I think I can get it to fall back to -O1 which safe when it fails the test , but I don't know how to get it to add a flag if it fails a test , only how to remove a flag if it fails a test, or we can just print a error message in configure to say to use -fno-strict-aliasing
I think a fallback to -O1 is best Jason ------------------------------------------------------------------------- I imagine the performance hit is minimal . We have a simple testcase int __attribute__((noinline)) foo(int i) { int *p = __builtin_malloc (4 * sizeof(int)); *p = 0; p[i] = 1; return *p; } extern void abort (void); int main() { if (foo(0) != 1) abort (); return 0; } So we can reject the bad debian gcc-4.3.2 only , but I'm not sure how to get it to select no-strict-aliasing if it fails the test. Jason ---------------------------- Thanks for the information. I am forwarding this to mpir-devel to see what Jason thinks about possibly disabling '-fstrict-aliasing' upstream for builds on Debian gcc 4.3.2. I wonder how much of a performance hit is involved. Bill. On 9 May 2012 16:23, leif <not.rea...@online.de> wrote: > Jeroen Demeyer wrote: >> >> On 2012-05-06 23:22, Alexander Dreyer wrote: >>> >>> Hi! >>> built and made ptestlong sucessfully on SuSE Enterprise 11. >>> >>> But I had to hack mpir to accept gcc 4.3.2 64bits: configure had >>> hard-coded not to build on this platform, because the original gcc 4.3.2 >>> was buggy, but mine is patched. (The mpir guys do not test the actual >>> bug, but just look for gcc's version.) >> >> >> I agree, hardcoding that bad version isn't a good idea. Also because >> MPIR might just work well enough to build GCC. > > > Turns out that apparently only Debian's GCC 4.3.2 is affected (not vanilla > FSF), and only if '-fstrict-aliasing' is enabled (which is the case with > '-O2'.) > > See: > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=524472 > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36765 > > > So we should probably patch MPIR's 'configure'. > > > -leif > > > -- > () The ASCII Ribbon Campaign > /\ Help Cure HTML E-Mail > > -- > You received this message because you are subscribed to the Google Groups > "sage-release" group. > To post to this group, send email to sage-rele...@googlegroups.com. > To unsubscribe from this group, send email to > sage-release+unsubscr...@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/sage-release?hl=en. > -- You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-devel@googlegroups.com. To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en. -- You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-devel@googlegroups.com. To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en. -- You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-devel@googlegroups.com. To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en.