mgorny (Michał Górny)
User

Projects

User does not belong to any projects.
User Since
Aug 15 2016, 6:00 AM (27 w, 5 d)

Recent Activity

Fri, Feb 24

mgorny committed rL296183: [test] Try to link -latomic to provide atomics when available.
[test] Try to link -latomic to provide atomics when available
Fri, Feb 24, 2:27 PM
mgorny closed D30083: [openmp] [test] Try to link -latomic to provide atomics when available by committing rL296183: [test] Try to link -latomic to provide atomics when available.
Fri, Feb 24, 2:27 PM
mgorny added a reviewer for D30025: [compiler-rt] [builtins] Fix building atomic.c with GCC: doug.gregor.

If I read the git correctly, the change that forbid defining builtins was initially made in rL64639. @doug.gregor, any chance you could help us over here? Is clang supposed to unconditionally reject those definitions, and are we supposed to always work-around it in compiler-rt, or should we consider adding some additional switch to allow them in clang?

Fri, Feb 24, 11:29 AM
mgorny added a comment to D30083: [openmp] [test] Try to link -latomic to provide atomics when available.

Le gentle ping.

Fri, Feb 24, 11:19 AM

Thu, Feb 23

mgorny added inline comments to D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.
Thu, Feb 23, 9:32 PM

Wed, Feb 22

mgorny committed rL295854: [Support] Provide linux/magic.h fallback for older kernels.
[Support] Provide linux/magic.h fallback for older kernels
Wed, Feb 22, 10:21 AM
mgorny closed D30261: [Support] Provide linux/magic.h fallback for older kernels by committing rL295854: [Support] Provide linux/magic.h fallback for older kernels.
Wed, Feb 22, 10:21 AM
mgorny created D30261: [Support] Provide linux/magic.h fallback for older kernels.
Wed, Feb 22, 9:12 AM

Sun, Feb 19

mgorny committed rL295621: [test] Fix finding LLDB tools when building stand-alone.
[test] Fix finding LLDB tools when building stand-alone
Sun, Feb 19, 2:23 PM
mgorny closed D29985: [lldb] [test] Fix finding LLDB tools when building stand-alone by committing rL295621: [test] Fix finding LLDB tools when building stand-alone.
Sun, Feb 19, 2:23 PM · Restricted Project
mgorny created D30155: [clang-tools-extra] [test] Fix clang library dir in LD_LIBRARY_PATH For stand-alone build.
Sun, Feb 19, 5:37 AM · Restricted Project

Fri, Feb 17

mgorny accepted D30015: Add arch-specific directory to search path.

LGTM modulo the test match split. But please wait for someone who has been longer here to confirm.

Fri, Feb 17, 2:39 PM
mgorny added inline comments to D30092: CMake: Clean up VersionFromVCS.cmake.
Fri, Feb 17, 7:28 AM
mgorny added inline comments to D30015: Add arch-specific directory to search path.
Fri, Feb 17, 2:12 AM
mgorny created D30083: [openmp] [test] Try to link -latomic to provide atomics when available.
Fri, Feb 17, 2:03 AM

Thu, Feb 16

mgorny added a comment to D30015: Add arch-specific directory to search path.

Thanks. The -L tests look good, -rpath is not perfect but I don't think you can improve it without additional changes to the Driver.

Thu, Feb 16, 11:37 PM
mgorny added inline comments to D30015: Add arch-specific directory to search path.
Thu, Feb 16, 2:14 PM
mgorny added a comment to D29985: [lldb] [test] Fix finding LLDB tools when building stand-alone.

Following @labath's suggestion, I've changed my approach a little. Now I'm using PATH lookup only for CC/CXX, while for remaining tools I'm using either llvm_tools_dir or lldb_tools_dir appropriately.

Thu, Feb 16, 1:03 PM · Restricted Project
mgorny updated the diff for D29985: [lldb] [test] Fix finding LLDB tools when building stand-alone.
Thu, Feb 16, 1:02 PM · Restricted Project
mgorny added a comment to D30025: [compiler-rt] [builtins] Fix building atomic.c with GCC.

