* Re: Notes on the merge
1998-04-04 11:01 Notes on the merge Jeffrey A Law
@ 1998-04-05 10:02 ` Franz Sirl
1998-04-05 21:29 ` Raja R Harinath
1 sibling, 0 replies; 3+ messages in thread
From: Franz Sirl @ 1998-04-05 10:02 UTC (permalink / raw)
To: law, egcs
At 17:31 Uhr -0000 04.04.1998, Jeffrey A Law wrote:
>Kenner made some pretty significant changes to how xm files are
>handled. Hopefully I got configure.in properly merged after these
>changes.
>
>Note in rs6000/sysv4.h, there's a include of linux.h that can
>probably disappear with a minor configure.in tweak.
>
>
>floating point conditional moves on the x86 are disabled and integer
>conditional moves have been simplified. Normally I wouldn't mention
>the simplified integer conditional moves, except we've had trouble
>with them in the past. Hopefully our fixes to jump.c make the
>simplification safe.
>
>
>Kenner made some changes to ABI_32 on the MIPS port. I've asked Jim
>to review them to make sure they're OK.
Hi,
on powerpc-unknown-linux-gnu everything compiled fine and during make
check-gcc I saw no regression.
I have included a gdb-backtrace of a SIGSEGV during compilation of one of
the newer tests (BTW, is the switch-1.c testcase I sent in OK?), looks like
a NULL-pointer dereference in rtlanal.c. The test failed before and after
the merge.
Bye,
Franz.
Failing test:
gcc/testsuite/gcc.c-torture/compile/980329-1.c -w -O2
-fomit-frame-pointer -finline-functions -funroll-all-loops
gdb-backtrace:
GNU gdb 4.16.86
Copyright 1997 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB. Type "show warranty" for details.
This GDB was configured as "powerpc-unknown-linux-gnu"...
Program terminated with signal 11, Segmentation fault.
#0 0x188df54 in single_set (insn=0x0) at ../../gcc/rtlanal.c:620
620 if (GET_RTX_CLASS (GET_CODE (insn)) != 'i')
(gdb) bt
#0 0x188df54 in single_set (insn=0x0) at ../../gcc/rtlanal.c:620
#1 0x18d0d88 in can_reverse_comparison_p (comparison=0x0, insn=0x1a6bba8)
at ../../gcc/jump.c:2886
#2 0x18d2180 in invert_exp (x=0x1a6bb78, insn=0x1a6bba8)
at ../../gcc/jump.c:3846
#3 0x18d224c in invert_exp (x=0x1a6bb68, insn=0x1a6bba8)
at ../../gcc/jump.c:3863
#4 0x18f3290 in copy_loop_body (copy_start=0x1a652b8, copy_end=0x1a653d0,
map=0x7fffcc70, exit_label=0x1a6baf0, last_iteration=0,
unroll_type=UNROLL_NAIVE, start_label=0x1a65278, loop_end=0x1a65538,
insert_before=0x1a65538, copy_notes_from=0x1a65538)
at ../../gcc/unroll.c:1936
#5 0x18f25f4 in unroll_loop (loop_end=0x1a65538, insn_count=712,
loop_start=0x1a652b8, end_insert_before=0x1a65550, strength_reduce_p=0)
at ../../gcc/unroll.c:1239
#6 0x18eb34c in strength_reduce (scan_start=0x1a65278, end=0x1a65538,
loop_top=0x0, insn_count=1, loop_start=0x1a652b8, loop_end=0x1a65538,
unroll_p=1) at ../../gcc/loop.c:3628
#7 0x18e7200 in scan_loop (loop_start=0x1a652b8, end=0x1a65538, nregs=178,
unroll_p=1) at ../../gcc/loop.c:1067
#8 0x18e6188 in loop_optimize (f=0x1a4e3c0, dumpfile=0x1a6bba8, unroll_p=1)
at ../../gcc/loop.c:521
#9 0x1828b0c in rest_of_compilation (decl=0x1a622a8)
at ../../gcc/toplev.c:3278
#10 0x181357c in finish_function (nested=0) at ../../gcc/c-decl.c:7110
#11 0x1802278 in yyparse () at c-parse.y:318
#12 0x1827690 in compile_file (name=0x0) at ../../gcc/toplev.c:2544
#13 0x182a8e0 in main (argc=13, argv=0x7ffffb14, envp=0x1a6bb30)
at ../../gcc/toplev.c:4361
#14 0x1801494 in __start1 (argc=13, argv=0x7ffffb14, envp=0x7ffffb4c,
auxvec=0x7ffffbc0, exitfn=0, stack_on_entry=0x1a30000)
at ../sysdeps/powerpc/elf/start.c:124
(gdb) q
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: Notes on the merge
1998-04-04 11:01 Notes on the merge Jeffrey A Law
1998-04-05 10:02 ` Franz Sirl
@ 1998-04-05 21:29 ` Raja R Harinath
1 sibling, 0 replies; 3+ messages in thread
From: Raja R Harinath @ 1998-04-05 21:29 UTC (permalink / raw)
To: egcs
Jeffrey A Law <law@cygnus.com> writes:
> Kenner made some pretty significant changes to how xm files are
> handled. Hopefully I got configure.in properly merged after these
> changes.
