, so that the user 
won't be overwhelmed with suggestions. I'm not sure that branch weights by 
themselves are sufficient for that use-case, but we should be able to come up 
with something that behaves reasonably well without too much trouble.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 211853.
paulkirth added a comment.

Add missing test for switch statements when the expected value is not a compile 
time constant.

Make sure that when the expected value cannot be evaluated, we do not issue any 
warnings or errors


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/CodeGenFunction.h
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch.c

Index: clang/test/Profile/misexpect-switch.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch.c
@@ -0,0 +1,68 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 1;
+  switch (__builtin_expect(condition, 0)) { //expected-warning {{Current PGO counters disagree with the use of __builtin_expect().}}
+  case 0:
+val += sum(arry, arry_size);
+break;
+  case 1:
+  case 2:
+  case 3:
+  case 4:
+val += random_sample(arry, arry_size);
+break;
+  default:
+__builtin_unreachable();
+  } // end switch
+}   // end inner_loop
+  } // end outer_loop
+
+  return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-nonconst.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-nonconst.c
@@ -0,0 +1,69 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+// expected-no-diagnostics
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 1;
+  switch (__builtin_expect(condition, rand())) {
+  case 0:
+val += sum(arry, arry_size);
+break;
+  case 1:
+  case 2:
+  case 3:
+  case 4:
+val += random_sample(arry, arry_size);
+break;
+  default:
+__builtin_unreachable();
+  } // end switch
+}   // end inner_loop
+  } // end outer_loop
+
+  return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-default.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-default.c
@@ -0,0 +1,68 @

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 212492.
paulkirth added a comment.

Update diff to have proper context on Phabricator


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/CodeGenFunction.h
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c

Index: clang/test/Profile/misexpect-switch.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch.c
@@ -0,0 +1,68 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 1;
+  switch (__builtin_expect(condition, 0)) { //expected-warning {{Current PGO counters disagree with the use of __builtin_expect().}}
+  case 0:
+val += sum(arry, arry_size);
+break;
+  case 1:
+  case 2:
+  case 3:
+  case 4:
+val += random_sample(arry, arry_size);
+break;
+  default:
+__builtin_unreachable();
+  } // end switch
+}   // end inner_loop
+  } // end outer_loop
+
+  return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-only-default-case.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-only-default-case.c
@@ -0,0 +1,62 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch-default-only.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+// expected-no-diagnostics
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main()
+{
+init_arry();
+int val = 0;
+
+int j, k;
+for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+unsigned condition = rand() % 1;
+switch (__builtin_expect(condition, 0)) {
+default:
+val += random_sample(arry, arry_size);
+break;
+}; // end switch
+}  // end inner_loop
+}  // end outer_loop
+
+return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-nonconst.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-nonconst.c
@@ -0,0 +1,69 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profd

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 213059.
paulkirth marked 2 inline comments as done.
paulkirth added a comment.

Fix typo in comments


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/CodeGenFunction.h
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c

Index: clang/test/Profile/misexpect-switch.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch.c
@@ -0,0 +1,68 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 1;
+  switch (__builtin_expect(condition, 0)) { //expected-warning {{Current PGO counters disagree with the use of __builtin_expect().}}
+  case 0:
+val += sum(arry, arry_size);
+break;
+  case 1:
+  case 2:
+  case 3:
+  case 4:
+val += random_sample(arry, arry_size);
+break;
+  default:
+__builtin_unreachable();
+  } // end switch
+}   // end inner_loop
+  } // end outer_loop
+
+  return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-only-default-case.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-only-default-case.c
@@ -0,0 +1,62 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch-default-only.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+// expected-no-diagnostics
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main()
+{
+init_arry();
+int val = 0;
+
+int j, k;
+for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+unsigned condition = rand() % 1;
+switch (__builtin_expect(condition, 0)) {
+default:
+val += random_sample(arry, arry_size);
+break;
+}; // end switch
+}  // end inner_loop
+}  // end outer_loop
+
+return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-nonconst.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-nonconst.c
@@ -0,0 +1,69 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// 

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth marked 2 inline comments as done.
paulkirth added inline comments.



Comment at: clang/lib/CodeGen/MisExpect.cpp:82
+  if (ExpectedTrueBranch) {
+const llvm::BranchProbability HotFunctionThreshold(1, 100);
+Scaled = HotFunctionThreshold.scale(CurrProfCount);

phosek wrote:
> Are these thresholds defined anywhere as constants?
These thresholds come from [[ 
https://github.com/llvm/llvm-project/blob/7ea131c20c113b78085301a1acd7b28884ec131e/llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp#L1084
 |  PGOInstrumentation.cpp:1084 ]]

I will update with a reference to the code that this comes from, but, as noted 
in the TODO we need a better heuristic. 



Comment at: clang/lib/CodeGen/MisExpect.cpp:140-146
+void EmitMisExpectWarning(const CallExpr *Call, CodeGenModule &CGM) {
+  SourceLocation ExprLoc = Call->getBeginLoc();
+  unsigned DiagID = CGM.getDiags().getCustomDiagID(
+  DiagnosticsEngine::Warning, "Current PGO counters disagree with "
+  "the use of __builtin_expect().");
+  CGM.getDiags().Report(ExprLoc, DiagID);
+}

lebedev.ri wrote:
> This is rather undescriptive.
> Can you output some more useful info?
Do you have a suggestion about what feedback would be more useful?

My initial thought with the somewhat generic message was to simply point out 
that this usage looked problematic, and let the developer investigate. I wasn't 
sure we wanted to expose details of the internal heuristic to the user by 
reporting the internal thresholds.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 213164.
paulkirth added a comment.

Address feedback from review

1. Fixed comments in License headers
2. removed excess {}
3. Changed conditional assignment to ternary operation
4. Moved local functions into anonymous namespace
5. Added reference to the origin of the threshold values in the clang source 
code
6. Updated warning to have more useful text
7. Fixed whitespace
8. Updated unqualified use of llvm::None
9. Updated tests with new warning text


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/CodeGenFunction.h
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c

Index: clang/test/Profile/misexpect-switch.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch.c
@@ -0,0 +1,68 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 1;
+  switch (__builtin_expect(condition, 0)) { // expected-warning-re {{Potential performance regression from use of __builtin_expect(): Annotation was correct on {{.+}}% of profiled executions.}}
+  case 0:
+val += sum(arry, arry_size);
+break;
+  case 1:
+  case 2:
+  case 3:
+  case 4:
+val += random_sample(arry, arry_size);
+break;
+  default:
+__builtin_unreachable();
+  } // end switch
+}   // end inner_loop
+  } // end outer_loop
+
+  return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-only-default-case.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-only-default-case.c
@@ -0,0 +1,62 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch-default-only.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+// expected-no-diagnostics
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main()
+{
+init_arry();
+int val = 0;
+
+int j, k;
+for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+unsigned condition = rand() % 1;
+switch (__builtin_expect(condition, 0)) {
+default:
+val += random_sample(arry, arry_size);
+break;
+}; // end switch
+}  // end inner_loop
+}  // end outer_loop
+
+return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 213167.
paulkirth added a comment.

Update documentation and fix typos


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/CodeGenFunction.h
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c

Index: clang/test/Profile/misexpect-switch.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch.c
@@ -0,0 +1,68 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 1;
+  switch (__builtin_expect(condition, 0)) { // expected-warning-re {{Potential performance regression from use of __builtin_expect(): Annotation was correct on {{.+}}% of profiled executions.}}
+  case 0:
+val += sum(arry, arry_size);
+break;
+  case 1:
+  case 2:
+  case 3:
+  case 4:
+val += random_sample(arry, arry_size);
+break;
+  default:
+__builtin_unreachable();
+  } // end switch
+}   // end inner_loop
+  } // end outer_loop
+
+  return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-only-default-case.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-only-default-case.c
@@ -0,0 +1,62 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch-default-only.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+// expected-no-diagnostics
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main()
+{
+init_arry();
+int val = 0;
+
+int j, k;
+for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+unsigned condition = rand() % 1;
+switch (__builtin_expect(condition, 0)) {
+default:
+val += random_sample(arry, arry_size);
+break;
+}; // end switch
+}  // end inner_loop
+}  // end outer_loop
+
+return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-nonconst.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-nonconst.c
@@ -0,0 +1,69 @@
+// Test that misexpect detects mis-ann

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 213379.
paulkirth edited the summary of this revision.
paulkirth added a comment.

Update threshold values to match those assigned when lowering __builtin_expect 
intrinsic.

I've modified the branch probability to match the probability assigned in 
LowerExpectIntrinsics.cpp

This is still debatable, but seems like a reasonable default. My next patch 
will make the probability threshold assignable through the command line, and 
extend the probability calculation into switch statements. This way users can 
determine how hot is hot enough, and otherwise get the same behavior that 
builtin expect will use anyway.

I also re applied a fix to a typo that I overwrote in an earlier patch.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/CodeGenFunction.h
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c

Index: clang/test/Profile/misexpect-switch.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch.c
@@ -0,0 +1,68 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 1;
+  switch (__builtin_expect(condition, 0)) { // expected-warning-re {{Potential performance regression from use of __builtin_expect(): Annotation was correct on {{.+}}% of profiled executions.}}
+  case 0:
+val += sum(arry, arry_size);
+break;
+  case 1:
+  case 2:
+  case 3:
+  case 4:
+val += random_sample(arry, arry_size);
+break;
+  default:
+__builtin_unreachable();
+  } // end switch
+}   // end inner_loop
+  } // end outer_loop
+
+  return 0;
+}
+
+int sum(int *buff, int size) {
+  int total = 0;
+  int i = 0;
+  for (i = 0; i < size; ++i) {
+total += buff[i];
+  }
+  return total;
+}
+
+int random_sample(int *buff, int size) {
+  int total = 0;
+  int i;
+  for (i = 0; i < size; ++i) {
+if (rand() % 5 == 0)
+  total += buff[i];
+  }
+
+  return total;
+}
Index: clang/test/Profile/misexpect-switch-only-default-case.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-only-default-case.c
@@ -0,0 +1,62 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch-default-only.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+// expected-no-diagnostics
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+void init_arry() {
+  int i;
+  for (i = 0; i < arry_size; ++i) {
+arry[i] = rand() % 10;
+  }
+}
+
+int main()
+{
+init_arry();
+int val = 0;
+
+int j, k;
+for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+unsigned condition = rand() % 1;
+switch (__builtin_expect(condition, 0)) {
+default:
+val += random_sample(arry, arry_size);
+break;
+}; // end switch
+}  // end inner_loop
+

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 213725.
paulkirth added a comment.

Use existing LLVM code for mapping case literals to their case arms.

This update refactors a great deal of the implementation and test code.

1. Removes the CaseMap data structure completely. LLVM already maintains a 
mapping of the constants to their case arm, so there is no reason to duplicate 
that logic. This also minimizes the changes impacting existing Clang/LLVM 
components.
2. Cleans up debug printing. Without the CaseMap printing debug output can be 
simplified.
3. Minimizes the code in the end-to-end tests & test profiles.
4. Improves formatting, white space, and comments.




Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c

Index: clang/test/Profile/misexpect-switch.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch.c
@@ -0,0 +1,41 @@
+// Test that misexpect detects mis-annotated switch statements
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+void init_arry();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 1;
+  switch (__builtin_expect(condition, 0)) { // expected-warning-re {{Potential performance regression from use of __builtin_expect(): Annotation was correct on {{.+}}% of profiled executions.}}
+  case 0:
+val += sum(arry, arry_size);
+break;
+  case 1:
+  case 2:
+  case 3:
+break;
+  default:
+val += random_sample(arry, arry_size);
+break;
+  } // end switch
+}   // end inner_loop
+  } // end outer_loop
+
+  return 0;
+}
Index: clang/test/Profile/misexpect-switch-only-default-case.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-only-default-case.c
@@ -0,0 +1,35 @@
+// Test that misexpect emits no warning when there is only one switch case
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch-default-only.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+// expected-no-diagnostics
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+void init_arry();
+
+const int inner_loop = 1000;
+const int outer_loop = 20;
+const int arry_size = 25;
+
+int arry[arry_size] = {0};
+
+int main() {
+  init_arry();
+  int val = 0;
+
+  int j, k;
+  for (j = 0; j < outer_loop; ++j) {
+for (k = 0; k < inner_loop; ++k) {
+  unsigned condition = rand() % 1;
+  switch (__builtin_expect(condition, 0)) {
+  default:
+val += random_sample(arry, arry_size);
+break;
+  }; // end switch
+}// end inner_loop
+  }  // end outer_loop
+
+  return 0;
+}
Index: clang/test/Profile/misexpect-switch-nonconst.c
===
--- /dev/null
+++ clang/test/Profile/misexpect-switch-nonconst.c
@@ -0,0 +1,43 @@
+// Test that misexpect emits no warning when switch condition is non-const
+
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+
+// expected-no-diagnostics
+int sum(int *buff, int size);
+int random_sample(int *buff, int size);
+int rand();
+void init_arry();
+
+const int inne

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth created this revision.
paulkirth added reviewers: phosek, leonardchan, jakehehrlich, mcgrathr.
Herald added subscribers: llvm-commits, cfe-commits, hiraditya, mgorny.
Herald added projects: clang, LLVM.
paulkirth added a parent revision: D65300: [clang] [CodeGen] clang-misexpect 
prototype for compiler warnings.

This patch contains a prototype of the basic functionality of clang-misexpect 
in the PGO pipeline. clang-misexpect is a proposed clang-tool that can report 
potentially incorrect usage of __builtin_expect() by comparing the developer's 
annotation against a collected PGO profile. A more detailed proposal and 
discussion appears on the CFE-dev mailing list 
(http://lists.llvm.org/pipermail/cfe-dev/2019-July/062971.html) and a prototype 
of the frontend changes appear here on 
phabricator(https://reviews.llvm.org/D65300)

We extend the work in https://reviews.llvm.org/D65300 by adding support for IR 
and sampling based profiles to the backend. We also supply a standalone 
clang-misexpect tool built on top of libTooling. The standalone tool uses a 
compile commands database and profiling data to issue the warnings enabled by 
the new checks in Clang and LLVM.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D66324

Files:
  clang-tools-extra/CMakeLists.txt
  clang-tools-extra/clang-misexpect/CMakeLists.txt
  clang-tools-extra/clang-misexpect/ClangMisExpect.cpp
  clang-tools-extra/clang-misexpect/ClangMisExpect.h
  clang-tools-extra/clang-misexpect/tool/CMakeLists.txt
  clang-tools-extra/clang-misexpect/tool/ClangMisExpectMain.cpp
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp

Index: llvm/lib/Transforms/Utils/MisExpect.cpp
===
--- /dev/null
+++ llvm/lib/Transforms/Utils/MisExpect.cpp
@@ -0,0 +1,105 @@
+//===--- MisExpect.cpp - Check Use of __builtin_expect() with PGO data ===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===--===//
+//
+// This contains code to emit warnings for potentially incorrect usage of
+// __builtin_expect(). This utility extracts the threshold values from metadata
+// associated with the instrumented Branch or Switch. The threshold values are
+// then used to determin if a warning would be emmited.
+//
+// MisExpect metadata is generated when llvm.expect intrinsics are lowered see
+// LowerExpectIntrinsic.cpp
+//
+//===--===//
+
+#include "llvm/Transforms/Utils/MisExpect.h"
+#include "llvm/ADT/Twine.h"
+#include "llvm/IR/DiagnosticInfo.h"
+#include "llvm/IR/Instruction.h"
+#include "llvm/IR/Instructions.h"
+#include "llvm/IR/LLVMContext.h"
+#include "llvm/Support/BranchProbability.h"
+#include "llvm/Support/Debug.h"
+#include "llvm/Support/FormatVariadic.h"
+#include 
+#include 
+
+#define DEBUG_TYPE "misexpect"
+
+namespace {
+
+llvm::Instruction *getOprndOrInst(llvm::Instruction *I) {
+  llvm::Instruction *Ret = nullptr;
+  if (auto B = llvm::dyn_cast(I)) {
+Ret = llvm::dyn_cast(B->getCondition());
+  } else if (auto S = llvm::dyn_cast(I)) {
+Ret = llvm::dyn_cast(S->getCondition());
+  }
+  return Ret ? Ret : I;
+}
+
+void emitMisexpectWarni

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

In D66324#1632653 , @lebedev.ri wrote:

> This is marked as child revision of D65300  
> but it seems like they both add
>  the same logic, just into different components, D65300 
>  to clang, this to llvm.
>  Is this duplication or are the diffs concurrent?


So this change extends and revises what is in D65300 
.  The diffs are somewhat concurrent. For 
example the backend changes rely on the frontend warning being available. i.e. 
diag::warn_profile_data_misexpect.  But yes the logic is mostly the same.

I'm also not terribly familiar with Phabricator, so maybe I should have 
uploaded a diff between my two revisions, rather than between the my changes 
and top of tree?

> Can D65300  be re-implemented to just use 
> this llvm code?

So it is possible to handle this completely in the backed, but the diagnostic 
output is not fantastic when using clang based instrumentation. In particular, 
we would need to emit the diagnostic in LowerExpectIntrisic.cpp by checking if 
the branch weight metadata already exists, and processing it there before it is 
overwritten. For some reason emitting the diagnostic at that point will not 
also emit the helpful source listing if there is a macro, even though a 
FullSourceLoc is available. This isn't an issue for either the handling in 
SampleProfile.cpp or in PGOInstrumentation.cpp.

In the end, I decided that it actually made sense for clang to handle the clang 
based instrumentation, and llvm to handle the IR/Sampling instrumentation.

> I'm also thinking the clang-misexpect *tool* should be a separate change,
>  it's misleading to have it here as it gives impression that it **has** to be 
> used,
>  which i'm not sure is so?

Sure, I can update this so that the standalone tool is in a separate change. 
The thought was to consolidate all the related changes needed for the 
standalone tool, but I see your point.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 215643.
paulkirth edited the summary of this revision.
paulkirth added a comment.
Herald added subscribers: dexonsmith, steven_wu, mehdi_amini.

