@@ -0,0 +1,79 @@
+; RUN: llc -mcpu=mvp -filetype=obj %s -mattr=+branch-hinting
-wasm-branch-prob-high=0x6000 -wasm-branch-prob-low=0x0 -o - | obj2yaml |
FileCheck %s
aengelke wrote:
Should test different values (multiple run lines, different check prefixes
@@ -55,6 +55,15 @@ using namespace llvm;
#define DEBUG_TYPE "asm-printer"
extern cl::opt WasmKeepRegisters;
+// values are divided by 1<<31 to calculate the probability
+static cl::opt WasmHighBranchProb(
+"wasm-branch-prob-high", cl::Hidden,
+cl::desc("lowest branch p
@@ -28,6 +33,9 @@ class LLVM_LIBRARY_VISIBILITY WebAssemblyAsmPrinter final :
public AsmPrinter {
WebAssemblyFunctionInfo *MFI;
bool signaturesEmitted = false;
+ // vec idx == local func_idx
+ std::vector> branchHints;
aengelke wrote:
Use SmallVector.
@@ -2562,8 +2565,22 @@ void
WebAssemblyCFGStackify::rewriteDepthImmediates(MachineFunction &MF) {
MO = MachineOperand::CreateImm(getDelegateDepth(Stack, MO.getMBB()));
else if (MI.getOpcode() == WebAssembly::RETHROW)
MO = MachineOperand::CreateImm(g
@@ -18,6 +18,11 @@
namespace llvm {
class WebAssemblyTargetStreamer;
+struct BranchHintRecord {
+ MCSymbol *func_sym;
aengelke wrote:
Naming: `FuncSym`
https://github.com/llvm/llvm-project/pull/146230
___
cfe-comm
@@ -18,6 +18,11 @@
namespace llvm {
class WebAssemblyTargetStreamer;
+struct BranchHintRecord {
+ MCSymbol *func_sym;
+ SmallVector, 0> hints;
aengelke wrote:
```suggestion
SmallVector, 0> Hints;
```
https://github.com/llvm/llvm-project/pull/146230
_
@@ -0,0 +1,79 @@
+; RUN: llc -mcpu=mvp -filetype=obj %s -mattr=+branch-hinting
-wasm-branch-prob-high=0x6000 -wasm-branch-prob-low=0x0 -o - | obj2yaml |
FileCheck %s
+
+; This test checks that branch weight metadata (!prof) is correctly translated
to webassembly branch hint
@@ -697,6 +738,34 @@ void WebAssemblyAsmPrinter::emitInstruction(const
MachineInstr *MI) {
WebAssemblyMCInstLower MCInstLowering(OutContext, *this);
MCInst TmpInst;
MCInstLowering.lower(MI, TmpInst);
+if (Subtarget->hasBranchHinting() &&
+MI->getOpcode(
@@ -697,6 +738,34 @@ void WebAssemblyAsmPrinter::emitInstruction(const
MachineInstr *MI) {
WebAssemblyMCInstLower MCInstLowering(OutContext, *this);
MCInst TmpInst;
MCInstLowering.lower(MI, TmpInst);
+if (Subtarget->hasBranchHinting() &&
+MI->getOpcode(
https://github.com/aengelke commented:
Some comments on the LLVM MC part. I'm not too familiar with the Wasm back-end
or LLD.
https://github.com/llvm/llvm-project/pull/146230
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.o
@@ -59,6 +67,7 @@ class LLVM_LIBRARY_VISIBILITY WebAssemblyAsmPrinter final :
public AsmPrinter {
void EmitProducerInfo(Module &M);
void EmitTargetFeatures(Module &M);
void EmitFunctionAttributes(Module &M);
+ void EmitBranchHintSection() const;
aengelk
https://github.com/aengelke edited
https://github.com/llvm/llvm-project/pull/146230
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/aengelke closed
https://github.com/llvm/llvm-project/pull/103025
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/aengelke closed
https://github.com/llvm/llvm-project/pull/103302
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/aengelke edited
https://github.com/llvm/llvm-project/pull/103302
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -109,7 +109,6 @@ initializeRecordStreamer(const Module &M,
MCContext MCCtx(TT, MAI.get(), MRI.get(), STI.get(), &SrcMgr);
std::unique_ptr MOFI(
T->createMCObjectFileInfo(MCCtx, /*PIC=*/false));
- MOFI->setSDKVersion(M.getSDKVersion());
aengelke wr
@@ -559,6 +559,9 @@ int main(int argc, char **argv) {
std::unique_ptr(CE), *STI));
if (NoExecStack)
Str->initSections(true, *STI);
+if (TheTriple.isOSBinFormatMachO() && TheTriple.isOSDarwin())
aengelke wrote:
No. I just put it in front o
https://github.com/aengelke updated
https://github.com/llvm/llvm-project/pull/103025
>From 90c48050f1e56613a4a68056e608de5fdc3e9cb6 Mon Sep 17 00:00:00 2001
From: Alexis Engelke
Date: Tue, 13 Aug 2024 09:08:15 +
Subject: [PATCH 1/2] [MC] Remove Darwin SDK/Version from ObjFileInfo
There's o
https://github.com/aengelke created
https://github.com/llvm/llvm-project/pull/103025
There's only a single user (MCMachOStreamer), so it makes more sense to move
the version emission to the source of the data.
(Not sure whether we want this, but I don't see benefits in storing this
informatio
aengelke wrote:
Is there a previous discussion (e.g., RFC) about this? Did you consider
alternative approaches? (E.g., just a spontaneous idea, adding a specially
named symbol for every jump table where the symbol size is the size of the jump
table; entry size (4/8) could be encoded in the nam
https://github.com/aengelke closed
https://github.com/llvm/llvm-project/pull/97396
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/aengelke edited
https://github.com/llvm/llvm-project/pull/97396
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/aengelke edited
https://github.com/llvm/llvm-project/pull/97396
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
aengelke wrote:
> What I feel the right initial question is, is this check actually making a
> significant difference to performance in real-world use cases?
raw_ostream::write shows up in my (downstream application) profiles at ~0.6%,
mostly through raw_svector_ostream, which is always unbuff
https://github.com/aengelke updated
https://github.com/llvm/llvm-project/pull/97396
>From 04b794a8a02f06803f96c40de042b13fd4b12c5f Mon Sep 17 00:00:00 2001
From: Alexis Engelke
Date: Mon, 1 Jul 2024 16:50:44 +0200
Subject: [PATCH 1/2] [Support] Remove raw_ostream::tie
Originally, tie was intro
https://github.com/aengelke created
https://github.com/llvm/llvm-project/pull/97396
Originally, tie was introduced by D81156 to flush stdout before writing to
stderr. 030897523 reverted this due to race conditions. Nonetheless, it does
cost performance, causing an extra check in the "cold" pat
26 matches
Mail list logo
