From: Borut R. <bor...@gm...> - 2012-06-24 05:20:11
|
No, it was not obvious to me. I thought that you asked Philipp to revert his changes, as it was obvious to me that the bug has to be fixed. Please apply the fix ASAP, I'll restart the builds if they already started. Sorry for my misunderstanding. Borut On 23. 06. 2012 23:46, Maarten Brock wrote: > Borut, > > In case it wasn't obvious enough, I await approval for > my proposed 'fix'. > > Maarten > >> Bugs item #3537338, was opened at 2012-06-23 04:56 >> Message generated for change (Comment added) made by maartenbrock >> You can respond by visiting: >> https://sourceforge.net/tracker/?func=detail&atid=100599&aid=3537338&group_id=599 >> >> Please note that this message will contain a full copy of the comment thread, >> including the initial issue submission, for this request, >> not just the latest update. >> Category: mcs51(8051) target >> Group: None >> Status: Open >> Resolution: None >> Priority: 8 >> Private: No >> Submitted By: Maarten Brock (maartenbrock) >> Assigned to: Maarten Brock (maartenbrock) >> Summary: Error 9 FATAL Compiler Internal Error >> >> Initial Comment: >> SDCC 3.2.0 Release Candidate generates: >> >> error 9: FATAL Compiler Internal Error in file '/home/sdcc-builder/build/sdcc-build/orig/sdcc/src/mcs51/gen.c' line number '8963' : code generator internal error >> Contact Author with source code >> >> for the code below that compiled fine with SDCC 3.1.0. I'm looking into it,but I think this must be fixed before the release. >> >> int globals[16]; >> >> int testBug(unsigned char index) >> { >> return *(globals+index); >> } >> >> >> ---------------------------------------------------------------------- >> >>> Comment By: Maarten Brock (maartenbrock) >> Date: 2012-06-23 13:00 >> >> Message: >> Well, the weird thing is that this comment is in both the mcs51 and the z80 >> port and in both genLeftShiftLiteral and genRightShiftLiteral but the >> assertion is only in two unrelated places. >> >> I suggest to remove the assertion in mcs51 now (before the release) as it >> breaks previously correctly generated code. Then keep this item open until >> the underlying problem is really fixed and the assertions can appear at all >> assumption comments. >> >> ---------------------------------------------------------------------- >> >> Comment By: Philipp Klaus Krause (spth) >> Date: 2012-06-23 08:16 >> >> Message: >> I had been hunting down another bug. After I finally had tracked it down, I >> noticed that in z80 port, in genRightShiftLiteral() in a place were there >> was a comment "/* I suppose that the left size >= result size */" had there >> been an assertion that had checked what the comment stated, I would have >> found the bug much earlier. So I added the assertion there. Then I noticed >> there was another such comment "/* I suppose that the left size >= result >> size */" in the cs51 port, genLeftShiftLiteral(). Adding the assertion >> caused no regression failures, so I thought it would be a good idea to add >> an assertion to check for what the comment states. >> >> Philipp >> >> >> ---------------------------------------------------------------------- >> >> Comment By: Maarten Brock (maartenbrock) >> Date: 2012-06-23 07:45 >> >> Message: >> This wassert was added by Philipp in #7628 to: >> genLeftShiftLiteral in mcs51/gen.c >> genRightShiftLiteral in z80/gen.c >> >> But I don't understand why only for LeftShift in mcs51 and only for >> RightShift in z80. >> >> And without the wassert the mcs51 generates correct code (for unsigned char >> index). So I was inclined to remove it again. Or move it to RightShift. But >> for signed char index the code is wrong. >> >> ---------------------------------------------------------------------- >> >> You can respond by visiting: >> https://sourceforge.net/tracker/?func=detail&atid=100599&aid=3537338&group_id=599 >> >> ------------------------------------------------------------------------------ >> Live Security Virtual Conference >> Exclusive live event will cover all the ways today's security and >> threat landscape has changed and how IT managers can respond. Discussions >> will include endpoint security, mobile security and the latest in malware >> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ >> _______________________________________________ >> sdcc-devel mailing list >> sdc...@li... >> https://lists.sourceforge.net/lists/listinfo/sdcc-devel >> > > > ------------------------------------------------------------------------------ > Live Security Virtual Conference > Exclusive live event will cover all the ways today's security and > threat landscape has changed and how IT managers can respond. Discussions > will include endpoint security, mobile security and the latest in malware > threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ > _______________________________________________ > sdcc-devel mailing list > sdc...@li... > https://lists.sourceforge.net/lists/listinfo/sdcc-devel > > |