[clang] [Clang] Improve error for `-fsanitize=function/kcfi -mexecute-only` incompatibility (PR #118816)

2024-12-05 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano updated 
https://github.com/llvm/llvm-project/pull/118816

From 94c933e0e7b7497e76c541f3ea7ed210349d2044 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Thu, 5 Dec 2024 14:46:08 +0100
Subject: [PATCH] [Clang] Improve error for `-fsanitize=function/kcfi
 -mexecute-only` incompatibility

The current error message when using the `-fsanitize=function -mexecute-only`
flags together points to the target triple as the reason that
`-fsanitize=function` is not allowed to be used, even when the function
sanitizer is otherwise supported on the target when not using
`-mexecute-only`.

The error message is improved to give `-mexecute-only` as the reason for
disallowing `-fsanitize=function` if it was passed to the driver.
---
 clang/lib/Driver/SanitizerArgs.cpp | 10 --
 clang/test/Driver/fsanitize.c  |  4 ++--
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/clang/lib/Driver/SanitizerArgs.cpp 
b/clang/lib/Driver/SanitizerArgs.cpp
index 1abfe8fd92807e..6a2d376e93d05b 100644
--- a/clang/lib/Driver/SanitizerArgs.cpp
+++ b/clang/lib/Driver/SanitizerArgs.cpp
@@ -418,8 +418,14 @@ SanitizerArgs::SanitizerArgs(const ToolChain &TC,
 Add & NotAllowedWithExecuteOnly & ~DiagnosedKinds) {
   if (DiagnoseErrors) {
 std::string Desc = describeSanitizeArg(Arg, KindsToDiagnose);
-D.Diag(diag::err_drv_argument_not_allowed_with)
-<< Desc << Triple.str();
+llvm::opt::Arg *A = Args.getLastArgNoClaim(
+options::OPT_mexecute_only, options::OPT_mno_execute_only);
+if (A && A->getOption().matches(options::OPT_mexecute_only))
+  D.Diag(diag::err_drv_argument_not_allowed_with)
+  << Desc << A->getAsString(Args);
+else
+  D.Diag(diag::err_drv_argument_not_allowed_with)
+  << Desc << Triple.str();
   }
   DiagnosedKinds |= KindsToDiagnose;
 }
diff --git a/clang/test/Driver/fsanitize.c b/clang/test/Driver/fsanitize.c
index 15f190165a7d73..a99bfb3e82ac95 100644
--- a/clang/test/Driver/fsanitize.c
+++ b/clang/test/Driver/fsanitize.c
@@ -1000,8 +1000,8 @@
 // RUN: not %clang --target=armv6t2-eabi -mexecute-only -fsanitize=kcfi %s 
-### 2>&1 | FileCheck %s --check-prefix=CHECK-UBSAN-KCFI
 // RUN: %clang --target=armv6t2-eabi -mexecute-only -fsanitize=undefined %s 
-### 2>&1 | FileCheck %s --check-prefix=CHECK-UBSAN-UNDEFINED-VPTR
 
-// CHECK-UBSAN-KCFI-DAG: error: invalid argument '-fsanitize=kcfi' not allowed 
with {{('x86_64-sie-ps5'|'armv6t2-unknown-unknown-eabi')}}
-// CHECK-UBSAN-FUNCTION-DAG: error: invalid argument '-fsanitize=function' not 
allowed with {{('x86_64-sie-ps5'|'armv6t2-unknown-unknown-eabi')}}
+// CHECK-UBSAN-KCFI-DAG: error: invalid argument '-fsanitize=kcfi' not allowed 
with {{('x86_64-sie-ps5'|'-mexecute-only')}}
+// CHECK-UBSAN-FUNCTION-DAG: error: invalid argument '-fsanitize=function' not 
allowed with {{('x86_64-sie-ps5'|'-mexecute-only')}}
 // CHECK-UBSAN-UNDEFINED-VPTR: 
"-fsanitize={{((alignment|array-bounds|bool|builtin|enum|float-cast-overflow|integer-divide-by-zero|nonnull-attribute|null|pointer-overflow|return|returns-nonnull-attribute|shift-base|shift-exponent|signed-integer-overflow|unreachable|vla-bound|vptr),?){18}"}}
 
 // * Test BareMetal toolchain sanitizer support *

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

[clang] [Clang] Improve error for `-fsanitize=function/kcfi -mexecute-only` incompatibility (PR #118816)

2024-12-05 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano created 
https://github.com/llvm/llvm-project/pull/118816

The current error message when using the `-fsanitize=function -mexecute-only` 
flags together points to the target triple as the reason that 
`-fsanitize=function` is not allowed to be used, even when the function 
sanitizer is otherwise supported on the target when not using `-mexecute-only`.

The error message is improved to give `-mexecute-only` as the reason for 
disallowing `-fsanitize=function` if it was passed to the driver.

From 691a26b6602a4f7c9f2ea5eaf27bb8d1a0be11a6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Thu, 5 Dec 2024 14:46:08 +0100
Subject: [PATCH] [Clang] Improve error for `-fsanitize=function/kcfi
 -mexecute-only` incompatibility

The current error message when using the `-fsanitize=function -mexecute-only`
flags together points to the target triple as the reason that
`-fsanitize=function` is not allowed to be used, even when the function
sanitizer is otherwise supported on the target when not using
`-mexecute-only`.

The error message is improved to give `-mexecute-only` as the reason for
disallowing `-fsanitize=function` if it was passed to the driver.
---
 clang/lib/Driver/SanitizerArgs.cpp | 10 --
 clang/test/Driver/fsanitize.c  |  4 ++--
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/clang/lib/Driver/SanitizerArgs.cpp 
b/clang/lib/Driver/SanitizerArgs.cpp
index 1abfe8fd92807e..6d87e25b49237e 100644
--- a/clang/lib/Driver/SanitizerArgs.cpp
+++ b/clang/lib/Driver/SanitizerArgs.cpp
@@ -418,8 +418,14 @@ SanitizerArgs::SanitizerArgs(const ToolChain &TC,
 Add & NotAllowedWithExecuteOnly & ~DiagnosedKinds) {
   if (DiagnoseErrors) {
 std::string Desc = describeSanitizeArg(Arg, KindsToDiagnose);
-D.Diag(diag::err_drv_argument_not_allowed_with)
-<< Desc << Triple.str();
+llvm::opt::Arg *A = Args.getLastArgNoClaim(
+options::OPT_mexecute_only, options::OPT_mno_execute_only);
+if (A && A->getOption().matches(options::OPT_mexecute_only))
+  D.Diag(diag::err_drv_argument_not_allowed_with)
+  << Desc << A->getAsString(Args);
+else
+  D.Diag(diag::err_drv_argument_not_allowed_with)
+  << Desc << Triple.str();
   }
   DiagnosedKinds |= KindsToDiagnose;
 }
diff --git a/clang/test/Driver/fsanitize.c b/clang/test/Driver/fsanitize.c
index 15f190165a7d73..a99bfb3e82ac95 100644
--- a/clang/test/Driver/fsanitize.c
+++ b/clang/test/Driver/fsanitize.c
@@ -1000,8 +1000,8 @@
 // RUN: not %clang --target=armv6t2-eabi -mexecute-only -fsanitize=kcfi %s 
-### 2>&1 | FileCheck %s --check-prefix=CHECK-UBSAN-KCFI
 // RUN: %clang --target=armv6t2-eabi -mexecute-only -fsanitize=undefined %s 
-### 2>&1 | FileCheck %s --check-prefix=CHECK-UBSAN-UNDEFINED-VPTR
 
