Re: [PATCH] D14737: Convert some ObjC msgSends to runtime calls

rjmccall added a comment. In http://reviews.llvm.org/D14737#293967, @pete wrote: > Added a couple of tests for retain returning types other than id. Returning > a pointer should still be converted to a call, while returning a non-pointer > such as float will get a message instead. > > I

Re: [PATCH] D14737: Convert some ObjC msgSends to runtime calls

rjmccall added a comment. The casts done by emitARCValueOperation will handle the input, but they don't quite handle the result properly. The right test case here is a method named "retain" that's declared to return something completely unrelated to its receiver type, e.g. @class A;

Re: [PATCH] D14796: Preserve exceptions information during calls code generation.

rjmccall added a comment. What I was thinking was something more along the lines of a little struct that stored either a Decl * or a FunctionType *, and you could change the TargetDecl argument to functions like EmitCall and ConstructAttributeList to that. Maybe call it something like

r253534 - Don't actually add the __unsafe_unretained qualifier in MRC;

Author: rjmccall Date: Wed Nov 18 20:28:03 2015 New Revision: 253534 URL: http://llvm.org/viewvc/llvm-project?rev=253534=rev Log: Don't actually add the __unsafe_unretained qualifier in MRC; driving a canonical difference between that and an unqualified type is a really bad idea when both are

r253533 - Fix the emission of ARC-style ivar layouts in the fragile runtime

Author: rjmccall Date: Wed Nov 18 20:27:55 2015 New Revision: 253533 URL: http://llvm.org/viewvc/llvm-project?rev=253533=rev Log: Fix the emission of ARC-style ivar layouts in the fragile runtime to start at the offset of the first ivar instead of the rounded-up end of the superclass. The latter

Re: [PATCH] D14737: Convert some ObjC msgSends to runtime calls

rjmccall added inline comments. Comment at: include/clang/Basic/ObjCRuntime.h:182 @@ +181,3 @@ +switch (getKind()) { +case FragileMacOSX: return false; +case MacOSX: return getVersion() >= VersionTuple(10, 10); I went ahead and asked Greg, and he

r253255 - Correctly handle type mismatches in the __weak copy/move-initialization

Author: rjmccall Date: Mon Nov 16 16:11:41 2015 New Revision: 253255 URL: http://llvm.org/viewvc/llvm-project?rev=253255=rev Log: Correctly handle type mismatches in the __weak copy/move-initialization peephole I added in r250916. rdar://23559789 Added: cfe/trunk/test/CodeGenObjC/arc-weak.m

r252971 - Remove -Wobjc-weak-compat; there isn't a compelling use case for this.

Author: rjmccall Date: Thu Nov 12 17:39:39 2015 New Revision: 252971 URL: http://llvm.org/viewvc/llvm-project?rev=252971=rev Log: Remove -Wobjc-weak-compat; there isn't a compelling use case for this. Modified: cfe/trunk/include/clang/Basic/DiagnosticSemaKinds.td

Re: [Diffusion] rL246882: Don't crash on a self-alias declaration

rjmccall added a subscriber: rjmccall. rjmccall added a comment. Hmm. That cast to GlobalAlias seems quite brittle anyway; it would be good to fix it (and diagnose the fact that forming the alias failed). But this change seems independently useful. Approved for 3.7. Users: hfinkel

Re: [PATCH] D12614: [OpenMP] Offloading descriptor registration and device codegen.

rjmccall added a comment. In http://reviews.llvm.org/D12614#284158, @sfantao wrote: > As for the structor variants, I am now using the complete variant to generate > the names of the kernels as you suggested. I didn't add any method to CXXABI > as that will require extra logic in ASTContext to

r252187 - After some discussion, promote -fobjc-weak to a driver option.

