[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n added inline comments.



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:597
+  // callsite violating the constraint.
+  if (checkCallerDoesNotAccessMemory() && !CB->doesNotAccessMemory()) {
+// Wait until we know we actually need it to do potentially expensive

goldstein.w.n wrote:
> nikic wrote:
> > For these you generally want to query `getMemoryEffects()` on the 
> > caller/callee once and then work on that representation, instead of doing 
> > separate queries for everything. This may allow more precise handling and 
> > likely reduces the compile-time impact, as doing these attribute lookups is 
> > quite expensive.
> Re-running compile time checks, will post revised numbers tomorrow.
No real impact on compile time.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n added inline comments.



Comment at: llvm/lib/Transforms/IPO/FunctionAttrs.cpp:1764
+  if (ICA.processFunction(F))
+Changed.insert(F);
+  }

Do we actually need to add the function to the changed list here? We aren't 
modifying the functions attributes, only callsites within the function.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n added inline comments.



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:597
+  // callsite violating the constraint.
+  if (checkCallerDoesNotAccessMemory() && !CB->doesNotAccessMemory()) {
+// Wait until we know we actually need it to do potentially expensive

nikic wrote:
> For these you generally want to query `getMemoryEffects()` on the 
> caller/callee once and then work on that representation, instead of doing 
> separate queries for everything. This may allow more precise handling and 
> likely reduces the compile-time impact, as doing these attribute lookups is 
> quite expensive.
Re-running compile time checks, will post revised numbers tomorrow.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n updated this revision to Diff 531233.
goldstein.w.n added a comment.

Remove unused check count. Use memory effects directly instead of
function/callbase attribute API


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

Files:
  clang/test/CodeGen/LoongArch/inline-asm-constraints.c
  clang/test/CodeGen/LoongArch/inline-asm-operand-modifiers.c
  clang/test/CodeGen/LoongArch/intrinsic-la32.c
  clang/test/CodeGen/LoongArch/intrinsic-la64.c
  clang/test/CodeGen/PowerPC/builtins-ppc-build-pair-mma.c
  clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c
  
clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond-64bit-only.c
  clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond.c
  clang/test/CodeGen/PowerPC/ppc64-inline-asm.c
  clang/test/CodeGen/RISCV/riscv-inline-asm.c
  clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vwrite-csr.c
  clang/test/CodeGen/X86/fma-builtins-constrained.c
  clang/test/CodeGen/X86/ms-x86-intrinsics.c
  clang/test/CodeGen/aarch64-bf16-ldst-intrinsics.c
  clang/test/CodeGen/aarch64-ls64-inline-asm.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sb.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1ub.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1b.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1h.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1w.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilege.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilegt.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilerw-bfloat.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilerw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilewr-bfloat.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilewr.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_dup_neonq.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_get_neonq.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_set_neonq.c
  clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c
  clang/test/CodeGen/attr-riscv-rvv-vector-bits-call.c
  clang/test/CodeGen/attr-riscv-rvv-vector-bits-cast.c
  clang/test/CodeGen/msp430-builtins.c
  clang/test/CodeGen/nofpclass.c
  clang/test/Headers/wasm.c
  llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h
  llvm/lib/Transforms/IPO/FunctionAttrs.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp
  llvm/test/Other/cgscc-devirt-iteration.ll
  llvm/test/Transforms/FunctionAttrs/nonnull.ll
  llvm/test/Transforms/FunctionAttrs/readattrs.ll
  llvm/test/Transforms/FunctionAttrs/willreturn-callsites.ll
  llvm/test/Transforms/MergeFunc/mergefunc-preserve-debug-info.ll
  llvm/test/Transforms/PhaseOrdering/X86/loop-idiom-vs-indvars.ll
  llvm/test/Transforms/PhaseOrdering/memset-tail.ll

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n added a comment.