-// CHECK-UBSAN-KCFI-DAG: error: invalid argument '-fsanitize=kcfi' not allowed 
with {{('x86_64-sie-ps5'|'armv6t2-unknown-unknown-eabi')}}
-// CHECK-UBSAN-FUNCTION-DAG: error: invalid argument '-fsanitize=function' not 
allowed with {{('x86_64-sie-ps5'|'armv6t2-unknown-unknown-eabi')}}
+// CHECK-UBSAN-KCFI-DAG: error: invalid argument '-fsanitize=kcfi' not allowed 
with {{('x86_64-sie-ps5'|'-mexecute-only')}}
+// CHECK-UBSAN-FUNCTION-DAG: error: invalid argument '-fsanitize=function' not 
allowed with {{('x86_64-sie-ps5'|'-mexecute-only')}}
 // CHECK-UBSAN-UNDEFINED-VPTR: 
"-fsanitize={{((alignment|array-bounds|bool|builtin|enum|float-cast-overflow|integer-divide-by-zero|nonnull-attribute|null|pointer-overflow|return|returns-nonnull-attribute|shift-base|shift-exponent|signed-integer-overflow|unreachable|vla-bound|vptr),?){18}"}}
 
 // * Test BareMetal toolchain sanitizer support *

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

[clang] [Clang] Add float type support to __builtin_reduce_add and __builtin_reduce_multipy (PR #120367)

2025-01-06 Thread Csanád Hajdú via cfe-commits 

Il-Capitano wrote:

Just to clarify: `@llvm.vector.reduce.fadd` does sequential reduction by 
default, so I don't see a point in doing that manually in the frontend. Your 
previous implementation had the same behaviour.

The inconsistency with the sequential approach is that Clang defines the 
`__builtin_reduce_*` operations to do recursive even-odd pairwise reduction 
(i.e. `(v[0] + v[1]) + (v[2] + v[3])` instead of `((v[0] + v[1]) + v[2]) + 
v[3])`), and since floating-poing addition is not associative, these two can 
have different results.

My suggestion was to not use the existing `__builtin_reduce_add` and `mul` 
builtins, but to define a new one that is defined to do sequential reduction, 
matching the behaviour of `@llvm.vector.reduce.fadd`, and one that is 
unordered, i.e. `@llvm.vector.reduce.fadd` with the `reassoc` fast-math flag 
set (in practice this will do the even-odd pairwise reduction, but there is a 
difference in generated code quality between doing that in the frontend, and 
the backend: https://godbolt.org/z/a4rd44Eza).

You can see the generated code difference of `@llvm.vector.reduce.fadd` with 
and without the `reassoc` flag here: https://godbolt.org/z/zeWjxrxo5.

https://github.com/llvm/llvm-project/pull/120367
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang] Add float type support to __builtin_reduce_add and __builtin_reduce_multipy (PR #120367)

2025-01-06 Thread Csanád Hajdú via cfe-commits 

Il-Capitano wrote:

`__builtin_reduce_add/mul` is defined to do _recursive even-odd pairwise 
reduction_ in Clang, and the `@llvm.vector.reduce.fadd` intrinsic doesn't do 
that.

I'm not sure if there's any benefit to this definition over a sequential 
reduction. The only thing I can think that could be affected is signed-integer 
overflow being UB, but that doesn't seem to be optimized for, or checked with 
UBSan: https://godbolt.org/z/TEsc86rf5.

Maybe, as RKSimon suggested, having a separate builtin for floating-point types 
could be better? Something like `__builtin_reduce_fadd/fmul` that does 
sequential reduction, and another one for unordered reduction?  
This builtin could also take a start value as the first argument to better 
match the LLVM intrinsic.

> Or maybe an alternative `__builtin_reduce_fastmath_fadd / fmul` that makes it 
> clear whats happening, and have them always emit the reassoc variants of the 
> reductions?

I think `__builtin_reduce_fadd_unordered` or something similar would be a 
better choice, since "fastmath" implies more restrictions/assumptions than 
`reassoc` (no nans, no signed zeros, etc.).

https://github.com/llvm/llvm-project/pull/120367
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [llvm] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-02-04 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano created 
https://github.com/llvm/llvm-project/pull/125688

Add support for the new SHF_AARCH64_PURECODE ELF section flag: 
https://github.com/ARM-software/abi-aa/pull/304

The general implementation follows the existing one for ARM targets. Simlarly 
to ARM targets, generating object files with the `SHF_AARCH64_PURECODE` flag 
set is enabled by the `-mexecute-only`/`-mpure-code` driver flag.

Depends on https://github.com/llvm/llvm-project/pull/125687. The changes in 
that PR are included here as well for now, but will be removed once that gets 
merged.

From 6ff1cc4abb39b0526cae22497b5147cae35a7d6f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Tue, 4 Feb 2025 14:32:58 +0100
Subject: [PATCH 1/2] [AArch64] Add support for SHF_AARCH64_PURECODE ELF
 section flag (1/3)

Add support for the new SHF_AARCH64_PURECODE ELF section flag: 
https://github.com/ARM-software/abi-aa/pull/304

The general implementation follows the existing one for ARM targets.
Generating object files with the `SHF_AARCH64_PURECODE` flag set is
enabled by the `+execute-only` target feature.
---
 llvm/include/llvm/BinaryFormat/ELF.h  |  5 ++-
 .../CodeGen/TargetLoweringObjectFileImpl.cpp  | 13 +++---
 llvm/lib/MC/MCParser/ELFAsmParser.cpp |  7 ++-
 llvm/lib/MC/MCSectionELF.cpp  |  3 ++
 llvm/lib/ObjectYAML/ELFYAML.cpp   |  3 ++
 llvm/lib/Target/AArch64/AArch64Features.td|  5 +++
 .../AArch64/AArch64TargetObjectFile.cpp   | 26 +++
 .../Target/AArch64/AArch64TargetObjectFile.h  |  6 +++
 .../AArch64/AArch64TargetTransformInfo.cpp| 17 +++-
 .../AArch64/AArch64TargetTransformInfo.h  |  2 +
 .../MCTargetDesc/AArch64ELFStreamer.cpp   | 18 
 .../CodeGen/AArch64/execute-only-section.ll   | 21 +
 .../execute-only-populated-text-section.s | 27 
 .../MC/ELF/AArch64/execute-only-section.s | 43 +++
 .../AArch64/execute-only-text-section-data.s  | 27 
 llvm/test/MC/ELF/section-flags-unknown.s  |  3 ++
 .../Inline/AArch64/inline-target-attr.ll  | 25 +++
 llvm/tools/llvm-readobj/ELFDumper.cpp |  8 
 18 files changed, 250 insertions(+), 9 deletions(-)
 create mode 100644 llvm/test/CodeGen/AArch64/execute-only-section.ll
 create mode 100644 
llvm/test/MC/ELF/AArch64/execute-only-populated-text-section.s
 create mode 100644 llvm/test/MC/ELF/AArch64/execute-only-section.s
 create mode 100644 llvm/test/MC/ELF/AArch64/execute-only-text-section-data.s

diff --git a/llvm/include/llvm/BinaryFormat/ELF.h 
b/llvm/include/llvm/BinaryFormat/ELF.h
index 8853c4a88b0b59..a5979d029a9eaf 100644
--- a/llvm/include/llvm/BinaryFormat/ELF.h
+++ b/llvm/include/llvm/BinaryFormat/ELF.h
@@ -1300,7 +1300,10 @@ enum : unsigned {
   SHF_MIPS_STRING = 0x8000,
 
   // Make code section unreadable when in execute-only mode
-  SHF_ARM_PURECODE = 0x2000
+  SHF_ARM_PURECODE = 0x2000,
+
+  // Section contains only program instructions and no program data.
+  SHF_AARCH64_PURECODE = 0x2000
 };
 