Author: rjmccall Date: Thu Nov 5 13:19:56 2015 New Revision: 252187 URL: http://llvm.org/viewvc/llvm-project?rev=252187=rev Log: After some discussion, promote -fobjc-weak to a driver option. rdar://problem/23415863 Added: cfe/trunk/test/Driver/objc-weak.m Modified:

Re: r251041 - Define weak and __weak to mean ARC-style weak references, even in MRC.

> On Nov 2, 2015, at 2:53 PM, Nico Weber wrote: > On Fri, Oct 30, 2015 at 5:55 PM, John McCall > wrote: > > On Oct 30, 2015, at 5:39 PM, Nico Weber > > wrote: > > Hi John, >

Re: r251041 - Define weak and __weak to mean ARC-style weak references, even in MRC.

> On Oct 30, 2015, at 5:39 PM, Nico Weber wrote: > Hi John, > > this breaks programs that use __weak and target 10.6, like so: > > $ cat test.m > #import > @interface I : NSObject { > __weak NSObject* foo_; > } > - (NSObject*)foo; > @end > > @implementation I > -

Re: [PATCH] D14149: __builtin_signbit fix for ppcf128 type

rjmccall added a comment. LGTM, thanks! Repository: rL LLVM http://reviews.llvm.org/D14149 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [PATCH] D12793: Three new overflow builtins with generic argument types

rjmccall added a comment. Sorry for the delay. Committed with a few minor tweaks in r251650. http://reviews.llvm.org/D12793 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

r251651 - Add support for __builtin_{add,sub,mul}_overflow.

Author: rjmccall Date: Thu Oct 29 15:48:01 2015 New Revision: 251651 URL: http://llvm.org/viewvc/llvm-project?rev=251651=rev Log: Add support for __builtin_{add,sub,mul}_overflow. Patch by David Grayson! Added: cfe/trunk/test/Sema/builtins-overflow.c Modified:

r251666 - Fix the emission of ARC ivar layouts in the non-fragile Mac runtime.

Author: rjmccall Date: Thu Oct 29 18:36:14 2015 New Revision: 251666 URL: http://llvm.org/viewvc/llvm-project?rev=251666=rev Log: Fix the emission of ARC ivar layouts in the non-fragile Mac runtime. My previous change in this area accidentally broke the rule when InstanceBegin was not a multiple

Re: [PATCH] D12614: [OpenMP] Offloading descriptor registration and device codegen.

rjmccall added a comment. In http://reviews.llvm.org/D12614#274349, @sfantao wrote: > Hi John, > > Thanks for the remark! > > In http://reviews.llvm.org/D12614#272354, @rjmccall wrote: > > > CurFuncDecl is supposed to be the enclosing user function. Things like > > outlined functions should be

Re: [PATCH] D14149: __builtin_signbit fix for ppcf128 type

rjmccall added a comment. Just a few comment suggestions, but functionally LGTM. Comment at: lib/CodeGen/CGBuiltin.cpp:246 @@ -244,1 +245,3 @@ +// On little-Endian, high-double will be in low part of i128. +// Therefore, on big-Endian we shift high part to low part.

r251469 - Add the ability to define "fake" arguments on attributes.

Author: rjmccall Date: Tue Oct 27 19:17:34 2015 New Revision: 251469 URL: http://llvm.org/viewvc/llvm-project?rev=251469=rev Log: Add the ability to define "fake" arguments on attributes. Fake arguments are automatically handled for serialization, cloning, and other representational tasks, but

r251496 - Refine r251469 to give better (and more localizable) diagnostics

Author: rjmccall Date: Wed Oct 28 00:03:19 2015 New Revision: 251496 URL: http://llvm.org/viewvc/llvm-project?rev=251496=rev Log: Refine r251469 to give better (and more localizable) diagnostics for all the reasons that ARC makes things implicitly unavailable. Modified:

Re: [PATCH] D13336: [MSVC] 'property' with an empty array in array subscript expression.