In D152226#4418978 , @goldstein.w.n 
wrote:

> In D152226#4418498 , @jdoerfert 
> wrote:
>
>> FWIW, I wish the time that went into this would have been to improve and 
>> enable the (lightweight) Attributor pass.
>> Anyway, I'm still doing that, and the latest numbers didn't look that bad. I 
>> will start to upstream stuff that I found.
>> At this point it might be good to understand if the (lightweight) Attributor 
>> would solve your problems or what would be left.
>
> With the way the `InferCallsiteAttrs` is written, it should be equally usable 
> in attributor as it is in FunctionAttrs. Guess my
> feeling was want to get this in s.t its actually usable. Once LW attributor 
> becomes usable should be easy to apply `InferCallsiteAttrs`
> there.

Guess I'm assuming this will be usable in the Attributor. That is the goal. If 
its not the case am happy to refactor so that it is.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n added a comment.

In D152226#4418498 , @jdoerfert wrote:

> FWIW, I wish the time that went into this would have been to improve and 
> enable the (lightweight) Attributor pass.
> Anyway, I'm still doing that, and the latest numbers didn't look that bad. I 
> will start to upstream stuff that I found.
> At this point it might be good to understand if the (lightweight) Attributor 
> would solve your problems or what would be left.

With the way the `InferCallsiteAttrs` is written, it should be equally usable 
in attributor as it is in FunctionAttrs. Guess my
feeling was want to get this in s.t its actually usable. Once LW attributor 
becomes usable should be easy to apply `InferCallsiteAttrs`
there.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Johannes Doerfert via Phabricator via cfe-commits]

jdoerfert added a comment.

FWIW, I wish the time that went into this would have been to improve and enable 
the (lightweight) Attributor pass.
Anyway, I'm still doing that, and the latest numbers didn't look that bad. I 
will start to upstream stuff that I found.
At this point it might be good to understand if the (lightweight) Attributor 
would solve your problems or what would be left.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Nikita Popov via Phabricator via cfe-commits]

nikic added inline comments.



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:597
+  // callsite violating the constraint.
+  if (checkCallerDoesNotAccessMemory() && !CB->doesNotAccessMemory()) {
+// Wait until we know we actually need it to do potentially expensive

For these you generally want to query `getMemoryEffects()` on the caller/callee 
once and then work on that representation, instead of doing separate queries 
for everything. This may allow more precise handling and likely reduces the 
compile-time impact, as doing these attribute lookups is quite expensive.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n updated this revision to Diff 530798.
goldstein.w.n added a comment.

fixes


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

Files:
  clang/test/CodeGen/LoongArch/inline-asm-constraints.c
  clang/test/CodeGen/LoongArch/inline-asm-operand-modifiers.c
  clang/test/CodeGen/LoongArch/intrinsic-la32.c
  clang/test/CodeGen/LoongArch/intrinsic-la64.c
  clang/test/CodeGen/PowerPC/builtins-ppc-build-pair-mma.c
  clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c
  
clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond-64bit-only.c
  clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond.c
  clang/test/CodeGen/PowerPC/ppc64-inline-asm.c
  clang/test/CodeGen/RISCV/riscv-inline-asm.c
  clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vwrite-csr.c
  clang/test/CodeGen/X86/fma-builtins-constrained.c
  clang/test/CodeGen/X86/ms-x86-intrinsics.c
  clang/test/CodeGen/aarch64-bf16-ldst-intrinsics.c
  clang/test/CodeGen/aarch64-ls64-inline-asm.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sb.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1ub.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1b.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1h.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1w.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilege.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilegt.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilerw-bfloat.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilerw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilewr-bfloat.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilewr.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_dup_neonq.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_get_neonq.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_set_neonq.c
  clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c
  clang/test/CodeGen/attr-riscv-rvv-vector-bits-call.c
  clang/test/CodeGen/attr-riscv-rvv-vector-bits-cast.c
  clang/test/CodeGen/msp430-builtins.c
  clang/test/CodeGen/nofpclass.c
  clang/test/Headers/wasm.c
  llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h
  llvm/lib/Transforms/IPO/FunctionAttrs.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp
  llvm/test/Other/cgscc-devirt-iteration.ll
  llvm/test/Transforms/FunctionAttrs/nonnull.ll
  llvm/test/Transforms/FunctionAttrs/readattrs.ll
  llvm/test/Transforms/FunctionAttrs/willreturn-callsites.ll
  llvm/test/Transforms/MergeFunc/mergefunc-preserve-debug-info.ll
  llvm/test/Transforms/PhaseOrdering/X86/loop-idiom-vs-indvars.ll
  llvm/test/Transforms/PhaseOrdering/memset-tail.ll

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n marked 8 inline comments as done.
goldstein.w.n added inline comments.



Comment at: llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h:28
+  // relatively high value is okay.
+  static constexpr unsigned kMaxChecks = UINT_MAX;
+

arsenm wrote:
> Does changing this to something meaningful change the compile time results?
lowered to 100, doesn't seem to help:
https://llvm-compile-time-tracker.com/compare.php?from=79feb6e78b818e33ec69abdc58c5f713d691554f&to=79e0b7e6f0547de9ad534c02869437160028af01&stat=instructions:u

Lowered to 10, doesn't seem to help either:
https://llvm-compile-time-tracker.com/compare.php?from=79feb6e78b818e33ec69abdc58c5f713d691554f&to=9e3f35bc8b6f0f261698ba1ce18229d1b8d9cdb6&stat=instructions%3Au

I'll drop unless you think there is a reason to keep it around.



Comment at: llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h:63
+  bool checkCallerHasFnAttr(Attribute::AttrKind Attr) const {
+return (CxtCB && CxtCB->hasFnAttr(Attr)) || Caller->hasFnAttribute(Attr);
+  };

arsenm wrote:
> Without looking at the rest of the context I would assume CxtB is always 
> valid if the class is valid and all the null checks should be dropped
Not quite, so `CxtB` is used to specify propagation of function attributes from 
a specific callsite. This may be useful in inline. i.e if we have:

foo -> bar -> baz
and
fiz -> bar -> buz

If `bar` in `foo` is getting inlined, we can propagate not only the function 
attributes of `bar`, but also the callsite attributes of that specific call to 
`bar` to any callsites (so `baz`).

At the moment its always null in fact.



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:343
+  // have derived from an argument. Finally, allocas/leaked mallocs in general
+  // are difficult (so we avoid them entirely). Callsites can arbitrarily
+  // store pointers in allocas for use later without violating a nocapture

arsenm wrote:
> 
I think callsite is regularly refererred to as singular word (in other files 
too) so prefer to keep as is. That okay?



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:357
+  // the caller will apply to the callsites throw. If the caller has a landing
+  // padd, its possible for the callsite to capture a pointer in a throw that
+  // is later cleared by the caller.

arsenm wrote:
> 
did 'pad' -> 'padd' but prefer to keep callsite as one word



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:692
+  else
+memset(&CurFnInfo, kMaybe, sizeof(CurFnInfo));
+  Caller = ParentFunc;

arsenm wrote:
> Not very C++y, use the default constructor?
If `PreserveCache` is true, then we can redo same function (non-sequentially) 
and get saved analysis. If you think this is meaningless and we will always do 
one-off functions I can change to just construct with a caller (although I 
think for inlining this might come up).


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein 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.
goldstein.w.n marked an inline comment as done.
Closed by commit rG4fa971ff62c3: [FunctionAttrs] Propagate some func/arg/ret 
attributes from caller to callsite… (authored by goldstein.w.n).

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

Files:
  clang/test/CodeGen/LoongArch/inline-asm-constraints.c
  clang/test/CodeGen/LoongArch/inline-asm-operand-modifiers.c
  clang/test/CodeGen/LoongArch/intrinsic-la32.c
  clang/test/CodeGen/LoongArch/intrinsic-la64.c
  clang/test/CodeGen/PowerPC/builtins-ppc-build-pair-mma.c
  clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c
  
clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond-64bit-only.c
  clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond.c
  clang/test/CodeGen/PowerPC/ppc64-inline-asm.c
  clang/test/CodeGen/RISCV/riscv-inline-asm.c
  clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vwrite-csr.c
  clang/test/CodeGen/X86/fma-builtins-constrained.c
  clang/test/CodeGen/X86/ms-x86-intrinsics.c
  clang/test/CodeGen/aarch64-bf16-ldst-intrinsics.c
  clang/test/CodeGen/aarch64-ls64-inline-asm.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sb.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1ub.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1b.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1h.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1w.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilege.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilegt.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilerw-bfloat.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilerw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilewr-bfloat.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilewr.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_dup_neonq.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_get_neonq.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_set_neonq.c
  clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c
  clang/test/CodeGen/attr-riscv-rvv-vector-bits-call.c
  clang/test/CodeGen/attr-riscv-rvv-vector-bits-cast.c
  clang/test/CodeGen/msp430-builtins.c
  clang/test/CodeGen/nofpclass.c
  clang/test/Headers/wasm.c
  llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h
  llvm/lib/Transforms/IPO/FunctionAttrs.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp
  llvm/test/Other/cgscc-devirt-iteration.ll
  llvm/test/Transforms/FunctionAttrs/nonnull.ll
  llvm/test/Transforms/FunctionAttrs/readattrs.ll
  llvm/test/Transforms/FunctionAttrs/willreturn-callsites.ll
  llvm/test/Transforms/MergeFunc/mergefunc-preserve-debug-info.ll
  llvm/test/Transforms/PhaseOrdering/X86/loop-idiom-vs-indvars.ll
  llvm/test/Transforms/PhaseOrdering/memset-tail.ll

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Matt Arsenault via Phabricator via cfe-commits]

arsenm added inline comments.



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:692
+  else
+memset(&CurFnInfo, kMaybe, sizeof(CurFnInfo));
+  Caller = ParentFunc;

Not very C++y, use the default constructor?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Matt Arsenault via Phabricator via cfe-commits]

arsenm added a comment.

Typo guranteed in message




Comment at: llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h:1
+//===- InferCallsiteAttrs.h - Propagate attributes to callsites 
---===//
+//

Missing C++ mode comment



Comment at: llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h:28
+  // relatively high value is okay.
+  static constexpr unsigned kMaxChecks = UINT_MAX;
+

Does changing this to something meaningful change the compile time results?



Comment at: llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h:63
+  bool checkCallerHasFnAttr(Attribute::AttrKind Attr) const {
+return (CxtCB && CxtCB->hasFnAttr(Attr)) || Caller->hasFnAttribute(Attr);
+  };

Without looking at the rest of the context I would assume CxtB is always valid 
if the class is valid and all the null checks should be dropped



Comment at: llvm/lib/Transforms/IPO/FunctionAttrs.cpp:72
 
+
 STATISTIC(NumMemoryAttr, "Number of functions with improved memory attribute");

Random whitespace change



Comment at: llvm/lib/Transforms/IPO/FunctionAttrs.cpp:1762
+  InferCallsiteAttrs ICA;
+  for (Function *F : Nodes.SCCNodes)
+if (F)

Braces



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:33
+static bool addCallsiteParamAttributes(CallBase *CB,
+   const SmallVector &ArgNos,
+   Attribute::AttrKind Attr) {

Use ArrayRef?



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:48
+// escape.
+static bool isCallUnkMallocLike(const Value *V) {
+  auto *MCB = dyn_cast(V);

Unk?



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:53
+
+  return MCB->returnDoesNotAlias();
+}

Don't mention malloc and just say noalias call?



Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:343
+  // have derived from an argument. Finally, allocas/leaked mallocs in general
+  // are difficult (so we avoid them entirely). Callsites can arbitrarily
+  // store pointers in allocas for use later without violating a nocapture





Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:345
+  // store pointers in allocas for use later without violating a nocapture
+  // gurantee by the caller, as the allocas are torn down at caller return.
+  // Likewise a leaked malloc would not be accessible outside of the caller,





Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:357
+  // the caller will apply to the callsites throw. If the caller has a landing
+  // padd, its possible for the callsite to capture a pointer in a throw that
+  // is later cleared by the caller.





Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:385
+  // attributes from.
+  for (unsigned I = 0; I < Caller->arg_size(); ++I)
+if (checkCallerHasParamAttr(I, Attribute::NoCapture))





Comment at: llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp:670
+  const Function *PF = BB->getParent();
+  assert(PF && "Function should never but null in this context");
+




Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n added a comment.

ping.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n updated this revision to Diff 529201.
goldstein.w.n added a comment.

Add header comments/license.
Use enum for kMaybe, kYes, kNo


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

Files:
  clang/test/CodeGen/LoongArch/inline-asm-constraints.c
  clang/test/CodeGen/LoongArch/inline-asm-operand-modifiers.c
  clang/test/CodeGen/LoongArch/intrinsic-la32.c
  clang/test/CodeGen/LoongArch/intrinsic-la64.c
  clang/test/CodeGen/PowerPC/builtins-ppc-build-pair-mma.c
  clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c
  
clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond-64bit-only.c
  clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond.c
  clang/test/CodeGen/PowerPC/ppc64-inline-asm.c
  clang/test/CodeGen/RISCV/riscv-inline-asm.c
  clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vwrite-csr.c
  clang/test/CodeGen/X86/fma-builtins-constrained.c
  clang/test/CodeGen/X86/ms-x86-intrinsics.c
  clang/test/CodeGen/aarch64-bf16-ldst-intrinsics.c
  clang/test/CodeGen/aarch64-ls64-inline-asm.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sb.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1ub.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1b.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1h.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1w.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilege.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilegt.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilerw-bfloat.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilerw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilewr-bfloat.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilewr.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_dup_neonq.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_get_neonq.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_set_neonq.c
  clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c
  clang/test/CodeGen/attr-riscv-rvv-vector-bits-call.c
  clang/test/CodeGen/attr-riscv-rvv-vector-bits-cast.c
  clang/test/CodeGen/msp430-builtins.c
  clang/test/CodeGen/nofpclass.c
  clang/test/Headers/wasm.c
  llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h
  llvm/lib/Transforms/IPO/FunctionAttrs.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp
  llvm/test/Other/cgscc-devirt-iteration.ll
  llvm/test/Transforms/FunctionAttrs/nonnull.ll
  llvm/test/Transforms/FunctionAttrs/readattrs.ll
  llvm/test/Transforms/FunctionAttrs/willreturn-callsites.ll
  llvm/test/Transforms/MergeFunc/mergefunc-preserve-debug-info.ll
  llvm/test/Transforms/PhaseOrdering/X86/loop-idiom-vs-indvars.ll
  llvm/test/Transforms/PhaseOrdering/memset-tail.ll

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Matt Arsenault via Phabricator via cfe-commits]

arsenm added inline comments.



Comment at: llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h:1
+#ifndef LLVM_TRANSFORMS_UTILS_INFERCALLSITEATTRS_H
+#define LLVM_TRANSFORMS_UTILS_INFERCALLSITEATTRS_H

Missing license header



Comment at: llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h:10-12
+  static constexpr uint8_t kMaybe = 0;
+  static constexpr uint8_t kYes = 1;
+  static constexpr uint8_t kNo = 2;

Why not a regular enum?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n added inline comments.



Comment at: libc/CMakeLists.txt:22
+# that clang will use it.
+if(LLVM_LIBC_MAKE_DISCOVERABLE)
+  if(LLVM_ENABLE_PER_TARGET_RUNTIME_DIR)

sivachandra wrote:
> Likely out of date and not required for this patch?
woops, left that in from the bug earlier.
Fixed.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n updated this revision to Diff 528707.
goldstein.w.n marked an inline comment as done.
goldstein.w.n added a comment.

Remove libc changes


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

Files:
  clang/test/CodeGen/LoongArch/inline-asm-constraints.c
  clang/test/CodeGen/LoongArch/inline-asm-operand-modifiers.c
  clang/test/CodeGen/LoongArch/intrinsic-la32.c
  clang/test/CodeGen/LoongArch/intrinsic-la64.c
  clang/test/CodeGen/PowerPC/builtins-ppc-build-pair-mma.c
  clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c
  
clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond-64bit-only.c
  clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond.c
  clang/test/CodeGen/PowerPC/ppc64-inline-asm.c
  clang/test/CodeGen/RISCV/riscv-inline-asm.c
  clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vwrite-csr.c
  clang/test/CodeGen/X86/fma-builtins-constrained.c
  clang/test/CodeGen/X86/ms-x86-intrinsics.c
  clang/test/CodeGen/aarch64-bf16-ldst-intrinsics.c
  clang/test/CodeGen/aarch64-ls64-inline-asm.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sb.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1ub.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1b.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1h.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1w.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilege.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilegt.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilerw-bfloat.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilerw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilewr-bfloat.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilewr.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_dup_neonq.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_get_neonq.c
  
clang/test/CodeGen/aarch64_neon_sve_bridge_intrinsics/acle_neon_sve_bridge_set_neonq.c
  clang/test/CodeGen/attr-arm-sve-vector-bits-cast.c
  clang/test/CodeGen/attr-riscv-rvv-vector-bits-call.c
  clang/test/CodeGen/attr-riscv-rvv-vector-bits-cast.c
  clang/test/CodeGen/msp430-builtins.c
  clang/test/CodeGen/nofpclass.c
  clang/test/Headers/wasm.c
  llvm/include/llvm/Transforms/Utils/InferCallsiteAttrs.h
  llvm/lib/Transforms/IPO/FunctionAttrs.cpp
  llvm/lib/Transforms/Utils/CMakeLists.txt
  llvm/lib/Transforms/Utils/InferCallsiteAttrs.cpp
  llvm/test/Other/cgscc-devirt-iteration.ll
  llvm/test/Transforms/FunctionAttrs/nonnull.ll
  llvm/test/Transforms/FunctionAttrs/readattrs.ll
  llvm/test/Transforms/FunctionAttrs/willreturn-callsites.ll
  llvm/test/Transforms/MergeFunc/mergefunc-preserve-debug-info.ll
  llvm/test/Transforms/PhaseOrdering/X86/loop-idiom-vs-indvars.ll
  llvm/test/Transforms/PhaseOrdering/memset-tail.ll

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Siva Chandra via Phabricator via cfe-commits]

sivachandra added inline comments.



Comment at: libc/CMakeLists.txt:22
+# that clang will use it.
+if(LLVM_LIBC_MAKE_DISCOVERABLE)
+  if(LLVM_ENABLE_PER_TARGET_RUNTIME_DIR)

Likely out of date and not required for this patch?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n added a comment.

llvm-test-suite passes with this patch enabled.
But note: **There currently some failing clang tests**.
Most of them look benign (just need to propagate a few attributes to a few 
thousand places),
but I was hoping there is a way to auto-regenerate them before going through 
laboriously by hand.
All the ones that can be auto regenerated I updated + a few by hand before 
thinking better of it.
Currently the list of failing tests in the llvm-project with all projects 
enabled using `ninja check-all` is:

  Failed Tests (52):
Clang :: CodeGen/SystemZ/builtins-systemz-zvector-constrained.c
Clang :: CodeGen/SystemZ/builtins-systemz-zvector.c
Clang :: CodeGen/SystemZ/builtins-systemz-zvector2-constrained.c
Clang :: CodeGen/SystemZ/builtins-systemz-zvector2.c
Clang :: CodeGen/SystemZ/builtins-systemz-zvector3-constrained.c
Clang :: CodeGen/SystemZ/builtins-systemz-zvector3.c
Clang :: CodeGen/SystemZ/systemz-inline-asm.c
Clang :: CodeGen/aarch64-neon-vcmla.c
Clang :: CodeGen/aarch64-sme-intrinsics/acle_sme_ld1.c
Clang :: CodeGen/aarch64-sme-intrinsics/acle_sme_ld1_vnum.c
Clang :: CodeGen/aarch64-sme-intrinsics/acle_sme_st1.c
Clang :: CodeGen/aarch64-sme-intrinsics/acle_sme_st1_vnum.c
Clang :: CodeGen/aarch64-sve-acle-__ARM_FEATURE_SVE_VECTOR_OPERATORS.c
Clang :: CodeGen/aarch64-sve-acle-__ARM_FEATURE_SVE_VECTOR_OPERATORS.cpp
Clang :: CodeGen/arm-cmse.c
Clang :: CodeGen/arm64-mte.c
Clang :: CodeGen/builtins-multiprecision.c
Clang :: CodeGen/builtins-wasm.c
Clang :: CodeGen/cfi-icall-cross-dso.c
Clang :: CodeGen/fp-contract-on-pragma.cpp
Clang :: CodeGen/fp-contract-pragma.cpp
Clang :: CodeGen/fp-strictfp-exp.cpp
Clang :: CodeGen/inline-asm-aarch64-flag-output.c
Clang :: CodeGen/inline-asm-x86-flag-output.c
Clang :: CodeGen/ms-intrinsics-other.c
Clang :: CodeGen/ms-intrinsics.c
Clang :: CodeGen/neon-crypto.c
Clang :: CodeGenCXX/RelativeVTablesABI/dynamic-cast.cpp
Clang :: CodeGenCXX/microsoft-abi-dynamic-cast.cpp
Clang :: CodeGenCXX/microsoft-abi-typeid.cpp
Clang :: CodeGenCXX/sizeof-unwind-exception.cpp
Clang :: CodeGenObjC/synchronized.m
Clang :: CodeGenObjCXX/exceptions-legacy.mm
Clang :: CodeGenOpenCL/builtins-amdgcn-dl-insts-gfx11.cl
Clang :: CodeGenOpenCL/builtins-amdgcn-dl-insts.cl
Clang :: CodeGenOpenCL/builtins-amdgcn-fp8.cl
Clang :: CodeGenOpenCL/builtins-amdgcn-gfx10.cl
Clang :: CodeGenOpenCL/builtins-amdgcn-gfx11.cl
Clang :: CodeGenOpenCL/builtins-amdgcn-gfx9.cl
Clang :: CodeGenOpenCL/builtins-amdgcn-mfma.cl
Clang :: CodeGenOpenCL/builtins-amdgcn-vi.cl
Clang :: CodeGenOpenCL/builtins-amdgcn-wmma-w32.cl
Clang :: CodeGenOpenCL/builtins-amdgcn-wmma-w64.cl
Clang :: CodeGenOpenCL/builtins-amdgcn.cl
Clang :: CodeGenOpenCL/builtins-f16.cl
Clang :: CodeGenOpenCL/builtins-generic-amdgcn.cl
Clang :: CodeGenOpenCL/single-precision-constant.cl
Clang :: Headers/__clang_hip_cmath.hip
Clang :: Headers/__clang_hip_math.hip
Clang :: Headers/__clang_hip_math_ocml_rounded_ops.hip
Clang :: Headers/ms-arm64-intrin.cpp
Clang :: OpenMP/bug57757.cpp

Is there a way to autoregen them, some of them need a few thousand changes 
(just adding `noundef` for the most part) that hopefully doesn't need to be 
done by hand.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D152226/new/

https://reviews.llvm.org/D152226

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

[PATCH] D152226: [FunctionAttrs] Propagate some func/arg/ret attributes from caller to callsite (WIP)

[Noah Goldstein via Phabricator via cfe-commits]

goldstein.w.n created this revision.
goldstein.w.n added reviewers: nikic, fhahn, jdoerfert, arsenm.
Herald added subscribers: libc-commits, luke, pmatos, asb, ormris, StephenFan, 
frasercrmck, luismarques, apazos, sameer.abuasal, pengfei, s.egerton, Jim, 
jocewei, PkmX, the_o, brucehoult, MartinMosbeck, rogfer01, edward-jones, 
zzheng, jrtc27, niosHD, sabuasal, simoncook, johnrusso, rbar, kbarton, 
hiraditya, sbc100, nemanjai.
Herald added projects: libc-project, All.
goldstein.w.n requested review of this revision.
Herald added subscribers: llvm-commits, cfe-commits, pcwang-thead, MaskRay, 
aheejin, wdng.
Herald added a reviewer: sstefan1.
Herald added projects: clang, LLVM.

This is the consolidation of D151644  and 
D151943  moved from
InstCombine to FunctionAttrs. This is based on discussion in the above
patches as well as D152081  (Attributor). 
This patch was written in a
way so it can have an immediate impact in currently active passes
(FunctionAttrs), but should be easy to port elsewhere (Attributor or
Inliner) if that makes more sense later on.

Some function attributes imply the attribute for all/some instructions
in the function. These attributes can be safely propagated to
callsites within the function that are missing the attribute. This can
be useful when 1) analyzing individual instructions in a function and

2. if the original caller is later inlined, as if the attributes are

not propagated, they will be lost.

This patch implements propagation in a new class/file
`InferCallsiteAttrs` which can hypothetically be included elsewhere.

At the moment this patch infers the following:

Function Attributes:

- mustprogress
- nofree
- willreturn
- All memory attributes (readnone, readonly, writeonly, argmem, etc...)
  - The memory attributes are only propagated IFF the set of pointers available 
to the callsite is the same as the set available outside the caller (i.e no 
local memory arguments from alloca or local malloc like functions).

Argument Attributes:

- noundef
- nonnull
- nofree
- readnone
- readonly
- writeonly
- nocapture
  - nocapture is only propagated IFF the set of pointers available to the 
callsite is the same as the set available outside the caller and its guranteed 
that between the callsite and function return, the state of any capture 
pointers will not change (so the nocaptured gurantee of the caller has been met 
by the instruction preceding the callsite and will not changed).

Argument are only propagated to callsite arguments that are also function
arguments, but not derived values.

Return Attributes:

- noundef
- nonnull

Return attributes are only propagated if the callsite's return value
is used as the caller's return and execution is guranteed to pass from
callsite to return.

The compile time hit of this for -O3 and -O3+thinLTO is ~[.02, .37]%
regression. Proper LTO, however, has more significant regressions (up
to 3.92%):
https://llvm-compile-time-tracker.com/compare.php?from=94407e1bba9807193afde61c56b6125c0fc0b1d1&to=79feb6e78b818e33ec69abdc58c5f713d691554f&stat=instructions:u


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D152226

Files:
  clang/test/CodeGen/LoongArch/inline-asm-constraints.c
  clang/test/CodeGen/LoongArch/inline-asm-operand-modifiers.c
  clang/test/CodeGen/LoongArch/intrinsic-la32.c
  clang/test/CodeGen/LoongArch/intrinsic-la64.c
  clang/test/CodeGen/PowerPC/builtins-ppc-build-pair-mma.c
  clang/test/CodeGen/PowerPC/builtins-ppc-pair-mma.c
  
clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond-64bit-only.c
  clang/test/CodeGen/PowerPC/builtins-ppc-xlcompat-LoadReseve-StoreCond.c
  clang/test/CodeGen/PowerPC/ppc64-inline-asm.c
  clang/test/CodeGen/RISCV/riscv-inline-asm.c
  clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vwrite-csr.c
  clang/test/CodeGen/X86/fma-builtins-constrained.c
  clang/test/CodeGen/X86/ms-x86-intrinsics.c
  clang/test/CodeGen/aarch64-bf16-ldst-intrinsics.c
  clang/test/CodeGen/aarch64-ls64-inline-asm.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sb.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1sw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1ub.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uh.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_ldnt1uw.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1b.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1h.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_stnt1w.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilege.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_whilegt.c
  clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_wh