On 2015-5-31 16:22, asmwarrior wrote: > On 2015-5-23 9:39, asmwarrior wrote: >> I just want to hunt the GCC bug: (big pch file will crash cc1plus.exe) >> 56926 – Crash (without ICE) while compiling Boost.Math - >> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56926 >> >> It turns out that build the GCC and G++ myself is too complex for me, so I >> would like to see if someone can supply a debug version of tool chain, so >> that I can run them under gdb to see where it crashed. >> >> BTW: I see that the latest LD can have separate debug file generated, so >> maybe, we can use it. >> >> Thanks. >> > Hi, with Kai's suggestion: https://sourceforge.net/p/mingw-w64/bugs/382/ > I guess that the hard limit value is 128M, and I would suggest a larger value > for this variable. See here: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56926#c16 > Any one would like to try a new build of gcc? > > There are some similar related discussion, see: > 1, mingw - cc1plus.exe crash when using large precompiled header file > http://stackoverflow.com/questions/10841306/cc1plus-exe-crash-when-using-large-precompiled-header-file > 2, MinGW-w64 - for 32 and 64 bit Windows / Mailing Lists > https://sourceforge.net/p/mingw-w64/mailman/message/30846624/ > > Thanks. >
Hi, good news, since I don't see any one rebuild a gcc chain by a larger pch_VA_max_size value recently. Today, I do it myself by modify the cc1plus.exe in a binary editor, just changed the values in the three referenced instructions. I changed the value from 128M to 512M. See details here: Comment 17 - https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56926#c17 Now, the modified cc1plus.exe never crash with a 200M pch file. So, the crash issue can to solved now!!! asmwarrior ------------------------------------------------------------------------------ _______________________________________________ Mingw-w64-public mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
