[PATCH] D42642: [CUDA] Detect installation in PATH

Hahnfeld updated this revision to Diff 131963. Hahnfeld added a comment. Check for `libdevice` in candidates from `PATH`. https://reviews.llvm.org/D42642 Files: include/clang/Driver/Options.td lib/Driver/ToolChains/Cuda.cpp test/Driver/Inputs/CUDA-nolibdevice/usr/local/cuda/bin/ptxas te

[PATCH] D42642: [CUDA] Detect installation in PATH

Hahnfeld added inline comments. Comment at: lib/Driver/ToolChains/Cuda.cpp:206 // -nocudalib hasn't been specified. -if (LibDeviceMap.empty() && !Args.hasArg(options::OPT_nocudalib)) +if (CheckLibDevice && LibDeviceMap.empty()) continue; tra w

[PATCH] D42642: [CUDA] Detect installation in PATH

Hahnfeld added a comment. In https://reviews.llvm.org/D42642#992137, @tra wrote: > I've thought a bit more about this and there's another quirk -- symlinks. > > What if we've found /usr/bin/ptxas and is a symlink pointing to the real > ptxas in the CUDA installation? If we add /usr to the list o

[PATCH] D42642: [CUDA] Detect installation in PATH

Hahnfeld updated this revision to Diff 132029. Hahnfeld added a comment. Follow symlinked `ptxas` executables. https://reviews.llvm.org/D42642 Files: include/clang/Driver/Options.td lib/Driver/ToolChains/Cuda.cpp test/Driver/Inputs/CUDA-nolibdevice/usr/local/cuda/bin/ptxas test/Driver/I

[PATCH] D42642: [CUDA] Detect installation in PATH

Hahnfeld marked 5 inline comments as done. Hahnfeld added a subscriber: sylvestre.ledru. Hahnfeld added inline comments. Comment at: lib/Driver/ToolChains/Cuda.cpp:96-105 + if (llvm::ErrorOr ptxas = + llvm::sys::findProgramByName("ptxas")) { +SmallString

[PATCH] D42642: [CUDA] Detect installation in PATH

Hahnfeld updated this revision to Diff 132103. Hahnfeld marked 3 inline comments as done. Hahnfeld added a comment. Add some `sysroot` arguments to new test to make sure it doesn't accidentally find CUDA installations in the system. https://reviews.llvm.org/D42642 Files: include/clang/Driver

[PATCH] D42642: [CUDA] Detect installation in PATH

This revision was automatically updated to reflect the committed changes. Closed by commit rL323848: [CUDA] Detect installation in PATH (authored by Hahnfeld, committed by ). Herald added a subscriber: llvm-commits. Changed prior to commit: https://reviews.llvm.org/D42642?vs=132103&id=132104#to

[PATCH] D42755: [libcxx] Fix last_write_time tests for filesystems that don't support very small times.

Hahnfeld requested changes to this revision. Hahnfeld added a comment. This revision now requires changes to proceed. Can you please add a summary that describes which filesystem this problem can be seen with? I think outside people can't access rdar tickets, so I can only guess that it's relate

[PATCH] D42800: Let CUDA toolchain support amdgpu target

Hahnfeld added a comment. Only commenting on parts that I'm a bit familiar with. In general, does it make sense to split this patch, are there different "stages" of support? Like 1) being able to compile an empty file, 2) generate optimized code, 3) allow using math functions? ==

[PATCH] D42755: [libcxx] Fix last_write_time tests for filesystems that don't support very small times.

Hahnfeld resigned from this revision. Hahnfeld added a comment. In https://reviews.llvm.org/D42755#995593, @vsapsai wrote: > Good suggestion for explaining the problem with `file_time_type::min()` in > more detail. Also PR35990 has > some details on

[PATCH] D42840: [docs] Fix duplicate arguments for JoinedAndSeparate

Hahnfeld created this revision. Hahnfeld added reviewers: rsmith, craig.topper. Herald added a subscriber: cfe-commits. We can't see how many arguments are in the meta var name, so just assume that is the right number. Repository: rC Clang https://reviews.llvm.org/D42840 Files: utils/Table