This code is working around something that's probably a clang bug. It would be better to fix the clang bug than add more complex workarounds.

Thu, Feb 16, 4:30 AM
mgorny added inline comments to D30015: Add arch-specific directory to search path.
Thu, Feb 16, 12:05 AM
mgorny added inline comments to D30015: Add arch-specific directory to search path.
Thu, Feb 16, 12:03 AM

Wed, Feb 15

mgorny created D30025: [compiler-rt] [builtins] Fix building atomic.c with GCC.
Wed, Feb 15, 11:14 PM
mgorny added a reviewer for D29985: [lldb] [test] Fix finding LLDB tools when building stand-alone: labath.
Wed, Feb 15, 10:27 AM · Restricted Project
mgorny added a comment to D29185: [lit] Allow llvm's build and test systems to support paths with spaces.

I would like to point out that if you proactively reject all paths with spaces, you wouldn't be able to add additional tests that would test support for paths with spaces ;-). While I agree that we should avoid spaces whenever possible, I think it would be reasonable to add tests for paths with spaces wherever we expect them to be well-supported and want to avoid regressions.

Wed, Feb 15, 5:04 AM
mgorny created D29985: [lldb] [test] Fix finding LLDB tools when building stand-alone.
Wed, Feb 15, 4:57 AM · Restricted Project
mgorny abandoned D28862: [compiler-rt] [test] Use approximate comparison on float types.

I've opened http://bugs.llvm.org/show_bug.cgi?id=31964 to track this further. I don't really have time to look into it in more detail at the moment.

Wed, Feb 15, 2:35 AM

Mon, Feb 13

mgorny added a comment to D28533: [llvm-config] Add minimal sanity tests for path options.

Ping.

Mon, Feb 13, 7:51 AM
mgorny added a comment to D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.

Le gentle ping.

Mon, Feb 13, 7:51 AM

Fri, Feb 10

mgorny created D29851: [clang-tools-extra] [test] Fix test dependencies when using installed tools.
Fri, Feb 10, 2:39 PM · Restricted Project

Wed, Feb 8

mgorny committed rL294508: [test] Fix zlib cond when building stand-alone, clean up.
[test] Fix zlib cond when building stand-alone, clean up
Wed, Feb 8, 12:20 PM
mgorny closed D29340: [lld] [test] Fix zlib cond when building stand-alone, clean up by committing rL294508: [test] Fix zlib cond when building stand-alone, clean up.
Wed, Feb 8, 12:20 PM · lld
mgorny committed rL294507: [test] Use LLD-specific binary&library dirs when building stand-alone.
[test] Use LLD-specific binary&library dirs when building stand-alone
Wed, Feb 8, 12:19 PM
mgorny closed D29335: [lld] [test] Use LLD-specific binary&library dirs when building stand-alone by committing rL294507: [test] Use LLD-specific binary&library dirs when building stand-alone.
Wed, Feb 8, 12:19 PM · lld
mgorny committed rL294438: [test] #ifdef new builtin tests for __arm__ platform.
[test] #ifdef new builtin tests for __arm__ platform
Wed, Feb 8, 4:14 AM
mgorny closed D29708: [compiler-rt] [test] #ifdef new builtin tests for __arm__ platform by committing rL294438: [test] #ifdef new builtin tests for __arm__ platform.
Wed, Feb 8, 4:14 AM
mgorny committed rL294431: [test] Fix hard_link_count test to account for fs with dir nlink==1.
[test] Fix hard_link_count test to account for fs with dir nlink==1
Wed, Feb 8, 2:09 AM
mgorny closed D29706: [libcxx] [test] Fix hard_link_count test to account for fs with dir nlink==1 by committing rL294431: [test] Fix hard_link_count test to account for fs with dir nlink==1.
Wed, Feb 8, 2:08 AM
mgorny created D29708: [compiler-rt] [test] #ifdef new builtin tests for __arm__ platform.
Wed, Feb 8, 2:05 AM
mgorny created D29706: [libcxx] [test] Fix hard_link_count test to account for fs with dir nlink==1.
Wed, Feb 8, 1:37 AM
mgorny committed rL294425: [test] Enable the strace_test only if strace is installed.
[test] Enable the strace_test only if strace is installed
Wed, Feb 8, 12:55 AM
mgorny closed D29628: [compiler-rt] [test] Enable the strace_test only if strace is installed by committing rL294425: [test] Enable the strace_test only if strace is installed.
Wed, Feb 8, 12:55 AM
mgorny added a comment to D29340: [lld] [test] Fix zlib cond when building stand-alone, clean up.