Here's what I needed to get it compiled on Solaris 2.5 (sparc-sun-solaris2.5).
Index: gcc/configure
===================================================================
RCS file: /egcs/carton/cvsfiles/egcs/gcc/configure,v
retrieving revision 1.77
diff -u -p -u -p -r1.77 egcs/gcc/configure
--- egcs/gcc/configure 17:04:23 1.77
+++ egcs/gcc/configure 22:44:16
@@ -4159,6 +4159,7 @@ for machine in $build $host $target; do
tm_file=sparc/sol2-sld.h
fi
xm_file="sparc/xm-sysv4.h sparc/xm-sol2.h"
+ xm_defines="USG POSIX"
tmake_file=sparc/t-sol2
xmake_file=sparc/x-sysv4
extra_parts="crt1.o crti.o crtn.o gcrt1.o gmon.o crtbegin.o crtend.o"
Index: gcc/configure.in
===================================================================
RCS file: /egcs/carton/cvsfiles/egcs/gcc/configure.in,v
retrieving revision 1.77
diff -u -p -u -p -r1.77 egcs/gcc/configure.in
--- egcs/gcc/configure.in 17:04:24 1.77
+++ egcs/gcc/configure.in 22:44:22
@@ -2476,6 +2476,7 @@ for machine in $build $host $target; do
tm_file=sparc/sol2-sld.h
fi
xm_file="sparc/xm-sysv4.h sparc/xm-sol2.h"
+ xm_defines="USG POSIX"
tmake_file=sparc/t-sol2
xmake_file=sparc/x-sysv4
extra_parts="crt1.o crti.o crtn.o gcrt1.o gmon.o crtbegin.o crtend.o"
Index: gcc/config/sparc/xm-sysv4.h
===================================================================
RCS file: /egcs/carton/cvsfiles/egcs/gcc/config/sparc/xm-sysv4.h,v
retrieving revision 1.1.1.2
diff -u -p -u -p -r1.1.1.2 egcs/gcc/config/sparc/xm-sysv4.h
--- egcs/gcc/config/sparc/xm-sysv4.h 05:29:15 1.1.1.2
+++ egcs/gcc/config/sparc/xm-sysv4.h 22:44:22
@@ -35,6 +35,8 @@ Boston, MA 02111-1307, USA. */
matters when cross-compiling. */
#define HOST_WORDS_BIG_ENDIAN 1
+#include "xm-siglist.h"
+
/* target machine dependencies.
tm.h is a symbolic link to the actual target specific file. */
#include "tm.h"
The last diff is to `sparc/xm-sysv4.h' rather than `sparc/xm-sol2.h':
`sparc/xm-sysv4.h' originally included `xm-svr4.h' (deleted by Kenner,
but still in egcs repo), which in turn had the defines `USG' & `POSIX',
and the contents of `xm-siglist.h'.
- Hari
--
Raja R Harinath ------------------------------ harinath@cs.umn.edu
"When all else fails, read the instructions." -- Cahn's Axiom
"Our policy is, when in doubt, do the right thing." -- Roy L Ash
^ permalink raw reply [flat|nested] 3+ messages in thread