Not sure if this is gcc related on not, first failed build since upgrade to gcc 3.3, error as follows- gcc -DHAVE_CONFIG_H -I. -I. -I../.. -I../../xmms -I/usr/include/gtk-1.2 -I/usr/include/glib-1.2 -I/usr/lib/glib/include -D_REENTRANT -I/usr/X11R6/include -DI386_ASSEM -I../../intl -I../.. -O2 -march=k6-2 -pipe -fomit-frame-pointer -Wall -Wpointer-arith -finline-functions -ffast-math -fomit-frame-pointer -funroll-all-loops -c layer3.c -fPIC -DPIC -o layer3.lo {standard input}: Assembler messages: {standard input}:4601: Error: value of ffffffffffffff7f too large for field of 1 bytes at 0000000000000b99 make[3]: *** [layer3.lo] Error 1 make[3]: Leaving directory `/var/tmp/portage/xmms-1.2.8-r2/work/xmms-1.2.8/Input/mpg123' make[2]: *** [all-recursive] Error 1
review Bug 24379
I'm sorry, how is bug #24379 relevant? I receive neither illegal instructions, nor am I in doubt about the arch of my cpu. As a matter of fact, that was the only package I had that DID fail to build w/3.3. Also of note is that my flags never failed to build any package for nigh on a year now, but all of a sudden 3.2 -> 3.3 causes a lone build to drop assembler messages and it's my flags? That's akin to saying that the toast went though the ceiling because I set my toaster settings wrong, when the fault is in the toaster itself. I disagree with the closing as invalid, because it is a legitimate bug, however if devs will not take a bug report because it contains -march=k6* (if if the sources getting built contain more radical flags that what I've assigned) I think marking them wontfix with a note to take it up with the gcc people would be more descriptive than "invalid". Also a note on -dev/-user letting people know you wont take their bugs seriously if submitted with k6 flags, though that might drop your prospective reporters/fixers by 20% or so...