A gentle ping.

Wed, Feb 8, 12:49 AM · lld
mgorny added a comment to D29335: [lld] [test] Use LLD-specific binary&library dirs when building stand-alone.

A gentle ping.

Wed, Feb 8, 12:49 AM · lld

Tue, Feb 7

mgorny created D29628: [compiler-rt] [test] Enable the strace_test only if strace is installed.
Tue, Feb 7, 4:21 AM

Mon, Feb 6

mgorny updated the diff for D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.

Removed the i386 branch. Now the i486+ are used unconditionally.

Mon, Feb 6, 2:52 PM
mgorny added inline comments to D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.
Mon, Feb 6, 10:11 AM
mgorny added a comment to D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.

At this point, I don't think there is any use on pretending that i386-as-default makes sense. So I would request that the i386 case should be made explicit or just dropped, with a preference for the latter.

Mon, Feb 6, 9:43 AM
mgorny added inline comments to D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.
Mon, Feb 6, 9:32 AM
mgorny added inline comments to D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.
Mon, Feb 6, 12:19 AM

Sun, Feb 5

mgorny updated the diff for D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.

CUDA: added the MaxAtomicPromoteWidth setting, and moved the CPU setting a little lower to ensure that it doesn't get called with null HostTarget.

Sun, Feb 5, 3:12 PM
mgorny added inline comments to D29555: [CMake] Fix pthread handling for out-of-tree builds.
Sun, Feb 5, 4:32 AM
mgorny updated the diff for D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.

Ok, this CUDA fix should be reasonable, i think. It simply assumes i586+ (i.e. all inline atomics enabled) for CUDA target builds. I seriously doubt it's technically possible that anyone will ever use CUDA on <i586 ;-).

Sun, Feb 5, 1:42 AM

Sat, Feb 4

mgorny added a comment to D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.

Could someone help me figure out what is the cause and correct solution to that failure? @jlebar?

You can see in NVPTXTargetInfo that we read properties from the host targetinfo so that we export the same macros. The problem here seems to be that we're mutating the x86 targetinfo after the nvptx targetinfo reads its properties.

Does that give you enough context to fix the problem?

Sat, Feb 4, 12:49 PM
mgorny created D29542: [TargetInfo] Adjust x86-32 atomic support to the CPU used.
Sat, Feb 4, 10:24 AM
mgorny added a comment to D28213: [Frontend] Correct values of ATOMIC_*_LOCK_FREE to match builtin.
In D28213#666269, @dim wrote:

What's the value of __atomic_always_lock_free(sizeof(long long), 0) for gcc and clang?

For gcc, it is always 0, for clang (I tested 3.4.1 through 4.0.0) it is always 1. Maybe that was always incorrect on 32-bit FreeBSD, then?

Sat, Feb 4, 1:12 AM

Fri, Feb 3

mgorny added a comment to D28213: [Frontend] Correct values of ATOMIC_*_LOCK_FREE to match builtin.
In D28213#665967, @dim wrote:

I don't think FreeBSD has lock-free 64 bit atomic operations on 32-bit x86. IIRC we already had some trouble before with clang emitting libcalls to __atomic_fetch_add_8 and friends, which then lead to linking errors.

See https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=216745, where this is now occurring with boost.

Fri, Feb 3, 12:52 PM
mgorny added a reviewer for D29335: [lld] [test] Use LLD-specific binary&library dirs when building stand-alone: beanz.
Fri, Feb 3, 9:14 AM · lld
mgorny added a reviewer for D29340: [lld] [test] Fix zlib cond when building stand-alone, clean up: beanz.
Fri, Feb 3, 9:14 AM · lld

Wed, Feb 1

mgorny added inline comments to D27405: Make sancov.py py3-compatible.
Wed, Feb 1, 12:34 AM

Tue, Jan 31

mgorny added inline comments to D27405: Make sancov.py py3-compatible.
Tue, Jan 31, 11:48 PM
mgorny added inline comments to D27404: Make asan_symbolize.py py3-compatible.
Tue, Jan 31, 11:41 PM
mgorny created D29340: [lld] [test] Fix zlib cond when building stand-alone, clean up.
Tue, Jan 31, 11:11 AM · lld
mgorny created D29335: [lld] [test] Use LLD-specific binary&library dirs when building stand-alone.
Tue, Jan 31, 10:10 AM · lld
mgorny committed rL293632: [cmake] Hint find_package() to prefer LLVM installed alongside clang.
[cmake] Hint find_package() to prefer LLVM installed alongside clang
Tue, Jan 31, 6:27 AM
mgorny closed D29304: [cmake] Hint find_package() to prefer LLVM installed alongside clang by committing rL293632: [cmake] Hint find_package() to prefer LLVM installed alongside clang.
Tue, Jan 31, 6:27 AM
mgorny committed rL293630: [cmake] Support running tests in stand-alone builds.
[cmake] Support running tests in stand-alone builds
Tue, Jan 31, 6:21 AM
mgorny closed D28750: [lld] [cmake] Support running tests in stand-alone builds by committing rL293630: [cmake] Support running tests in stand-alone builds.
Tue, Jan 31, 6:21 AM

Mon, Jan 30

mgorny created D29304: [cmake] Hint find_package() to prefer LLVM installed alongside clang.
Mon, Jan 30, 1:55 PM
mgorny added a comment to D28750: [lld] [cmake] Support running tests in stand-alone builds.

Le ping.

Mon, Jan 30, 10:12 AM

Jan 26 2017

mgorny updated the diff for D25402: [Driver] Pass -lunwind along with compiler-rt when necessary on Linux.

Here's a v2. It turns out that you also need -lunwind when using to link C programs with -static -rtlib=compiler-rt. I've also disabled the changes for Android targets.

Jan 26 2017, 1:49 AM

Jan 25 2017

mgorny committed rL293078: [cmake] Fix -rpath-link in stand-alone build.
[cmake] Fix -rpath-link in stand-alone build
Jan 25 2017, 11:44 AM
mgorny closed D29099: [lld] [cmake] Fix -rpath-link in stand-alone build by committing rL293078: [cmake] Fix -rpath-link in stand-alone build.
Jan 25 2017, 11:44 AM · lld
mgorny accepted D28908: [Compiler-rt] Broken compiler-rt CMake configuring on Windows.

Well, it doesn't seem to break anything for me on Linux and it resembles the code below, so I guess it's good to go.

Jan 25 2017, 11:05 AM
mgorny added a comment to D29126: [cmake] Remove VERSION property from executable targets.

Technically the patch looks sane; however, I suggest you wait for some input from the others. As for Gentoo, I don't think we ever really cared for LLDB being versioned — only clang.

Jan 25 2017, 10:55 AM
mgorny committed rL293053: [test] Add HAVE_LIBZ to canonicalized booleans.
[test] Add HAVE_LIBZ to canonicalized booleans
Jan 25 2017, 5:43 AM
mgorny committed rL293052: [test] Port clang tests to canonicalized booleans.
[test] Port clang tests to canonicalized booleans
Jan 25 2017, 5:23 AM
mgorny closed D28529: [test] Port clang tests to canonicalized booleans by committing rL293052: [test] Port clang tests to canonicalized booleans.
Jan 25 2017, 5:23 AM
mgorny added reviewers for D29099: [lld] [cmake] Fix -rpath-link in stand-alone build: compnerd, phosek.
Jan 25 2017, 4:11 AM · lld

Jan 24 2017

mgorny added a comment to D29099: [lld] [cmake] Fix -rpath-link in stand-alone build.

I've filed https://llvm.org/bugs/show_bug.cgi?id=31745 to track it for 4.0.0 backport.

Jan 24 2017, 2:47 PM · lld
mgorny created D29099: [lld] [cmake] Fix -rpath-link in stand-alone build.
Jan 24 2017, 2:44 PM · lld
mgorny resigned from D25304: cmake: Set the proper rpath in add_llvm_executable and llvm_add_library.
Jan 24 2017, 2:41 PM
mgorny accepted D28894: [Support] Use O_CLOEXEC only when declared.

Builds fine for me.

Jan 24 2017, 1:42 AM
mgorny added a comment to D29007: [lld][cmake] Fix BUILD_SHARED_LIBS installation.

Oh, and I thought that it was a feature since lld was linked statically anyway! ;-D

Jan 24 2017, 12:10 AM

Jan 23 2017

mgorny added a comment to D28894: [Support] Use O_CLOEXEC only when declared.

Haha, I thought you meant you'll do it since you had the code already :-)

Jan 23 2017, 4:04 AM

Jan 21 2017

mgorny committed rL292729: [test] Fix page address logic in clear_cache_test.
[test] Fix page address logic in clear_cache_test
Jan 21 2017, 2:06 PM
mgorny closed D28849: [compiler-rt] [test] Fix page address logic in clear_cache_test by committing rL292729: [test] Fix page address logic in clear_cache_test.
Jan 21 2017, 2:06 PM

Jan 20 2017

mgorny updated the diff for D28849: [compiler-rt] [test] Fix page address logic in clear_cache_test.

Fixed the missing change. Also created a local variable to avoid calling the sysconf three times.

Jan 20 2017, 3:43 PM

Jan 19 2017

mgorny added a reviewer for D28908: [Compiler-rt] Broken compiler-rt CMake configuring on Windows: compnerd.

@compnerd, could you share your Windows knowledge? ;-)

Jan 19 2017, 10:04 AM
mgorny added a comment to D28894: [Support] Use O_CLOEXEC only when declared.

Yeah, I think that'll work for me.

Jan 19 2017, 2:18 AM
mgorny created D28894: [Support] Use O_CLOEXEC only when declared.
Jan 19 2017, 1:33 AM
mgorny added a comment to D28862: [compiler-rt] [test] Use approximate comparison on float types.

Just to be clear, the values also differ per -O0 vs -O2. I'm not sure if we can even reliably figure that out.

Jan 19 2017, 12:19 AM
mgorny updated the diff for D28849: [compiler-rt] [test] Fix page address logic in clear_cache_test.

Updated to use page size logic on POSIX and Windows systems. I haven't tested the latter, though.

Jan 19 2017, 12:14 AM

Jan 18 2017

mgorny added inline comments to D28855: [CMake] Copy per-component `required_libraries` into `LINK_COMPONENTS`. NFC..
Jan 18 2017, 11:49 AM
mgorny created D28862: [compiler-rt] [test] Use approximate comparison on float types.
Jan 18 2017, 8:23 AM
mgorny created D28849: [compiler-rt] [test] Fix page address logic in clear_cache_test.
Jan 18 2017, 12:56 AM

Jan 17 2017

mgorny added a comment to D28574: [scudo] Refactor of CRC32 and ARM runtime CRC32 detection.

Still LGTM. I presume you have tested the final version ;-).

Jan 17 2017, 2:11 PM
mgorny added inline comments to D28558: [CMake] Ignore hidden/metadata files when adding lit check-* testsuites.
Jan 17 2017, 2:06 PM
mgorny committed rL292257: [test] [builtins] Remove obsolete/UB tests in __fixuns?fdi based.
[test] [builtins] Remove obsolete/UB tests in __fixuns?fdi based
Jan 17 2017, 1:19 PM
mgorny closed D28146: [compiler-rt] [test] [builtins] Remove obsolete/UB tests in __fixuns?fdi based by committing rL292257: [test] [builtins] Remove obsolete/UB tests in __fixuns?fdi based.
Jan 17 2017, 1:19 PM