<html><head><style type="text/css"><!-- DIV {margin:0px;} --></style></head><body><div style="font-family:arial, helvetica, sans-serif;font-size:12pt"><DIV><FONT size=2>Since my code does change the way the debug info is emitted it is likely to be my change, but i would have expected it to fail in stage 1 in that case. It is strange that slightly different debug info should cause a miscompile.<BR><BR>- Jan<BR></FONT></DIV>
<DIV style="FONT-FAMILY: arial, helvetica, sans-serif; FONT-SIZE: 12pt"><BR>
<DIV style="FONT-FAMILY: Courier New, monaco, monospace, sans-serif; FONT-SIZE: 10pt"><FONT size=2 face=Tahoma>
<HR SIZE=1>
<B><SPAN style="FONT-WEIGHT: bold">From:</SPAN></B> Duncan Sands <baldrick@free.fr><BR><B><SPAN style="FONT-WEIGHT: bold">To:</SPAN></B> NAKAMURA Takumi <geek4civic@gmail.com><BR><B><SPAN style="FONT-WEIGHT: bold">Cc:</SPAN></B> llvm-testresults <llvm-testresults@cs.uiuc.edu>; Daniel Dunbar <daniel@zuster.org>; Jan Sjodin <jan_sjodin@yahoo.com>; Peter Collingbourne <peter@pcc.me.uk><BR><B><SPAN style="FONT-WEIGHT: bold">Sent:</SPAN></B> Sat, March 12, 2011 7:43:03 AM<BR><B><SPAN style="FONT-WEIGHT: bold">Subject:</SPAN></B> Re: [llvm-testresults] buildbot failure in llvm on llvm-gcc-i386-linux-selfhost<BR></FONT><BR>On 12/03/11 12:11, NAKAMURA Takumi wrote:<BR>>> http://google1.osuosl.org:8011/builders/llvm-gcc-i386-linux-selfhost/builds/1540<BR>> <BR>> It seems it has been failing on since then. Why?<BR>> <BR>> ...Takumi(fine)<BR><BR>The llvm-x86_64-linux-checks builder is failing in the same
way: these tests<BR>are failing in stage2:<BR> 2006-11-30-pubnames.cpp<BR> 2010-08-31-byvalarg.cpp<BR>They do not fail in stage 1, suggesting a miscompile. The first failures were<BR>here:<BR><A href="http://google1.osuosl.org:8011/builders/llvm-gcc-i386-linux-selfhost/builds/1540" target=_blank>http://google1.osuosl.org:8011/builders/llvm-gcc-i386-linux-selfhost/builds/1540</A><BR>http://google1.osuosl.org:8011/builders/llvm-x86_64-linux-checks/builds/557<BR><BR>The set of commits for the llvm-x86_64-linux-checks builder is a strict superset<BR>of those implicated by the llvm-gcc-i386-linux-selfhost builder. Eliminating<BR>those involving clang, this leaves two commits. I think the first one, by<BR>ddunbar, can be eliminated: it was reverting a recent change, so is unlikely<BR>to have introduced a miscompilation, and the original patch has since been<BR>reapplied without fixing the buildbots. So that leaves jsjodin's
change which<BR>I am about to try reverting.<BR><BR>Ciao, Duncan.<BR><BR>Changed by: ddunbar<BR>Changed at: Fri 11 Mar 2011 11:35:18<BR>Revision: 127477<BR>Changed files:<BR><BR>lib/Transforms/Scalar/CodeGenPrepare.cpp<BR>test/CodeGen/ARM/constants.ll<BR>test/CodeGen/CellSPU/jumptable.ll<BR>test/CodeGen/Thumb/dyn-stackalloc.ll<BR>test/CodeGen/Thumb2/2009-08-06-SpDecBug.ll<BR>test/CodeGen/Thumb2/2009-09-28-ITBlockBug.ll<BR>test/CodeGen/Thumb2/2010-11-22-EpilogueBug.ll<BR>test/CodeGen/Thumb2/cross-rc-coalescing-2.ll<BR>test/CodeGen/X86/2008-04-16-ReMatBug.ll<BR>test/CodeGen/X86/MachineSink-PHIUse.ll<BR>test/CodeGen/X86/personality.ll<BR>test/CodeGen/X86/pr3366.ll<BR>test/CodeGen/X86/sext-i1.ll<BR>test/CodeGen/X86/zext-extract_subreg.ll<BR>test/MC/ARM/simple-encoding.ll<BR>test/Transforms/CodeGenPrepare/basic.ll<BR>Comments:<BR>Revert r127459, "Optimize trivial branches in CodeGenPrepare, which often get<BR>created from the", it broke some GCC test suite
tests.<BR>Properties:<BR>Changed by: jsjodin<BR>Changed at: Fri 11 Mar 2011 11:41:18<BR>Revision: 127478<BR>Changed files:<BR><BR>include/llvm/MC/MCSection.h<BR>include/llvm/MC/MCSectionELF.h<BR>lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp<BR>Comments:<BR>Remove optimization emitting a reference insted of label difference, since it can create more relocations. Removed isBaseAddressKnownZero method, because it is no longer used.<BR></DIV></DIV></div></body></html>