Removes standalone clang-misexpect from revision


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang-tools-extra/CMakeLists.txt
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/ThinLTO/X86/lazyload_metadata.ll
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll

Index: llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
===
--- llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
+++ llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
@@ -138,7 +138,7 @@
   %conv1 = sext i32 %conv to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv1, i64 0)
   %tobool = icmp ne i64 %expval, 0
-; CHECK: !prof !1
+; CHECK: !prof !2
 ; CHECK-NOT: @llvm.expect
   br i1 %tobool, label %if.then, label %if.end
 
@@ -165,7 +165,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv, i64 1)
-; CHECK: !prof !2
+; CHECK: !prof !4
 ; CHECK-NOT: @llvm.expect
   switch i64 %expval, label %sw.epilog [
 i64 1, label %sw.bb
@@ -194,7 +194,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv, i64 1)
-; CHECK: !prof !3
+; CHECK: !prof !5
 ; CHECK-NOT: @llvm.expect
   switch i64 %expval, label %sw.epilog [
 i64 2, label %sw.bb
@@ -278,7 +278,7 @@
   %t7 = call i64 @llvm.expect.i64(i64 %t6, i64 0)
   %t8 = icmp ne i64 %t7, 0
   %t9 = select i1 %t8, i32 1, i32 2
-; CHECK: select{{.*}}, !prof !1
+; CHECK: select{{.*}}, !prof !2
   ret i32 %t9
 }
 
@@ -286,6 +286,6 @@
 declare i1 @llvm.expect.i1(i1, i1) nounwind readnone
 
 ; CHECK: !0 = !{!"branch_weights", i32 2000, i32 1}
-; CHECK: !1 = !{!"branch_weights", i32 1, i32 2000}
-; CHECK: !2 = !{!"branch_weights", i32 1, i32 2000, i32 1}
-; CHECK: !3 = !{!"branch_weights", i32 2000, i32 1, i32 1}
+; CHECK: !2 = !{!"branch_weights", i32 1, i32 2000}
+; CHECK: !4 = !{!"branch_weights", i32 1, i32 2000, i32 1}
+; CHECK: !5 = !{!"branch_weights", i32 2000, i32 1, i32 1}
Index: llvm/test/ThinLTO/X86/lazyload_metadata.ll
===
--- llvm/test/ThinLTO/X86/lazyload_metadata.ll
+++ llvm/test/ThinLTO/X86/lazyload_metadata.ll
@@ -10,13 +10,13 @@
 ; RUN: llvm-lto -thinlto-action=import %t2.bc -thinlto-index=%t3.bc \
 ; RUN:  -o /dev/null -stats \
 ; RUN:  2>&1 | FileCheck %s -check-prefix=LAZY
-; LAZY: 61 bitcode-reader  - Number of Metadata records loaded
+; LAZY: 63 bitcode-reader  - Number of Metadata records loaded
 ; LAZY: 2 bitcode-reader  - Number of MDStrings loaded
 
 ; RUN: llvm-lto -thinlto-action=import %t2.bc -thinlto-index=%t3.bc \
 ; RUN:  -o /dev/null -disable-ondemand-mds-loading -stats \
 ; RUN:  2>&1 | FileCheck %s -check-prefix=NOTLAZY
-; NOTLAZY: 70 bitcode-reader  - Number of Metadata records loaded
+; NOTLAZY: 72 bitcode-reader  - Number of Metadata records loaded
 ; NOTLAZY: 7 bitcode-reader  - Number of MDStrings loaded
 
 
Index: llvm/lib/Transforms/Utils/MisExpect.cpp
===
--- /dev/null
+++ llvm/lib/Transform

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 215646.
paulkirth added a comment.

Remove reference to clang-misexpect from CMakeLists.txt


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/CGStmt.cpp
  clang/lib/CodeGen/CMakeLists.txt
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/CodeGen/MisExpect.cpp
  clang/lib/CodeGen/MisExpect.h
  clang/lib/Driver/ToolChains/Clang.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/ThinLTO/X86/lazyload_metadata.ll
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll

Index: llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
===
--- llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
+++ llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
@@ -138,7 +138,7 @@
   %conv1 = sext i32 %conv to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv1, i64 0)
   %tobool = icmp ne i64 %expval, 0
-; CHECK: !prof !1
+; CHECK: !prof !2
 ; CHECK-NOT: @llvm.expect
   br i1 %tobool, label %if.then, label %if.end
 
@@ -165,7 +165,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv, i64 1)
-; CHECK: !prof !2
+; CHECK: !prof !4
 ; CHECK-NOT: @llvm.expect
   switch i64 %expval, label %sw.epilog [
 i64 1, label %sw.bb
@@ -194,7 +194,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv, i64 1)
-; CHECK: !prof !3
+; CHECK: !prof !5
 ; CHECK-NOT: @llvm.expect
   switch i64 %expval, label %sw.epilog [
 i64 2, label %sw.bb
@@ -278,7 +278,7 @@
   %t7 = call i64 @llvm.expect.i64(i64 %t6, i64 0)
   %t8 = icmp ne i64 %t7, 0
   %t9 = select i1 %t8, i32 1, i32 2
-; CHECK: select{{.*}}, !prof !1
+; CHECK: select{{.*}}, !prof !2
   ret i32 %t9
 }
 
@@ -286,6 +286,6 @@
 declare i1 @llvm.expect.i1(i1, i1) nounwind readnone
 
 ; CHECK: !0 = !{!"branch_weights", i32 2000, i32 1}
-; CHECK: !1 = !{!"branch_weights", i32 1, i32 2000}
-; CHECK: !2 = !{!"branch_weights", i32 1, i32 2000, i32 1}
-; CHECK: !3 = !{!"branch_weights", i32 2000, i32 1, i32 1}
+; CHECK: !2 = !{!"branch_weights", i32 1, i32 2000}
+; CHECK: !4 = !{!"branch_weights", i32 1, i32 2000, i32 1}
+; CHECK: !5 = !{!"branch_weights", i32 2000, i32 1, i32 1}
Index: llvm/test/ThinLTO/X86/lazyload_metadata.ll
===
--- llvm/test/ThinLTO/X86/lazyload_metadata.ll
+++ llvm/test/ThinLTO/X86/lazyload_metadata.ll
@@ -10,13 +10,13 @@
 ; RUN: llvm-lto -thinlto-action=import %t2.bc -thinlto-index=%t3.bc \
 ; RUN:  -o /dev/null -stats \
 ; RUN:  2>&1 | FileCheck %s -check-prefix=LAZY
-; LAZY: 61 bitcode-reader  - Number of Metadata records loaded
+; LAZY: 63 bitcode-reader  - Number of Metadata records loaded
 ; LAZY: 2 bitcode-reader  - Number of MDStrings loaded
 
 ; RUN: llvm-lto -thinlto-action=import %t2.bc -thinlto-index=%t3.bc \
 ; RUN:  -o /dev/null -disable-ondemand-mds-loading -stats \
 ; RUN:  2>&1 | FileCheck %s -check-prefix=NOTLAZY
-; NOTLAZY: 70 bitcode-reader  - Number of Metadata records loaded
+; NOTLAZY: 72 bitcode-reader  - Number of Metadata records loaded
 ; NOTLAZY: 7 bitcode-reader  - Number of MDStrings loaded
 
 
Index: llvm/lib/Transforms/Utils/MisExpect.cpp
===
--- /dev/null
+++ llvm/lib/Transforms/Utils/MisExpect.cpp
@@ -0,0 +1,105 @@
+//===--- MisExpect.cpp - Check Use of __builtin_expect() with PGO data ===//
+//
+// Part of

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.



> So it is possible to handle this completely in the backed, but the diagnostic 
> output is not fantastic when using clang based instrumentation. In 
> particular, we would need to emit the diagnostic in LowerExpectIntrisic.cpp 
> by checking if the branch weight metadata already exists, and processing it 
> there before it is overwritten. For some reason emitting the diagnostic at 
> that point will not also emit the helpful source listing if there is a macro, 
> even though a FullSourceLoc is available. This isn't an issue for either the 
> handling in SampleProfile.cpp or in PGOInstrumentation.cpp.

Actually, I made a mistake evaluating the diagnostic output. I redid my test 
for handling everything the Clang based instrumentation in 
LowerExpectIntrinsic.cpp and the output now looks as expected, so I must have 
had a logic error somewhere.

I think this can be handled 100% in the backend.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 215734.
paulkirth added a comment.

Remove frontend components of clang-misexpect in favor of backend 
implementations


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/ThinLTO/X86/lazyload_metadata.ll
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll

Index: llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
===
--- llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
+++ llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
@@ -138,7 +138,7 @@
   %conv1 = sext i32 %conv to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv1, i64 0)
   %tobool = icmp ne i64 %expval, 0
-; CHECK: !prof !1
+; CHECK: !prof !2
 ; CHECK-NOT: @llvm.expect
   br i1 %tobool, label %if.then, label %if.end
 
@@ -165,7 +165,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv, i64 1)
-; CHECK: !prof !2
+; CHECK: !prof !4
 ; CHECK-NOT: @llvm.expect
   switch i64 %expval, label %sw.epilog [
 i64 1, label %sw.bb
@@ -194,7 +194,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv, i64 1)
-; CHECK: !prof !3
+; CHECK: !prof !5
 ; CHECK-NOT: @llvm.expect
   switch i64 %expval, label %sw.epilog [
 i64 2, label %sw.bb
@@ -278,7 +278,7 @@
   %t7 = call i64 @llvm.expect.i64(i64 %t6, i64 0)
   %t8 = icmp ne i64 %t7, 0
   %t9 = select i1 %t8, i32 1, i32 2
-; CHECK: select{{.*}}, !prof !1
+; CHECK: select{{.*}}, !prof !2
   ret i32 %t9
 }
 
@@ -286,6 +286,6 @@
 declare i1 @llvm.expect.i1(i1, i1) nounwind readnone
 
 ; CHECK: !0 = !{!"branch_weights", i32 2000, i32 1}
-; CHECK: !1 = !{!"branch_weights", i32 1, i32 2000}
-; CHECK: !2 = !{!"branch_weights", i32 1, i32 2000, i32 1}
-; CHECK: !3 = !{!"branch_weights", i32 2000, i32 1, i32 1}
+; CHECK: !2 = !{!"branch_weights", i32 1, i32 2000}
+; CHECK: !4 = !{!"branch_weights", i32 1, i32 2000, i32 1}
+; CHECK: !5 = !{!"branch_weights", i32 2000, i32 1, i32 1}
Index: llvm/test/ThinLTO/X86/lazyload_metadata.ll
===
--- llvm/test/ThinLTO/X86/lazyload_metadata.ll
+++ llvm/test/ThinLTO/X86/lazyload_metadata.ll
@@ -10,13 +10,13 @@
 ; RUN: llvm-lto -thinlto-action=import %t2.bc -thinlto-index=%t3.bc \
 ; RUN:  -o /dev/null -stats \
 ; RUN:  2>&1 | FileCheck %s -check-prefix=LAZY
-; LAZY: 61 bitcode-reader  - Number of Metadata records loaded
+; LAZY: 63 bitcode-reader  - Number of Metadata records loaded
 ; LAZY: 2 bitcode-reader  - Number of MDStrings loaded
 
 ; RUN: llvm-lto -thinlto-action=import %t2.bc -thinlto-index=%t3.bc \
 ; RUN:  -o /dev/null -disable-ondemand-mds-loading -stats \
 ; RUN:  2>&1 | FileCheck %s -check-prefix=NOTLAZY
-; NOTLAZY: 70 bitcode-reader  - Number of Metadata records loaded
+; NOTLAZY: 72 bitcode-reader  - Number of Metadata records loaded
 ; NOTLAZY: 7 bitcode-reader  - Number of MDStrings loaded
 
 
Index: llvm/lib/Transforms/Utils/MisExpect.cpp
===
--- /dev/null
+++ llvm/lib/Transforms/Utils/MisExpect.cpp
@@ -0,0 +1,125 @@
+//===--- MisExpect.cpp - Check Use of __builtin_expect() with PGO data ===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===--===

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 215938.
paulkirth edited the summary of this revision.
paulkirth added a comment.

- Update CodeGenOptions to remove -fmisexpect
- Access the LLVMContext directly
- Add -line-tables-only for more accurate diagnostics

Fixes various issues with tests and cleans up some residual code from the 
original frontend implementation


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/lib/Transforms/Utils/MisExpect.cpp

Index: llvm/lib/Transforms/Utils/MisExpect.cpp
===
--- llvm/lib/Transforms/Utils/MisExpect.cpp
+++ llvm/lib/Transforms/Utils/MisExpect.cpp
@@ -114,7 +114,7 @@
   mdconst::dyn_extract(MD->getOperand(i));
   RealWeights[i - 1] = Value->getZExtValue();
 }
-misexpect::verifyMisExpect(&I, RealWeights, I.getParent()->getParent()->getContext());
+misexpect::verifyMisExpect(&I, RealWeights, I.getContext());
   }
 }
   }
Index: clang/test/Profile/misexpect-switch.c
===
--- clang/test/Profile/misexpect-switch.c
+++ clang/test/Profile/misexpect-switch.c
@@ -1,7 +1,7 @@
 // Test that misexpect detects mis-annotated switch statements
 
 // RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
-// RUN: %clang_cc1 %s -O2 -o - -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect
+// RUN: %clang_cc1 %s -O2 -o - -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect -debug-info-kind=line-tables-only
 
 int sum(int *buff, int size);
 int random_sample(int *buff, int size);
Index: clang/test/Profile/misexpect-switch-only-default-case.c
===
--- clang/test/Profile/misexpect-switch-only-default-case.c
+++ clang/test/Profile/misexpect-switch-only-default-case.c
@@ -1,7 +1,7 @@
 // Test that misexpect emits no warning when there is only one switch case
 
 // RUN: llvm-profdata merge %S/Inputs/misexpect-switch-default-only.proftext -o %t.profdata
-// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -fmisexpect
+// RUN: %clang_cc1 %s -O2 -o - -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect -debug-info-kind=line-tables-only
 
 // expected-no-diagnostics
 int sum(int *buff, int size);
Index: clang/test/Profile/misexpect-switch-default.c
===
--- clang/test/Profile/misexpect-switch-default.c
+++ clang/test/Profile/misexpect-switch-default.c
@@ -1,7 +1,7 @@
 // Test that misexpect detects mis-annotated switch statements for default case
 
-// RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
-// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect
+// RUN: llvm-profdata merge %S/Inputs/misexpect-switch-default.proftext -o %t.profdata
+// RUN: %clang_cc1 %s -O2 -o - -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect -debug-info-kind=line-tables-only
 
 int sum(int *buff, int size);
 int random_sample(int *buff, int size);
@@ -17,26 +17,24 @@
 int main() {
   init_arry();
   int val = 0;
-
-  int j, k;
-  for (j = 0; j < outer_loop; ++j) {
-for (k = 0; k < inner_loop; ++k) {
-  unsigned condition = rand() % 5;
-  switch (__builtin_expect(condition, 6)) { // expected-warning-re {{Potential performance regression from use of __builtin_expect(): Annotation was correct on {{.+}}% of profiled executions.}}
-  case 0:
-val += sum(arry, arry_size);
-break;
-  case 1:
-  case 2:
-  case 3:
-  case 4:
-val += random_sample(arry, arry_size);
-break;
-  default:
-__builtin_unreachable();
-  } // end switch
-}   // end inner_loop
-  } // end outer_loop
+  int j;
+  for (j = 0; j < outer_loop * inner_loop; ++j) {
+unsigned condition = rand() % 5;
+switch (__builtin_expect(condition, 6)) { // expected-warning-re {{Potential performance regression from use of __builtin_expect(): Annotation was correct on {{.+}}% of profiled executions.}}
+case 0:
+  val += sum(arry, arry_size);
+  break;
+case 1:
+case 2:
+case 3:
+  break;
+case 4:
+  val += random_sample(arry, arry_size);
+  break;
+default:
+  __builtin_unreachable();
+} // end switch
+  }   // end outer_loop
 
   return 0;
 }
Index: clang/test/Profile/misexpect-branch.c
=

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 215940.
paulkirth added a comment.

Fix missing context in prior diff


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-no-warning-without-flag.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/ThinLTO/X86/lazyload_metadata.ll
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll

Index: llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
===
--- llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
+++ llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
@@ -138,7 +138,7 @@
   %conv1 = sext i32 %conv to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv1, i64 0)
   %tobool = icmp ne i64 %expval, 0
-; CHECK: !prof !1
+; CHECK: !prof !2
 ; CHECK-NOT: @llvm.expect
   br i1 %tobool, label %if.then, label %if.end
 
@@ -165,7 +165,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv, i64 1)
-; CHECK: !prof !2
+; CHECK: !prof !4
 ; CHECK-NOT: @llvm.expect
   switch i64 %expval, label %sw.epilog [
 i64 1, label %sw.bb
@@ -194,7 +194,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv, i64 1)
-; CHECK: !prof !3
+; CHECK: !prof !5
 ; CHECK-NOT: @llvm.expect
   switch i64 %expval, label %sw.epilog [
 i64 2, label %sw.bb
@@ -278,7 +278,7 @@
   %t7 = call i64 @llvm.expect.i64(i64 %t6, i64 0)
   %t8 = icmp ne i64 %t7, 0
   %t9 = select i1 %t8, i32 1, i32 2
-; CHECK: select{{.*}}, !prof !1
+; CHECK: select{{.*}}, !prof !2
   ret i32 %t9
 }
 
@@ -286,6 +286,6 @@
 declare i1 @llvm.expect.i1(i1, i1) nounwind readnone
 
 ; CHECK: !0 = !{!"branch_weights", i32 2000, i32 1}
-; CHECK: !1 = !{!"branch_weights", i32 1, i32 2000}
-; CHECK: !2 = !{!"branch_weights", i32 1, i32 2000, i32 1}
-; CHECK: !3 = !{!"branch_weights", i32 2000, i32 1, i32 1}
+; CHECK: !2 = !{!"branch_weights", i32 1, i32 2000}
+; CHECK: !4 = !{!"branch_weights", i32 1, i32 2000, i32 1}
+; CHECK: !5 = !{!"branch_weights", i32 2000, i32 1, i32 1}
Index: llvm/test/ThinLTO/X86/lazyload_metadata.ll
===
--- llvm/test/ThinLTO/X86/lazyload_metadata.ll
+++ llvm/test/ThinLTO/X86/lazyload_metadata.ll
@@ -10,13 +10,13 @@
 ; RUN: llvm-lto -thinlto-action=import %t2.bc -thinlto-index=%t3.bc \
 ; RUN:  -o /dev/null -stats \
 ; RUN:  2>&1 | FileCheck %s -check-prefix=LAZY
-; LAZY: 61 bitcode-reader  - Number of Metadata records loaded
+; LAZY: 63 bitcode-reader  - Number of Metadata records loaded
 ; LAZY: 2 bitcode-reader  - Number of MDStrings loaded
 
 ; RUN: llvm-lto -thinlto-action=import %t2.bc -thinlto-index=%t3.bc \
 ; RUN:  -o /dev/null -disable-ondemand-mds-loading -stats \
 ; RUN:  2>&1 | FileCheck %s -check-prefix=NOTLAZY
-; NOTLAZY: 70 bitcode-reader  - Number of Metadata records loaded
+; NOTLAZY: 72 bitcode-reader  - Number of Metadata records loaded
 ; NOTLAZY: 7 bitcode-reader  - Number of MDStrings loaded
 
 
Index: llvm/lib/Transforms/Utils/MisExpect.cpp
===
--- /dev/null
+++ llvm/lib/Transforms/Utils/MisExpect.cpp
@@ -0,0 +1,125 @@
+//===--- MisExpect.cpp - Check Use of __builtin_expect() with PGO data ===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===--===//
+//
+// This contains c

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 219617.
paulkirth added a comment.