rjmccall added inline comments. Comment at: include/clang/AST/ExprCXX.h:689 @@ +688,3 @@ +/// indices. In this case, i=p->x[a][b] will be turned into i=p->GetX(a, b), and +/// p->x[a][b] = i will be turned into p->PutX(a, b, i). +class MSPropertySubscriptExpr : public Expr {

r251384 - Be more conservative about diagnosing "incorrect" uses of __weak:

Author: rjmccall Date: Mon Oct 26 23:54:50 2015 New Revision: 251384 URL: http://llvm.org/viewvc/llvm-project?rev=251384=rev Log: Be more conservative about diagnosing "incorrect" uses of __weak: allow them to be written in certain kinds of user declaration and diagnose on the use-site instead.

r251041 - Define weak and __weak to mean ARC-style weak references, even in MRC.

Author: rjmccall Date: Thu Oct 22 13:38:17 2015 New Revision: 251041 URL: http://llvm.org/viewvc/llvm-project?rev=251041=rev Log: Define weak and __weak to mean ARC-style weak references, even in MRC. Previously, __weak was silently accepted and ignored in MRC mode. That makes this a potentially

Re: [PATCH] D13959: Fix crash in EmitDeclMetadata mode

rjmccall added a comment. LGTM. Repository: rL LLVM http://reviews.llvm.org/D13959 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

r250918 - Unify the ObjC entrypoint caches.

Author: rjmccall Date: Wed Oct 21 13:06:43 2015 New Revision: 250918 URL: http://llvm.org/viewvc/llvm-project?rev=250918=rev Log: Unify the ObjC entrypoint caches. Modified: cfe/trunk/lib/CodeGen/CGCall.cpp cfe/trunk/lib/CodeGen/CGObjC.cpp cfe/trunk/lib/CodeGen/CodeGenFunction.h

r250919 - Fix and stylize the emission of GC/ARC ivar and GC block layout strings.

Author: rjmccall Date: Wed Oct 21 13:06:47 2015 New Revision: 250919 URL: http://llvm.org/viewvc/llvm-project?rev=250919=rev Log: Fix and stylize the emission of GC/ARC ivar and GC block layout strings. Specifically, handle under-aligned object references (by explicitly ignoring them, because

r250916 - In ARC, peephole the initialization of a __weak variable with

Author: rjmccall Date: Wed Oct 21 13:06:31 2015 New Revision: 250916 URL: http://llvm.org/viewvc/llvm-project?rev=250916=rev Log: In ARC, peephole the initialization of a __weak variable with a value loaded from a __weak variable into a call to objc_copyWeak or objc_moveWeak. Modified:

Re: [PATCH] D12614: [OpenMP] Offloading descriptor registration and device codegen.

rjmccall added a comment. CurFuncDecl is supposed to be the enclosing user function. Things like outlined functions should be getting stored in CurCodeDecl; that's how it's done for blocks and lambdas. http://reviews.llvm.org/D12614 ___

Re: [PATCH] D13954: CodeGen: Fix LLVM assertion if Swift and Clang emit Objective-C class reference in same LLVM module

rjmccall added a comment. It just occurred to me that there is a way to test this in Clang with the asm-label extension: int Foo_class asm("OBJC_CLASS_$_Foo"); Of course, you'll have to actually use it from somewhere, or define it, in order for it to actually show up in the IR and cause a

r250955 - Enable ARC on the fragile runtime.

Author: rjmccall Date: Wed Oct 21 17:06:03 2015 New Revision: 250955 URL: http://llvm.org/viewvc/llvm-project?rev=250955=rev Log: Enable ARC on the fragile runtime. This is almost entirely a matter of just flipping a switch. 99% of the runtime support is available all the way back to when it

Re: [PATCH] D13336: [MSVC] 'property' with an empty array in array subscript expression.

rjmccall added a comment. Needs more tests. 1. Dependent template instantiation. 2. Non-dependent template instantiation. 3. Indexes which are themselves pseudo-objects. 4. Templated getters and setters. 5. Non-POD by-value argument types. Comment at:

Re: [PATCH] D13336: [MSVC] 'property' with an empty array in array subscript expression.

rjmccall added a comment. I agree with Reid that you should not be adding a DenseMap to Sema for this. Just build a SubscriptExpr for the syntactic form and have it yield an expression of pseudo-object type; or you can make your own AST node for it if that makes things easier.

