--- Additional Comments From mustafa at il dot ibm dot com 2005-05-17
15:05 ---
Janis, can you try this patch?
Index: modulo-sched.c
===
RCS file: /cvs/gcc/gcc/gcc/modulo-sched.c,v
retrieving revision 1.29
diff -c -p
--- Additional Comments From mustafa at il dot ibm dot com 2005-05-04
12:31 ---
Is seems like this is not an SMS bug, sixtrack is failing for me with -m64 -O2
without -fmodulo-sched.
Jania, have you checked that and have a different results?
--
http://gcc.gnu.org/bugzilla
--- Additional Comments From mustafa at il dot ibm dot com 2005-03-17
11:53 ---
The following patch should fix the Segmentation fault in gap (from SPEC2000)
mentioned in comment 14. This patch is combined with the
patch from comment 13. Janis can you try it out.
Index: ddg.c
--- Additional Comments From mustafa at il dot ibm dot com 2005-03-17
07:30 ---
I get the ICE in schedule_insns at sched-rgn.c:2549 also in bootstrap with -O2
-fmodulo-sched.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20177
--- Additional Comments From mustafa at il dot ibm dot com 2005-03-16
17:45 ---
For some reason the REG_DEAD is not the cause of the failure
it is the fact that the SMSed basic-block wasn't mark dirty for
update_life_info that come after it. doing so fixes the failure
even
--- Additional Comments From mustafa at il dot ibm dot com 2005-03-16
12:05 ---
After a bit more debugging I found out that the error caused due to copying the
REG_DEAD note with the instructions when we are generating the prologues and
epilogues in SMS. The REG_DEAD is correct in the
--- Additional Comments From mustafa at il dot ibm dot com 2005-03-16
09:36 ---
I suppose that the REG_DEAD for 136 in insn 65 is correct, because the next insn
is a DEF of 136. So the problem here is that 136 is not in the liveout of BB 7.
I can guess that insn 97 that defines 136 is
--- Additional Comments From mustafa at il dot ibm dot com 2005-03-13
07:11 ---
Compiling the gcse_test.c on powerpc-apple-darwin7.6.0 with the following
options:
-O3 -mcpu=G5 -fprofile-generate
causes the following ICE :
gcse_test.c: In function 'alloc_and_load_edges_and_swi
--- Additional Comments From mustafa at il dot ibm dot com 2005-03-13
07:09 ---
Created an attachment (id=8381)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=8381&action=view)
gcse_test.c
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20450
--
Summary: ICE in postreload-gcse
Product: gcc
Version: 4.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: rtl-optimization
AssignedTo: mustafa at il dot ibm dot com
ReportedBy: mustafa at il dot ibm dot com
--- Additional Comments From mustafa at il dot ibm dot com 2004-12-24
12:13 ---
Doing loop header copying in RTL works around this problem, doing so
gain us improvements of 3.9% for SPECINT on a G5 machine and a 66%
for gcc benchmark.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id
11 matches
Mail list logo