Add a new profdata file to use w/ misexpect-nonconst-switch.c

ASAN exposed an issue where a function may hash the same even if the number of 
counters is different. This means that when a PGO profile is read in, it is 
possible for an out of bounds write to occur due to a mismatch in the number of 
statements that require profile counters, and the number of counters 
initialized from the profile that was read in.

This change should allow the misexpect patch to reland w/o issue.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/ThinLTO/X86/lazyload_metadata.ll
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,293 @@
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; New PM
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.c.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=CORRECT
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:5: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:5: Potential performance re

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth marked an inline comment as done.
paulkirth added inline comments.



Comment at: compiler-rt/trunk/lib/profile/xxhash.h:41-42
+
+#include "llvm/ADT/ArrayRef.h"
+#include "llvm/ADT/StringRef.h"
+

chandlerc wrote:
> Sorry folks, but you can't do this.
> 
> You can't depend on ADT from compiler-rt currently.
> 
> There are at least two problems here:
> 
> First problem is that this pollutes the profile library with symbols from 
> ADT. That really doesn't seem reasonable without *significant* and invasive 
> changes to ADT. Otherwise building LLVM and linking it with the profile 
> library will create an ODR violation (imagine different assert levels or 
> different versions of LLVM buing built and the host toolchain).
> 
> 
> Second, and much more critically, we haven't gotten to 100% relicensed on 
> ADT, so it is critical that we not depend on it from runtime libraries.
> 
> Third, a lot of this code seems to use old license headers. Please do not add 
> any code like this to LLVM, and instead use the new LLVM license for all new 
> code.
> 
> For now, this patch (and any related patches) need to be reverted until these 
> are addressed. Especially the license issues.
Sorry, this looks like a mismerge somehow. My patch should't have anything from 
compiler-rt. I think maybe a local change got rolled in when Petr landed my 
patch.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

In D66324#1665784 , @lebedev.ri wrote:

> In D66324#1665773 , @gribozavr wrote:
>
> > Reverted in r371598.
> >
> > Another concern that I have is cross-compilation. LLVM's ADT is not set up 
> > to be cross-compiled like the rest of compiler-rt is.
>
>
> Uhm, i have a better question still - why xxhash is even there? it's not used 
> in the diff, and was not reviewed, it just magically appeared in the last 
> update of the patch:
>  https://reviews.llvm.org/D66324?vs=219617&id=219645


It shouldn't have. My last change only modified a test file to not trigger a 
memory corruption bug. I'm filing a bug for that, but am waiting on a login for 
the bug tracker.

What I suspect happened is that when this landed some local changes got sucked 
in. I'm terribly sorry, it wasn't my intention to add anything significant to 
the patch.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 219728.
paulkirth added a comment.

Revert mismerge when landing.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/ThinLTO/X86/lazyload_metadata.ll
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,293 @@
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; New PM
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.c.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=CORRECT
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:5: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:5: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:5: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:5: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:5: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:2

[PATCH] D67253: clang-misexpect: a standalone tool for verifying the use of __builtin_expect with PGO data

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 220239.
paulkirth added a comment.

Addresses problems running the standalone tool w/ the libTooling executors.

When using the CodeGenAction and setting LLVM backend options, I found several 
places where data races occurred. This seems like a more significant 
architectural issue than mitigating access to a few global variables. To avoid 
these issues I've locked the executor concurrency to 1, i.e. single threaded. 
This prevents any data races when the executor is configuring the backends for 
each compiler invocation.

I've included a python script based on the run-clang-tidy.py that tidy uses. 
This allows the standalone tool to take advantage of parallellism without 
running into the data races between threads.

I have also added a new checking mechanism to ensure that PGO profiles and the 
command line options are compatible.

Lastly, I've included new documentation and tests for the standalone tool.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D67253/new/

https://reviews.llvm.org/D67253

Files:
  clang-tools-extra/CMakeLists.txt
  clang-tools-extra/clang-misexpect/CMakeLists.txt
  clang-tools-extra/clang-misexpect/ClangMisExpect.cpp
  clang-tools-extra/clang-misexpect/ClangMisExpect.h
  clang-tools-extra/clang-misexpect/tool/CMakeLists.txt
  clang-tools-extra/clang-misexpect/tool/ClangMisExpectMain.cpp
  clang-tools-extra/clang-misexpect/tool/run-clang-misexpect.py
  clang-tools-extra/docs/clang-misexpect.rst
  clang-tools-extra/test/CMakeLists.txt
  clang-tools-extra/test/clang-misexpect/Inputs/basic.proftext
  clang-tools-extra/test/clang-misexpect/Inputs/clean.c
  clang-tools-extra/test/clang-misexpect/basic.cpp
  clang-tools-extra/test/lit.cfg.py
  clang/cmake/caches/Fuchsia-stage2.cmake
  clang/test/Profile/misexpect-switch-nonconst.c
  llvm/lib/Transforms/Utils/MisExpect.cpp

Index: llvm/lib/Transforms/Utils/MisExpect.cpp
===
--- llvm/lib/Transforms/Utils/MisExpect.cpp
+++ llvm/lib/Transforms/Utils/MisExpect.cpp
@@ -43,7 +43,8 @@
 static cl::opt PGOWarnMisExpect(
 "pgo-warn-misexpect", cl::init(false), cl::Hidden,
 cl::desc("Use this option to turn on/off "
- "warnings about incorrect usage of llvm.expect intrinsics."));
+ "warnings about incorrect usage of llvm.expect intrinsics."),
+cl::ZeroOrMore);
 
 } // namespace llvm
 
Index: clang/test/Profile/misexpect-switch-nonconst.c
===
--- clang/test/Profile/misexpect-switch-nonconst.c
+++ clang/test/Profile/misexpect-switch-nonconst.c
@@ -1,7 +1,7 @@
 // Test that misexpect emits no warning when switch condition is non-const
 
 // RUN: llvm-profdata merge %S/Inputs/misexpect-switch-nonconst.proftext -o %t.profdata
-// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify
+// RUN: %clang_cc1 %s -O2 -o - -disable-llvm-passes -emit-llvm -fprofile-instrument-use-path=%t.profdata -verify -Wmisexpect
 
 // expected-no-diagnostics
 int sum(int *buff, int size);
Index: clang/cmake/caches/Fuchsia-stage2.cmake
===
--- clang/cmake/caches/Fuchsia-stage2.cmake
+++ clang/cmake/caches/Fuchsia-stage2.cmake
@@ -207,6 +207,7 @@
   LTO
   clang-apply-replacements
   clang-doc
+  clang-misexpect
   clang-format
   clang-resource-headers
   clang-include-fixer
Index: clang-tools-extra/test/lit.cfg.py
===
--- clang-tools-extra/test/lit.cfg.py
+++ clang-tools-extra/test/lit.cfg.py
@@ -144,6 +144,12 @@
 ('%run_clang_tidy',
  '%s %s' % (python_exec, run_clang_tidy)) )
 
+run_clang_misexpect = os.path.join(
+config.test_source_root, "..", "clang-misexpect", "tool", "run-clang-misexpect.py")
+config.substitutions.append(
+('%run_clang_misexpect',
+ '%s %s' % (python_exec, run_clang_misexpect)) )
+
 clangd_benchmarks_dir = os.path.join(os.path.dirname(config.clang_tools_dir),
  "tools", "clang", "tools", "extra",
  "clangd", "benchmarks")
Index: clang-tools-extra/test/clang-misexpect/basic.cpp
===
--- /dev/null
+++ clang-tools-extra/test/clang-misexpect/basic.cpp
@@ -0,0 +1,41 @@
+// Test that clang-misexpect can run over a compilation database
+
+// Now create a directory with a compilation database file and ensure we don't
+// use it after failing to parse commands from the command line:
+//
+// RUN: mkdir -p %T/misexpect/
+// RUN: echo '[{"directory": "%/T/misexpect/","arguments": ["clang++","-O2","-c","%/T/misexpect/basic.cpp"], "file": "basic.cpp"},{"directory": "%/T/misexpect/","command": "clang -c %/T/misexpect/clean.c", "file": "%/T/misexpect/clean.c"}]' > %T/misexpect/compile_comm

[PATCH] D67253: clang-misexpect: a standalone tool for verifying the use of __builtin_expect with PGO data

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

In D67253#1670569 , @lebedev.ri wrote:

> Layering feels weird here.
>  Can this be implemented as/limited to just a
>  `run-clang-misexpect.py` wrapper over clang itself?
>  That would be best IMHO.


I discussed the concurrency issue with some folks who work on libTooling, 
notably Sam McCall & Dmitri Gribenko. This was the approach they suggested I 
follow. LibTooling also provides some nice mechanisms for curating compiler 
options, which is possible, but less than ideal to reimplement in a python 
script. There are probably more benefits that escape me at the moment, but that 
was the first one I thought of.

Out of curiosity, if the concurrency issue was fixed in the compiler and the 
python script was removed/deprecated, would you still feel the same way?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D67253/new/

https://reviews.llvm.org/D67253



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D67253: clang-misexpect: a standalone tool for verifying the use of __builtin_expect with PGO data

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

In D67253#1670681 , @lebedev.ri wrote:

> Re concurrency - you had standalone `LLVMContext` for each thread, right?


I believe there was, but this is just whatever happens when libtooling creates 
and executes a compiler invocation. It seems to me that there is some global 
state that ends up being shared between threads. I wasn’t able to fully trace 
the issue, but it at least partly involves the processing of backend options. I 
believe there is more, that needs to be addressed but this is probably a good 
place for me to start. Thanks for the suggestion.

> 
> 
> In D67253#1670667 , @paulkirth wrote:
> 
>> In D67253#1670569 , @lebedev.ri 
>> wrote:
>>
>> > Layering feels weird here.
>> >  Can this be implemented as/limited to just a
>> >  `run-clang-misexpect.py` wrapper over clang itself?
>> >  That would be best IMHO.
>>
>>
>> I discussed the concurrency issue with some folks who work on libTooling, 
>> notably Sam McCall & Dmitri Gribenko. This was the approach they suggested I 
>> follow. LibTooling also provides some nice mechanisms for curating compiler 
>> options, which is possible, but less than ideal to reimplement in a python 
>> script. There are probably more benefits that escape me at the moment, but 
>> that was the first one I thought of.
>>
>> Out of curiosity, if the concurrency issue was fixed in the compiler and the 
>> python script was removed/deprecated, would you still feel the same way?
> 
> 
> I was actually talking/thinking about the opposite direction, only having the 
> .py wrapper, no standalone tool;
>  so the opposite solution (no .py, only the tool) is tangential/has different 
> direction.

Right, I was asking if you still see this as an issue if the python script 
wasn’t necessary. I take from this comment that it would not make you feel 
differently.

I think ultimately keeping the implementation in c++ makes the most sense and 
can evolve with the rest of libtooling.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D67253/new/

https://reviews.llvm.org/D67253



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added inline comments.



Comment at: llvm/trunk/include/llvm/IR/DiagnosticInfo.h:79
+  DK_FirstPluginKind,
+  DK_MisExpect
 };

jrtc27 wrote:
> Is this not going to clash with the first caller to 
> `getNextAvailablePluginDiagnosticKind`? `DK_FirstPluginKind` is special and 
> shouldn't have anything after it.
hmm, I wasn't aware of that. Is this documented somewhere? Also no tests seem 
to break from this, so that might be another area to address.

I'll upload a patch for this that moves the flag, and adds a note about the 
ordering.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth marked an inline comment as done.
paulkirth added inline comments.



Comment at: llvm/trunk/include/llvm/IR/DiagnosticInfo.h:79
+  DK_FirstPluginKind,
+  DK_MisExpect
 };

paulkirth wrote:
> jrtc27 wrote:
> > Is this not going to clash with the first caller to 
> > `getNextAvailablePluginDiagnosticKind`? `DK_FirstPluginKind` is special and 
> > shouldn't have anything after it.
> hmm, I wasn't aware of that. Is this documented somewhere? Also no tests seem 
> to break from this, so that might be another area to address.
> 
> I'll upload a patch for this that moves the flag, and adds a note about the 
> ordering.
Please see D67648 for future reference of this defect.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D131306: [llvm][misexpect] Track provenance of branch weights

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.



In D131306#3756009 , @tejohnson wrote:

> @davidxl @xur for review and thoughts.
>
> I'm a little wary of requiring that both pieces of metadata be carried 
> together, as that seems very brittle to maintain in the compiler. What would 
> happen if the MD_expected didn't get propagated by some pass along with the 
> MD_prof? I think you would get a false negative, which I suppose is better 
> than a false positive. An alternative, that I guess would require more 
> extensive changes, is to add an additional item to the "branch_weights" list 
> (would need to be obviously distinguishable by type from the list of weights 
> since that can be variable though).

Agreed. This isn't my preferred solution, but it seemed far less invasive than 
changing the format of profiling metadata. Originally, I had looked into adding 
a provenance field to the metadata. It required changes to every test that has 
branch weights, and I balked at submitting that. I'm also a little wary of 
making a heavily used metadata type larger. I guess there isn't a lot of 
difference, but the external metadata is optional, and we could remove it after 
checking is complete.

Unfortunately, I just don't think there is a clean solution here. Either we 
make an invasive change to the metadata format, or we deal w/ updating 2 pieces 
of metadata everywhere.  I'm just very unsure about which is the right tradeoff.

> Patch needs tests showing uses of the new metadata, and some documentation in 
> LangRef (i.e. near https://llvm.org/docs/LangRef.html#prof-metadata).

Yes, testing and documentation is something I plan to improve, but I wanted to 
get some feedback on this approach before investing too heavily.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131306/new/

https://reviews.llvm.org/D131306

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D131306: [llvm][misexpect] Track provenance of branch weights

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

In D131306#3756087 , @tejohnson wrote:

> Well I was thinking the extra field would be optional as well and could be 
> removed. But understood that this requires more changes (although maybe not 
> if it is optional, and after your recent changes to centralize some of the 
> prof metadata handling in the compiler).

Hmm, I don't think I considered that a field in the metadata could be optional. 
Do you mean something like this?

  !{!"branch_weights", !10, i32 1717986918, i32 429496731}

where `!10` is just some optional metadata, and we'd just ensure things that 
parse the MD_prof data skip it correctly? Given that we've mostly consolidated 
how branch weights are extracted and manipulated that might only require a 
limited number of updates to the code and tests.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131306/new/

https://reviews.llvm.org/D131306

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D131306: [llvm][misexpect] Track provenance of branch weights

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

I think that may be a better approach. Let me investigate a bit and see what 
the exact tradeoffs are.

If anyone else has a suggestion or thought on another way to handle this, I'm 
all ears.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131306/new/

https://reviews.llvm.org/D131306

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D133732: [clang-doc] Support default args for functions.

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Thanks for the patch. This is mostly LGTM modulo a few nits.

My one question is regarding documentation. Do you think this needs to be 
described in the clang-tools-extra/doc/clang-doc.rst? And are there any changes 
in workflow or tool usage that we should document? I assume not, since this 
just changes the YAML format. Is that correct?




Comment at: clang-tools-extra/clang-doc/Serialize.cpp:300
 
-auto& member = I.Members.emplace_back(
+auto &member = I.Members.emplace_back(
 F->getTypeSourceInfo()->getType().getAsString(), F->getNameAsString(),

nit: can we avoid introducing unrelated formatting changes, here and elsewhere 
in the patch?



Comment at: clang-tools-extra/clang-doc/Serialize.cpp:324-329
+  } else {
+FieldInfo = &I.Params.emplace_back(P->getOriginalType().getAsString(),
+   P->getNameAsString());
   }
+} else {
+  FieldInfo = &I.Params.emplace_back(P->getOriginalType().getAsString(),

can we combine these conditions somehow, given that they do the same thing?

Maybe this? WDYT?

```
if(!FieldInfo)
  FieldInfo = &I.Params.emplace_back(P->getOriginalType().getAsString(), ...
```




Comment at: clang-tools-extra/clang-doc/Serialize.cpp:452
 
-  ASTContext& Context = D->getASTContext();
+  ASTContext &Context = D->getASTContext();
   // TODO investigate whether we can use ASTContext::getCommentForDecl instead

nit: same here



Comment at: clang-tools-extra/clang-doc/Serialize.cpp:460
   Comment->setAttached();
-  if (comments::FullComment* fc = Comment->parse(Context, nullptr, D)) {
+  if (comments::FullComment *fc = Comment->parse(Context, nullptr, D)) {
 I.Description.emplace_back();

nit: ditto


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D133732/new/

https://reviews.llvm.org/D133732

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 418025.
paulkirth added a comment.

Add missing GenerateArgs call to propagate flags to the backend outside of cc1


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/docs/MisExpect.rst
  clang/docs/ReleaseNotes.rst
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/CodeGenOptions.h
  clang/include/clang/Basic/DiagnosticDriverKinds.td
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-threshold.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,285 @@
+; Test misexpect diagnostics handle swich statements, and report source locations correctly
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; DISABLED-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; DISABLED-NOT: remark: misexpect-switch.c:26:30: Potential performa

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added inline comments.



Comment at: clang/lib/Frontend/CompilerInvocation.cpp:1554
+  Twine(*Opts.DiagnosticsMisExpectTolerance), SA);
+
   for (StringRef Sanitizer : serializeSanitizerKinds(Opts.SanitizeRecover))

I really don't understand why this step is necessary, or why in my local builds 
omitting the call to `GenerateArgs` works at all. I only arrived at this change 
by noticing other options do the same, and this seemed to fix the issue with 
tests using the repro instructions from the precommit bots.

It seems a bit strange that Clang parses the option, stores it to a 
`CodeGenOption` then puts it back as a string argument to be parsed again later 
and put into the same data structure. Is this a result of an earlier 
architecture in Clang? if so, should we reconsider its design?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 418280.
paulkirth added a comment.