[PATCH] D42841: [docs] Improve help for OpenMP options

Hahnfeld created this revision. Hahnfeld added reviewers: ABataev, gtbercea. Herald added subscribers: cfe-commits, guansong. - Add HelpText for -fopenmp so that it appears in clang --help. - Hide -fopenmp-relocatable-target from there. - Hide -fno-openmp-simd, other options only list the positive

[PATCH] D42841: [docs] Improve help for OpenMP options

Hahnfeld updated this revision to Diff 132571. Hahnfeld edited the summary of this revision. https://reviews.llvm.org/D42841 Files: include/clang/Driver/Options.td Index: include/clang/Driver/Options.td === --- include/clang/Driv

[PATCH] D42920: [CUDA] Fix test cuda-external-tools.cu

Hahnfeld created this revision. Hahnfeld added reviewers: jlebar, tra. Herald added a subscriber: cfe-commits. Hahnfeld added a dependent revision: D42921: [CUDA] Add option to generate relocatable device code. This didn't verify the CHECK prefix before! Repository: rC Clang https://reviews.

[PATCH] D42922: [CUDA] Register relocatable GPU binaries

Hahnfeld created this revision. Hahnfeld added reviewers: jlebar, tra, hfinkel. Herald added a subscriber: cfe-commits. nvcc generates a unique registration function for each object file that contains relocatable device code. Unique names are achieved with a module id that is also reflected in the

[PATCH] D42921: [CUDA] Add option to generate relocatable device code

Hahnfeld created this revision. Hahnfeld added reviewers: jlebar, tra. Herald added a subscriber: cfe-commits. Hahnfeld added a dependency: D42920: [CUDA] Fix test cuda-external-tools.cu. Hahnfeld added a reviewer: hfinkel. As a first step, pass '-c/--compile-only' to ptxas so that it doesn't comp

[PATCH] D42923: [CUDA] Allow external variables in separate compilation

Hahnfeld created this revision. Hahnfeld added reviewers: jlebar, tra, hfinkel. Herald added a subscriber: cfe-commits. According to the CUDA Programming Guide this is prohibited in whole program compilation mode. This makes sense because external references cannot be satisfied in that mode anyway

[PATCH] D42922: [CUDA] Register relocatable GPU binaries

Hahnfeld planned changes to this revision. Hahnfeld added a comment. I didn't write tests for this yet, but I wanted to get some early feedback on this and show what I have in mind. Comment at: lib/CodeGen/CGCUDANV.cpp:330-331 // the GPU side. for (const std::string &Gpu

[PATCH] D30015: [OpenMP] Add arch-specific directory to search path

Hahnfeld added a comment. Why is this only for OpenMP? I imagine this should be done for **all** runtime libraries https://reviews.llvm.org/D30015 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinf

[PATCH] D30015: [OpenMP] Add arch-specific directory to search path

Hahnfeld added a subscriber: mgorny. Hahnfeld added a comment. Yes, that's the current state. We had some discussion on cfe-dev on that matter: http://lists.llvm.org/pipermail/cfe-dev/2017-January/052512.html Could you slightly adapt your patch so that other runtime libraries can follow this id

[PATCH] D30015: [OpenMP] Add arch-specific directory to search path

Hahnfeld added a comment. I still think this should not be done for OpenMP only and hence move out of `addOpenMPRuntime`. (Why the heck are there //two// places to add `-lomp`?!? I'll clean that up later...) https://reviews.llvm.org/D30015 ___ cf

[PATCH] D30015: [OpenMP] Add arch-specific directory to search path

Hahnfeld added a comment. In https://reviews.llvm.org/D30015#678294, @pirama wrote: > I am fine to do this more generally, but not sure of the scope. Should this > be always added or only if a runtime (sanitizer or openmp or xray) is > requested? I think always. From my point of view, this w

[PATCH] D30015: [OpenMP] Add arch-specific directory to search path

Hahnfeld added a comment. Maybe something like the following (without tests): diff --git a/include/clang/Driver/ToolChain.h b/include/clang/Driver/ToolChain.h index ffb0d60..0f3507d 100644 --- a/include/clang/Driver/ToolChain.h +++ b/include/clang/Driver/ToolChain.h @@ -299,6 +299,11 @

[PATCH] D30015: [OpenMP] Add arch-specific directory to search path

Hahnfeld added a comment. Please adapt the title and summary for the more general changes this has evolved to. Comment at: lib/Driver/Tools.cpp:284 +// If we are not cross-compling, add '-rpath' with architecture-specific +// library path so libraries lined from this p

[PATCH] D30015: [OpenMP] Add arch-specific directory to search path

Hahnfeld added inline comments. Comment at: test/Driver/arch-specific-libdir-rpath.c:6 +// -rpath only gets added during native compilation +// REQUIRES: native +// mgorny wrote: > Hahnfeld wrote: > > pirama wrote: > > > pirama wrote: > > > > I feel this test is

[PATCH] D30087: [Driver] Unify linking of OpenMP runtime

Hahnfeld created this revision. Herald added a subscriber: emaste. This enables libomptarget for Darwin, FreeBSD and NetBSD, NFCI otherwise. While at it, extend test for FreeBSD and check for -lrt iff on Linux. https://reviews.llvm.org/D30087 Files: lib/Driver/Tools.cpp test/Driver/fopenmp.

[PATCH] D30088: [OpenMP] Remove barriers at cancel and cancellation point

Hahnfeld created this revision. This resolves a deadlock with the cancel directive when there is no explicit cancellation point. In that case, the implicit barrier acts as cancellation point. After removing the barrier after cancel, the now unmatched barrier for the explicit cancellation point has

[PATCH] D30091: [OpenMP] Fix cancellation point in task with no cancel

Hahnfeld created this revision. With tasks, the cancel may happen in another task. This has a different region info which means that we can't find it here. https://reviews.llvm.org/D30091 Files: lib/CodeGen/CGOpenMPRuntime.cpp test/OpenMP/cancellation_point_codegen.cpp Index: test/OpenMP/

[PATCH] D30087: [Driver] Unify linking of OpenMP runtime

Hahnfeld added inline comments. Comment at: lib/Driver/Tools.cpp:8683 if (!Args.hasArg(options::OPT_nostdlib, options::OPT_nodefaultlibs)) -addOpenMPRuntime(CmdArgs, getToolChain(), Args); +addOpenMPRuntime(CmdArgs, getToolChain(), Args, JA); pirama

[PATCH] D30088: [OpenMP] Remove barriers at cancel and cancellation point

This revision was automatically updated to reflect the committed changes. Closed by commit rL295473: [OpenMP] Remove barriers at cancel and cancellation point (authored by Hahnfeld). Changed prior to commit: https://reviews.llvm.org/D30088?vs=88872&id=88921#toc Repository: rL LLVM https://r

[PATCH] D30091: [OpenMP] Fix cancellation point in task with no cancel

This revision was automatically updated to reflect the committed changes. Closed by commit rL295474: [OpenMP] Fix cancellation point in task with no cancel (authored by Hahnfeld). Changed prior to commit: https://reviews.llvm.org/D30091?vs=88878&id=88922#toc Repository: rL LLVM https://revi

[PATCH] D30135: [OpenMP] Generate better diagnostics for cancel and cancellation point

Hahnfeld created this revision. Handle errors related to a specific directive before checking the nesting: The specific checks may validate required arguments and give more helpful messages, especially when the nesting depends on those arguments. This change requires some minor adaptions to the t

[PATCH] D30135: [OpenMP] Generate better diagnostics for cancel and cancellation point

Hahnfeld added a comment. In https://reviews.llvm.org/D30135#681354, @ABataev wrote: > Not sure that this is better because at first, we need to be sure that this > nesting is allowed. Why do we need to perform some additional analysis if > nesting is not allowed at all? `CheckNestingOfRegion

[PATCH] D30135: [OpenMP] Generate better diagnostics for cancel and cancellation point

Hahnfeld updated this revision to Diff 89099. Hahnfeld edited the summary of this revision. Hahnfeld added a comment. new static function `CheckCancelRegion` https://reviews.llvm.org/D30135 Files: lib/Sema/SemaOpenMP.cpp test/OpenMP/cancel_messages.cpp test/OpenMP/cancellation_point_messa

[PATCH] D30135: [OpenMP] Generate better diagnostics for cancel and cancellation point

Hahnfeld added inline comments. Comment at: lib/Sema/SemaOpenMP.cpp:1959 +static bool CheckCancelRegion(Sema &SemaRef, OpenMPDirectiveKind CurrentRegion, + OpenMPDirectiveKind CancelRegion, ABataev wrote: > Should be `checkCancelReg

[PATCH] D30135: [OpenMP] Generate better diagnostics for cancel and cancellation point

Hahnfeld added inline comments. Comment at: lib/Sema/SemaOpenMP.cpp:1968-1973 + if (CancelRegion != OMPD_parallel && CancelRegion != OMPD_for && + CancelRegion != OMPD_sections && CancelRegion != OMPD_taskgroup) { +SemaRef.Diag(StartLoc, diag::err_omp_wrong_cancel_regio

[PATCH] D30135: [OpenMP] Generate better diagnostics for cancel and cancellation point

Hahnfeld updated this revision to Diff 89106. Hahnfeld marked 3 inline comments as done. Hahnfeld edited the summary of this revision. Hahnfeld added a comment. Address review comment's and apply new naming style to checkNestingOfRegions https://reviews.llvm.org/D30135 Files: lib/Sema/SemaOpe

[PATCH] D30087: [Driver] Unify linking of OpenMP runtime

Hahnfeld updated this revision to Diff 89110. Hahnfeld added a comment. rebase https://reviews.llvm.org/D30087 Files: lib/Driver/Tools.cpp test/Driver/fopenmp.c Index: test/Driver/fopenmp.c === --- test/Driver/fopenmp.c +++ te

[PATCH] D30214: [Driver] Search for libc++ headers in ResourceDir

Hahnfeld created this revision. https://reviews.llvm.org/D30214 Files: lib/Driver/ToolChains.cpp Index: lib/Driver/ToolChains.cpp === --- lib/Driver/ToolChains.cpp +++ lib/Driver/ToolChains.cpp @@ -4642,6 +4642,7 @@ std::string

[PATCH] D30135: [OpenMP] Generate better diagnostics for cancel and cancellation point

This revision was automatically updated to reflect the committed changes. Closed by commit rL295808: [OpenMP] Generate better diagnostics for cancel and cancellation point (authored by Hahnfeld). Changed prior to commit: https://reviews.llvm.org/D30135?vs=89106&id=89326#toc Repository: rL LL

[PATCH] D30087: [Driver] Unify linking of OpenMP runtime

Hahnfeld added a comment. Ping https://reviews.llvm.org/D30087 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D30214: [Driver] Search for libc++ headers in ResourceDir

Hahnfeld added a comment. Ping https://reviews.llvm.org/D30214 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D30214: [Driver] Search for libc++ headers in ResourceDir

Hahnfeld added a comment. In https://reviews.llvm.org/D30214#690010, @jroelofs wrote: > libc++ headers should not be installed in the resource dir. They are currently not by default. But with https://reviews.llvm.org/D30015 for runtime libraries, this may become a place for headers packaged wi

[PATCH] D30087: [Driver] Unify linking of OpenMP runtime

Hahnfeld updated this revision to Diff 90295. Hahnfeld marked an inline comment as done. Hahnfeld added a comment. Reword comment https://reviews.llvm.org/D30087 Files: lib/Driver/Tools.cpp test/Driver/fopenmp.c Index: test/Driver/fopenmp.c =

[PATCH] D30214: [Driver] Search for libc++ headers in ResourceDir

Hahnfeld updated this revision to Diff 90296. Hahnfeld edited the summary of this revision. Hahnfeld added a comment. Adapt and add tests. https://reviews.llvm.org/D30214 Files: lib/Driver/ToolChains.cpp test/Driver/Inputs/resource_dir_libcxx/include/c++/v1/.keep test/Driver/Inputs/resour

[PATCH] D30214: [Driver] Search for libc++ headers in ResourceDir

Hahnfeld abandoned this revision. Hahnfeld added a comment. http://lists.llvm.org/pipermail/cfe-dev/2017-January/052512.html suggested to install runtime libraries (like libomp, but I think libc++ in certain cases may fall under the same category) to ResourceDir and I wanted the headers to be c

[PATCH] D30087: [Driver] Unify linking of OpenMP runtime

Hahnfeld updated this revision to Diff 90987. Hahnfeld added a comment. Rebase for recent refactoring and ping. https://reviews.llvm.org/D30087 Files: lib/Driver/ToolChains/CommonArgs.cpp lib/Driver/ToolChains/CommonArgs.h lib/Driver/ToolChains/Darwin.cpp lib/Driver/ToolChains/FreeBSD.c

[PATCH] D30700: [Driver] Always add arch-specific-subdir to -rpath

Hahnfeld added a comment. No build system will ever set `-frtlib-add-rpath` to enable this "feature". I'm for keeping this opt-out until we have configuration files to set this by default. Making it opt-in would weaken its main reason of existence: Not to break simple binaries for the user, and

[PATCH] D30733: [Driver] Add arch-specific rpath for libc++

Hahnfeld created this revision. Herald added a reviewer: EricWF. libc++ may be installed there as a runtime library. https://reviews.llvm.org/D30733 Files: lib/Driver/ToolChain.cpp Index: lib/Driver/ToolChain.cpp === --- lib/Dr

[PATCH] D30733: [Driver] Add arch-specific rpath for libc++

Hahnfeld marked an inline comment as done. Hahnfeld added a comment. In https://reviews.llvm.org/D30733#697108, @jroelofs wrote: > As I said on https://reviews.llvm.org/D30214, it is inappropriate to be > installing libc++ in the resource directory... please **do not** do that. Can you give re

[PATCH] D25263: [Driver] Allow setting the default linker during build

Hahnfeld added inline comments. Comment at: lib/Driver/ToolChain.cpp:721-724 + +const char *ToolChain::getDefaultLinker() const { + return CLANG_DEFAULT_LINKER; +} sfertile wrote: > Hahnfeld wrote: > > I think this could go into the header > The CLANG_DEFAULT_LI

[PATCH] D26244: [Driver] Prefer libraries installed next to Clang over those from GCC

Hahnfeld added a comment. Ping! https://reviews.llvm.org/D26244 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D25669: [Driver] Simplify ToolChain::GetCXXStdlibType (NFC)

Hahnfeld added a comment. Ping! https://reviews.llvm.org/D25669 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D25263: [Driver] Allow setting the default linker during build

Hahnfeld accepted this revision. Hahnfeld added a comment. This revision is now accepted and ready to land. LGTM with one nit Comment at: lib/Driver/ToolChain.cpp:362 + return UseLinker; + } else if (UseLinker == "ld") { +// If we're passed the argument ld, then use w

[PATCH] D25669: [Driver] Simplify ToolChain::GetCXXStdlibType (NFC)

This revision was automatically updated to reflect the committed changes. Closed by commit rL289422: [Driver] Simplify ToolChain::GetCXXStdlibType (NFC) (authored by Hahnfeld). Changed prior to commit: https://reviews.llvm.org/D25669?vs=74948&id=81053#toc Repository: rL LLVM https://reviews

[PATCH] D30087: [Driver] Unify linking of OpenMP runtime. NFCI.

This revision was automatically updated to reflect the committed changes. Closed by commit rL300689: [Driver] Unify linking of OpenMP runtime. NFCI. (authored by Hahnfeld). Changed prior to commit: https://reviews.llvm.org/D30087?vs=94372&id=95733#toc Repository: rL LLVM https://reviews.llv

[PATCH] D29904: [OpenMP] Prevent emission of exception handling code when using OpenMP to offload to NVIDIA devices.

Hahnfeld accepted this revision. Hahnfeld added a comment. This revision is now accepted and ready to land. LGTM unless Alexey still has objections Repository: rL LLVM https://reviews.llvm.org/D29904 ___ cfe-commits mailing list cfe-commits@lists

[PATCH] D29658: [OpenMP] Customize CUDA-based tool chain selection

Hahnfeld accepted this revision. Hahnfeld added a comment. This revision is now accepted and ready to land. LGTM with one small note Comment at: lib/Driver/Driver.cpp:564 + auto &CudaTC = + ToolChains[TT.str() + "/" + HostTC->getTriple().normalize(

[PATCH] D32735: [libcxxabi][CMake] Set library dir when built under runtimes/

Hahnfeld created this revision. Herald added a subscriber: mgorny. Herald added a reviewer: EricWF. This will put libraries into the build root's lib/ directory by default. https://reviews.llvm.org/D32735 Files: CMakeLists.txt Index: CMakeLists.txt ==

[PATCH] D32736: [libcxx][CMake] Set library dir when built under runtimes/

Hahnfeld created this revision. Herald added a subscriber: mgorny. This will put libraries into the build root's lib/ directory by default. https://reviews.llvm.org/D32736 Files: CMakeLists.txt Index: CMakeLists.txt === --- CMa

[PATCH] D32736: [libcxx][CMake] Set library dir when built under runtimes/

This revision was automatically updated to reflect the committed changes. Closed by commit rL302117: [libcxx][CMake] Set library dir when built under runtimes/ (authored by Hahnfeld). Changed prior to commit: https://reviews.llvm.org/D32736?vs=97403&id=97776#toc Repository: rL LLVM https://

[PATCH] D32735: [libcxxabi][CMake] Set library dir when built under runtimes/

This revision was automatically updated to reflect the committed changes. Closed by commit rL302118: [libcxxabi][CMake] Set library dir when built under runtimes/ (authored by Hahnfeld). Changed prior to commit: https://reviews.llvm.org/D32735?vs=97402&id=9#toc Repository: rL LLVM https

[PATCH] D29647: [OpenMP] Extend CLANG target options with device offloading kind.

Hahnfeld added a comment. One minor drive-by comment. I think there is still one outstanding from Justin... Comment at: lib/Driver/ToolChains/Cuda.cpp:368 + assert(!GpuArch.empty() && "Must have an explicit GPU arch."); + std::string LibDeviceFile; You can

[PATCH] D98902: [Clang][OpenMP][NVPTX] Fixed failure in openmp-offload-gpu.c if the system has CUDA

Hahnfeld added a comment. Can we get this fixed somehow? It's annoying that there is a test failure in Clang without building the OpenMP runtime, just because I have CUDA installed on my machine... Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D989

[PATCH] D108456: [CUDA] Fix static device variables with -fgpu-rdc

Hahnfeld created this revision. Hahnfeld added reviewers: tra, yaxunl. Hahnfeld requested review of this revision. Herald added a reviewer: jdoerfert. Herald added subscribers: cfe-commits, sstefan1. Herald added a project: clang. NVPTX does not allow dots in the identifier, so ptxas errors out wi

[PATCH] D85223: [CUDA][HIP] Support accessing static device variable in host code for -fgpu-rdc

Hahnfeld added inline comments. Comment at: clang/lib/CodeGen/CodeGenModule.cpp:6265-6268 +void CodeGenModule::printPostfixForExternalizedStaticVar( +llvm::raw_ostream &OS) const { + OS << ".static." << getContext().getCUIDHash(); +} I've tried to use this w

[PATCH] D108456: [CUDA] Fix static device variables with -fgpu-rdc

Hahnfeld added inline comments. Comment at: clang/lib/CodeGen/CodeGenModule.cpp:6450 llvm::raw_ostream &OS) const { - OS << ".static." << getContext().getCUIDHash(); + OS << "__static__" << getContext().getCUIDHash(); } tra wrote: > I would expect NVPTXA

[PATCH] D108456: [CUDA] Fix static device variables with -fgpu-rdc

This revision was automatically updated to reflect the committed changes. Closed by commit rGea08c4cd1c08: [CUDA] Fix static device variables with -fgpu-rdc (authored by Hahnfeld). Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D108456/new/ https://re

[PATCH] D107049: [clang-repl] Re-implement clang-interpreter as a test case.

Hahnfeld added a comment. In D107049#2984835 , @v.g.vassilev wrote: > It looks like we hit https://bugs.llvm.org/show_bug.cgi?id=49692 I agree. Unfortunately, as described in the report, this is something that needs to be fixed on Apples side first.

<    1   2   3   4   5   6