Today

compnerd accepted D28971: [Bultin][ARM] Make aeabi_memset be Thumb1 compatible.

That muls requirement that Rd and Rm be the same is dumb.

Sat, Jan 21, 8:48 AM
compnerd accepted D28849: [compiler-rt] [test] Fix page address logic in clear_cache_test.

Would be nice to clang-format those lines, but its no worse than before.

Sat, Jan 21, 8:44 AM
compnerd committed rL292723: X86: swap EBP, ESP on !APPLE.
X86: swap EBP, ESP on !APPLE
Sat, Jan 21, 8:34 AM
compnerd committed rL292722: DWARF: allow enabling tracing at runtime.
DWARF: allow enabling tracing at runtime
Sat, Jan 21, 8:34 AM
compnerd committed rL292721: DWARF: convert error logs to _LIBUNWIND_LOG.
DWARF: convert error logs to _LIBUNWIND_LOG
Sat, Jan 21, 8:34 AM
compnerd committed rL292720: config: clean up some of the macro definition.
config: clean up some of the macro definition
Sat, Jan 21, 8:34 AM
compnerd committed rL292719: rename OtherAddressSpace to RemoteAddressSpace; NFC.
rename OtherAddressSpace to RemoteAddressSpace; NFC
Sat, Jan 21, 8:33 AM
alexfh_ edited reviewers for D28973: Supresses misc-move-constructor-init warning for const fields., added: alexfh; removed: alexfh_.
Sat, Jan 21, 8:03 AM
dim created D28981: Use GNU-style attributes for several __throw_XXX() functions.
Sat, Jan 21, 8:02 AM
a.sidorin added a comment to D26753: ASTImporter: improve support for C++ templates.

I got it. I have hard-coded paths in CHECK-lines so these tests are passed on my machine but not on other. Thank you Kareem!

Sat, Jan 21, 7:27 AM
EricWF added a comment to D28785: Split exception.cpp and new.cpp implementation into different files for different runtimes.

While I love this direction (the original version really was an unintelligible pile of code), I really think that this change may be taking on too much. Why not split it up first and do nothing else. We could do the MS ABI implementation in a subsequent change. This would improve the code and would not be gated on the MS ABI changes.

I agree this review is taking on too much, it started out much smaller and I tried to avoid expanding it, but in the end I had three options:

A) Regress and remove all support for MSVC, this would break the windows build. (at least in exception.cpp and new.cpp).
B) Implement incorrect versions of support/runtime/<header>_msvc.ipp based on w/e we currently have, just to keep Windows building.
C) Implement correct versions of support/runtime/<header>_msvc.ipp.

I choose (C) since I didn't want to regress Windows, or spend time implementing incorrect <header>_msvc.ipp versions.
However I'm willing to try and shrink this down if you think that would be better.

Sat, Jan 21, 7:15 AM
streit committed rL292718.
Sat, Jan 21, 6:54 AM
EricWF committed rL292717: Revert accidentally changes which reverted r292582.
Revert accidentally changes which reverted r292582
Sat, Jan 21, 6:53 AM
streit committed rL292716.
Sat, Jan 21, 6:51 AM
firolino added a comment to D27621: [clang-tidy] check to find declarations declaring more than one name.
  • nothing special, just went through some open comments and marked them as Done.
Sat, Jan 21, 6:35 AM
vpykhtin updated the summary of D28980: Fix typo in GCNSchedStrategy.
Sat, Jan 21, 6:31 AM
vpykhtin updated the summary of D28980: Fix typo in GCNSchedStrategy.
Sat, Jan 21, 6:29 AM
streit committed rL292715.
Sat, Jan 21, 6:27 AM
vpykhtin created D28980: Fix typo in GCNSchedStrategy.
Sat, Jan 21, 6:25 AM
firolino added a comment to D27621: [clang-tidy] check to find declarations declaring more than one name.

@aaron.ballman I am going to implement CppCore, Cert and Everything and test it on some projects and provide the results next week, to find out which one to set Default.

Sat, Jan 21, 6:24 AM
firolino added inline comments to D27621: [clang-tidy] check to find declarations declaring more than one name.
Sat, Jan 21, 6:17 AM
firolino added inline comments to D27621: [clang-tidy] check to find declarations declaring more than one name.
Sat, Jan 21, 6:15 AM
rengolin accepted D28971: [Bultin][ARM] Make aeabi_memset be Thumb1 compatible.

All encodings seem to be Thumb1, not just ARMv6M. LGTM. Thanks!

Sat, Jan 21, 4:53 AM
a.sidorin added a comment to D28952: [analyzer] Add new Z3 constraint manager backend.

Amazing work, Dominic. That's what I wanted to test for long time. But, personally, I'm not happy with massive changes in tests.

  1. I don't think that we need to change run line for tests if they pass with both managers. These changes are pretty noisy,
  2. If Z3 is optional, we cannot enforce its usage.
  3. How testing will be performed without Z3?
Sat, Jan 21, 4:02 AM
EricWF added a comment to D28397: [lld][cmake] Fix LLVM_LINK_LLVM_DYLIB build.

TL;DR This broke lld.

I was trying to build an RPM for LLVM/Clang 4.0.0 RC1 toolchain and noticed issues, which at first looks a caused by this change.

With this change you are no more installing required shared libraries:
error: Failed dependencies:

liblldCOFF.so.4()(64bit) is needed by external+llvm+4.0.0-cms-1-1.x86_64
liblldDriver.so.4()(64bit) is needed by external+llvm+4.0.0-cms-1-1.x86_64
liblldELF.so.4()(64bit) is needed by external+llvm+4.0.0-cms-1-1.x86_64

I checked CMake install logs and they were not listed as installed.

Looking at packages (thus installed) lld binary it depends on them.

Sat, Jan 21, 3:52 AM
bryant added a comment to D28855: [CMake] Copy per-component `required_libraries` into `LINK_COMPONENTS`. NFC..
  • Further update is_llvm_target to is_omitted_target_lib. The former checks whether a component is part of a known target...
Sat, Jan 21, 2:19 AM
bryant updated the diff for D28855: [CMake] Copy per-component `required_libraries` into `LINK_COMPONENTS`. NFC..
  • Fix is_llvm_target's logic. Previously, it checked a component name (X86Utils, LanaiInstPrinter, so on) against fully qualified libnames (LLVMX86Utils, LLVMLanaiInstPrinter), which is incorrect. This is exactly the fix in D28869 by Chris.
Sat, Jan 21, 2:15 AM
grosser added a comment to D24716: [Polly] DeLICM/DePRE (WIP).

Hi Michael,

Sat, Jan 21, 2:00 AM · Restricted Project
craig.topper created D28979: [X86] Add demanded elts support for the inputs to pclmul intrinsic.
Sat, Jan 21, 1:03 AM

Yesterday

lhames committed rL292714: [Orc][RPC] Add 'removeHandler' and 'clearHandlers' methods to RPC endpoints..
[Orc][RPC] Add 'removeHandler' and 'clearHandlers' methods to RPC endpoints.
Fri, Jan 20, 11:57 PM
grimar added a comment to D28951: [ELF] - Linkerscripts: ignore CONSTRUCTORS in output section declaration..

Are you sure that this CONSTRUCTORS is trying to be a linker keyword? The nearby DATA_DATA is actually a macro expanded from http://src.illumos.org/source/xref/linux-master/include/asm-generic/vmlinux.lds.h#205
Is the CONSTRUCTORS still in file after it has been preprocessed? (I'm guessing so since otherwise this patch wouldn't fix the build, but just wanted to be sure).
Remember to keep in mind that these linker scripts all get preprocessed by the C preprocessor, so remember to look at the preprocessed linker script instead of the original source.

Fri, Jan 20, 11:52 PM
rgov added a comment to D28955: [analyzer] Enable support for symbolic extension/truncation.

Are all the changes here related to the extension/truncation support, for instance the changes to test/Analysis/malloc.c? Can you move misc. cleanup changes to another review?

Fri, Jan 20, 11:12 PM
ctopper committed rL292713: [X86] Don't allow commuting to form phsub operations..
[X86] Don't allow commuting to form phsub operations.
Fri, Jan 20, 11:10 PM
ctopper committed rL292712: [X86] Add test cases that show bad commuting being allowed to create a phsub….
[X86] Add test cases that show bad commuting being allowed to create a phsub…
Fri, Jan 20, 11:10 PM
rgov added a comment to D28952: [analyzer] Add new Z3 constraint manager backend.

Do you think you could upload the patch omitting all of the test case changes for now? Maybe include one as an example but it seems to be just adding %z3_cc1 so we don't need to see all of them right now.

Fri, Jan 20, 11:04 PM
mehdi_amini created D28978: [ThinLTO] Add an auto-hide feature.
Fri, Jan 20, 11:00 PM
rgov awarded D28952: [analyzer] Add new Z3 constraint manager backend a Love token.
Fri, Jan 20, 10:51 PM
majnemer added inline comments to D23563: [RISCV 6/10] Add basic RISCVAsmParser.
Fri, Jan 20, 10:34 PM
mehdi_amini committed rL292711: Add missing dependency to "Module Summary Analysis" pass.
Add missing dependency to "Module Summary Analysis" pass
Fri, Jan 20, 10:12 PM
chandlerc updated the diff for D28848: [PM] Teach LoopUnroll to update the LPM infrastructure as it unrolls loops..

::sigh:: *Actually* remove the unnecessary loop-simplify runs.

Fri, Jan 20, 9:46 PM