Fix check for tolerance diagnostic since we now use GenerateArgs the same way 
as hotness threshold.

Since it has a default value, we need to also ensure that it uses a pattern 
similar to other options.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/docs/MisExpect.rst
  clang/docs/ReleaseNotes.rst
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/CodeGenOptions.h
  clang/include/clang/Basic/DiagnosticDriverKinds.td
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-threshold.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,285 @@
+; Test misexpect diagnostics handle swich statements, and report source locations correctly
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 418328.
paulkirth added a comment.

rebase


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/docs/MisExpect.rst
  clang/docs/ReleaseNotes.rst
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/CodeGenOptions.h
  clang/include/clang/Basic/DiagnosticDriverKinds.td
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-threshold.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,285 @@
+; Test misexpect diagnostics handle swich statements, and report source locations correctly
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; DISABLED-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; DISABLED-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was cor

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

@tejohnson thanks for pointing me to the document. I knew it had something to 
do w/ CC1 but missed that this was well documented.

Is there anything else that needs to be done, or do you think this is good to 
land again?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

This revision was automatically updated to reflect the committed changes.
Closed by commit rG2add3fbd976d: [misexpect] Re-implement MisExpect Diagnostics 
(authored by paulkirth).

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/docs/MisExpect.rst
  clang/docs/ReleaseNotes.rst
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/CodeGenOptions.h
  clang/include/clang/Basic/DiagnosticDriverKinds.td
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-threshold.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,285 @@
+; Test misexpect diagnostics handle swich statements, and report source locations correctly
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; DISABLED-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; DISABLED-NOT: remark: misexpect-switch.c:2

[PATCH] D122623: [docs][misexpect] Fix malformed table in docs

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 418735.
paulkirth added a comment.
Herald added a project: clang.
Herald added a subscriber: cfe-commits.

update table in clang docs too


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D122623/new/

https://reviews.llvm.org/D122623

Files:
  clang/docs/MisExpect.rst
  llvm/docs/MisExpect.rst


Index: llvm/docs/MisExpect.rst
===
--- llvm/docs/MisExpect.rst
+++ llvm/docs/MisExpect.rst
@@ -62,7 +62,7 @@
 
++--+
 | Profile Type   | Description 
 |
 
++==+
-| Frontend   | Profiling instrumentation added during compilation by the 
frontend, i.e. ``clang``|
+| Frontend   | Profiling instrumentation added during compilation by the 
frontend, i.e. ``clang``   |
 
++--+
 | IR | Profiling instrumentation added during by the LLVM backend  
 |
 
++--+
Index: clang/docs/MisExpect.rst
===
--- clang/docs/MisExpect.rst
+++ clang/docs/MisExpect.rst
@@ -64,7 +64,7 @@
 
++--+
 | Profile Type   | Description 
 |
 
++==+
-| Frontend   | Profiling instrumentation added during compilation by the 
frontend, i.e. ``clang``|
+| Frontend   | Profiling instrumentation added during compilation by the 
frontend, i.e. ``clang``   |
 
++--+
 | IR | Profiling instrumentation added during by the LLVM backend  
 |
 
++--+


Index: llvm/docs/MisExpect.rst
===
--- llvm/docs/MisExpect.rst
+++ llvm/docs/MisExpect.rst
@@ -62,7 +62,7 @@
 ++--+
 | Profile Type   | Description  |
 ++==+
-| Frontend   | Profiling instrumentation added during compilation by the frontend, i.e. ``clang``|
+| Frontend   | Profiling instrumentation added during compilation by the frontend, i.e. ``clang``   |
 ++--+
 | IR | Profiling instrumentation added during by the LLVM backend   |
 ++--+
Index: clang/docs/MisExpect.rst
===
--- clang/docs/MisExpect.rst
+++ clang/docs/MisExpect.rst
@@ -64,7 +64,7 @@
 ++--+
 | Profile Type   | Description  |
 ++==+
-| Frontend   | Profiling instrumentation added during compilation by the frontend, i.e. ``clang``|
+| Frontend   | Profiling instrumentation added during compilation by the frontend, i.e. ``clang``   |
 ++--+
 | IR | Profiling instrumentation added during by the LLVM backend   |
 ++--+
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D122623: [docs][misexpect][NFC] Fix malformed table in docs

[Paul Kirth via Phabricator via cfe-commits]

This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
Closed by commit rGa427e18896de: [docs][misexpect][NFC] Fix malformed table in 
docs (authored by paulkirth).

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D122623/new/

https://reviews.llvm.org/D122623

Files:
  clang/docs/MisExpect.rst
  llvm/docs/MisExpect.rst


Index: llvm/docs/MisExpect.rst
===
--- llvm/docs/MisExpect.rst
+++ llvm/docs/MisExpect.rst
@@ -62,7 +62,7 @@
 
++--+
 | Profile Type   | Description 
 |
 
++==+
-| Frontend   | Profiling instrumentation added during compilation by the 
frontend, i.e. ``clang``|
+| Frontend   | Profiling instrumentation added during compilation by the 
frontend, i.e. ``clang``   |
 
++--+
 | IR | Profiling instrumentation added during by the LLVM backend  
 |
 
++--+
Index: clang/docs/MisExpect.rst
===
--- clang/docs/MisExpect.rst
+++ clang/docs/MisExpect.rst
@@ -64,7 +64,7 @@
 
++--+
 | Profile Type   | Description 
 |
 
++==+
-| Frontend   | Profiling instrumentation added during compilation by the 
frontend, i.e. ``clang``|
+| Frontend   | Profiling instrumentation added during compilation by the 
frontend, i.e. ``clang``   |
 
++--+
 | IR | Profiling instrumentation added during by the LLVM backend  
 |
 
++--+


Index: llvm/docs/MisExpect.rst
===
--- llvm/docs/MisExpect.rst
+++ llvm/docs/MisExpect.rst
@@ -62,7 +62,7 @@
 ++--+
 | Profile Type   | Description  |
 ++==+
-| Frontend   | Profiling instrumentation added during compilation by the frontend, i.e. ``clang``|
+| Frontend   | Profiling instrumentation added during compilation by the frontend, i.e. ``clang``   |
 ++--+
 | IR | Profiling instrumentation added during by the LLVM backend   |
 ++--+
Index: clang/docs/MisExpect.rst
===
--- clang/docs/MisExpect.rst
+++ clang/docs/MisExpect.rst
@@ -64,7 +64,7 @@
 ++--+
 | Profile Type   | Description  |
 ++==+
-| Frontend   | Profiling instrumentation added during compilation by the frontend, i.e. ``clang``|
+| Frontend   | Profiling instrumentation added during compilation by the frontend, i.e. ``clang``   |
 ++--+
 | IR | Profiling instrumentation added during by the LLVM backend   |
 ++--+
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D122636: [llvm][clang][NFC] Add missing references to MisExpect in TOC

[Paul Kirth via Phabricator via cfe-commits]

paulkirth created this revision.
Herald added a subscriber: arphaman.
Herald added a project: All.
paulkirth requested review of this revision.
Herald added projects: clang, LLVM.
Herald added subscribers: llvm-commits, cfe-commits.

Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D122636

Files:
  clang/docs/index.rst
  llvm/docs/UserGuides.rst


Index: llvm/docs/UserGuides.rst
===
--- llvm/docs/UserGuides.rst
+++ llvm/docs/UserGuides.rst
@@ -44,6 +44,7 @@
MemorySSA
MergeFunctions
MCJITDesignAndImplementation
+   MisExpect
ORCv2
OpaquePointers
JITLink
Index: clang/docs/index.rst
===
--- clang/docs/index.rst
+++ clang/docs/index.rst
@@ -42,6 +42,7 @@
SourceBasedCodeCoverage
Modules
MSVCCompatibility
+   MisExpect
OpenCLSupport
OpenMPSupport
SYCLSupport


Index: llvm/docs/UserGuides.rst
===
--- llvm/docs/UserGuides.rst
+++ llvm/docs/UserGuides.rst
@@ -44,6 +44,7 @@
MemorySSA
MergeFunctions
MCJITDesignAndImplementation
+   MisExpect
ORCv2
OpaquePointers
JITLink
Index: clang/docs/index.rst
===
--- clang/docs/index.rst
+++ clang/docs/index.rst
@@ -42,6 +42,7 @@
SourceBasedCodeCoverage
Modules
MSVCCompatibility
+   MisExpect
OpenCLSupport
OpenMPSupport
SYCLSupport
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

reverted. I will fix this tomorrow. Sorry for the trouble


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 418924.
paulkirth added a comment.
Herald added a subscriber: arphaman.

Fix missing reference in TOC & fix formatting of tables


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/docs/MisExpect.rst
  clang/docs/ReleaseNotes.rst
  clang/docs/index.rst
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/CodeGenOptions.h
  clang/include/clang/Basic/DiagnosticDriverKinds.td
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/docs/UserGuides.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-threshold.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,285 @@
+; Test misexpect diagnostics handle swich statements, and report source locations correctly
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; DISABLED-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; DI

[PATCH] D122636: [llvm][clang][NFC] Add missing references to MisExpect in TOC

[Paul Kirth via Phabricator via cfe-commits]

paulkirth abandoned this revision.
paulkirth added a comment.

Since I had to revert, these changes will be incorporated into the main patch 
when it lands again


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D122636/new/

https://reviews.llvm.org/D122636

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 419181.
paulkirth added a comment.

Rebase


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/docs/MisExpect.rst
  clang/docs/ReleaseNotes.rst
  clang/docs/index.rst
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/CodeGenOptions.h
  clang/include/clang/Basic/DiagnosticDriverKinds.td
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/docs/UserGuides.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-threshold.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,285 @@
+; Test misexpect diagnostics handle swich statements, and report source locations correctly
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; DISABLED-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; DISABLED-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from us

[PATCH] D119996: [safestack] Support safestack in stack size diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 419222.
paulkirth added a comment.

Rebase


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D119996/new/

https://reviews.llvm.org/D119996

Files:
  clang/test/Frontend/stack-usage-safestack.c
  llvm/include/llvm/CodeGen/MachineFrameInfo.h
  llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
  llvm/lib/CodeGen/MachineFunction.cpp
  llvm/lib/CodeGen/PrologEpilogInserter.cpp
  llvm/lib/CodeGen/SafeStack.cpp
  llvm/test/Transforms/SafeStack/ARM/debug.ll

Index: llvm/test/Transforms/SafeStack/ARM/debug.ll
===
--- llvm/test/Transforms/SafeStack/ARM/debug.ll
+++ llvm/test/Transforms/SafeStack/ARM/debug.ll
@@ -10,8 +10,8 @@
 ; void Capture(char*x);
 ; void f() { char c[16]; Capture(c); }
 
-; CHECK: !35 = !DILocation(line: 3, column: 11, scope: !17, inlinedAt: !36)
-; CHECK: !36 = distinct !DILocation(line: 6, scope: !27)
+; CHECK: !36 = !DILocation(line: 3, column: 11, scope: !17, inlinedAt: !37)
+; CHECK: !37 = distinct !DILocation(line: 6, scope: !27)
 
 @addr = common local_unnamed_addr global i8*** null, align 4, !dbg !0
 
Index: llvm/lib/CodeGen/SafeStack.cpp
===
--- llvm/lib/CodeGen/SafeStack.cpp
+++ llvm/lib/CodeGen/SafeStack.cpp
@@ -48,6 +48,7 @@
 #include "llvm/IR/IntrinsicInst.h"
 #include "llvm/IR/Intrinsics.h"
 #include "llvm/IR/MDBuilder.h"
+#include "llvm/IR/Metadata.h"
 #include "llvm/IR/Module.h"
 #include "llvm/IR/Type.h"
 #include "llvm/IR/Use.h"
@@ -642,6 +643,13 @@
   // FIXME: no need to update BasePointer in leaf functions.
   unsigned FrameSize = alignTo(SSL.getFrameSize(), StackAlignment);
 
+  MDBuilder MDB(F.getContext());
+  SmallVector Data;
+  Data.push_back(MDB.createString("unsafe-stack-size"));
+  Data.push_back(MDB.createConstant(ConstantInt::get(Int32Ty, FrameSize)));
+  MDNode *MD = MDTuple::get(F.getContext(), Data);
+  F.setMetadata(LLVMContext::MD_annotation, MD);
+
   // Update shadow stack pointer in the function epilogue.
   IRB.SetInsertPoint(BasePointer->getNextNode());
 
Index: llvm/lib/CodeGen/PrologEpilogInserter.cpp
===
--- llvm/lib/CodeGen/PrologEpilogInserter.cpp
+++ llvm/lib/CodeGen/PrologEpilogInserter.cpp
@@ -283,6 +283,9 @@
 assert(!Failed && "Invalid warn-stack-size fn attr value");
 (void)Failed;
   }
+  if (MF.getFunction().hasFnAttribute(Attribute::SafeStack)) {
+StackSize += MFI.getUnsafeStackSize();
+  }
   if (StackSize > Threshold) {
 DiagnosticInfoStackSize DiagStackSize(F, StackSize, Threshold, DS_Warning);
 F.getContext().diagnose(DiagStackSize);
Index: llvm/lib/CodeGen/MachineFunction.cpp
===
--- llvm/lib/CodeGen/MachineFunction.cpp
+++ llvm/lib/CodeGen/MachineFunction.cpp
@@ -107,6 +107,27 @@
   llvm_unreachable("Invalid machine function property");
 }
 
+void setUnsafeStackSize(const Function &F, MachineFrameInfo &FrameInfo) {
+  if (!F.hasFnAttribute(Attribute::SafeStack))
+return;
+
+  auto *Existing =
+  dyn_cast_or_null(F.getMetadata(LLVMContext::MD_annotation));
+
+  if (!Existing || Existing->getNumOperands() != 2)
+return;
+
+  auto *MetadataName = "unsafe-stack-size";
+  if (auto &N = Existing->getOperand(0)) {
+if (cast(N.get())->getString() == MetadataName) {
+  if (auto &Op = Existing->getOperand(1)) {
+auto Val = mdconst::extract(Op)->getZExtValue();
+FrameInfo.setUnsafeStackSize(Val);
+  }
+}
+  }
+}
+
 // Pin the vtable to this file.
 void MachineFunction::Delegate::anchor() {}
 
@@ -175,6 +196,8 @@
   /*ForcedRealign=*/CanRealignSP &&
   F.hasFnAttribute(Attribute::StackAlignment));
 
+  setUnsafeStackSize(F, *FrameInfo);
+
   if (F.hasFnAttribute(Attribute::StackAlignment))
 FrameInfo->ensureMaxAlignment(*F.getFnStackAlign());
 
Index: llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
===
--- llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
+++ llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
@@ -1186,7 +1186,8 @@
   OutStreamer->SwitchSection(StackSizeSection);
 
   const MCSymbol *FunctionSymbol = getFunctionBegin();
-  uint64_t StackSize = FrameInfo.getStackSize();
+  uint64_t StackSize =
+  FrameInfo.getStackSize() + FrameInfo.getUnsafeStackSize();
   OutStreamer->emitSymbolValue(FunctionSymbol, TM.getProgramPointerSize());
   OutStreamer->emitULEB128IntValue(StackSize);
 
@@ -1201,7 +1202,8 @@
 return;
 
   const MachineFrameInfo &FrameInfo = MF.getFrameInfo();
-  uint64_t StackSize = FrameInfo.getStackSize();
+  uint64_t StackSize =
+  FrameInfo.getStackSize() + FrameInfo.getUnsafeStackSize();
 
   if (StackUsageStream == nullptr) {
 std::error_code EC;
Index: llvm/include/llvm/CodeGen/MachineFrameInfo.h

[PATCH] D119996: [safestack] Support safestack in stack size diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 419493.
paulkirth added a comment.
Herald added a subscriber: pengfei.

Update tests.

- Dont rely on size of `int` based on platform
- Add checking to backend tests for warn-stack-size to ensure the behavior is 
consistant when safestack is enabled


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D119996/new/

https://reviews.llvm.org/D119996

Files:
  clang/test/Frontend/stack-usage-safestack.c
  llvm/include/llvm/CodeGen/MachineFrameInfo.h
  llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
  llvm/lib/CodeGen/MachineFunction.cpp
  llvm/lib/CodeGen/PrologEpilogInserter.cpp
  llvm/lib/CodeGen/SafeStack.cpp
  llvm/test/CodeGen/X86/warn-stack.ll
  llvm/test/Transforms/SafeStack/ARM/debug.ll

Index: llvm/test/Transforms/SafeStack/ARM/debug.ll
===
--- llvm/test/Transforms/SafeStack/ARM/debug.ll
+++ llvm/test/Transforms/SafeStack/ARM/debug.ll
@@ -10,8 +10,8 @@
 ; void Capture(char*x);
 ; void f() { char c[16]; Capture(c); }
 
-; CHECK: !35 = !DILocation(line: 3, column: 11, scope: !17, inlinedAt: !36)
-; CHECK: !36 = distinct !DILocation(line: 6, scope: !27)
+; CHECK: !36 = !DILocation(line: 3, column: 11, scope: !17, inlinedAt: !37)
+; CHECK: !37 = distinct !DILocation(line: 6, scope: !27)
 
 @addr = common local_unnamed_addr global i8*** null, align 4, !dbg !0
 
Index: llvm/test/CodeGen/X86/warn-stack.ll
===
--- llvm/test/CodeGen/X86/warn-stack.ll
+++ llvm/test/CodeGen/X86/warn-stack.ll
@@ -21,4 +21,17 @@
   ret void
 }
 
+; Ensure that warn-stack-size also considers the size of the unsafe stack.
+; With safestack enabled the machine stack size is well below 80, but the
+; combined stack size of the machine stack and unsafe stack will exceed the
+; warning threshold
+
+; CHECK: warning: stack frame size (120) exceeds limit (80) in function 'warn_safestack'
+define void @warn_safestack() nounwind ssp safestack "warn-stack-size"="80" {
+entry:
+  %buffer = alloca [80 x i8], align 1
+  %arraydecay = getelementptr inbounds [80 x i8], [80 x i8]* %buffer, i64 0, i64 0
+  call void @doit(i8* %arraydecay) nounwind
+  ret void
+}
 declare void @doit(i8*)
Index: llvm/lib/CodeGen/SafeStack.cpp
===
--- llvm/lib/CodeGen/SafeStack.cpp
+++ llvm/lib/CodeGen/SafeStack.cpp
@@ -48,6 +48,7 @@
 #include "llvm/IR/IntrinsicInst.h"
 #include "llvm/IR/Intrinsics.h"
 #include "llvm/IR/MDBuilder.h"
+#include "llvm/IR/Metadata.h"
 #include "llvm/IR/Module.h"
 #include "llvm/IR/Type.h"
 #include "llvm/IR/Use.h"
@@ -642,6 +643,13 @@
   // FIXME: no need to update BasePointer in leaf functions.
   unsigned FrameSize = alignTo(SSL.getFrameSize(), StackAlignment);
 
+  MDBuilder MDB(F.getContext());
+  SmallVector Data;
+  Data.push_back(MDB.createString("unsafe-stack-size"));
+  Data.push_back(MDB.createConstant(ConstantInt::get(Int32Ty, FrameSize)));
+  MDNode *MD = MDTuple::get(F.getContext(), Data);
+  F.setMetadata(LLVMContext::MD_annotation, MD);
+
   // Update shadow stack pointer in the function epilogue.
   IRB.SetInsertPoint(BasePointer->getNextNode());
 
Index: llvm/lib/CodeGen/PrologEpilogInserter.cpp
===
--- llvm/lib/CodeGen/PrologEpilogInserter.cpp
+++ llvm/lib/CodeGen/PrologEpilogInserter.cpp
@@ -283,6 +283,9 @@
 assert(!Failed && "Invalid warn-stack-size fn attr value");
 (void)Failed;
   }
+  if (MF.getFunction().hasFnAttribute(Attribute::SafeStack)) {
+StackSize += MFI.getUnsafeStackSize();
+  }
   if (StackSize > Threshold) {
 DiagnosticInfoStackSize DiagStackSize(F, StackSize, Threshold, DS_Warning);
 F.getContext().diagnose(DiagStackSize);
Index: llvm/lib/CodeGen/MachineFunction.cpp
===
--- llvm/lib/CodeGen/MachineFunction.cpp
+++ llvm/lib/CodeGen/MachineFunction.cpp
@@ -107,6 +107,27 @@
   llvm_unreachable("Invalid machine function property");
 }
 
+void setUnsafeStackSize(const Function &F, MachineFrameInfo &FrameInfo) {
+  if (!F.hasFnAttribute(Attribute::SafeStack))
+return;
+
+  auto *Existing =
+  dyn_cast_or_null(F.getMetadata(LLVMContext::MD_annotation));
+
+  if (!Existing || Existing->getNumOperands() != 2)
+return;
+
+  auto *MetadataName = "unsafe-stack-size";
+  if (auto &N = Existing->getOperand(0)) {
+if (cast(N.get())->getString() == MetadataName) {
+  if (auto &Op = Existing->getOperand(1)) {
+auto Val = mdconst::extract(Op)->getZExtValue();
+FrameInfo.setUnsafeStackSize(Val);
+  }
+}
+  }
+}
+
 // Pin the vtable to this file.
 void MachineFunction::Delegate::anchor() {}
 
@@ -175,6 +196,8 @@
   /*ForcedRealign=*/CanRealignSP &&
   F.hasFnAttribute(Attribute::StackAlignment));
 
+  setUnsafeStackSize(F, *FrameInfo);
+
   if (F.hasFnAttrib

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
Closed by commit rG46774df30715: [misexpect] Re-implement MisExpect Diagnostics 
(authored by paulkirth).

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/docs/MisExpect.rst
  clang/docs/ReleaseNotes.rst
  clang/docs/index.rst
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/CodeGenOptions.h
  clang/include/clang/Basic/DiagnosticDriverKinds.td
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/docs/UserGuides.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-threshold.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,285 @@
+; Test misexpect diagnostics handle swich statements, and report source locations correctly
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

@jgorbe Sorry for the trouble. Thank you for the backtrace and the revert. 
Indeed, it seems like I've missed an assumption w.r.t.  over/underflow, and 
will have to sort that out before re-landing.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D130807: [InstrProf] Add the omitprofile attribute

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Do you expect the difference between `noprofile` and `omitprofile` to be 
confusing to users? I can certainly see how users could be confused by what the 
difference is between `noprofile` and `omitprofile` ...

Since what you want to communicate is "never profile"(which `noprofile` can 
probably communicate as is) and "never profile this, but allow inlining of 
profiled functions" (which I'm not sure `omitprofile` communicates), then maybe 
there is a more obvious way we could communicate that? Maybe `neverprofile` and 
`nodirectprofile` are more descriptive names? I don't love the idea of changing 
an existing attribute name, but we can transparently upgrade existing uses for 
backwards compatibility if we have to.  What do you think?

I also think this will need to need to be supported in LLVM's passes, right? So 
the instrumentation passes in `PGOInstrumentation.cpp` (I can't remember if 
'InstrProfiling.cpp` will need this too), and probably the Inlining passes too, 
right?  I assume those will be follow up patches, but I don't see them in the 
current stack. Is that an accurate assumption?

BTW I like the direction of these patches, I think adding the ability to 
differentiate these cases will add a lot of value to the profiling runtimes + 
coverage. :)


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D130807/new/

https://reviews.llvm.org/D130807

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D130807: [InstrProf] Add the omitprofile attribute

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.



> I agree that `omitprofile` probably doesn't communicate the correct meaning 
> very well. I like `nodirectprofile` because it hopefully implies that 
> indirect profiles are allowed. I'm wondering if anyone else has suggestions.
>
> Is there any precedent for renaming function attributes? My assumption was 
> that users would add the `noprofile` attribute to source mostly for 
> correctness concerns and `noprofile` is easy to remember and already in use, 
> so I'd rather not change it.

I agree with you that it's probably best to avoid changing the spelling of 
`noprofile`, but since you're making a change in the area, we may as well 
discuss if 1) that is still the best name, and 2) if we want to keep it.

I don't know if we've ever changed the spelling of a `FnAttribute` before. IIRC 
we've dropped them, and changed their format e.g., from being a boolean to 
taking a value or an enum. At least that's my memory, which seems to correspond 
to the autoupgrade stuff in `BitcodeReader`. But I think the project's stance 
on this(like most things) is that "if there's a good reason, then we should do 
it." At least if we can maintain our stated compatibility guarantees. This is a 
place where I think we should get additional feedback before proceeding though.

It might be worth asking about this on discourse. Even if we don't want to make 
such a change in these patches, it may be good for the project to have a policy 
(or at least guidelines) about changing attributes.

> I actually do have a change in `PGOInstrumentation.cpp` that skips profiling 
> functions with this new attribute. As far as I know, this stack has all the 
> necessary code changes.

I can't believe I missed that. I looked three times before I commented, and 
blew right passed it every time. That's what happens when I look at code late 
on a Friday...


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D130807/new/

https://reviews.llvm.org/D130807

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D131306: [llvm][misexpect] Track provenance of branch weights

[Paul Kirth via Phabricator via cfe-commits]

paulkirth created this revision.
paulkirth added a reviewer: tejohnson.
Herald added subscribers: ormris, okura, kuter, hiraditya.
Herald added a project: All.
paulkirth requested review of this revision.
Herald added a reviewer: jdoerfert.
Herald added a reviewer: sstefan1.
Herald added projects: clang, LLVM.
Herald added subscribers: llvm-commits, cfe-commits.

MisExpect needs to know if a branch weight intrinsic originates from an
llvm.expect intrinsic.

This patch allows us to track that provenance by adding a new
metadata type that can be moved in concert with the existing branch
weights. The new metadata is copied whenever branch weights are copied,
and is removed when new branch weights are added.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D131306

Files:
  clang/test/CodeGenCXX/attr-likelihood-if-branch-weights.cpp
  clang/test/CodeGenCXX/attr-likelihood-if-vs-builtin-expect.cpp
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/lib/Transforms/IPO/ArgumentPromotion.cpp
  llvm/lib/Transforms/IPO/Attributor.cpp
  llvm/lib/Transforms/IPO/DeadArgumentElimination.cpp
  llvm/lib/Transforms/IPO/LowerTypeTests.cpp
  llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
  llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/Local.cpp
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
  llvm/test/Transforms/LowerExpectIntrinsic/expect-with-probability.ll

Index: llvm/test/Transforms/LowerExpectIntrinsic/expect-with-probability.ll
===
--- llvm/test/Transforms/LowerExpectIntrinsic/expect-with-probability.ll
+++ llvm/test/Transforms/LowerExpectIntrinsic/expect-with-probability.ll
@@ -138,7 +138,7 @@
   %conv1 = sext i32 %conv to i64
   %expval = call i64 @llvm.expect.with.probability.i64(i64 %conv1, i64 0, double 8.00e-01)
   %tobool = icmp ne i64 %expval, 0
-; CHECK: !prof !1
+; CHECK: !prof !2
 ; CHECK-NOT: @llvm.expect.with.probability
   br i1 %tobool, label %if.then, label %if.end
 
@@ -165,7 +165,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.with.probability.i64(i64 %conv, i64 2, double 8.00e-01)
-; CHECK: !prof !2
+; CHECK: !prof !3
 ; CHECK-NOT: @llvm.expect.with.probability
   switch i64 %expval, label %sw.epilog [
 i64 1, label %sw.bb
@@ -194,7 +194,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.with.probability.i64(i64 %conv, i64 1, double 8.00e-01)
-; CHECK: !prof !3
+; CHECK: !prof !4
 ; CHECK-NOT: @llvm.expect.with.probability
   switch i64 %expval, label %sw.epilog [
 i64 2, label %sw.bb
@@ -278,7 +278,7 @@
   %t7 = call i64 @llvm.expect.with.probability.i64(i64 %t6, i64 0, double 8.00e-01)
   %t8 = icmp ne i64 %t7, 0
   %t9 = select i1 %t8, i32 1, i32 2
-; CHECK: select{{.*}}, !prof !1
+; CHECK: select{{.*}}, !prof !2
   ret i32 %t9
 }
 
@@ -286,6 +286,6 @@
 declare i1 @llvm.expect.with.probability.i1(i1, i1, double) nounwind readnone
 
 ; CHECK: !0 = !{!"branch_weights", i32 1717986918, i32 429496731}
-; CHECK: !1 = !{!"branch_weights", i32 429496731, i32 1717986918}
-; CHECK: !2 = !{!"branch_weights", i32 214748366, i32 214748366, i32 1717986918}
-; CHECK: !3 = !{!"branch_weights", i32 1717986918, i32 214748366, i32 214748366}
+; CHECK: !2 = !{!"branch_weights", i32 429496731, i32 1717986918}
+; CHECK: !3 = !{!"branch_weights", i32 214748366, i32 214748366, i32 1717986918}
+; CHECK: !4 = !{!"branch_weights", i32 1717986918, i32 214748366, i32 214748366}
Index: llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
===
--- llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
+++ llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
@@ -138,7 +138,7 @@
   %conv1 = sext i32 %conv to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv1, i64 0)
   %tobool = icmp ne i64 %expval, 0
-; CHECK: !prof !1
+; CHECK: !prof !2
 ; CHECK-NOT: @llvm.expect
   br i1 %tobool, label %if.then, label %if.end
 
@@ -165,7 +165,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv, i64 2)
-; CHECK: !prof !2
+; CHECK: !prof !3
 ; CHECK-NOT: @llvm.expect
   switch i64 %expval, label %sw.epilog [
 i64 1, label %sw.bb
@@ -194,7 +194,7 @@
   %tmp = load i32, i32* %x.addr, align 4
   %conv = sext i32 %tmp to i64
   %expval = call i64 @llvm.expect.i64(i64 %conv, i64 1)
-; CHECK: !prof !3
+; CHECK: !prof !4
 ; CHECK-NOT: @llvm.expect
   switch i64 %expval, label %sw.epilog [
 i64 2, label %sw.bb
@@ -278,7 +278,7 @@
   %t7 = call i64 @llvm.expect.i64(i64 %t6, i64 0)
   %t8 = icmp ne i64 %t7, 0
   %t9 = select i1 %t8, i32 1, i32 2
-; CHECK: select{{.*}}, !prof !1
+; CHECK: select{{.*}}, !prof !2
   ret i32 %t9
 }
 
@@

[PATCH] D131298: [clang-doc] Read docstrings for record members

[Paul Kirth via Phabricator via cfe-commits]

paulkirth requested changes to this revision.
paulkirth added a comment.
This revision now requires changes to proceed.

I agree w/ @phosek on unit testing.

Additionally, while we don't have many of them in clang-doc, an end-to-end test 
would also be welcome.  In this case, it will also probably help w/ review, 
since we can see how clang-doc output is changing as part of the review 
process.  To be clear: I'm not going to block acceptance on the end-to-end 
tests, since that may be asking too much. It would just be helpful.

I've left more direct feedback inline.




Comment at: clang-tools-extra/clang-doc/BitcodeReader.cpp:354
+template <> llvm::Expected getCommentInfo(MemberTypeInfo *I) {
+  I->Description.emplace_back();
+  return &I->Description.back();

So, I see that this uses the same pattern as in other `getCommentInfo(...)` 
API's, but I'm a bit confused by this.

Do we always grow the vector whenever `getCommentInfo(...)` is  called? I would 
expect a `get...` function to be `const` and just return data. This grows the 
`Description` vector on every call, which seems like an odd choice. The pattern 
is also pervasive in BitcodeReader.cpp.

@phosek is this intentional? If `Description` exceeds capacity and reallocs on 
`emplace_back`, then any reference it had returned would be invalidated, right? 
Or am I missing something here re: the various `*TypeInfos` and how clang-doc 
uses them?



Comment at: clang-tools-extra/clang-doc/Serialize.cpp:439
+static void populateMemberTypeInfo(MemberTypeInfo &I, const FieldDecl *D) {
+  ASTContext& Context = D->getASTContext();
+  RawComment *Comment = Context.getRawCommentForDeclNoCache(D);

I think we need to at least assert that `D != nullptr`. While there is only one 
caller now, its likely there may be more in the future. Plus `nullptr` is deref 
is never fun to debug.  

If getting a `nullptr` is expected to be a common input, then an early return 
is also fine.

If you prefer, a few places in the codebase do both to maintain correct 
behavior when asserts are disabled. Often they have a form along these lines:

```
if(!D){
  assert(false && "Some error message");
  return;
}
```

Any of these would be fine, and I don't have a preference between them.




Comment at: clang-tools-extra/clang-doc/Serialize.cpp:440
+  ASTContext& Context = D->getASTContext();
+  RawComment *Comment = Context.getRawCommentForDeclNoCache(D);
+  if (!Comment)

If the goal is to get the FullComment, then doesn't `getCommentForDecl(...) ` 
handle that? This also seems like an area where we'd want to use caching if we 
could.



Comment at: clang-tools-extra/clang-doc/YAMLGenerator.cpp:195
 IO.mapOptional("Access", I.Access, clang::AccessSpecifier::AS_none);
+IO.mapOptional("Description", I.Description);
   }

Can we get a comment describing this?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131298/new/

https://reviews.llvm.org/D131298

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D131298: [clang-doc] Read docstrings for record members

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added inline comments.



Comment at: clang-tools-extra/clang-doc/Serialize.cpp:439
+static void populateMemberTypeInfo(MemberTypeInfo &I, const FieldDecl *D) {
+  ASTContext& Context = D->getASTContext();
+  RawComment *Comment = Context.getRawCommentForDeclNoCache(D);

brettw wrote:
> paulkirth wrote:
> > I think we need to at least assert that `D != nullptr`. While there is only 
> > one caller now, its likely there may be more in the future. Plus `nullptr` 
> > is deref is never fun to debug.  
> > 
> > If getting a `nullptr` is expected to be a common input, then an early 
> > return is also fine.
> > 
> > If you prefer, a few places in the codebase do both to maintain correct 
> > behavior when asserts are disabled. Often they have a form along these 
> > lines:
> > 
> > ```
> > if(!D){
> >   assert(false && "Some error message");
> >   return;
> > }
> > ```
> > 
> > Any of these would be fine, and I don't have a preference between them.
> > 
> I'm fine adding an assert to unblock the review.
> 
> But clang-doc is full of pointers and none of them are asserted on. Is this 
> one special in some way?  To me, it seems weird to assert on this one random 
> pointer. Is there some policy about this?
> 
> I follow the opposite theory that asserts on null pointers all over the place 
> clutter the code and random null parameters are one of the easiest types of 
> crashes to debug.
There's nothing special regarding this particular pointer, but such checks are 
common in other parts of LLVM.

We use asserts liberally throughout the LLVM codebase 
(https://llvm.org/docs/CodingStandards.html#assert-liberally) and asserting 
that a pointer is valid is common. 

The other thing to keep in mind is that debug builds of LLVM are > 20GB, and 
enabling asserts is a more desirable choice for most of our developers. That 
probably isn't as bad for clang-doc, but I'd rather err on the side of caution.

While I agree that right now it may be a bit strange to have a single assert,  
hopefully overtime the surrounding code will start to use such checks in a way 
that is more consistent with the rest of the codebase.

But like I said, an early return is also fine IMO.




Comment at: clang-tools-extra/clang-doc/YAMLGenerator.cpp:195
 IO.mapOptional("Access", I.Access, clang::AccessSpecifier::AS_none);
+IO.mapOptional("Description", I.Description);
   }

brettw wrote:
> paulkirth wrote:
> > Can we get a comment describing this?
> I don't think this needs a comment, a substantial part of this file is just 
> adding these mappings between YAML keys and field names. The one above it has 
> a comment only because of the non-obvious default value. If you feel 
> strongly, please let me know what the comment should say.
Now that I see this in context, I agree that it doesn't need a comment.



Comment at: clang-tools-extra/unittests/clang-doc/SerializeTest.cpp:197
   Method.IsMethod = true;
+
   ExpectedRecordWithMethod.ChildFunctions.emplace_back(std::move(Method));

?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131298/new/

https://reviews.llvm.org/D131298

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D131298: [clang-doc] Read docstrings for record members

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

LGTM, modulo the small fixes I've left inline




Comment at: clang-tools-extra/clang-doc/Serialize.cpp:439
+static void populateMemberTypeInfo(MemberTypeInfo &I, const FieldDecl *D) {
+  assert(D);
+  ASTContext& Context = D->getASTContext();

Can you add a String message to the assert?

This is a fairly standard example of the usage w/in LLVM:
```assert(LCS && RCS && "Expect non-null FunctionSamples");```
Taken from 
https://github.com/llvm/llvm-project/blob/30bbb73bb448910f791088bfc3154e752d42241a/llvm/lib/Transforms/IPO/SampleProfile.cpp#L400





Comment at: clang-tools-extra/clang-doc/Serialize.cpp:440
+  ASTContext& Context = D->getASTContext();
+  RawComment *Comment = Context.getRawCommentForDeclNoCache(D);
+  if (!Comment)

brettw wrote:
> paulkirth wrote:
> > If the goal is to get the FullComment, then doesn't `getCommentForDecl(...) 
> > ` handle that? This also seems like an area where we'd want to use caching 
> > if we could.
> I have no idea how this works, I copied this snipped from 
> MapASTVisitor::getComment in Mapper.cpp
That's fine, but lets add a TODO here to consider the other API, so we track 
this.



Comment at: clang-tools-extra/unittests/clang-doc/SerializeTest.cpp:166
+  ExpectedE.Members.emplace_back("int", "value", AccessSpecifier::AS_public);
+  //ExpectedE.Members.back().Description.push_back(MakeOneLineCommentInfo(" 
Some docs"));
   CheckRecordInfo(&ExpectedE, E);

brettw wrote:
> I spent quite some time trying to get this to work and I'm stuck. If I copy 
> the contents of the raw string above into a file and run clang-doc on it, I 
> get a comment in the YAML output. But in this test I never get a comment and 
> I don't know how this parsing environment is different than the "real" one.
> 
> I'm very confused by the result of `ExtractInfosFromCode` function in the 
> first place. What are the 1st and 6th item which are never checked? Why does 
> it emit two duplicate records for the `E` struct, one (Infos[2]) with a data 
> member and no function and one (Infos[3]) with no data member and a function 
> in it?
> 
> Any suggestions here would be appreciated.
So I also wanted to understand why. `MakeOneLinecCmmentIInfo()` is not used 
elsewhere in clang-doc, and I don't think it was ever tested.

Looking at the body it doesn't use it's input parameter at all. 

There may also be an issue w/ how it constructs the `CommentInfo`.

Leaving this as a TODO is fine. I don't think this is something we're going to 
solve in this patch, but if you can also add a TODO to `MakeOneLine 
CommentInfo`, it would be appreciated.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131298/new/

https://reviews.llvm.org/D131298

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D131298: [clang-doc] Read docstrings for record members

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Thanks. Can we get the small changes to the assert + the TODO in Serialize.cpp? 
I'll be happy to land this for you after that.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131298/new/

https://reviews.llvm.org/D131298

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 414230.
paulkirth added a comment.

Consolodate common code, clarify documentation, and address comments


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,290 @@
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; New PM
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.c.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=CORRECT
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth marked 6 inline comments as done.
paulkirth added inline comments.



Comment at: clang/lib/CodeGen/CodeGenAction.cpp:1203
 
+  if (CodeGenOpts.MisExpect) {
+Ctx.setMisExpectWarningRequested(true);

tejohnson wrote:
> Out of curiosity, since I am less familiar with clang than llvm, when is this 
> path taken vs where this is done at line 336 in HandleTranslationUnit?
> 
> Also, it might be nice to have this code placed either consistently before or 
> after the OptRecordFile handling in these two locations, so it is easier to 
> compare.
That was a good catch, I had forgotten to remove this after patching in line 
366. I moved that a bit a few lines lower to fit more nicely, and removed this 
one entirely.



Comment at: llvm/docs/MisExpect.rst:2
+===
+Misexpect
+===

tejohnson wrote:
> I don't see a mention of -Wmisexpect in the document. Should there be 
> user-facing clang documentation, this seems more specific to the LLVM 
> implementation?
I will update the patch to include similar documentation this for clang.

Additionally, my understanding is that documentation for warnings is 
auto-generated from the tablegen, so that at least will be available 
automatically.





Comment at: llvm/docs/MisExpect.rst:22
+The MisExpect checks in the LLVM backend follow a simple procedure: if the
+profiling counter associated with an instruction using the ``llvm.expect``
+intrinsic was too low along the expected path, then it emits a diagnostic

tejohnson wrote:
> Suggest using "profile weight" not profiling counter since we don't really 
> have profile counters associated with instructions (rather with edges in the 
> MST, at least in the IR PGO).
> 
> Also, doesn't the profile weight being too low only handle the LIKELY case? 
> What about something being marked UNLIKELY with a hot/high profile weight? 
> edit: After looking through the implementation, I think the comparison only 
> being done in this direction is based on the way it is implemented, but it is 
> unclear from this documentation here how the comparison is handling things 
> being off in either direction.
Maybe my mental model is off here, but doesn't the `llvm.expect` intrinsic mark 
a specific value as the 'likely' or expected value? So if you want to mark a 
branch as unlikely, you're essentially marking the other half of it as 'hot'.

We could change the comparison to compare all parts of the branch weights too, 
but the case of an 'unlikely' branch being too hot in my model is captured by 
the 'likely' branch becoming too 'cold'.

If that is incorrect, I'd really appreciate some guidance on how to model this 
more accurately.



Comment at: llvm/docs/MisExpect.rst:39
+``-unlikely-branch-weight`` LLVM options. During verification, if the
+profile count is less than the calculated threshold, then we will emit a
+remark or warning detailing a potential performance regression. The

tejohnson wrote:
> s/count/weight/
> 
> Also, similar to my earlier comment, what about an expect UNLIKELY branch 
> with a high profile weight?
I think I've reworded this to clarify things a bit more, but let me know if it 
still needs some polish.



Comment at: llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp:109
 
+  SI.setCondition(ArgValue);
+

tejohnson wrote:
> Is there a reason why this line has changed ordering w.r.t. setMetadata?
I think I actually restored this to what it was before the original MisExpect 
patch, but there is no need for that, so I've removed it.



Comment at: llvm/lib/Transforms/Utils/MisExpect.cpp:170
+
+  if (ProfileCount < ScaledThreshold)
+emitMisexpectDiagnostic(I, Ctx, ProfileCount, CaseTotal);

tejohnson wrote:
> Is this too strict - i.e. what if they are off by only a small amount?
Maybe. I don't know what a good default fudge factor would be, though. Maybe 
use 5%, and expose a flag that can override it?

Another approach is that the values used by `llvm.expect` can be changed via 
clang flags, which will update the weights checked against. 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

MisExpect was originally intended to be quite strict, so that developers would 
audit their code and re-evaluate the correctness of their annotations,or if 
they were needed at all.

I think I'm still of a mind that getting flagged by MisExpect indicates that a 
different annotation would be more beneficial, such as 
`llvm.expect.with.probability`.

Regardless, I agree its good to give users an option to relax the checking when 
they want to, so I will add an option that allows them to specify a scaling 
factor for the threshold.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 415990.
paulkirth added a comment.

Address comments.

Added clang documentation.

Fixed various typos.

Updated tests per comments for clang and LLVM.

Removed dead class definition.

Removed clamping from tolerance parser.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/docs/MisExpect.rst
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/CodeGenOptions.h
  clang/include/clang/Basic/DiagnosticDriverKinds.td
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/include/llvm/Transforms/Utils/MisExpectToleranceParser.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-threshold.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,286 @@
+; Test misexpect diagnostics handle swich statements, and report source locations correctly
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.c.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=CORRECT
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:30: Potential performance r

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth marked 10 inline comments as done.
paulkirth added inline comments.



Comment at: clang/test/Profile/misexpect-branch.c:25
+  int x = 0;
+  if (likely(rando % (outer_loop * inner_loop) == 0)) { // exact-warning-re 
{{Potential performance regression from use of __builtin_expect(): Annotation 
was correct on {{.+}}% ({{[0-9]+ / [0-9]+}}) of profiled executions.}}
+x = baz(rando);

tejohnson wrote:
> Can you add a case where an unlikely() is wrong?
I've added one, but I'm not sure it tests anything meaningfully, since an 
unlikely path being too hot is really the same check as a likely path being too 
cold. 



Comment at: llvm/docs/MisExpect.rst:2
+===
+Misexpect
+===

tejohnson wrote:
> paulkirth wrote:
> > tejohnson wrote:
> > > I don't see a mention of -Wmisexpect in the document. Should there be 
> > > user-facing clang documentation, this seems more specific to the LLVM 
> > > implementation?
> > I will update the patch to include similar documentation this for clang.
> > 
> > Additionally, my understanding is that documentation for warnings is 
> > auto-generated from the tablegen, so that at least will be available 
> > automatically.
> > 
> > 
> Should the clang documentation already be added to this patch? I couldn't 
> find it.
In the last update the clang documentation was a bit too close to the llvm 
documentation. I think they are distinct enough now to be separately useful, 
but maybe it's better to merge the two?



Comment at: llvm/docs/MisExpect.rst:54
+
+.. option:: -pgo-warn-misexpect
+

tejohnson wrote:
> Looking at the code, the -pgo-warn-misexpect seems to be useful for internal 
> LLVM testing via 'opt', to mimic -Wmisexpect, so it probably should be 
> documented as such.
Yes, its useful to test llvm, but shouldn't `opt` be usable in the same way? To 
me it seems useful for opt to support the warning directly too, but I'm happy 
to defer here if you think that's confusing or shouldn't be the case.



Comment at: llvm/lib/Transforms/Utils/MisExpect.cpp:78
+
+Instruction *getOprndOrInst(Instruction *I) {
+  assert(I != nullptr && "MisExpect target Instruction cannot be nullptr");

tejohnson wrote:
> Suggest renaming to something more intuitive. E.g. getInstCondition? 
That's definitely a better name. Thanks for the suggestion.



Comment at: llvm/lib/Transforms/Utils/MisExpect.cpp:188
+  auto Tollerance = getMisExpectTollerance(I.getContext());
+  if (Tollerance < 0)
+Tollerance = 0;

tejohnson wrote:
> afaict this handling duplicates what is done in parseTolleranceOption when 
> the tolerance comes in from the clang option. Can you just do this handling 
> once, here?
This is now the only place we check/modify the value.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 416356.
paulkirth marked 4 inline comments as done.
paulkirth added a comment.

Fixup remaining outstanding issues

- fix typos
- add misexpect to clang release notes
- restore new pm style opt invocations in llvm tests
- move single header function into implementation file
- clarify comment in test file


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/docs/MisExpect.rst
  clang/docs/ReleaseNotes.rst
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/CodeGenOptions.h
  clang/include/clang/Basic/DiagnosticDriverKinds.td
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-threshold.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,285 @@
+; Test misexpect diagnostics handle swich statements, and report source locations correctly
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annot

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth marked 5 inline comments as done.
paulkirth added inline comments.



Comment at: llvm/docs/MisExpect.rst:54
+
+.. option:: -pgo-warn-misexpect
+

tejohnson wrote:
> paulkirth wrote:
> > tejohnson wrote:
> > > Looking at the code, the -pgo-warn-misexpect seems to be useful for 
> > > internal LLVM testing via 'opt', to mimic -Wmisexpect, so it probably 
> > > should be documented as such.
> > Yes, its useful to test llvm, but shouldn't `opt` be usable in the same 
> > way? To me it seems useful for opt to support the warning directly too, but 
> > I'm happy to defer here if you think that's confusing or shouldn't be the 
> > case.
> Oh I'm not disagreeing with having the internal option and using it for opt, 
> that's very useful. My comment was in the context of not having the 
> user-facing clang documentation with the clang driver level option (which you 
> since added). I was just suggesting you add a note that this option is an 
> alternative to the clang option for use when e.g. testing via opt. Since 
> there is now separate clang documentation I think it is less important now.
Thanks for the clarification.



Comment at: llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll:3
+
+; RUN: opt < %s -lower-expect -pgo-instr-use 
-pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect 
-pass-remarks=misexpect 2>&1 | FileCheck %s
+

tejohnson wrote:
> tejohnson wrote:
> > The old PM is gone in the optimization pipeline, so these old PM lines in 
> > this and other tests and the comment about the new PM can go away.
> I see you removed the NewPM-style invocations and kept the legacy ones - 
> afaict opt will now convert the legacy style pass invocations to the new PM, 
> but I think it is preferable to use the New PM style invocations.
yes, I misread your earlier comment.  I have restored the New PM style opt 
invocations, and removed the legacy PM ones.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
Closed by commit rGe7749d4713a5: [misexpect] Re-implement MisExpect Diagnostics 
(authored by paulkirth).

Changed prior to commit:
  https://reviews.llvm.org/D115907?vs=416356&id=416360#toc

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

Files:
  clang/docs/MisExpect.rst
  clang/docs/ReleaseNotes.rst
  clang/include/clang/Basic/CodeGenOptions.def
  clang/include/clang/Basic/CodeGenOptions.h
  clang/include/clang/Basic/DiagnosticDriverKinds.td
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/include/clang/Driver/Options.td
  clang/lib/CodeGen/BackendUtil.cpp
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch-nonconst.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch-unpredictable.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/docs/MisExpect.rst
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/LLVMContext.h
  llvm/include/llvm/Target/TargetOptions.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/LLVMContext.cpp
  llvm/lib/IR/LLVMContextImpl.h
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-threshold.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-stripped.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch-unpredictable.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,285 @@
+; Test misexpect diagnostics handle swich statements, and report source locations correctly
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:30: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00% (0 / 8112) of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:30: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:30: Potential performance regression from use of the llvm.expect intrinsic: Annotation was correct on 0.00%

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

These pass for me locally, so I'm reverting for now and will dig into this 
tomorrow.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D123345: Treat `std::move`, `forward`, and `move_if_noexcept` as builtins.

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Hi, this is also breaking Fuchsia's clang CI builders 
(https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-linux-x64/b8816531831869640417/overview).
 If this will be hard to address, would you mind reverting until a patch is 
ready?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D123345/new/

https://reviews.llvm.org/D123345

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D115907: [misexpect] Re-implement MisExpect Diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Ah, thanks. I've had to try to re-land this so many times, I wanted to be sure 
the pre-submit was looking OK after rebasing. or at least be sure it didn't 
look like it was failing from any of my changes.

w.r.t. `clamp`, keeping it compatible was my intent.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115907/new/

https://reviews.llvm.org/D115907

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D123840: [Clang][Sema] Fix invalid redefinition error in if/switch/for statement

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Hi I think this patch is the root cause of 
https://github.com/llvm/llvm-project/issues/54968

The issue first appears in our builders here:
https://luci-milo.appspot.com/ui/p/fuchsia/builders/ci/clang_toolchain.ci.core.x64-release/b8816803488685972465/overview

based on the blamelist 
(https://luci-milo.appspot.com/ui/p/fuchsia/builders/ci/clang_toolchain.ci.core.x64-release/b8816803488685972465/blamelist)
 I'm fairly certain that this change is responsible.

I'm guessing that there is some additional case the new branch condition is 
missing that causes the assertion failure to manifest.

If this will be hard to fix, can you revert this patch until one is ready?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D123840/new/

https://reviews.llvm.org/D123840

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D119996: [safestack] Support safestack in stack size diagnostics

[Paul Kirth via Phabricator via cfe-commits]

This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
paulkirth marked an inline comment as done.
Closed by commit rG61e36e87df1a: [safestack] Support safestack in stack size 
diagnostics (authored by paulkirth).

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D119996/new/

https://reviews.llvm.org/D119996

Files:
  clang/test/Frontend/stack-usage-safestack.c
  llvm/include/llvm/CodeGen/MachineFrameInfo.h
  llvm/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
  llvm/lib/CodeGen/MachineFunction.cpp
  llvm/lib/CodeGen/PrologEpilogInserter.cpp
  llvm/lib/CodeGen/SafeStack.cpp
  llvm/test/CodeGen/X86/warn-stack.ll
  llvm/test/Transforms/SafeStack/ARM/debug.ll

Index: llvm/test/Transforms/SafeStack/ARM/debug.ll
===
--- llvm/test/Transforms/SafeStack/ARM/debug.ll
+++ llvm/test/Transforms/SafeStack/ARM/debug.ll
@@ -10,8 +10,8 @@
 ; void Capture(char*x);
 ; void f() { char c[16]; Capture(c); }
 
-; CHECK: !35 = !DILocation(line: 3, column: 11, scope: !17, inlinedAt: !36)
-; CHECK: !36 = distinct !DILocation(line: 6, scope: !27)
+; CHECK: !36 = !DILocation(line: 3, column: 11, scope: !17, inlinedAt: !37)
+; CHECK: !37 = distinct !DILocation(line: 6, scope: !27)
 
 @addr = common local_unnamed_addr global i8*** null, align 4, !dbg !0
 
Index: llvm/test/CodeGen/X86/warn-stack.ll
===
--- llvm/test/CodeGen/X86/warn-stack.ll
+++ llvm/test/CodeGen/X86/warn-stack.ll
@@ -21,4 +21,17 @@
   ret void
 }
 
+; Ensure that warn-stack-size also considers the size of the unsafe stack.
+; With safestack enabled the machine stack size is well below 80, but the
+; combined stack size of the machine stack and unsafe stack will exceed the
+; warning threshold
+
+; CHECK: warning: stack frame size (120) exceeds limit (80) in function 'warn_safestack'
+define void @warn_safestack() nounwind ssp safestack "warn-stack-size"="80" {
+entry:
+  %buffer = alloca [80 x i8], align 1
+  %arraydecay = getelementptr inbounds [80 x i8], [80 x i8]* %buffer, i64 0, i64 0
+  call void @doit(i8* %arraydecay) nounwind
+  ret void
+}
 declare void @doit(i8*)
Index: llvm/lib/CodeGen/SafeStack.cpp
===
--- llvm/lib/CodeGen/SafeStack.cpp
+++ llvm/lib/CodeGen/SafeStack.cpp
@@ -48,6 +48,7 @@
 #include "llvm/IR/IntrinsicInst.h"
 #include "llvm/IR/Intrinsics.h"
 #include "llvm/IR/MDBuilder.h"
+#include "llvm/IR/Metadata.h"
 #include "llvm/IR/Module.h"
 #include "llvm/IR/Type.h"
 #include "llvm/IR/Use.h"
@@ -633,6 +634,13 @@
   // FIXME: no need to update BasePointer in leaf functions.
   unsigned FrameSize = alignTo(SSL.getFrameSize(), StackAlignment);
 
+  MDBuilder MDB(F.getContext());
+  SmallVector Data;
+  Data.push_back(MDB.createString("unsafe-stack-size"));
+  Data.push_back(MDB.createConstant(ConstantInt::get(Int32Ty, FrameSize)));
+  MDNode *MD = MDTuple::get(F.getContext(), Data);
+  F.setMetadata(LLVMContext::MD_annotation, MD);
+
   // Update shadow stack pointer in the function epilogue.
   IRB.SetInsertPoint(BasePointer->getNextNode());
 
Index: llvm/lib/CodeGen/PrologEpilogInserter.cpp
===
--- llvm/lib/CodeGen/PrologEpilogInserter.cpp
+++ llvm/lib/CodeGen/PrologEpilogInserter.cpp
@@ -283,6 +283,9 @@
 assert(!Failed && "Invalid warn-stack-size fn attr value");
 (void)Failed;
   }
+  if (MF.getFunction().hasFnAttribute(Attribute::SafeStack)) {
+StackSize += MFI.getUnsafeStackSize();
+  }
   if (StackSize > Threshold) {
 DiagnosticInfoStackSize DiagStackSize(F, StackSize, Threshold, DS_Warning);
 F.getContext().diagnose(DiagStackSize);
Index: llvm/lib/CodeGen/MachineFunction.cpp
===
--- llvm/lib/CodeGen/MachineFunction.cpp
+++ llvm/lib/CodeGen/MachineFunction.cpp
@@ -107,6 +107,27 @@
   llvm_unreachable("Invalid machine function property");
 }
 
+void setUnsafeStackSize(const Function &F, MachineFrameInfo &FrameInfo) {
+  if (!F.hasFnAttribute(Attribute::SafeStack))
+return;
+
+  auto *Existing =
+  dyn_cast_or_null(F.getMetadata(LLVMContext::MD_annotation));
+
+  if (!Existing || Existing->getNumOperands() != 2)
+return;
+
+  auto *MetadataName = "unsafe-stack-size";
+  if (auto &N = Existing->getOperand(0)) {
+if (cast(N.get())->getString() == MetadataName) {
+  if (auto &Op = Existing->getOperand(1)) {
+auto Val = mdconst::extract(Op)->getZExtValue();
+FrameInfo.setUnsafeStackSize(Val);
+  }
+}
+  }
+}
+
 // Pin the vtable to this file.
 void MachineFunction::Delegate::anchor() {}
 
@@ -175,6 +196,8 @@
   /*ForcedRealign=*/CanRealignSP &&
   F.hasFnAttribute(Attribute::StackAlignment));
 
+  setUnsafeStackSize(F, *FrameInfo);
+
   if (F.hasFnAttribut

[PATCH] D119996: [safestack] Support safestack in stack size diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

In D119996#3465735 , @vvereschaka 
wrote:

> Hi @paulkirth,
>
> using of specific triple within `stack-usage-safestack.c` test causes a 
> failure for the compilers, which don't support these triples (arm/aarch64 in 
> my case).
> Such as:
>
>   error: unable to create target: 'No available targets are compatible with 
> triple "i386-apple-darwin"'
>
> see more details in 
> https://lab.llvm.org/buildbot/#/builders/119/builds/8169/steps/9/logs/FAIL__Clang__stack-usage-safestack_c
>  result for the failed build.
>
> would you fix the test by removing these triples from the command line or by 
> isolating this test for specific target with `// REQUIRES:` directive?

Oh, that's surprising. I followed the procedure from another test, so I'm 
surprised that this is failing when that one is not, but maybe I missed the 
`REQUIRES` directive.  I can probably get a change out fairly quickly, but if 
its blocking you, feel free to revert this and we can re-land later.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D119996/new/

https://reviews.llvm.org/D119996

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D124203: [clang][safestack] Remove triple from stack usage test

[Paul Kirth via Phabricator via cfe-commits]

paulkirth created this revision.
Herald added a project: All.
paulkirth requested review of this revision.
Herald added a project: clang.
Herald added a subscriber: cfe-commits.

Supplying the target triple caused breakeages for compilers that don't support
the supplied triple.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D124203

Files:
  clang/test/Frontend/stack-usage-safestack.c


Index: clang/test/Frontend/stack-usage-safestack.c
===
--- clang/test/Frontend/stack-usage-safestack.c
+++ clang/test/Frontend/stack-usage-safestack.c
@@ -1,12 +1,10 @@
 /// Check that stack frame size warnings behave the same when safe stack is 
enabled
 
-// REQUIRES: x86-registered-target
+// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck 
--check-prefix=REGULAR %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck 
--check-prefix=IGNORE %s
 
-// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 
2>&1 | FileCheck --check-prefix=REGULAR %s
-// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 
2>&1 | FileCheck --check-prefix=IGNORE %s
-
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 
-triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=SAFESTACK %s
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 
-triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 2>&1 
| FileCheck --check-prefix=SAFESTACK %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 
2>&1 | FileCheck --check-prefix=IGNORE %s
 
 extern void init(char *buf, int size);
 extern int use_buf(char *buf, int size);


Index: clang/test/Frontend/stack-usage-safestack.c
===
--- clang/test/Frontend/stack-usage-safestack.c
+++ clang/test/Frontend/stack-usage-safestack.c
@@ -1,12 +1,10 @@
 /// Check that stack frame size warnings behave the same when safe stack is enabled
 
-// REQUIRES: x86-registered-target
+// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck --check-prefix=REGULAR %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck --check-prefix=IGNORE %s
 
-// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=REGULAR %s
-// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
-
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=SAFESTACK %s
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck --check-prefix=SAFESTACK %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck --check-prefix=IGNORE %s
 
 extern void init(char *buf, int size);
 extern int use_buf(char *buf, int size);
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D119996: [safestack] Support safestack in stack size diagnostics

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

@vvereschaka I have a fix out in D124203 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D119996/new/

https://reviews.llvm.org/D119996

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D124203: [clang][safestack] Remove triple from stack usage test

[Paul Kirth via Phabricator via cfe-commits]

This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
Closed by commit rG414f84ba29d9: [clang][safestack] Remove triple from stack 
usage test (authored by paulkirth).

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D124203/new/

https://reviews.llvm.org/D124203

Files:
  clang/test/Frontend/stack-usage-safestack.c


Index: clang/test/Frontend/stack-usage-safestack.c
===
--- clang/test/Frontend/stack-usage-safestack.c
+++ clang/test/Frontend/stack-usage-safestack.c
@@ -1,12 +1,10 @@
 /// Check that stack frame size warnings behave the same when safe stack is 
enabled
 
-// REQUIRES: x86-registered-target
+// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck 
--check-prefix=REGULAR %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck 
--check-prefix=IGNORE %s
 
-// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 
2>&1 | FileCheck --check-prefix=REGULAR %s
-// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 
2>&1 | FileCheck --check-prefix=IGNORE %s
-
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 
-triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=SAFESTACK %s
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 
-triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 2>&1 
| FileCheck --check-prefix=SAFESTACK %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 
2>&1 | FileCheck --check-prefix=IGNORE %s
 
 extern void init(char *buf, int size);
 extern int use_buf(char *buf, int size);


Index: clang/test/Frontend/stack-usage-safestack.c
===
--- clang/test/Frontend/stack-usage-safestack.c
+++ clang/test/Frontend/stack-usage-safestack.c
@@ -1,12 +1,10 @@
 /// Check that stack frame size warnings behave the same when safe stack is enabled
 
-// REQUIRES: x86-registered-target
+// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck --check-prefix=REGULAR %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck --check-prefix=IGNORE %s
 
-// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=REGULAR %s
-// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
-
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=SAFESTACK %s
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck --check-prefix=SAFESTACK %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck --check-prefix=IGNORE %s
 
 extern void init(char *buf, int size);
 extern int use_buf(char *buf, int size);
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D124203: [clang][safestack] Remove triple from stack usage test

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

unfortunately it breaks many other tests, so we probably need to revert this


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D124203/new/

https://reviews.llvm.org/D124203

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D124210: Revert "[clang][safestack] Remove triple from stack usage test"

[Paul Kirth via Phabricator via cfe-commits]

paulkirth created this revision.
Herald added a project: All.
paulkirth requested review of this revision.
Herald added a project: clang.
Herald added a subscriber: cfe-commits.

This reverts commit 414f84ba29d96c8cbbe198cfc022146e4582cbef 
.

Requires deirectives had already been added to the test, and CC1 tests
should use a target triple.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D124210

Files:
  clang/test/Frontend/stack-usage-safestack.c


Index: clang/test/Frontend/stack-usage-safestack.c
===
--- clang/test/Frontend/stack-usage-safestack.c
+++ clang/test/Frontend/stack-usage-safestack.c
@@ -1,10 +1,12 @@
 /// Check that stack frame size warnings behave the same when safe stack is 
enabled
 
-// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck 
--check-prefix=REGULAR %s
-// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck 
--check-prefix=IGNORE %s
+// REQUIRES: x86-registered-target
 
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 2>&1 
| FileCheck --check-prefix=SAFESTACK %s
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 
2>&1 | FileCheck --check-prefix=IGNORE %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 
2>&1 | FileCheck --check-prefix=REGULAR %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 
2>&1 | FileCheck --check-prefix=IGNORE %s
+
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 
-triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=SAFESTACK %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 
-triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
 
 extern void init(char *buf, int size);
 extern int use_buf(char *buf, int size);


Index: clang/test/Frontend/stack-usage-safestack.c
===
--- clang/test/Frontend/stack-usage-safestack.c
+++ clang/test/Frontend/stack-usage-safestack.c
@@ -1,10 +1,12 @@
 /// Check that stack frame size warnings behave the same when safe stack is enabled
 
-// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck --check-prefix=REGULAR %s
-// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck --check-prefix=IGNORE %s
+// REQUIRES: x86-registered-target
 
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck --check-prefix=SAFESTACK %s
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck --check-prefix=IGNORE %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=REGULAR %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
+
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=SAFESTACK %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
 
 extern void init(char *buf, int size);
 extern int use_buf(char *buf, int size);
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D124210: Revert "[clang][safestack] Remove triple from stack usage test"

[Paul Kirth via Phabricator via cfe-commits]

This revision was not accepted when it landed; it landed in state "Needs 
Review".
This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
Closed by commit rGb8786413d800: Revert "[clang][safestack] Remove triple 
from stack usage test" (authored by paulkirth).

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D124210/new/

https://reviews.llvm.org/D124210

Files:
  clang/test/Frontend/stack-usage-safestack.c


Index: clang/test/Frontend/stack-usage-safestack.c
===
--- clang/test/Frontend/stack-usage-safestack.c
+++ clang/test/Frontend/stack-usage-safestack.c
@@ -1,10 +1,12 @@
 /// Check that stack frame size warnings behave the same when safe stack is 
enabled
 
-// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck 
--check-prefix=REGULAR %s
-// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck 
--check-prefix=IGNORE %s
+// REQUIRES: x86-registered-target
 
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 2>&1 
| FileCheck --check-prefix=SAFESTACK %s
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 
2>&1 | FileCheck --check-prefix=IGNORE %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 
2>&1 | FileCheck --check-prefix=REGULAR %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 
2>&1 | FileCheck --check-prefix=IGNORE %s
+
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 
-triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=SAFESTACK %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 
-triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
 
 extern void init(char *buf, int size);
 extern int use_buf(char *buf, int size);


Index: clang/test/Frontend/stack-usage-safestack.c
===
--- clang/test/Frontend/stack-usage-safestack.c
+++ clang/test/Frontend/stack-usage-safestack.c
@@ -1,10 +1,12 @@
 /// Check that stack frame size warnings behave the same when safe stack is enabled
 
-// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck --check-prefix=REGULAR %s
-// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck --check-prefix=IGNORE %s
+// REQUIRES: x86-registered-target
 
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - 2>&1 | FileCheck --check-prefix=SAFESTACK %s
-// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - 2>&1 | FileCheck --check-prefix=IGNORE %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=REGULAR %s
+// RUN: %clang_cc1 %s -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
+
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=48 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=SAFESTACK %s
+// RUN: %clang_cc1 %s -fsanitize=safe-stack -fwarn-stack-size=1060 -S -o - -triple=i386-apple-darwin 2>&1 | FileCheck --check-prefix=IGNORE %s
 
 extern void init(char *buf, int size);
 extern int use_buf(char *buf, int size);
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D118428: [clang-cl] Support the /JMC flag

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Hi,

We have two failing test cases on Fuchsia's clang canary builder on Windows x64.

LLVM :: Instrumentation/JustMyCode/jmc-instrument-x86.ll 
LLVM :: Instrumentation/JustMyCode/jmc-instrument.ll

First seen here: 
https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-windows-x64/b8822587673277278177/overview

These are JustMyCode tests, added in this patch, and it appears these tests may 
need to be adjusted.

You can find the full output in the linked builders, but here is a sample 
output from one of the tests. It seems to me like the lit file may just need to 
be adjusted slightly?

  Script:
  --
  : 'RUN: at line 1';   c:\b\s\w\ir\x\w\staging\llvm_build\bin\opt.exe 
-jmc-instrument -S < 
C:\b\s\w\ir\x\w\llvm-llvm-project\llvm\test\Instrumentation\JustMyCode\jmc-instrument-x86.ll
 | c:\b\s\w\ir\x\w\staging\llvm_build\bin\filecheck.exe 
C:\b\s\w\ir\x\w\llvm-llvm-project\llvm\test\Instrumentation\JustMyCode\jmc-instrument-x86.ll
  --
  Exit Code: 1
  
  Command Output (stdout):
  --
  $ ":" "RUN: at line 1"
  $ "c:\b\s\w\ir\x\w\staging\llvm_build\bin\opt.exe" "-jmc-instrument" "-S"
  $ "c:\b\s\w\ir\x\w\staging\llvm_build\bin\filecheck.exe" 
"C:\b\s\w\ir\x\w\llvm-llvm-project\llvm\test\Instrumentation\JustMyCode\jmc-instrument-x86.ll"
  # command stderr:
  
C:\b\s\w\ir\x\w\llvm-llvm-project\llvm\test\Instrumentation\JustMyCode\jmc-instrument-x86.ll:5:10:
 error: CHECK: expected string not found in input
  ; CHECK: @"_A85D9D03_x@c" = internal unnamed_addr global i8 1, section 
".msvcjmc", align 1, !dbg !0
   ^
  :6:34: note: scanning from here
  $_JustMyCode_Default = comdat any
   ^
  :8:1: note: possible intended match here
  @"_A8764FDD_x@c" = internal unnamed_addr global i8 1, section ".msvcjmc", 
align 1, !dbg !0
  ^
  
  Input file: 
  Check file: 
C:\b\s\w\ir\x\w\llvm-llvm-project\llvm\test\Instrumentation\JustMyCode\jmc-instrument-x86.ll
  
  -dump-input=help explains the following input dump.
  
  Input was:
  <<
 1: ; ModuleID = '' 
 2: source_filename = "" 
 3: target datalayout = 
"e-m:x-p:32:32-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32-a:0:32-S32"
 
 4: target triple = "i386-pc-windows-msvc" 
 5:  
 6: $_JustMyCode_Default = comdat any 
  check:5'0  X error: no match found
 7:  
  check:5'0 ~
 8: @"_A8764FDD_x@c" = internal unnamed_addr global i8 1, section 
".msvcjmc", align 1, !dbg !0 
  check:5'0 
~~~
  check:5'1 ?   
possible intended match
 9: @llvm.used = appending global [1 x i8*] [i8* bitcast (void 
(i8*)* @_JustMyCode_Default to i8*)], section "llvm.metadata" 
  check:5'0 
~
10:  
  check:5'0 ~
11: define void @w1() #0 !dbg !10 { 
  check:5'0 
12:  call x86_fastcallcc void @__CheckForDebuggerJustMyCode(i8* 
inreg noundef @"_A8764FDD_x@c") 
  check:5'0 

13:  ret void 
  check:5'0 ~~
 .
 .
 .
  >>
  
  error: command failed with exit status: 1
  
  --

If fixing the test will take a long time, can you revert until one is ready?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D118428/new/

https://reviews.llvm.org/D118428

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D119574: [clang] Expose -fprofile-use in clang-cl

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Hi,

I think we hit a test failure due to this patch in Fuchsia's Clang canary for 
Windows. We've had a lot of breakages on Windows today, so this was hidden 
until 0574b5fc 
 landed.

The failing test is: Clang :: Driver/cl-options.c

The failing bot can be found here: 
https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-windows-x64/b8822483567668224209/overview


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D119574/new/

https://reviews.llvm.org/D119574

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D119574: [clang] Expose -fprofile-use in clang-cl

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Sorry, that bot is green now as of 437d4e01fe4c057509dff30efd560049ad07bc99 
, so it 
looks like it was just bad timing that led me to suspect your commit, since it 
touched the test file.

I'm not sure why the bots test output is so limited, they usually point to the 
failing assert or provide the output from lit. But either way, sorry for the 
trouble.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D119574/new/

https://reviews.llvm.org/D119574

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D130226: [clang-doc] Default to Standalone executor and improve documentation

[Paul Kirth via Phabricator via cfe-commits]

paulkirth accepted this revision.
paulkirth added a comment.
This revision is now accepted and ready to land.

LGTM w/ the caveat that I think we should update the release notes, since we're 
changing the default behavior of a command line tool. I've also left a few 
small suggestions the wording.




Comment at: clang-tools-extra/clang-doc/tool/ClangDocMain.cpp:188
 
-  if (!Exec) {
-llvm::errs() << toString(Exec.takeError()) << "\n";
+  Example usage for files without flags:
+





Comment at: clang-tools-extra/clang-doc/tool/ClangDocMain.cpp:192
+
+  Example usage for a project using compile commands database:
+





Comment at: clang-tools-extra/docs/clang-doc.rst:28
 
-By default, the tool will run on all files listed in the given compile commands
-database:
+The tool will process listed files by default:
 





Comment at: clang-tools-extra/docs/clang-doc.rst:34
 
-The tool can also be used on a single file or multiple files if a build path is
-passed with the ``-p`` flag.
+The tool can be also used with compile commands database:
 





Comment at: clang-tools-extra/docs/clang-doc.rst:77
+
+Example usage for files without flags:
+





Comment at: clang-tools-extra/docs/clang-doc.rst:81
+
+Example usage for a project using compile commands database:
+




Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D130226/new/

https://reviews.llvm.org/D130226

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D130279: [clang-doc] Add check for pointer validity

[Paul Kirth via Phabricator via cfe-commits]

paulkirth created this revision.
paulkirth added reviewers: phosek, abrachet.
Herald added a project: All.
paulkirth requested review of this revision.
Herald added a project: clang-tools-extra.
Herald added a subscriber: cfe-commits.

clang-doc would SEGV when running over the Fuchsia code base.
This patch adds a check to avoid dereferencing potentially null pointers
in the Values vector. These pointers were either never valid or had been
invalidated when the underlying pointer in std::unique_ptr was moved from,
hence making it nullptr within the vector.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D130279

Files:
  clang-tools-extra/clang-doc/Representation.cpp


Index: clang-tools-extra/clang-doc/Representation.cpp
===
--- clang-tools-extra/clang-doc/Representation.cpp
+++ clang-tools-extra/clang-doc/Representation.cpp
@@ -33,7 +33,7 @@
 template 
 llvm::Expected>
 reduce(std::vector> &Values) {
-  if (Values.empty())
+  if (Values.empty() || !Values[0])
 return llvm::createStringError(llvm::inconvertibleErrorCode(),
"no value to reduce");
   std::unique_ptr Merged = std::make_unique(Values[0]->USR);
@@ -95,7 +95,7 @@
 // Dispatch function.
 llvm::Expected>
 mergeInfos(std::vector> &Values) {
-  if (Values.empty())
+  if (Values.empty() || !Values[0])
 return llvm::createStringError(llvm::inconvertibleErrorCode(),
"no info values to merge");
 


Index: clang-tools-extra/clang-doc/Representation.cpp
===
--- clang-tools-extra/clang-doc/Representation.cpp
+++ clang-tools-extra/clang-doc/Representation.cpp
@@ -33,7 +33,7 @@
 template 
 llvm::Expected>
 reduce(std::vector> &Values) {
-  if (Values.empty())
+  if (Values.empty() || !Values[0])
 return llvm::createStringError(llvm::inconvertibleErrorCode(),
"no value to reduce");
   std::unique_ptr Merged = std::make_unique(Values[0]->USR);
@@ -95,7 +95,7 @@
 // Dispatch function.
 llvm::Expected>
 mergeInfos(std::vector> &Values) {
-  if (Values.empty())
+  if (Values.empty() || !Values[0])
 return llvm::createStringError(llvm::inconvertibleErrorCode(),
"no info values to merge");
 
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D130231: [WIP][clang-doc] Improve the Markdown output

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Overall I like the direction of the patch, I know this is WIP, but I've left 
some small question inline regarding scope and work tracking.




Comment at: clang-tools-extra/clang-doc/MDGenerator.cpp:23-52
+static std::string genEscaped(StringRef Name) {
+  std::string Buffer;
+  llvm::raw_string_ostream Stream(Buffer);
+  for (unsigned I = 0, E = Name.size(); I != E; ++I) {
+unsigned char C = Name[I];
+switch (C) {
+  case '\\':

This isn't something I expect to get addressed in this patch, but I feel like 
I've seen this same pattern of escaping characters repeated throughout the 
codebase. Maybe we should add a TODO regarding it? I'm also fine with ignoring 
this.



Comment at: clang-tools-extra/clang-doc/MDGenerator.cpp:110-113
+// TODO: @return is a block command and should be included in the 
"Returns" table.
+// TODO: @see block commands should be grouped and rendered as "See Also" 
section.
+// TODO: Figure out handling for @brief.
+// TODO: What other block commands need special handling?

Are these TODOs (and the others throughout the patch) something you plan to 
address in this patch before landing? or are they future improvements? If the 
latter, maybe we should file an issue on github reference it here?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D130231/new/

https://reviews.llvm.org/D130231

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D130279: [clang-doc] Add check for pointer validity

[Paul Kirth via Phabricator via cfe-commits]

This revision was automatically updated to reflect the committed changes.
Closed by commit rG30360d88d422: [clang-doc] Add check for pointer validity 
(authored by paulkirth).

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D130279/new/

https://reviews.llvm.org/D130279

Files:
  clang-tools-extra/clang-doc/Representation.cpp


Index: clang-tools-extra/clang-doc/Representation.cpp
===
--- clang-tools-extra/clang-doc/Representation.cpp
+++ clang-tools-extra/clang-doc/Representation.cpp
@@ -33,7 +33,7 @@
 template 
 llvm::Expected>
 reduce(std::vector> &Values) {
-  if (Values.empty())
+  if (Values.empty() || !Values[0])
 return llvm::createStringError(llvm::inconvertibleErrorCode(),
"no value to reduce");
   std::unique_ptr Merged = std::make_unique(Values[0]->USR);
@@ -95,7 +95,7 @@
 // Dispatch function.
 llvm::Expected>
 mergeInfos(std::vector> &Values) {
-  if (Values.empty())
+  if (Values.empty() || !Values[0])
 return llvm::createStringError(llvm::inconvertibleErrorCode(),
"no info values to merge");
 


Index: clang-tools-extra/clang-doc/Representation.cpp
===
--- clang-tools-extra/clang-doc/Representation.cpp
+++ clang-tools-extra/clang-doc/Representation.cpp
@@ -33,7 +33,7 @@
 template 
 llvm::Expected>
 reduce(std::vector> &Values) {
-  if (Values.empty())
+  if (Values.empty() || !Values[0])
 return llvm::createStringError(llvm::inconvertibleErrorCode(),
"no value to reduce");
   std::unique_ptr Merged = std::make_unique(Values[0]->USR);
@@ -95,7 +95,7 @@
 // Dispatch function.
 llvm::Expected>
 mergeInfos(std::vector> &Values) {
-  if (Values.empty())
+  if (Values.empty() || !Values[0])
 return llvm::createStringError(llvm::inconvertibleErrorCode(),
"no info values to merge");
 
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D120727: [libc++] Overhaul how we select the ABI library

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Hi, we're seeing some failures in Fuchsia's Clang CI. Our runtimes build seems 
to be unable to find `cxxabi.h`.

The failing bot can be found here: 
https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-linux-x64/b8814278370664903633/overview


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D120727/new/

https://reviews.llvm.org/D120727

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D125570: [CMake] Disable libedit in Fuchsia toolchain

[Paul Kirth via Phabricator via cfe-commits]

paulkirth accepted this revision.
paulkirth added a comment.
This revision is now accepted and ready to land.

LGTM.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D125570/new/

https://reviews.llvm.org/D125570

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D123534: [dwarf] Emit a DIGlobalVariable for constant strings.

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Hi, we're seeing a breakage in Fuchsia's clang CI for x64 windows that I think 
is related to this patch.

https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-windows-x64/b8813962058917346337/overview

We see a test failure in Clang :: CodeGen/debug-info-variables.c, which this 
patch modifies.

If this will be hard to fix, would you mind reverting until a fix is ready?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D123534/new/

https://reviews.llvm.org/D123534

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D123534: [dwarf] Emit a DIGlobalVariable for constant strings.

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

Thanks for the quick response.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D123534/new/

https://reviews.llvm.org/D123534

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D123534: [dwarf] Emit a DIGlobalVariable for constant strings.

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

We noticed when building Fuchsia using ToT LLVM that CGo has trouble with the 
new changes to DWARF introduced in this patch. It appears as if CGo is confused 
because it found a `DW_TAG_variable` that doesn't have a name, which is allowed 
by the DWARF standard.

This  patch seems like a good change to me and is well thought out and tested, 
but I think this may cause issues in the wider ecosystem as people interact 
with it. It may simply be a problem w/ Go, and a bug fix there is all that is 
necessary,  but there is also the chance that other compilers and toolchains 
will have similar issues.

I want to be clear that I don't necessarily think a revert is in order, but we 
should take the opportunity to determine the right course of action.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D123534/new/

https://reviews.llvm.org/D123534

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D126224: Add DWARF string debug to clang release notes.

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a subscriber: mcgrathr.
paulkirth added a comment.

@rnk the standard even has an example of this exact behavior, so I think it's 
hard to say its //wrong// for LLVM to do this, but that may be more gentle. I'm 
going to defer to more expert opinions here, however, and loop in @mcgrathr.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D126224/new/

https://reviews.llvm.org/D126224

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D126224: Add DWARF string debug to clang release notes.

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

I'm going to try to summarize an offline discussion w/ @mcgrathr about this 
here:

There are some other considerations to think about w.r.t. emitting names for 
non-source language constructs, as would be the case here. In fact, DWARF 
already handles this case: the "this" pointer in C++. Despite the fact that it 
may not be directly used in the source program, it is tagged as artificial, and 
gets a name, so that programmers can reference it within the debugger.  
Debuggers don't know anything special about C++,  the compiler just emitted a  
DWARF variable that has the `DW_AT_name` of "this". They just reference the 
name, and allow users to access it as if it were another variable.

The point is that the `DW_AT_name` has a special meaning to debuggers, which 
may be problematic if we try to emit a name when this isn't really a source 
level construct and isn't something that should be accessed like another 
variable. Generating a name for these items makes them addressable, and is 
//probably// more problematic for downstream consumers of DWARF than omitting 
the name.  Generally, DWARF consumers probably shouldn't be relying on things 
like the name always being there. The standard is pretty clear that many fields 
are optional, i.e., they //may// be there or they //may not// be there.

It's probably fine to gate the new DWARF items behind a compiler flag if there 
is concern about incompatible consumers. It could even be on by default, but at 
least it would be possible to opt out of the new behavior. You may even be able 
to gate it based on the DWARF version. That isn't strictly correct, but is 
offered as a possible pragmatic convenience on the rationale that concern about 
incompatible consumers probably applies to "old" consumers and consumers new 
enough to handle v5 could be assumed new/well-maintained enough to either 
already handle, or quickly be fixed to handle, nameless `DW_TAG_variable`.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D126224/new/

https://reviews.llvm.org/D126224

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D126224: Add DWARF string debug to clang release notes.

[Paul Kirth via Phabricator via cfe-commits]

paulkirth accepted this revision.
paulkirth added a comment.
This revision is now accepted and ready to land.

LGTM.




Comment at: clang/docs/ReleaseNotes.rst:447
+  structures *without* a ``DW_AT_name`` field, which is valid DWARF, but may
+  lead to assertion failures in some software.
+

nit: its likely not an assertion failure, but just invalid code.

It's also fine w/ me to word this differently, or ignore the suggestion.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D126224/new/

https://reviews.llvm.org/D126224

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 216274.
paulkirth added a comment.
Herald added a subscriber: ormris.

Update LLVM based implementation

Adds LLVM based tests
Cleans up existing tests in clang
Adds new llvm flag -pgo-warn-misexpect
Adds llvm flag when -Wmisexpect enabled in frontend
Adds new debug output 
Fixes error when calculating & scaling branch weights
Improves fidelity of diagnostic outpus from SwithInst
Removes Remark output from MisExpecDiagHandler & uses ORE directly


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/ThinLTO/X86/lazyload_metadata.ll
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,293 @@
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; New PM
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.c.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=CORRECT
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:5: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:5: Potential performance regression from use of __builtin_expect(): Annotation was correct on 0.00% of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:5: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:5: Potential performance regression from 

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a reviewer: nickdesaulniers.
paulkirth added a comment.

Add Nick Desaulniers as an additional reviewer


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 217282.
paulkirth marked 3 inline comments as done.
paulkirth added a comment.

Address Code Review

- Give better names to extracted constants
- Remove extraneous call to getValue()
- Make integers const
- Remove extra braces
- Move NOps into branch
- Use std::plus


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/ThinLTO/X86/lazyload_metadata.ll
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,293 @@
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; New PM
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.c.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=CORRECT
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:5: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:5: Potential performance regression from use of __builtin_expect(): Annotation was correct on 0.00% of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:5: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:5: Potential performance regression from use of __builtin_expect(): Annotation was correct on 0.00% of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:5: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:5:

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth marked 5 inline comments as done.
paulkirth added inline comments.



Comment at: llvm/lib/Transforms/Utils/MisExpect.cpp:108
+  auto *L = 
mdconst::dyn_extract(MisExpectData->getOperand(2));
+  auto *U = 
mdconst::dyn_extract(MisExpectData->getOperand(3));
+

nickdesaulniers wrote:
> What's C, L, and U?
These are the constant integers for:
C = Index into the branch weight vector
L = likely branch weight
U = unlikely branch weight

The patch renames these to be a bit more transparent.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 217285.
paulkirth marked an inline comment as done.
paulkirth added a comment.

Revert to use of auto when extracting metadata


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/ThinLTO/X86/lazyload_metadata.ll
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,293 @@
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; New PM
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.c.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=CORRECT
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:5: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:5: Potential performance regression from use of __builtin_expect(): Annotation was correct on 0.00% of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:5: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:5: Potential performance regression from use of __builtin_expect(): Annotation was correct on 0.00% of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:5: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:5: Potential performance regression from use of __builtin_expect(): Annotation was correct on 0.00% of profiled executions.
+
+; DISABLED-NOT:

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth marked 2 inline comments as done.
paulkirth added inline comments.



Comment at: llvm/lib/Transforms/Utils/MisExpect.cpp:123
+  const uint64_t CaseTotal = std::accumulate(
+  Weights.begin(), Weights.end(), (uint64_t)0, std::plus());
+  const int NumUnlikelyTargets = Weights.size() - 1;

nickdesaulniers wrote:
> s/(uint64_t)0/0ULL/ might be simpler, but I don't know if `long long == 
> uint64_t` on every host Clang can be built for...
I'm not sure if that is true, either. 

I think its safe to follow the convention of existing LLVM code doing more or 
less the same thing, though, right?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth updated this revision to Diff 217479.
paulkirth added a comment.

Removes unused constructor, reformat code, and remove braces


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324

Files:
  clang/include/clang/Basic/DiagnosticFrontendKinds.td
  clang/include/clang/Basic/DiagnosticGroups.td
  clang/lib/CodeGen/CodeGenAction.cpp
  clang/lib/CodeGen/CodeGenFunction.cpp
  clang/lib/Frontend/CompilerInvocation.cpp
  clang/test/Profile/Inputs/misexpect-branch-nonconst-expect-arg.proftext
  clang/test/Profile/Inputs/misexpect-branch.proftext
  clang/test/Profile/Inputs/misexpect-switch-default-only.proftext
  clang/test/Profile/Inputs/misexpect-switch-default.proftext
  clang/test/Profile/Inputs/misexpect-switch.proftext
  clang/test/Profile/misexpect-branch-cold.c
  clang/test/Profile/misexpect-branch-nonconst-expected-val.c
  clang/test/Profile/misexpect-branch.c
  clang/test/Profile/misexpect-switch-default.c
  clang/test/Profile/misexpect-switch-nonconst.c
  clang/test/Profile/misexpect-switch-only-default-case.c
  clang/test/Profile/misexpect-switch.c
  llvm/include/llvm/IR/DiagnosticInfo.h
  llvm/include/llvm/IR/FixedMetadataKinds.def
  llvm/include/llvm/IR/MDBuilder.h
  llvm/include/llvm/Transforms/Utils/MisExpect.h
  llvm/lib/IR/DiagnosticInfo.cpp
  llvm/lib/IR/MDBuilder.cpp
  llvm/lib/Transforms/IPO/SampleProfile.cpp
  llvm/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
  llvm/lib/Transforms/Scalar/LowerExpectIntrinsic.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/MisExpect.cpp
  llvm/test/ThinLTO/X86/lazyload_metadata.ll
  llvm/test/Transforms/LowerExpectIntrinsic/basic.ll
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-branch.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch-correct.proftext
  llvm/test/Transforms/PGOProfile/Inputs/misexpect-switch.proftext
  llvm/test/Transforms/PGOProfile/misexpect-branch-correct.ll
  llvm/test/Transforms/PGOProfile/misexpect-branch.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch-default.ll
  llvm/test/Transforms/PGOProfile/misexpect-switch.ll

Index: llvm/test/Transforms/PGOProfile/misexpect-switch.ll
===
--- /dev/null
+++ llvm/test/Transforms/PGOProfile/misexpect-switch.ll
@@ -0,0 +1,293 @@
+
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch.proftext -o %t.profdata
+; RUN: llvm-profdata merge %S/Inputs/misexpect-switch-correct.proftext -o %t.c.profdata
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; New PM
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -S 2>&1 | FileCheck %s --check-prefix=WARNING
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=REMARK
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=BOTH
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.profdata -S 2>&1 | FileCheck %s --check-prefix=DISABLED
+
+; RUN: opt < %s -lower-expect -pgo-instr-use -pgo-test-profile-file=%t.c.profdata -S -pgo-warn-misexpect -pass-remarks=misexpect 2>&1 | FileCheck %s --check-prefix=CORRECT
+; RUN: opt < %s -passes="function(lower-expect),pgo-instr-use" -pgo-test-profile-file=%t.c.profdata -pgo-warn-misexpect -pass-remarks=misexpect -S 2>&1 | FileCheck %s --check-prefix=CORRECT
+
+; WARNING-DAG: warning: misexpect-switch.c:26:5: 0.00%
+; WARNING-NOT: remark: misexpect-switch.c:26:5: Potential performance regression from use of __builtin_expect(): Annotation was correct on 0.00% of profiled executions.
+
+; REMARK-NOT: warning: misexpect-switch.c:26:5: 0.00%
+; REMARK-DAG: remark: misexpect-switch.c:26:5: Potential performance regression from use of __builtin_expect(): Annotation was correct on 0.00% of profiled executions.
+
+; BOTH-DAG: warning: misexpect-switch.c:26:5: 0.00%
+; BOTH-DAG: remark: misexpect-switch.c:26:5: Potential performance regression from use of __builtin_expect(): Annotation was correct on 0.00% of profiled executions.
+
+; DISABLED-NOT: warning: misexpect-switch.c:2

[PATCH] D66324: clang-misexpect: Profile Guided Validation of Performance Annotations in LLVM

[Paul Kirth via Phabricator via cfe-commits]

paulkirth marked 6 inline comments as done.
paulkirth added inline comments.



Comment at: llvm/include/llvm/IR/DiagnosticInfo.h:1009
+public:
+  DiagnosticInfoMisExpect(const Function &Fn, const Twine &Msg,
+  const DiagnosticLocation &Loc = DiagnosticLocation(),

nickdesaulniers wrote:
> Is this constructor used anywhere in this patch?
No, sorry, I changed this interface and failed to remove the constructor.



Comment at: llvm/lib/IR/DiagnosticInfo.cpp:378
+ Inst->getDebugLoc()),
+  Msg(Msg) {}
+

nickdesaulniers wrote:
> Why is this constructor defined out of line, when the other is in the header? 
>  Seems like they should either both be in the header, or both be out of line, 
> with the goal being consistency.
I removed the other constructor, which was unused. 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D66324/new/

https://reviews.llvm.org/D66324



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D65300: [clang] [CodeGen] clang-misexpect prototype for compiler warnings

[Paul Kirth via Phabricator via cfe-commits]

paulkirth added a comment.

In D65300#1647746 , @lebedev.ri wrote:

> Please can you clarify hat's the current layout of these patches?
>  Is this patch required, or is it superseded by D66324 
>  (and thus should be abandoned)? 
>  I'd like to begin reviewing, but i don't understand where to start.


Sorry for the confusion. D66324  should 
supersede this patch. It re-implements everything in the backend, and adds 
support for IR & Sample based profiles, so I think it is safe to ignore this 
patch for now.

Is there something I should do to mark/change the patches?

Also, thanks for the feedback.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D65300/new/

https://reviews.llvm.org/D65300



___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits