date:20230118

[PATCH] D141811: [clang-format] Allow trailing return types in macros

2023-01-18 Thread Emilia Dreamer via Phabricator via cfe-commits

rymiel added a comment.

I suppose this means `auto a = (b) -> c;` is also technically broken, actually, 
and there's no easy fix for that.

I did have an idea for adding an additional pass taking place somewhere in 
`TokenAnnotator::calculateFormattingInformation`, which would detect trailing 
return types using a bit more context, as a potential fix for 
https://github.com/llvm/llvm-project/issues/41495. I suppose I could make that 
theoretically replace all of this AutoFound logic. I should actually try to see 
if my idea would work at all though.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D141811/new/

https://reviews.llvm.org/D141811

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

[PATCH] D141908: [C++20][Modules] Handle defaulted and deleted functions in header units.

2023-01-18 Thread Iain Sandoe via Phabricator via cfe-commits

iains updated this revision to Diff 490383.
iains added a comment.

rebased.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D141908/new/

https://reviews.llvm.org/D141908

Files:
  clang/lib/Sema/SemaDecl.cpp
  clang/test/CXX/module/module.import/p6.cpp


Index: clang/test/CXX/module/module.import/p6.cpp
===
--- clang/test/CXX/module/module.import/p6.cpp
+++ clang/test/CXX/module/module.import/p6.cpp
@@ -28,3 +28,11 @@
 static const int value = 43; 
 };
 
+void deleted_fn_ok (void) = delete;
+
+struct S {
+   ~S() noexcept(false) = default;
+private:
+  S(S&);
+};
+S::S(S&) = default;
Index: clang/lib/Sema/SemaDecl.cpp
===
--- clang/lib/Sema/SemaDecl.cpp
+++ clang/lib/Sema/SemaDecl.cpp
@@ -15254,9 +15254,15 @@
   }
 
   // C++ [module.import/6] external definitions are not permitted in header
-  // units.
+  // units.  Deleted and Defaulted functions are implicitly inline (but the
+  // inline state is not set at this point, so check the BodyKind explicitly).
+  // FIXME: Consider an alternate location for the test where the inlined()
+  // state is complete.
   if (getLangOpts().CPlusPlusModules && currentModuleIsHeaderUnit() &&
-  FD->getFormalLinkage() == Linkage::ExternalLinkage && !FD->isInlined()) {
+  FD->getFormalLinkage() == Linkage::ExternalLinkage &&
+  !FD->isInvalidDecl() && BodyKind != FnBodyKind::Delete &&
+  BodyKind != FnBodyKind::Default && !FD->isInlined()) {
+assert(FD->isThisDeclarationADefinition());
 Diag(FD->getLocation(), diag::err_extern_def_in_header_unit);
 FD->setInvalidDecl();
   }


Index: clang/test/CXX/module/module.import/p6.cpp
===
--- clang/test/CXX/module/module.import/p6.cpp
+++ clang/test/CXX/module/module.import/p6.cpp
@@ -28,3 +28,11 @@
 static const int value = 43; 
 };
 
+void deleted_fn_ok (void) = delete;
+
+struct S {
+   ~S() noexcept(false) = default;
+private:
+  S(S&);
+};
+S::S(S&) = default;
Index: clang/lib/Sema/SemaDecl.cpp
===
--- clang/lib/Sema/SemaDecl.cpp
+++ clang/lib/Sema/SemaDecl.cpp
@@ -15254,9 +15254,15 @@
   }
 
   // C++ [module.import/6] external definitions are not permitted in header
-  // units.
+  // units.  Deleted and Defaulted functions are implicitly inline (but the
+  // inline state is not set at this point, so check the BodyKind explicitly).
+  // FIXME: Consider an alternate location for the test where the inlined()
+  // state is complete.
   if (getLangOpts().CPlusPlusModules && currentModuleIsHeaderUnit() &&
-  FD->getFormalLinkage() == Linkage::ExternalLinkage && !FD->isInlined()) {
+  FD->getFormalLinkage() == Linkage::ExternalLinkage &&
+  !FD->isInvalidDecl() && BodyKind != FnBodyKind::Delete &&
+  BodyKind != FnBodyKind::Default && !FD->isInlined()) {
+assert(FD->isThisDeclarationADefinition());
 Diag(FD->getLocation(), diag::err_extern_def_in_header_unit);
 FD->setInvalidDecl();
   }
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D141811: [clang-format] Allow trailing return types in macros

2023-01-18 Thread Emilia Dreamer via Phabricator via cfe-commits

rymiel planned changes to this revision.
rymiel added a comment.

In D141811#4055485 , 
@HazardyKnusperkeks wrote:

> What about `decltype(auto)`?

Turns out this is a problem even without this patch:

`decltype(auto) a = (b) -> c;`

I'm sure this could somehow be valid c++ syntax, but it does have to be at the 
top scope. I suppose I could fix it in this patch.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D141811/new/

https://reviews.llvm.org/D141811

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

[PATCH] D142046: [BPF][clang] Ignore stack protector options for BPF target

2023-01-18 Thread Yonghong Song via Phabricator via cfe-commits

yonghong-song added a comment.

@compnerd could you also take a look at this patch?

First, some background about this patch. The reason of this patch is due to:

https://lore.kernel.org/bpf/CAOFdcFPnHEc2qd-=c+hdk4ntjjfbhsf4r-g7pdjtrbat6mu...@mail.gmail.com/

Further the following link has details about hardened gentoo and modified clang
default flags:

https://wiki.gentoo.org/wiki/Hardened_Gentoo

Unfortunately, we don't want -fstack-protector for bpf target as bpf kernel
verifier will do stack verification.
This patch tries to ignore -fstack-protector. NVPTX arch has the same need of
ignoring -fstack-protector.
This patch presented a different message than NVPTX. This patch:

clang-16: warning: ignoring '-fstack-protector' option as it is not currently
supported for target 'bpf' [-Woption-ignored]

NVPTX

clang-16: warning: argument unused during compilation: '-fstack-protector'
[-Wunused-command-line-argument]

Does clang community has a preference for the above two formats?

Let us say we do implement this patch and merged into clang, then using gentoo
clang compiler, we will hit
the warning:

clang-16: warning: ignoring '-fstack-protector' option as it is not currently
supported for target 'bpf' [-Woption-ignored]

I suspect gentoo community might complain. I don't think we should add
-Wno-option-ignored since this may cause
unintended issues with compiler command line arguments.
Without -Wno-option-ignored, gentoo community might wants to append
-fno-stack-protector to disable stack-protector,
but with the current implementation, we might actually trigger another warning.

1 2 3 >

1 - 100 of 228 matches

Mail list logo