[LLVMbugs] [Bug 15798] New: Regression(r178606): Clang runs out of registers when building libyuv's row_posix for 32bit with -fPIC

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=15798

            Bug ID: 15798
           Summary: Regression(r178606): Clang runs out of registers when
                    building libyuv's row_posix for 32bit with -fPIC
           Product: clang
           Version: unspecified
          Hardware: PC
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: Driver
          Assignee: unassignedclangbugs at nondot.org
          Reporter: nicolasweber at gmx.de
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

Created attachment 10387
  --> http://llvm.org/bugs/attachment.cgi?id=10387&action=edit
Repro case out.ii

thakis at yearofthelinuxdesktop:/usr/local/google/chrome/src$
third_party/llvm-build/Release+Asserts/bin/clang --versionclang version 3.3
(trunk 178606)
Target: x86_64-unknown-linux-gnu
Thread model: posix
thakis at yearofthelinuxdesktop:/usr/local/google/chrome/src$
third_party/llvm-build/Release+Asserts/bin/clang  -c out.ii -m32 -fPIC
../../third_party/libyuv/source/row_posix.cc:1825:5: error: ran out of
registers during register allocation
    "movdqa    %[kShuffleMaskARGBToRGB24_0],%%xmm5 \n"
    ^
../../third_party/libyuv/source/row_posix.cc:1880:5: error: ran out of
registers during register allocation
    "movdqa    %[kShuffleMaskARGBToRAW_0],%%xmm5 \n"
    ^
2 errors generated.


Used to work one revision earlier:

thakis at yearofthelinuxdesktop:/usr/local/google/chrome/src$
third_party/llvm-build/Release+Asserts/bin/clang --version
clang version 3.3 (trunk 178605)
Target: x86_64-unknown-linux-gnu
Thread model: posix
thakis at yearofthelinuxdesktop:/usr/local/google/chrome/src$
third_party/llvm-build/Release+Asserts/bin/clang  -c out.ii -m32 -fPIC
# works



I suppose the workaround is to pass -momit-leaf-frame-pointer, which
effectively reverts r178606 user-side, but the change description doesn't say
why the change was done and the change does regress functionality, so maybe it
should be undone as well.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20130419/f0ab7ab8/attachment.html>