Re: [PATCH] D13582: [DEBUG INFO] Emit debug info for type used in explicit cast only.

rjmccall added a comment. That looks great, thank you. http://reviews.llvm.org/D13582 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [PATCH] D6700: Diagnose UnresolvedLookupExprs that resolve to instance members in static methods

rjmccall added inline comments. Comment at: lib/Sema/TreeTransform.h:9135 @@ +9134,3 @@ +if (NamedDecl *D = R.getAsSingle()) { + D = D->getUnderlyingDecl(); + if (isa(D) || isa(D) || getAsSingle already looks through to the underlying decl.

Re: [PATCH] D6700: Diagnose UnresolvedLookupExprs that resolve to instance members in static methods

rjmccall added a comment. As a more general comment, I believe the rule is that we try to always make a MemberExpr/UnresolvedMemberExpr whenever there *might* be a base, but that the resulting distinction between an implicit-base UnresolvedMemberExpr and an UnresolvedLookupExpr is not actually

Re: [PATCH] D13285: Fix for bug 24196: clang fails on assertion on complex doubles multiplication when EH is enabled

rjmccall added a comment. This is an inappropriate fix for this problem. If these runtime functions can never throw, which seems to be the case, you should create the function type with a no-throw exception specification, which will make EmitCall emit the call with a CallInst.

Re: [PATCH] D13582: [DEBUG INFO] Emit debug info for type used in explicit cast only.

> On Oct 12, 2015, at 10:50 AM, David Blaikie wrote: > +John, author of the original patch - in case it's an obvious mistake. I > haven't looked at John's change yet & compared it to the intended behavior, > etc. Will do so soon and/or if John doesn't see something at a

Re: [PATCH] D13325: Fix crash in codegen on casting to `bool &`.

rjmccall added a comment. LGTM. http://reviews.llvm.org/D13325 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

r248862 - Don't crash when a reserved global placement operator new is paired

Author: rjmccall Date: Tue Sep 29 18:55:17 2015 New Revision: 248862 URL: http://llvm.org/viewvc/llvm-project?rev=248862=rev Log: Don't crash when a reserved global placement operator new is paired with a non-reserved operator delete in a new-expression. Modified:

r248775 - Honor the casted-to alignment of an explicit cast even when

Author: rjmccall Date: Mon Sep 28 23:37:40 2015 New Revision: 248775 URL: http://llvm.org/viewvc/llvm-project?rev=248775=rev Log: Honor the casted-to alignment of an explicit cast even when Sema thinks the cast is a no-op, as it does when (e.g.) the only thing that changes is an alignment

Re: [PATCH] D12793: Three new overflow builtins with generic argument types

rjmccall added a comment. LGTM, thanks. http://reviews.llvm.org/D12793 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [PATCH] D12793: Three new security overflow builtins with generic argument types

rjmccall added a comment. X and Y aren't unreasonable for the operands, although you could also use "left" and "right" (or LHS/RHS), especially since it's significant for subtraction. R is short for "result" and should be spelled out. E is presumably short for "encompassing", but that is not

Re: [PATCH] D12793: Three new security overflow builtins with generic argument types

rjmccall added a comment. Yes, the main result is defined to be the true value mod 2^k even when overflow occurs. We do use the intrinsics to implement the existing fixed-width GCC builtins, which are meant to be useful not just for security checking, but for implementing arbitrary-precision

r248436 - Forbid qualifiers on ObjC generic parameters and arguments, but

Author: rjmccall Date: Wed Sep 23 17:14:21 2015 New Revision: 248436 URL: http://llvm.org/viewvc/llvm-project?rev=248436=rev Log: Forbid qualifiers on ObjC generic parameters and arguments, but silently ignore them on arguments when they're provided indirectly (.e.g behind a template argument or

