This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Merging Graphite to trunk
On Thu, Jul 30, 2009 at 03:33:46PM -0500, Sebastian Pop wrote:
> Hi,
>
> I attached a new set of patches that fix the comments from Richi and
> Joseph.
>
> The previous patch-set did passes regstrap with some extra tests
> failing like this:
>
> FAIL: gcc.dg/tree-ssa/20050314-1.c scan-tree-dump-times lim "Moving
> statement" 1: dump file does not exist
>
> and
>
> FAIL: gcc.dg/vect/dump-tree-dceloop-pr26359.c scan-tree-dump-times
> dceloop2 "Deleting : vect_" 0: dump file does not exist
>
> This is linked to the fact that we now schedule one more pass of lim
> and dceloop after the graphite pass. I updated the pattern matching
> in these testcases from:
>
> /* { dg-final { scan-tree-dump-times "Deleting : vect_" 0 "dceloop2" } } */
> /* { dg-final { cleanup-tree-dump "dceloop\[1-2\]" } } */
>
> to
>
> /* { dg-final { scan-tree-dump-times "Deleting : vect_" 0 "dceloop3" } } */
> /* { dg-final { cleanup-tree-dump "dceloop\[1-3\]" } } */
>
> see 0016-Fix-patterns-in-lim-and-dceloop-testcases.patch
> With this patch tree-ssa.exp and vect.exp are clean, and the remaining
> fails different than on trunk are only in the graphite testsuite:
> interchange-8.c and all the force-parallel-*.c testcases are failing
> on my tester with the following error message:
>
> xgcc: libgomp.spec: No such file or directory
> compiler exited with status 1
> output is:
> xgcc: libgomp.spec: No such file or directory
>
> FAIL: gcc.dg/graphite/graphite_autopar/force-parallel-1.c (test for
> excess errors)
> Excess errors:
> xgcc: libgomp.spec: No such file or directory
>
> UNRESOLVED: gcc.dg/graphite/graphite_autopar/force-parallel-1.c
> compilation failed to produce executable
>
> Any idea on how to fix this?
>
> Thanks,
> Sebastian
Sebastian,
When I try to build gcc trunk from 20090725 with the 1662 graphite
patches on x86_64-apple-darwin, the build fails with...
gcc -c -g -fkeep-inline-functions -DIN_GCC -W -Wall -Wwrite-strings -Wcast-qual -Wstrict-prototypes -Wmissing-prototypes -Wmissing-format-attribute -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlength-strings -Wold-style-definition -Wc++-compat -fno-common -DHAVE_CONFIG_H -I. -I. -I../../gcc-4.5-20090725/gcc -I../../gcc-4.5-20090725/gcc/. -I../../gcc-4.5-20090725/gcc/../include -I../../gcc-4.5-20090725/gcc/../libcpp/include -I/sw/include -I../../gcc-4.5-20090725/gcc/../libdecnumber -I../../gcc-4.5-20090725/gcc/../libdecnumber/dpd -I../libdecnumber -I/sw/include -I/sw/include -DCLOOG_PPL_BACKEND -I/sw/include ../../gcc-4.5-20090725/gcc/graphite-clast-to-gimple.c -o graphite-clast-to-gimple.o
../../gcc-4.5-20090725/gcc/graphite-clast-to-gimple.c: In function âgraphite_loop_normal_formâ:
../../gcc-4.5-20090725/gcc/graphite-clast-to-gimple.c:1134: warning: passing argument 2 of âcanonicalize_loop_ivsâ from incompatible pointer type
../../gcc-4.5-20090725/gcc/graphite-clast-to-gimple.c:1134: error: too few arguments to function âcanonicalize_loop_ivsâ
make[3]: *** [graphite-clast-to-gimple.o] Error 1
make[2]: *** [all-stage1-gcc] Error 2
make[1]: *** [stage1-bubble] Error 2
make: *** [all] Error 2
Also it would help to know which revision you created a given set
of patches against. Current gcc trunk doesn't cleanly take
0013-New-implementation-of-Graphite.patch for example.
Jack