The following testcase errors while reloading the asms with -O3 -m32 or
-O3 -m32 -DOMIT_FRAME_POINTER -fomit-frame-pointer since r139993 (before that
it worked even with IRA, after that -fno-ira cured it while we still had it).
I agree mplayer pushes the limits a little bit too far by having so many "m"
and "=m" constraints, on the other side e.g. in baz when it is inlined none of
the asms actually need any registers to reload them, all the mems are actually
s[012345] + some small constant, so on i386 all of them can be expanded as
s0+16 and similar. In the *.optimized dump they are still constants, wonder
why during expand we decide to force all the MEM addresses into registers.
--
Summary: [4.4 Regression] Reload failure on mplayer from SVN
Product: gcc
Version: 4.4.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: rtl-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: jakub at gcc dot gnu dot org
GCC target triplet: i686-linux
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39543
