This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
cris-axis-elf on trunk broken; structure member assignments
- From: Hans-Peter Nilsson <hp at bitrange dot com>
- To: gcc at gcc dot gnu dot org
- Date: Sun, 4 Jan 2004 07:26:13 -0500 (EST)
- Subject: cris-axis-elf on trunk broken; structure member assignments
I see massive regressions (~4000) for cris-axis-elf due to some change in
the time LAST_UPDATED Tue Dec 30 20:46:23 GMT 2003 to
Fri Jan 2 16:23:20 GMT 2004
Looking at gcc.c-torture/execute/20000706-5.c (-O2; -O0, -O1 and -Os fail
too), it seems as structure member assignments are not emitted, maybe
related to the structure being passed by value (in source; really by ref).
The a.a, a.b and a.c assignments in main are gone, not even there in the
initial rtl dump.
I also see 340 regressions for mmix-knuth-mmixware (results to be sent).
Due to lots of tests failing due to *timeout*, no other targets
have been checked (it takes a few days for a testsuite run to
timeout). I'm now checking v850-elf, h8300-hms, m32r-elf and
mn10300-elf or at least up to the first in that list that
exhibits the same problem.
I'll investigate, not ruling out an update in the middle of some
commit. Still, maybe someone has a clue that'd help a quick
resolution.
brgds, H-P