 // Section Group Flags
diff --git a/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp 
b/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
index 3c2c7c8c9fed69..6cfebca9a4d576 100644
--- a/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
+++ b/llvm/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
@@ -547,7 +547,7 @@ static unsigned getELFSectionType(StringRef Name, 
SectionKind K) {
   return ELF::SHT_PROGBITS;
 }
 
-static unsigned getELFSectionFlags(SectionKind K) {
+static unsigned getELFSectionFlags(SectionKind K, const Triple &T) {
   unsigned Flags = 0;
 
   if (!K.isMetadata() && !K.isExclude())
@@ -559,9 +559,12 @@ static unsigned getELFSectionFlags(SectionKind K) {
   if (K.isText())
 Flags |= ELF::SHF_EXECINSTR;
 
-  if (K.isExecuteOnly())
+  if ((T.isARM() || T.isThumb()) && K.isExecuteOnly())
 Flags |= ELF::SHF_ARM_PURECODE;
 
+  if (T.isAArch64() && K.isExecuteOnly())
+Flags |= ELF::SHF_AARCH64_PURECODE;
+
   if (K.isWriteable())
 Flags |= ELF::SHF_WRITE;
 
@@ -840,7 +843,7 @@ static MCSection *selectExplicitSectionGlobal(const 
GlobalObject *GO,
   // Infer section flags from the section name if we can.
   Kind = getELFKindForNamedSection(SectionName, Kind);
 
-  unsigned Flags = getELFSectionFlags(Kind);
+  unsigned Flags = getELFSectionFlags(Kind, TM.getTargetTriple());
   auto [Group, IsComdat, ExtraFlags] = getGlobalObjectInfo(GO, TM);
   Flags |= ExtraFlags;
 
@@ -947,7 +950,7 @@ static MCSection *selectELFSectionForGlobal(
 
 MCSection *TargetLoweringObjectFileELF::SelectSectionForGlobal(
 const GlobalObject *GO, SectionKind Kind, const TargetMachine &TM) const {
-  unsigned Flags = getELFSectionFlags(Kind);
+  unsigned Flags = getELFSectionFlags(Kind, TM.getTargetTriple());
 
   // If we have -ffunction-section or -fdata-section then we should emit the
   // global value to a uniqued section specifically for it.
@@ -967,7 +

[clang] [llvm] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-02-05 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano converted_to_draft 
https://github.com/llvm/llvm-project/pull/125688
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang] Improve error for `-fsanitize=function/kcfi -mexecute-only` incompatibility (PR #118816)

2024-12-11 Thread Csanád Hajdú via cfe-commits 

Il-Capitano wrote:

Thanks for reviewing!

I updated the code to use `err_drv_unsupported_opt_for_target`, and added test 
cases for the `-mpure-code` alias of `-mexecute-only` also.

https://github.com/llvm/llvm-project/pull/118816
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang] Improve error for `-fsanitize=function/kcfi -mexecute-only` incompatibility (PR #118816)

2024-12-11 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano updated 
https://github.com/llvm/llvm-project/pull/118816

From 94c933e0e7b7497e76c541f3ea7ed210349d2044 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Thu, 5 Dec 2024 14:46:08 +0100
Subject: [PATCH 1/2] [Clang] Improve error for `-fsanitize=function/kcfi
 -mexecute-only` incompatibility

The current error message when using the `-fsanitize=function -mexecute-only`
flags together points to the target triple as the reason that
`-fsanitize=function` is not allowed to be used, even when the function
sanitizer is otherwise supported on the target when not using
`-mexecute-only`.

The error message is improved to give `-mexecute-only` as the reason for
disallowing `-fsanitize=function` if it was passed to the driver.
---
 clang/lib/Driver/SanitizerArgs.cpp | 10 --
 clang/test/Driver/fsanitize.c  |  4 ++--
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/clang/lib/Driver/SanitizerArgs.cpp 
b/clang/lib/Driver/SanitizerArgs.cpp
index 1abfe8fd92807e..6a2d376e93d05b 100644
--- a/clang/lib/Driver/SanitizerArgs.cpp
+++ b/clang/lib/Driver/SanitizerArgs.cpp
@@ -418,8 +418,14 @@ SanitizerArgs::SanitizerArgs(const ToolChain &TC,
 Add & NotAllowedWithExecuteOnly & ~DiagnosedKinds) {
   if (DiagnoseErrors) {
 std::string Desc = describeSanitizeArg(Arg, KindsToDiagnose);
-D.Diag(diag::err_drv_argument_not_allowed_with)
-<< Desc << Triple.str();
+llvm::opt::Arg *A = Args.getLastArgNoClaim(
+options::OPT_mexecute_only, options::OPT_mno_execute_only);
+if (A && A->getOption().matches(options::OPT_mexecute_only))
+  D.Diag(diag::err_drv_argument_not_allowed_with)
+  << Desc << A->getAsString(Args);
+else
+  D.Diag(diag::err_drv_argument_not_allowed_with)
+  << Desc << Triple.str();
   }
   DiagnosedKinds |= KindsToDiagnose;
 }
diff --git a/clang/test/Driver/fsanitize.c b/clang/test/Driver/fsanitize.c
index 15f190165a7d73..a99bfb3e82ac95 100644
--- a/clang/test/Driver/fsanitize.c
+++ b/clang/test/Driver/fsanitize.c
@@ -1000,8 +1000,8 @@
 // RUN: not %clang --target=armv6t2-eabi -mexecute-only -fsanitize=kcfi %s 
-### 2>&1 | FileCheck %s --check-prefix=CHECK-UBSAN-KCFI
 // RUN: %clang --target=armv6t2-eabi -mexecute-only -fsanitize=undefined %s 
-### 2>&1 | FileCheck %s --check-prefix=CHECK-UBSAN-UNDEFINED-VPTR
 
-// CHECK-UBSAN-KCFI-DAG: error: invalid argument '-fsanitize=kcfi' not allowed 
with {{('x86_64-sie-ps5'|'armv6t2-unknown-unknown-eabi')}}
-// CHECK-UBSAN-FUNCTION-DAG: error: invalid argument '-fsanitize=function' not 
allowed with {{('x86_64-sie-ps5'|'armv6t2-unknown-unknown-eabi')}}
+// CHECK-UBSAN-KCFI-DAG: error: invalid argument '-fsanitize=kcfi' not allowed 
with {{('x86_64-sie-ps5'|'-mexecute-only')}}
+// CHECK-UBSAN-FUNCTION-DAG: error: invalid argument '-fsanitize=function' not 
allowed with {{('x86_64-sie-ps5'|'-mexecute-only')}}
 // CHECK-UBSAN-UNDEFINED-VPTR: 
"-fsanitize={{((alignment|array-bounds|bool|builtin|enum|float-cast-overflow|integer-divide-by-zero|nonnull-attribute|null|pointer-overflow|return|returns-nonnull-attribute|shift-base|shift-exponent|signed-integer-overflow|unreachable|vla-bound|vptr),?){18}"}}
 
 // * Test BareMetal toolchain sanitizer support *

From 7c352152d8d90bd02ca11071365b71af8f6678c1 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Wed, 11 Dec 2024 11:41:56 +0100
Subject: [PATCH 2/2] Use `err_drv_unsupported_opt_for_target` and add test for
 `-mpure-code`