Re: [PATCH] D10881: [Sema] Catch a case when 'volatile' qualifier is dropped while binding

rjmccall added a comment. This is outside of my expertise, but I've asked Doug Gregor to take a look. http://reviews.llvm.org/D10881 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [PATCH] D12793: Three new security overflow builtins with generic argument types

rjmccall added a comment. Thanks for doing this; this is a great start. Comment at: docs/LanguageExtensions.rst:1720 @@ -1712,1 +1719,3 @@ +being stored there, and the function returns 1. The behavior of these builtins +is well-defined for all argument values.

r247597 - Fix a nasty bug with the partial destruction of nested arrays;

Author: rjmccall Date: Mon Sep 14 13:57:08 2015 New Revision: 247597 URL: http://llvm.org/viewvc/llvm-project?rev=247597=rev Log: Fix a nasty bug with the partial destruction of nested arrays; it escaped notice because it's only used for heterogeneous initialization. rdar://21397946 Modified:

Re: [PATCH] D12743: [CodeGen] Teach SimplifyPersonality about the updated LandingPadInst

rjmccall added a comment. In http://reviews.llvm.org/D12743#244375, @vsk wrote: > Ah, ok. We have some objective-c++ code which calls into a boost routine > which throws an exception. That results in an undefined reference to > ___objc_personality_v0, because the boost library we linked

r247482 - When comparing two block captures for layout, don't crash

Author: rjmccall Date: Fri Sep 11 17:00:51 2015 New Revision: 247482 URL: http://llvm.org/viewvc/llvm-project?rev=247482=rev Log: When comparing two block captures for layout, don't crash if they have the same alignment and one was 'this'. Fixes PR24780. Modified:

Re: [PATCH] D11361: [OpenMP] Target directive host codegen

rjmccall added a comment. Sorry for putting off the final review on this; I was heads-down trying to get the alignment patch done. It's looking good; obviously you'll need to update it to work with Addresses properly, but hopefully that won't be too much of a problem. When you do, maybe you

Re: [PATCH] D12743: [CodeGen] Teach SimplifyPersonality about the updated LandingPadInst

rjmccall added a comment. LGTM, thanks. http://reviews.llvm.org/D12743 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

r247207 - Fix access control for lookups using the Microsoft __super extension.

Author: rjmccall Date: Wed Sep 9 18:04:17 2015 New Revision: 247207 URL: http://llvm.org/viewvc/llvm-project?rev=247207=rev Log: Fix access control for lookups using the Microsoft __super extension. rdar://22464808 Added: cfe/trunk/test/SemaCXX/microsoft-super.cpp Modified:

r247209 - ARC: Fix the precise-lifetime suppression of returns_inner_pointer

Author: rjmccall Date: Wed Sep 9 18:37:17 2015 New Revision: 247209 URL: http://llvm.org/viewvc/llvm-project?rev=247209=rev Log: ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 Modified:

r247228 - Don't crash when emitting a block under returns_nonnull.

Author: rjmccall Date: Wed Sep 9 19:57:46 2015 New Revision: 247228 URL: http://llvm.org/viewvc/llvm-project?rev=247228=rev Log: Don't crash when emitting a block under returns_nonnull. rdar://22071955 Added: cfe/trunk/test/CodeGen/sanitize-blocks.c Modified:

r246991 - When building the alloca for a local variable, set its name

Author: rjmccall Date: Tue Sep 8 04:18:30 2015 New Revision: 246991 URL: http://llvm.org/viewvc/llvm-project?rev=246991=rev Log: When building the alloca for a local variable, set its name separately from building the instruction so that it's preserved even in -Asserts builds. Employ C++'s

r246988 - Remove unnecessary braces; this resolves against a

Author: rjmccall Date: Tue Sep 8 03:57:00 2015 New Revision: 246988 URL: http://llvm.org/viewvc/llvm-project?rev=246988=rev Log: Remove unnecessary braces; this resolves against a single-pointer overload instead of the ArrayRef one. Modified: cfe/trunk/lib/CodeGen/CGBuilder.h Modified:

Re: r246991 - When building the alloca for a local variable, set its name

> On Sep 8, 2015, at 10:52 AM, Chandler Carruth wrote: > None of my performance concerns were relevant to this change FWIW. > > I think the reason that this got "fixed" was because people had a tendancy to > *rely* on the name downstream when we made it always present. =/

Re: r246985 - Compute and preserve alignment more faithfully in IR-generation.

> On Sep 8, 2015, at 6:19 PM, Steven Wu wrote: > CreateElementBitcast doesn’t seem to do what the name suggested. If you give > it VTy, it doesn’t grab the element type to generate bitcast. Is this > by-design? If so, I need to do this: > PtrOp0 =

Re: r246991 - When building the alloca for a local variable, set its name

> On Sep 8, 2015, at 8:25 AM, David Blaikie <dblai...@gmail.com> wrote: > On Tue, Sep 8, 2015 at 2:18 AM, John McCall via cfe-commits > <cfe-commits@lists.llvm.org <mailto:cfe-commits@lists.llvm.org>> wrote: > Author: rjmccall > Date: Tue Sep 8 04:18:30 2015 >

Re: [PATCH] D11297: PR17829: Functions declared extern C with a name matching a mangled C++ function are allowed

rjmccall added a comment. Yes, please make it an error. And the obvious test changes are fine. :) http://reviews.llvm.org/D11297 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [PATCH] D12389: Conditionalize X86 Darwin MaxVectorAlign on the presence of AVX.

rjmccall added a comment. Okay, and you're doing the avx512 part of this in a separate commit? LGTM. http://reviews.llvm.org/D12389 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [PATCH] D12390: Also enable the avx/avx512 ABIs for i386, not just x86_64.

rjmccall added a comment. This gives no-MMX priority over turning on SSE, which sounds like a major change in behavior to me. There are definitely clients that specifically never want to use MMX but do care deeply about SSE; my understanding is that modern microarchitectures heavily punish

Re: [PATCH] D11297: PR17829: Functions declared extern C with a name matching a mangled C++ function are allowed

rjmccall added inline comments. Comment at: lib/CodeGen/CodeGenModule.h:354 @@ +353,3 @@ + /// call). + llvm::DenseSetGlobalDecl ExplicitDefinitions; + andreybokhanko wrote: Checking that a GlobalDecl is not in ExplicitDefinitions yet is actually required to

Re: [PATCH] D12390: Also enable the avx/avx512 ABIs for i386, not just x86_64.

rjmccall added a comment. In http://reviews.llvm.org/D12390#234458, @ab wrote: Unless I'm misunderstanding, I believe this has much less impact than you're thinking; there are three cases: - x86_64: no change (-mno-mmx is guarded by x86) - x86, with -mno-mmx: no change (because previously,

Re: [PATCH] D12312: Emiting invariant.group.barrier and adding -fstrict-vptrs

rjmccall added inline comments. Comment at: include/clang/Driver/Options.td:990 @@ -988,2 +989,3 @@ value range; +def fstrict_vptrs: Flag[-], fstrict-vptrs, Groupf_Group, Flags[CC1Option]; def fstrict_overflow : Flag[-], fstrict-overflow, Groupf_Group;

Re: [PATCH] D12312: Emiting invariant.group.barrier and adding -fstrict-vptrs

rjmccall added inline comments. Comment at: lib/CodeGen/CGClass.cpp:1279 @@ +1278,3 @@ + if (CGM.getCodeGenOpts().StrictVPtrs BaseVPtrsInitialized) +CXXThisValue = Builder.CreateInvariantGroupBarrier(LoadCXXThis()); + Prazek wrote: rjmccall wrote: Prazek

Re: [PATCH] D12312: Emiting invariant.group.barrier and adding -fstrict-vptrs

rjmccall added a comment. You should add a test that actually checks that your feature works. Comment at: lib/CodeGen/CGClass.cpp:1279 @@ +1278,3 @@ + if (CGM.getCodeGenOpts().StrictVPtrs BaseVPtrsInitialized) +CXXThisValue =

Re: [PATCH] D11297: PR17829: Functions declared extern C with a name matching a mangled C++ function are allowed

rjmccall added a comment. This looks generally like what I'm looking for, thanks! Some comments. Comment at: lib/CodeGen/CodeGenModule.cpp:1129 @@ +1128,3 @@ +if (GV GV != GetGlobalValue(getMangledName(D))) + continue; + This is a pretty expensive

r245771 - When building a pseudo-object assignment, and the RHS is

Author: rjmccall Date: Fri Aug 21 19:35:27 2015 New Revision: 245771 URL: http://llvm.org/viewvc/llvm-project?rev=245771view=rev Log: When building a pseudo-object assignment, and the RHS is a contextually-typed expression that semantic analysis will probably need to invasively rewrite, don't

Re: [PATCH] D12128: Generating available_externally vtables bugfix

rjmccall added a comment. LGTM. http://reviews.llvm.org/D12128 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

r245514 - Fix the layout of bitfields in ms_struct unions: their

Author: rjmccall Date: Wed Aug 19 17:42:36 2015 New Revision: 245514 URL: http://llvm.org/viewvc/llvm-project?rev=245514view=rev Log: Fix the layout of bitfields in ms_struct unions: their alignment is ignored, and they always allocate a complete storage unit. Also, change the dumping of AST

Re: [PATCH] D11859: Generating vptr assume loads

rjmccall added a comment. Just a couple tweaks and then LGTM. Comment at: lib/CodeGen/CGClass.cpp:1833 @@ +1832,3 @@ + // unless we are calling base constructor - we don't want to generating + // assumption loads for not completed because vptr may still change. + if

Re: [PATCH] D11859: Generating vptr assume loads

rjmccall added inline comments. Comment at: lib/CodeGen/CGClass.cpp:1862 @@ +1861,3 @@ + for (const VPtr Vptr : getVTablePointers(ClassDecl)) +if (CGM.getCXXABI().requiresVPtrInitialization(Vptr)) + EmitVTableAssumptionLoad(Vptr, This); No, it only

Re: [PATCH] D11859: Generating vptr assume loads

rjmccall added a comment. In http://reviews.llvm.org/D11859#225384, @Prazek wrote: In http://reviews.llvm.org/D11859#225025, @rjmccall wrote: Mostly LGTM. Are you going to emit assumptions for vbptrs in a separate patch? I wasn't planning to. I am focusing now on upgrading GVN for

Re: [PATCH] D11859: Generating vptr assume loads

rjmccall added a comment. Mostly LGTM. Are you going to emit assumptions for vbptrs in a separate patch? Comment at: lib/CodeGen/CGCXXABI.h:349-357 @@ -348,1 +348,11 @@ + /// Checks if ABI requires extra virtual offset for vtable field. + virtual bool +

Re: [PATCH] D11297: PR17829: Functions declared extern C with a name matching a mangled C++ function are allowed

rjmccall added a comment. In http://reviews.llvm.org/D11297#223622, @andreybokhanko wrote: John, Thank you for the quick reply! Let me make sure I understand what you said, using my test as an example (BTW, sorry if this is a dumb question -- I asked our local Clang experts, but no-one

Re: [PATCH] D11297: PR17829: Functions declared extern C with a name matching a mangled C++ function are allowed

rjmccall added a comment. You only have one attempt to define the function here; I don't see the problem. Recall that I said to add a flag to getOrCreateLLVMFunction that says whether the caller intends to define the function. The rule should be that only callers that pass true should be

<    2   3   4   5   6   7