---
 clang/lib/Driver/SanitizerArgs.cpp |  2 +-
 clang/test/Driver/fsanitize.c  | 22 ++
 2 files changed, 15 insertions(+), 9 deletions(-)

diff --git a/clang/lib/Driver/SanitizerArgs.cpp 
b/clang/lib/Driver/SanitizerArgs.cpp
index 6a2d376e93d05b..e77857930996b2 100644
--- a/clang/lib/Driver/SanitizerArgs.cpp
+++ b/clang/lib/Driver/SanitizerArgs.cpp
@@ -424,7 +424,7 @@ SanitizerArgs::SanitizerArgs(const ToolChain &TC,
   D.Diag(diag::err_drv_argument_not_allowed_with)
   << Desc << A->getAsString(Args);
 else
-  D.Diag(diag::err_drv_argument_not_allowed_with)
+  D.Diag(diag::err_drv_unsupported_opt_for_target)
   << Desc << Triple.str();
   }
   DiagnosedKinds |= KindsToDiagnose;
diff --git a/clang/test/Driver/fsanitize.c b/clang/test/Driver/fsanitize.c
index a99bfb3e82ac95..bb692b2aeea1d3 100644
--- a/clang/test/Driver/fsanitize.c
+++ b/clang/test/Driver/fsanitize.c
@@ -989,19 +989,25 @@
 // RUN: not %clang --target=x86_64-linux-gnu -fsanitize=undefined,function 
-mcmodel=large %s -### 2>&1 | FileCheck %s 
--check-prefix=CHECK-UBSAN-FUNCTION-CODE-MODEL
 // CHECK-UBSAN-FUNCTION-CODE-MODEL: error: invalid argument 
'-fsanitize=function' only allowed with '-mcmodel=small'
 
-// RUN: not %clang --target=x86

[clang] [Clang] Improve error for `-fsanitize=function/kcfi -mexecute-only` incompatibility (PR #118816)

2024-12-10 Thread Csanád Hajdú via cfe-commits 

Il-Capitano wrote:

@MaskRay Can I ask you to review/add appropriate reviewers for this patch?

https://github.com/llvm/llvm-project/pull/118816
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang] Improve error for `-fsanitize=function/kcfi -mexecute-only` incompatibility (PR #118816)

2024-12-07 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano edited 
https://github.com/llvm/llvm-project/pull/118816
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang] Improve error for `-fsanitize=function/kcfi -mexecute-only` incompatibility (PR #118816)

2024-12-07 Thread Csanád Hajdú via cfe-commits 


@@ -418,8 +418,14 @@ SanitizerArgs::SanitizerArgs(const ToolChain &TC,
 Add & NotAllowedWithExecuteOnly & ~DiagnosedKinds) {
   if (DiagnoseErrors) {
 std::string Desc = describeSanitizeArg(Arg, KindsToDiagnose);
-D.Diag(diag::err_drv_argument_not_allowed_with)
-<< Desc << Triple.str();
+llvm::opt::Arg *A = Args.getLastArgNoClaim(
+options::OPT_mexecute_only, options::OPT_mno_execute_only);
+if (A && A->getOption().matches(options::OPT_mexecute_only))
+  D.Diag(diag::err_drv_argument_not_allowed_with)
+  << Desc << A->getAsString(Args);
+else
+  D.Diag(diag::err_drv_argument_not_allowed_with)
+  << Desc << Triple.str();

Il-Capitano wrote:

Maybe `diag::err_drv_unsupported_opt_for_target` would be a better diagnostic 
in this case? That would produce
```
error: unsupported option '-fsanitize=function' for target ''
```

https://github.com/llvm/llvm-project/pull/118816
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang] Add float type support to __builtin_reduce_add and __builtin_reduce_multipy (PR #120367)

2024-12-25 Thread Csanád Hajdú via cfe-commits 


@@ -4274,12 +4274,37 @@ RValue CodeGenFunction::EmitBuiltinExpr(const 
GlobalDecl GD, unsigned BuiltinID,
 *this, E, GetIntrinsicID(E->getArg(0)->getType()), "rdx.min"));
   }
 
-  case Builtin::BI__builtin_reduce_add:
+  case Builtin::BI__builtin_reduce_add: {
+// Note: vector_reduce_fadd takes two arguments a
+// scalar start value and a vector. That would mean to
+// correctly call it we would need emitBuiltinWithOneOverloadedType<2>
+// To keep the  builtin sema behavior the same despite type we will
+// popululate vector_reduce_fadd scalar value with a 0.
+if (E->getArg(0)->getType()->hasFloatingRepresentation()) {
+  Value *X = EmitScalarExpr(E->getArg(0));
+  auto EltTy = X->getType()->getScalarType();
+  Value *Seed = ConstantFP::get(EltTy, 0);
+  return RValue::get(Builder.CreateIntrinsic(
+  /*ReturnType=*/EltTy, llvm::Intrinsic::vector_reduce_fadd,
+  ArrayRef{Seed, X}, nullptr, "rdx.fadd"));
+}
+assert(E->getArg(0)->getType()->hasIntegerRepresentation());
 return RValue::get(emitBuiltinWithOneOverloadedType<1>(
 *this, E, llvm::Intrinsic::vector_reduce_add, "rdx.add"));
-  case Builtin::BI__builtin_reduce_mul:
+  }
+  case Builtin::BI__builtin_reduce_mul: {
+if (E->getArg(0)->getType()->hasFloatingRepresentation()) {
+  Value *X = EmitScalarExpr(E->getArg(0));
+  auto EltTy = X->getType()->getScalarType();
+  Value *Seed = ConstantFP::get(EltTy, 0);

Il-Capitano wrote:

Use `1.0` as the initial value.

https://github.com/llvm/llvm-project/pull/120367
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang] Add float type support to __builtin_reduce_add and __builtin_reduce_multipy (PR #120367)

2024-12-25 Thread Csanád Hajdú via cfe-commits 


@@ -4274,12 +4274,37 @@ RValue CodeGenFunction::EmitBuiltinExpr(const 
GlobalDecl GD, unsigned BuiltinID,
 *this, E, GetIntrinsicID(E->getArg(0)->getType()), "rdx.min"));
   }
 
-  case Builtin::BI__builtin_reduce_add:
+  case Builtin::BI__builtin_reduce_add: {
+// Note: vector_reduce_fadd takes two arguments a
+// scalar start value and a vector. That would mean to
+// correctly call it we would need emitBuiltinWithOneOverloadedType<2>
+// To keep the  builtin sema behavior the same despite type we will
+// popululate vector_reduce_fadd scalar value with a 0.
+if (E->getArg(0)->getType()->hasFloatingRepresentation()) {
+  Value *X = EmitScalarExpr(E->getArg(0));
+  auto EltTy = X->getType()->getScalarType();
+  Value *Seed = ConstantFP::get(EltTy, 0);

Il-Capitano wrote:

Use `-0.0` as the initial value, as that's the neutral additive float, i.e. 
`-0.0 + x == x`. This can have an effect on code generation, see for example 
https://github.com/rust-lang/rust/issues/130028.

https://github.com/llvm/llvm-project/pull/120367
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang] Add float type support to __builtin_reduce_add and __builtin_reduce_multipy (PR #120367)

2024-12-25 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano edited 
https://github.com/llvm/llvm-project/pull/120367
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][Docs] Add release note for execute-only support on AArch64 (PR #134802)

2025-04-08 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano updated 
https://github.com/llvm/llvm-project/pull/134802

From c48168df95781b8fe5d6f72c9bdcbe1ad6a94fe5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Tue, 8 Apr 2025 09:47:46 +0200
Subject: [PATCH 1/2] [Clang][Docs] Add release note for execute-only support
 on AArch64

---
 clang/docs/ReleaseNotes.rst | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index e671183522565..2ca9b0e4b0ae1 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -190,6 +190,9 @@ Modified Compiler Flags
 
 - The compiler flag `-fbracket-depth` default value is increased from 256 to 
2048. (#GH94728)
 
+- The ``-mexecute-only`` and ``-mpure-code`` flags are now accepted for 
AArch64 targets,
+  allowing the generation of binaries with executable-only code sections. 
(#GH125688)
+
 Removed Compiler Flags
 -
 

From e3a97bca7622846ad1b55829c8242dc91fa55ee7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Tue, 8 Apr 2025 11:54:30 +0200
Subject: [PATCH 2/2] Add note in "Arm and AArch64 support" section too

---
 clang/docs/ReleaseNotes.rst | 6 --
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 2ca9b0e4b0ae1..17e18eae9e36b 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -190,8 +190,7 @@ Modified Compiler Flags
 
 - The compiler flag `-fbracket-depth` default value is increased from 256 to 
2048. (#GH94728)
 
-- The ``-mexecute-only`` and ``-mpure-code`` flags are now accepted for 
AArch64 targets,
-  allowing the generation of binaries with executable-only code sections. 
(#GH125688)
+- The ``-mexecute-only`` and ``-mpure-code`` flags are now accepted for 
AArch64 targets. (#GH125688)
 
 Removed Compiler Flags
 -
@@ -474,6 +473,9 @@ X86 Support
 Arm and AArch64 Support
 ^^^
 
+- For AArch64, added support for generating executable-only code sections by 
using the
+  ``-mexecute-only`` or ``-mpure-code`` compiler flags. (#GH125688)
+
 Android Support
 ^^^
 

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

[clang] [Clang][Docs] Add release note for execute-only support on AArch64 (PR #134802)

2025-04-08 Thread Csanád Hajdú via cfe-commits 

Il-Capitano wrote:

> There is a `Arm and AArch64 Support` section too, is it worth adding a note 
> there either instead or as well? It could maybe just be a note that points 
> back to the Modified Compiler Flags section.

Thanks, I didn't notice that section.

I simplified the note in "Modified Compiler Flags" and added one in "Arm and 
AArch64 Support".

https://github.com/llvm/llvm-project/pull/134802
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][Docs] Add release note for execute-only support on AArch64 (PR #134802)

2025-04-08 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano created 
https://github.com/llvm/llvm-project/pull/134802

None

From c48168df95781b8fe5d6f72c9bdcbe1ad6a94fe5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Tue, 8 Apr 2025 09:47:46 +0200
Subject: [PATCH] [Clang][Docs] Add release note for execute-only support on
 AArch64

---
 clang/docs/ReleaseNotes.rst | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index e671183522565..2ca9b0e4b0ae1 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -190,6 +190,9 @@ Modified Compiler Flags
 
 - The compiler flag `-fbracket-depth` default value is increased from 256 to 
2048. (#GH94728)
 
+- The ``-mexecute-only`` and ``-mpure-code`` flags are now accepted for 
AArch64 targets,
+  allowing the generation of binaries with executable-only code sections. 
(#GH125688)
+
 Removed Compiler Flags
 -
 

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

[clang] [Clang][Docs] Add release note for execute-only support on AArch64 (PR #134802)

2025-04-17 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano closed 
https://github.com/llvm/llvm-project/pull/134802
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][Docs] Add release note for execute-only support on AArch64 (PR #134802)

2025-04-17 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano updated 
https://github.com/llvm/llvm-project/pull/134802

From f41bcb085ac07ed5a05c1246a13048896e63d54c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Tue, 8 Apr 2025 09:47:46 +0200
Subject: [PATCH 1/2] [Clang][Docs] Add release note for execute-only support
 on AArch64

---
 clang/docs/ReleaseNotes.rst | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 0ea8498351c24..c99033fae4d39 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -219,6 +219,9 @@ Modified Compiler Flags
 
 - `-Wpadded` option implemented for the `x86_64-windows-msvc` target. Fixes 
#61702
 
+- The ``-mexecute-only`` and ``-mpure-code`` flags are now accepted for 
AArch64 targets,
+  allowing the generation of binaries with executable-only code sections. 
(#GH125688)
+
 Removed Compiler Flags
 -
 

From 87fec09c9fbff9637979465c0286ba2a56866995 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Tue, 8 Apr 2025 11:54:30 +0200
Subject: [PATCH 2/2] Add note in "Arm and AArch64 support" section too

---
 clang/docs/ReleaseNotes.rst | 6 --
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index c99033fae4d39..0a46bbea96f6c 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -219,8 +219,7 @@ Modified Compiler Flags
 
 - `-Wpadded` option implemented for the `x86_64-windows-msvc` target. Fixes 
#61702
 
-- The ``-mexecute-only`` and ``-mpure-code`` flags are now accepted for 
AArch64 targets,
-  allowing the generation of binaries with executable-only code sections. 
(#GH125688)
+- The ``-mexecute-only`` and ``-mpure-code`` flags are now accepted for 
AArch64 targets. (#GH125688)
 
 Removed Compiler Flags
 -
@@ -572,6 +571,9 @@ Arm and AArch64 Support
 
 -  Support for __ptrauth type qualifier has been added.
 
+- For AArch64, added support for generating executable-only code sections by 
using the
+  ``-mexecute-only`` or ``-mpure-code`` compiler flags. (#GH125688)
+
 Android Support
 ^^^
 

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

[clang] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-02-18 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano updated 
https://github.com/llvm/llvm-project/pull/125688

From a25845dcb484514c83022a21ee924e31a39882f4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Tue, 4 Feb 2025 14:51:55 +0100
Subject: [PATCH 1/2] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF
 section flag (2/3)

Add support for the new SHF_AARCH64_PURECODE ELF section flag: 
https://github.com/ARM-software/abi-aa/pull/304

The general implementation follows the existing one for ARM targets.
Simlarly to ARM targets, generating object files with the
`SHF_AARCH64_PURECODE` flag set is enabled by the
`-mexecute-only`/`-mpure-code` driver flag.
---
 clang/include/clang/Driver/Options.td|  4 ++--
 clang/lib/Driver/ToolChains/Arch/AArch64.cpp | 11 +
 clang/test/CodeGen/aarch64-execute-only.c| 24 
 clang/test/Driver/aarch64-execute-only.c |  6 +
 clang/test/Driver/fsanitize.c|  8 ++-
 5 files changed, 50 insertions(+), 3 deletions(-)
 create mode 100644 clang/test/CodeGen/aarch64-execute-only.c
 create mode 100644 clang/test/Driver/aarch64-execute-only.c

diff --git a/clang/include/clang/Driver/Options.td 
b/clang/include/clang/Driver/Options.td
index 5ad187926e710..0c8c4e5e0fae4 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -4692,9 +4692,9 @@ def mno_long_calls : Flag<["-"], "mno-long-calls">, 
Group,
   HelpText<"Restore the default behaviour of not generating long calls">;
 } // let Flags = [TargetSpecific]
 def mexecute_only : Flag<["-"], "mexecute-only">, Group,
-  HelpText<"Disallow generation of data access to code sections (ARM only)">;
+  HelpText<"Disallow generation of data access to code sections (AArch64/ARM 
only)">;
 def mno_execute_only : Flag<["-"], "mno-execute-only">, 
Group,
-  HelpText<"Allow generation of data access to code sections (ARM only)">;
+  HelpText<"Allow generation of data access to code sections (AArch64/ARM 
only)">;
 let Flags = [TargetSpecific] in {
 def mtp_mode_EQ : Joined<["-"], "mtp=">, Group, 
Values<"soft,cp15,tpidrurw,tpidruro,tpidrprw,el0,el1,el2,el3,tpidr_el0,tpidr_el1,tpidr_el2,tpidr_el3,tpidrro_el0">,
   HelpText<"Thread pointer access method. "
diff --git a/clang/lib/Driver/ToolChains/Arch/AArch64.cpp 
b/clang/lib/Driver/ToolChains/Arch/AArch64.cpp
index 1e2ac4e501baf..1248fea50f9de 100644
--- a/clang/lib/Driver/ToolChains/Arch/AArch64.cpp
+++ b/clang/lib/Driver/ToolChains/Arch/AArch64.cpp
@@ -332,6 +332,17 @@ void aarch64::getAArch64TargetFeatures(const Driver &D,
   } else if (Triple.isOSOpenBSD())
 Features.push_back("+strict-align");
 
+  // Generate execute-only output (no data access to code sections).
+  // This only makes sense for the compiler, not for the assembler.
+  if (!ForAS) {
+if (Arg *A = Args.getLastArg(options::OPT_mexecute_only,
+ options::OPT_mno_execute_only)) {
+  if (A->getOption().matches(options::OPT_mexecute_only)) {
+Features.push_back("+execute-only");
+  }
+}
+  }
+
   if (Args.hasArg(options::OPT_ffixed_x1))
 Features.push_back("+reserve-x1");
 
diff --git a/clang/test/CodeGen/aarch64-execute-only.c 
b/clang/test/CodeGen/aarch64-execute-only.c
new file mode 100644
index 0..d885e954166f6
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-execute-only.c
@@ -0,0 +1,24 @@
+// RUN: %clang -target aarch64 -### %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-NO-EXECUTE-ONLY
+
+// RUN: %clang -target aarch64 -### -mexecute-only %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-EXECUTE-ONLY
+
+// RUN: %clang -target aarch64 -### -mexecute-only -mno-execute-only %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-NO-EXECUTE-ONLY
+
+
+// -mpure-code flag for GCC compatibility
+// RUN: %clang -target aarch64 -### %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-NO-EXECUTE-ONLY
+
+// RUN: %clang -target aarch64 -### -mpure-code %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-EXECUTE-ONLY
+
+// RUN: %clang -target aarch64 -### -mpure-code -mno-pure-code %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-NO-EXECUTE-ONLY
+
+// CHECK-NO-EXECUTE-ONLY-NOT: "+execute-only"
+// CHECK-EXECUTE-ONLY: "+execute-only"
+
+void a() {}
diff --git a/clang/test/Driver/aarch64-execute-only.c 
b/clang/test/Driver/aarch64-execute-only.c
new file mode 100644
index 0..67c6d4a5eb066
--- /dev/null
+++ b/clang/test/Driver/aarch64-execute-only.c
@@ -0,0 +1,6 @@
+// RUN: %clang --sysroot=%S/Inputs -c -fdriver-only -Werror 
--target=aarch64-unknown-linux-gnu \
+// RUN:   -mexecute-only %s 2>&1 | count 0
+
+// RUN: %clang -### --target=aarch64-unknown-linux-gnu -x assembler 
-mexecute-only %s -c -### 2>&1 \
+// RUN:| FileCheck %s --check-prefix=CHECK-NO-EXECUTE-ONLY-ASM
+// CHECK-NO-EXECUTE-ONLY-ASM: warning: argument unused during compilation: 
'-mexecute-only'
diff --git a/clang/test/Driver/fsaniti

[clang] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-02-18 Thread Csanád Hajdú via cfe-commits 


@@ -0,0 +1,6 @@
+// RUN: %clang --sysroot=%S/Inputs -c -fdriver-only -Werror 
--target=aarch64-unknown-linux-gnu \

Il-Capitano wrote:

There must have been a good reason I removed the `!ForMultilib` check, but I 
can't remember now...

You're right, the check is needed for bare metal targets. As is, the test would 
pass with `--target=aarch64-none-elf`, but that's because there's no 
`sysroot/aarch64-none-elf/multilib.yaml` file found. However I was able to test 
for this by passing the `--multi-lib-config=%S/Inputs/multilib/empty.yaml` flag.

https://github.com/llvm/llvm-project/pull/125688
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-03-10 Thread Csanád Hajdú via cfe-commits 

Il-Capitano wrote:

Ping.

https://github.com/llvm/llvm-project/pull/125688
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-02-28 Thread Csanád Hajdú via cfe-commits 

Il-Capitano wrote:

@jansvoboda11 You're tagged as the compiler options maintainer for Clang, so 
this may be of interest to you as well.

https://github.com/llvm/llvm-project/pull/125688
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-02-14 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano updated 
https://github.com/llvm/llvm-project/pull/125688

From a25845dcb484514c83022a21ee924e31a39882f4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Tue, 4 Feb 2025 14:51:55 +0100
Subject: [PATCH] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF
 section flag (2/3)

Add support for the new SHF_AARCH64_PURECODE ELF section flag: 
https://github.com/ARM-software/abi-aa/pull/304

The general implementation follows the existing one for ARM targets.
Simlarly to ARM targets, generating object files with the
`SHF_AARCH64_PURECODE` flag set is enabled by the
`-mexecute-only`/`-mpure-code` driver flag.
---
 clang/include/clang/Driver/Options.td|  4 ++--
 clang/lib/Driver/ToolChains/Arch/AArch64.cpp | 11 +
 clang/test/CodeGen/aarch64-execute-only.c| 24 
 clang/test/Driver/aarch64-execute-only.c |  6 +
 clang/test/Driver/fsanitize.c|  8 ++-
 5 files changed, 50 insertions(+), 3 deletions(-)
 create mode 100644 clang/test/CodeGen/aarch64-execute-only.c
 create mode 100644 clang/test/Driver/aarch64-execute-only.c

diff --git a/clang/include/clang/Driver/Options.td 
b/clang/include/clang/Driver/Options.td
index 5ad187926e710..0c8c4e5e0fae4 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -4692,9 +4692,9 @@ def mno_long_calls : Flag<["-"], "mno-long-calls">, 
Group,
   HelpText<"Restore the default behaviour of not generating long calls">;
 } // let Flags = [TargetSpecific]
 def mexecute_only : Flag<["-"], "mexecute-only">, Group,
-  HelpText<"Disallow generation of data access to code sections (ARM only)">;
+  HelpText<"Disallow generation of data access to code sections (AArch64/ARM 
only)">;
 def mno_execute_only : Flag<["-"], "mno-execute-only">, 
Group,
-  HelpText<"Allow generation of data access to code sections (ARM only)">;
+  HelpText<"Allow generation of data access to code sections (AArch64/ARM 
only)">;
 let Flags = [TargetSpecific] in {
 def mtp_mode_EQ : Joined<["-"], "mtp=">, Group, 
Values<"soft,cp15,tpidrurw,tpidruro,tpidrprw,el0,el1,el2,el3,tpidr_el0,tpidr_el1,tpidr_el2,tpidr_el3,tpidrro_el0">,
   HelpText<"Thread pointer access method. "
diff --git a/clang/lib/Driver/ToolChains/Arch/AArch64.cpp 
b/clang/lib/Driver/ToolChains/Arch/AArch64.cpp
index 1e2ac4e501baf..1248fea50f9de 100644
--- a/clang/lib/Driver/ToolChains/Arch/AArch64.cpp
+++ b/clang/lib/Driver/ToolChains/Arch/AArch64.cpp
@@ -332,6 +332,17 @@ void aarch64::getAArch64TargetFeatures(const Driver &D,
   } else if (Triple.isOSOpenBSD())
 Features.push_back("+strict-align");
 
+  // Generate execute-only output (no data access to code sections).
+  // This only makes sense for the compiler, not for the assembler.
+  if (!ForAS) {
+if (Arg *A = Args.getLastArg(options::OPT_mexecute_only,
+ options::OPT_mno_execute_only)) {
+  if (A->getOption().matches(options::OPT_mexecute_only)) {
+Features.push_back("+execute-only");
+  }
+}
+  }
+
   if (Args.hasArg(options::OPT_ffixed_x1))
 Features.push_back("+reserve-x1");
 
diff --git a/clang/test/CodeGen/aarch64-execute-only.c 
b/clang/test/CodeGen/aarch64-execute-only.c
new file mode 100644
index 0..d885e954166f6
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-execute-only.c
@@ -0,0 +1,24 @@
+// RUN: %clang -target aarch64 -### %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-NO-EXECUTE-ONLY
+
+// RUN: %clang -target aarch64 -### -mexecute-only %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-EXECUTE-ONLY
+
+// RUN: %clang -target aarch64 -### -mexecute-only -mno-execute-only %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-NO-EXECUTE-ONLY
+
+
+// -mpure-code flag for GCC compatibility
+// RUN: %clang -target aarch64 -### %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-NO-EXECUTE-ONLY
+
+// RUN: %clang -target aarch64 -### -mpure-code %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-EXECUTE-ONLY
+
+// RUN: %clang -target aarch64 -### -mpure-code -mno-pure-code %s 2>&1 \
+// RUN:| FileCheck %s -check-prefix CHECK-NO-EXECUTE-ONLY
+
+// CHECK-NO-EXECUTE-ONLY-NOT: "+execute-only"
+// CHECK-EXECUTE-ONLY: "+execute-only"
+
+void a() {}
diff --git a/clang/test/Driver/aarch64-execute-only.c 
b/clang/test/Driver/aarch64-execute-only.c
new file mode 100644
index 0..67c6d4a5eb066
--- /dev/null
+++ b/clang/test/Driver/aarch64-execute-only.c
@@ -0,0 +1,6 @@
+// RUN: %clang --sysroot=%S/Inputs -c -fdriver-only -Werror 
--target=aarch64-unknown-linux-gnu \
+// RUN:   -mexecute-only %s 2>&1 | count 0
+
+// RUN: %clang -### --target=aarch64-unknown-linux-gnu -x assembler 
-mexecute-only %s -c -### 2>&1 \
+// RUN:| FileCheck %s --check-prefix=CHECK-NO-EXECUTE-ONLY-ASM
+// CHECK-NO-EXECUTE-ONLY-ASM: warning: argument unused during compilation: 
'-mexecute-only'
diff --git a/clang/test/Driver/fsanitize.c

[clang] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-02-14 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano edited 
https://github.com/llvm/llvm-project/pull/125688
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-02-14 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano ready_for_review 
https://github.com/llvm/llvm-project/pull/125688
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-02-14 Thread Csanád Hajdú via cfe-commits 

Il-Capitano wrote:

I've rebased this patch onto main, since 
https://github.com/llvm/llvm-project/pull/125687 has been merged, so it's now 
ready for review.

https://github.com/llvm/llvm-project/pull/125688
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[clang] [Clang][AArch64] Add support for SHF_AARCH64_PURECODE ELF section flag (2/3) (PR #125688)

2025-02-24 Thread Csanád Hajdú via cfe-commits 

Il-Capitano wrote:

@smithp35 @davemgreen @paulwalker-arm @jh7370 
Pinging to ask for more reviews on this, please.

https://github.com/llvm/llvm-project/pull/125688
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[libunwind] [libunwind] Add CMake option to enable execute-only code generation on AArch64 (PR #140554)

2025-05-19 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano created 
https://github.com/llvm/llvm-project/pull/140554

For a full toolchain supporting execute-only code generation the runtime 
libraries also need to be pre-compiled with it enabled. For libunwind this can 
now be enabled with the `LIBUNWIND_EXECUTE_ONLY_CODE` CMake option during build 
configuration.

The build option can only be enabled for a runtimes build of libunwind, because 
a recent version of Clang is needed to correctly compile assembly files with 
execute-only code support.

Related RFC: 
https://discourse.llvm.org/t/rfc-execute-only-code-support-for-runtime-libraries-on-aarch64/86180

From c69f17edaceb95154e25f0e9062363d88498a775 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Mon, 19 May 2025 12:29:43 +0200
Subject: [PATCH] [libunwind] Add CMake option to enable execute-only code
 generation on AArch64

For a full toolchain supporting execute-only code generation the runtime
libraries also need to be pre-compiled with it enabled. For libunwind
this can now be enabled with the `LIBUNWIND_EXECUTE_ONLY_CODE` CMake
option during build configuration.

The build option can only be enabled for a runtimes build of libunwind,
because a recent version of Clang is needed to correctly compile
assembly files with execute-only code support.

Related RFC: 
https://discourse.llvm.org/t/rfc-execute-only-code-support-for-runtime-libraries-on-aarch64/86180
---
 libunwind/CMakeLists.txt   | 18 ++
 libunwind/src/UnwindRegistersRestore.S |  2 ++
 libunwind/src/UnwindRegistersSave.S|  2 ++
 3 files changed, 22 insertions(+)

diff --git a/libunwind/CMakeLists.txt b/libunwind/CMakeLists.txt
index e27f3c2e2fc17..75195fe022d4b 100644
--- a/libunwind/CMakeLists.txt
+++ b/libunwind/CMakeLists.txt
@@ -55,6 +55,7 @@ option(LIBUNWIND_USE_FRAME_HEADER_CACHE "Cache frame headers 
for unwinding. Requ
 option(LIBUNWIND_REMEMBER_HEAP_ALLOC "Use heap instead of the stack for 
.cfi_remember_state." OFF)
 option(LIBUNWIND_INSTALL_HEADERS "Install the libunwind headers." ON)
 option(LIBUNWIND_ENABLE_FRAME_APIS "Include libgcc-compatible frame apis." OFF)
+option(LIBUNWIND_EXECUTE_ONLY_CODE "Compile libunwind as execute-only." OFF)
 
 set(LIBUNWIND_LIBDIR_SUFFIX "${LLVM_LIBDIR_SUFFIX}" CACHE STRING
 "Define suffix of library directory name (32/64)")
@@ -109,6 +110,11 @@ endif()
 option(LIBUNWIND_HIDE_SYMBOLS
   "Do not export any symbols from the static library." 
${LIBUNWIND_DEFAULT_HIDE_SYMBOLS})
 
+if (LIBUNWIND_EXECUTE_ONLY_CODE AND NOT LLVM_RUNTIMES_BUILD)
+  message(SEND_ERROR "LIBUNWIND_EXECUTE_ONLY_CODE is only supported "
+ "for runtimes build of libunwind.")
+endif()
+
 # If toolchain is FPXX, we switch to FP64 to save the full FPRs. See:
 # 
https://web.archive.org/web/20180828210612/https://dmz-portal.mips.com/wiki/MIPS_O32_ABI_-_FR0_and_FR1_Interlinking
 check_symbol_exists(__mips_hard_float "" __MIPSHF)
@@ -330,6 +336,18 @@ if (C_SUPPORTS_COMMENT_LIB_PRAGMA)
   endif()
 endif()
 
+if (LIBUNWIND_EXECUTE_ONLY_CODE)
+  add_compile_flags_if_supported(-mexecute-only)
+  if (NOT CXX_SUPPORTS_MEXECUTE_ONLY_FLAG)
+add_compile_flags_if_supported(-mpure-code)
+if (NOT CXX_SUPPORTS_MPURE_CODE_FLAG)
+  message(SEND_ERROR
+"Compiler doesn't support -mexecute-only or -mpure-code option!")
+endif()
+  endif()
+  add_compile_definitions(_LIBUNWIND_EXECUTE_ONLY_CODE)
+endif()
+
 
#===
 # Setup Source Code
 
#===
diff --git a/libunwind/src/UnwindRegistersRestore.S 
b/libunwind/src/UnwindRegistersRestore.S
index 5e199188945df..2a0e81a9794eb 100644
--- a/libunwind/src/UnwindRegistersRestore.S
+++ b/libunwind/src/UnwindRegistersRestore.S
@@ -16,6 +16,8 @@
 
 #if defined(_AIX)
   .toc
+#elif defined(__aarch64__) && defined(__ELF__) && 
defined(_LIBUNWIND_EXECUTE_ONLY_CODE)
+  .section .text,"axy",@progbits,unique,0
 #else
   .text
 #endif
diff --git a/libunwind/src/UnwindRegistersSave.S 
b/libunwind/src/UnwindRegistersSave.S
index 5139a551ad245..1ff14cf272b61 100644
--- a/libunwind/src/UnwindRegistersSave.S
+++ b/libunwind/src/UnwindRegistersSave.S
@@ -16,6 +16,8 @@
 
 #if defined(_AIX)
 .toc
+#elif defined(__aarch64__) && defined(__ELF__) && 
defined(_LIBUNWIND_EXECUTE_ONLY_CODE)
+.section .text,"axy",@progbits,unique,0
 #else
 .text
 #endif

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

[libunwind] [libunwind] Add CMake option to enable execute-only code generation on AArch64 (PR #140554)

2025-06-11 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano edited 
https://github.com/llvm/llvm-project/pull/140554
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[libunwind] [libunwind] Add CMake option to enable execute-only code generation on AArch64 (PR #140554)

2025-06-11 Thread Csanád Hajdú via cfe-commits 

https://github.com/Il-Capitano updated 
https://github.com/llvm/llvm-project/pull/140554

From c69f17edaceb95154e25f0e9062363d88498a775 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Mon, 19 May 2025 12:29:43 +0200
Subject: [PATCH 1/2] [libunwind] Add CMake option to enable execute-only code
 generation on AArch64

For a full toolchain supporting execute-only code generation the runtime
libraries also need to be pre-compiled with it enabled. For libunwind
this can now be enabled with the `LIBUNWIND_EXECUTE_ONLY_CODE` CMake
option during build configuration.

The build option can only be enabled for a runtimes build of libunwind,
because a recent version of Clang is needed to correctly compile
assembly files with execute-only code support.

Related RFC: 
https://discourse.llvm.org/t/rfc-execute-only-code-support-for-runtime-libraries-on-aarch64/86180
---
 libunwind/CMakeLists.txt   | 18 ++
 libunwind/src/UnwindRegistersRestore.S |  2 ++
 libunwind/src/UnwindRegistersSave.S|  2 ++
 3 files changed, 22 insertions(+)

diff --git a/libunwind/CMakeLists.txt b/libunwind/CMakeLists.txt
index e27f3c2e2fc17..75195fe022d4b 100644
--- a/libunwind/CMakeLists.txt
+++ b/libunwind/CMakeLists.txt
@@ -55,6 +55,7 @@ option(LIBUNWIND_USE_FRAME_HEADER_CACHE "Cache frame headers 
for unwinding. Requ
 option(LIBUNWIND_REMEMBER_HEAP_ALLOC "Use heap instead of the stack for 
.cfi_remember_state." OFF)
 option(LIBUNWIND_INSTALL_HEADERS "Install the libunwind headers." ON)
 option(LIBUNWIND_ENABLE_FRAME_APIS "Include libgcc-compatible frame apis." OFF)
+option(LIBUNWIND_EXECUTE_ONLY_CODE "Compile libunwind as execute-only." OFF)
 
 set(LIBUNWIND_LIBDIR_SUFFIX "${LLVM_LIBDIR_SUFFIX}" CACHE STRING
 "Define suffix of library directory name (32/64)")
@@ -109,6 +110,11 @@ endif()
 option(LIBUNWIND_HIDE_SYMBOLS
   "Do not export any symbols from the static library." 
${LIBUNWIND_DEFAULT_HIDE_SYMBOLS})
 
+if (LIBUNWIND_EXECUTE_ONLY_CODE AND NOT LLVM_RUNTIMES_BUILD)
+  message(SEND_ERROR "LIBUNWIND_EXECUTE_ONLY_CODE is only supported "
+ "for runtimes build of libunwind.")
+endif()
+
 # If toolchain is FPXX, we switch to FP64 to save the full FPRs. See:
 # 
https://web.archive.org/web/20180828210612/https://dmz-portal.mips.com/wiki/MIPS_O32_ABI_-_FR0_and_FR1_Interlinking
 check_symbol_exists(__mips_hard_float "" __MIPSHF)
@@ -330,6 +336,18 @@ if (C_SUPPORTS_COMMENT_LIB_PRAGMA)
   endif()
 endif()
 
+if (LIBUNWIND_EXECUTE_ONLY_CODE)
+  add_compile_flags_if_supported(-mexecute-only)
+  if (NOT CXX_SUPPORTS_MEXECUTE_ONLY_FLAG)
+add_compile_flags_if_supported(-mpure-code)
+if (NOT CXX_SUPPORTS_MPURE_CODE_FLAG)
+  message(SEND_ERROR
+"Compiler doesn't support -mexecute-only or -mpure-code option!")
+endif()
+  endif()
+  add_compile_definitions(_LIBUNWIND_EXECUTE_ONLY_CODE)
+endif()
+
 
#===
 # Setup Source Code
 
#===
diff --git a/libunwind/src/UnwindRegistersRestore.S 
b/libunwind/src/UnwindRegistersRestore.S
index 5e199188945df..2a0e81a9794eb 100644
--- a/libunwind/src/UnwindRegistersRestore.S
+++ b/libunwind/src/UnwindRegistersRestore.S
@@ -16,6 +16,8 @@
 
 #if defined(_AIX)
   .toc
+#elif defined(__aarch64__) && defined(__ELF__) && 
defined(_LIBUNWIND_EXECUTE_ONLY_CODE)
+  .section .text,"axy",@progbits,unique,0
 #else
   .text
 #endif
diff --git a/libunwind/src/UnwindRegistersSave.S 
b/libunwind/src/UnwindRegistersSave.S
index 5139a551ad245..1ff14cf272b61 100644
--- a/libunwind/src/UnwindRegistersSave.S
+++ b/libunwind/src/UnwindRegistersSave.S
@@ -16,6 +16,8 @@
 
 #if defined(_AIX)
 .toc
+#elif defined(__aarch64__) && defined(__ELF__) && 
defined(_LIBUNWIND_EXECUTE_ONLY_CODE)
+.section .text,"axy",@progbits,unique,0
 #else
 .text
 #endif

From debdd954146a2d564c95be5bc565a53a884f3069 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Csan=C3=A1d=20Hajd=C3=BA?= 
Date: Wed, 11 Jun 2025 11:13:11 +0200
Subject: [PATCH 2/2] Simplify CMake configuration based on RFC feedback

---
 libunwind/CMakeLists.txt | 16 +---
 1 file changed, 1 insertion(+), 15 deletions(-)

diff --git a/libunwind/CMakeLists.txt b/libunwind/CMakeLists.txt
index 75195fe022d4b..48a29a2a7ed9c 100644
--- a/libunwind/CMakeLists.txt
+++ b/libunwind/CMakeLists.txt
@@ -55,7 +55,6 @@ option(LIBUNWIND_USE_FRAME_HEADER_CACHE "Cache frame headers 
for unwinding. Requ
 option(LIBUNWIND_REMEMBER_HEAP_ALLOC "Use heap instead of the stack for 
.cfi_remember_state." OFF)
 option(LIBUNWIND_INSTALL_HEADERS "Install the libunwind headers." ON)
 option(LIBUNWIND_ENABLE_FRAME_APIS "Include libgcc-compatible frame apis." OFF)
-option(LIBUNWIND_EXECUTE_ONLY_CODE "Compile libunwind as execute-only." OFF)
 
 set(LIBUNWIND_LIBDIR_SUFFIX "${LLVM_LIBDIR_SUFFIX}" CACHE STRING
 "Define suffix of library directory name (32/64)")

[libunwind] [libunwind] Add CMake option to enable execute-only code generation on AArch64 (PR #140554)

2025-06-11 Thread Csanád Hajdú via cfe-commits 

Il-Capitano wrote:

I simplified the configuration based on the discussion in the RFC. C and C++ 
compiler flags will be handled by the change in 
https://github.com/llvm/llvm-project/pull/143698.

https://github.com/llvm/llvm-project/pull/140554
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits