[Bug target/92665] [AArch64] low lanes select not optimized out for vmlal intrinsics
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92665 --- Comment #7 from Sebastian Pop --- Hi Andrew, have you committed the fix for this?
[Bug rtl-optimization/92665] New: [AArch64] low lanes select not optimized out for vmlal intrinsics
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92665
Bug ID: 92665
Summary: [AArch64] low lanes select not optimized out for vmlal
intrinsics
Product: gcc
Version: unknown
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: rtl-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: spop at gcc dot gnu.org
Target Milestone: ---
With gcc as of today I see dup instructions that could be optimized out:
$ cat red.c
#include "arm_neon.h"
int32x4_t fun(int32x4_t a, int16x8_t b, int16x8_t c) {
a = vmlal_s16(a, vget_low_s16(b), vget_low_s16(c));
a = vmlal_high_s16(a, b, c);
return a;
}
$ gcc -O3 -S -o- red.c
fun:
dup d3, v1.d[0]
dup d4, v2.d[0]
smlal v0.4s,v3.4h,v4.4h
smlal2 v0.4s,v1.8h,v2.8h
ret
$ clang -O3 -S -o- red.c
fun:
smlal v0.4s, v1.4h, v2.4h
smlal2 v0.4s, v1.8h, v2.8h
ret
[Bug target/92692] Saving off the callee saved register between ldxr/stxr (caused by shrink wrapping improvements)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92692
Sebastian Pop changed:
What|Removed |Added
CC||spop at gcc dot gnu.org
--- Comment #21 from Sebastian Pop ---
It looks like this hunk from the trunk version of the patch is missing on gcc-9
branch:
diff --git a/gcc/config/aarch64/atomics.md b/gcc/config/aarch64/atomics.md
index cabcc58f1a0..1458bc00095 100644
--- a/gcc/config/aarch64/atomics.md
+++ b/gcc/config/aarch64/atomics.md
@@ -104,7 +104,7 @@
(clobber (match_scratch:SI 7 "=&r"))]
""
"#"
- "&& reload_completed"
+ "&& epilogue_completed"
[(const_int 0)]
{
aarch64_split_compare_and_swap (operands);
With this hunk applied my bootstrap passes on the gcc-9 branch on an
aarch64-linux graviton2.
Without this hunk I see an error in thread sanitizers.
I also have checked gcc-8 release branch and it seems that the patch is not
missing any hunks in that branch.
Could somebody apply the missing hunk to the gcc-9 release branch? Thanks!
[Bug target/92692] Saving off the callee saved register between ldxr/stxr (caused by shrink wrapping improvements)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92692 --- Comment #23 from Sebastian Pop --- > I don't see anything like that on the gcc-9 branch - are you sure you don't > have an outstanding change somehow? You are right, a part of the -moutline-atomics patch that I am working on backporting to branch 9 added that change.
[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230 --- Comment #24 from Sebastian Pop 2010-12-16 22:30:35 UTC --- *** Bug 45406 has been marked as a duplicate of this bug. ***
[Bug middle-end/45406] ICE: in rename_uses, at sese.c:534
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45406 Sebastian Pop changed: What|Removed |Added Status|WAITING |RESOLVED Resolution||DUPLICATE --- Comment #7 from Sebastian Pop 2010-12-16 22:30:35 UTC --- Fixed. As previously mentioned, this is a dup of PR45230. *** This bug has been marked as a duplicate of bug 45230 ***
[Bug middle-end/46924] [4.6 Regression] ICE: SIGSEGV in rewrite_commutative_reductions_out_of_ssa_loop (graphite-poly.h:450) with -fgraphite-identity -ffast-math -fno-tree-loop-im
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46924 --- Comment #3 from Sebastian Pop 2010-12-16 22:54:24 UTC --- Author: spop Date: Thu Dec 16 22:54:17 2010 New Revision: 167962 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=167962 Log: Fix PR46924: Do not detect reductions outside the current SESE region. 2010-12-16 Sebastian Pop PR tree-optimization/46924 * graphite-sese-to-poly.c (detect_commutative_reduction): Do not detect reductions outside the current SESE region. * sese.h (stmt_in_sese_p): New. (defined_in_sese_p): Call stmt_in_sese_p. * gcc.dg/graphite/pr46924.c: New. Added: trunk/gcc/testsuite/gcc.dg/graphite/pr46924.c Modified: trunk/gcc/ChangeLog trunk/gcc/graphite-sese-to-poly.c trunk/gcc/sese.h trunk/gcc/testsuite/ChangeLog
[Bug middle-end/46924] [4.6 Regression] ICE: SIGSEGV in rewrite_commutative_reductions_out_of_ssa_loop (graphite-poly.h:450) with -fgraphite-identity -ffast-math -fno-tree-loop-im
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46924 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #4 from Sebastian Pop 2010-12-16 22:55:14 UTC --- Fixed.
[Bug tree-optimization/45552] [graphite] ICE in sese_loop_depth, at sese.h:172
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45552 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||DUPLICATE --- Comment #5 from Sebastian Pop 2010-12-17 22:23:21 UTC --- This is a duplicate of PR45758 that has been fixed on trunk. I backported this change to the 4.5 branch and that fixes the testcase. I will commit the fix after regstrap. *** This bug has been marked as a duplicate of bug 45758 ***
[Bug middle-end/45758] [4.6 Regression] ICE in expr_invariant_in_loop_p
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45758 Sebastian Pop changed: What|Removed |Added CC||tolkiendili at gmail dot ||com --- Comment #11 from Sebastian Pop 2010-12-17 22:23:21 UTC --- *** Bug 45552 has been marked as a duplicate of this bug. ***
[Bug tree-optimization/47002] [4.6 Regression] segmentation fault in find_uses_to_rename_use
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47002 --- Comment #2 from Sebastian Pop 2010-12-20 05:23:29 UTC --- I cannot reproduce the error on amd64-linux with "./cc1 -O3 bug.i" Please report the exact flags and the architecture. However I can see several unrelated memory leaks (i.e., not the one reported) with valgrind --leak-check=full ./cc1 -O3 bug.i
[Bug tree-optimization/47002] [4.6 Regression] segmentation fault in find_uses_to_rename_use
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47002 --- Comment #3 from Sebastian Pop 2010-12-20 05:25:00 UTC --- I tested tr...@168000.
[Bug tree-optimization/47021] New: graphite branch fails to bootstrap
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47021
Summary: graphite branch fails to bootstrap
Product: gcc
Version: 4.6.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
AssignedTo: unassig...@gcc.gnu.org
ReportedBy: s...@gcc.gnu.org
Last message error on the auto-tester is while building the libstdc++:
libtool: compile:
/n/16/grosser/daily_git_builds/obj/2010_12_18_00_03_09/./gcc/xgcc
-shared-libgcc -B/n/16/grosser/daily_git_builds/obj/2010_12_18_00_03_09/./gcc
-nostdinc++
-L/n/16/grosser/daily_git_builds/obj/2010_12_18_00_03_09/x86_64-unknown-linux-gnu/32/libstdc++-v3/src
-L/n/16/grosser/daily_git_builds/obj/2010_12_18_00_03_09/x86_64-unknown-linux-gnu/32/libstdc++-v3/src/.libs
-B/n/16/grosser/daily_git_builds/install/2010_12_18_00_03_09/x86_64-unknown-linux-gnu/bin/
-B/n/16/grosser/daily_git_builds/install/2010_12_18_00_03_09/x86_64-unknown-linux-gnu/lib/
-isystem
/n/16/grosser/daily_git_builds/install/2010_12_18_00_03_09/x86_64-unknown-linux-gnu/include
-isystem
/n/16/grosser/daily_git_builds/install/2010_12_18_00_03_09/x86_64-unknown-linux-gnu/sys-include
-m32
-I/n/16/grosser/daily_git_builds/obj/2010_12_18_00_03_09/x86_64-unknown-linux-gnu/32/libstdc++-v3/include/x86_64-unknown-linux-gnu
-I/n/16/grosser/daily_git_builds/obj/2010_12_18_00_03_09/x86_64-unknown-linux-gnu/32/libstdc++-v3/include
-I/n/16/grosser/daily_git_builds/tmp_src/2010_12_18_00_03_09/libstdc++-v3/libsupc++
-fno-implicit-templates -Wall -Wextra -Wwrite-strings -Wcast-qual
-fdiagnostics-show-location=once -ffunction-sections -fdata-sections -g -O2
-D_GNU_SOURCE -m32 -std=gnu++0x -c
/n/16/grosser/daily_git_builds/tmp_src/2010_12_18_00_03_09/libstdc++-v3/src/debug.cc
-fPIC -DPIC -o .libs/debug.o
/n/16/grosser/daily_git_builds/tmp_src/2010_12_18_00_03_09/libstdc++-v3/src/debug.cc:
In function ‘__gnu_cxx::__mutex& {anonymous}::get_safe_base_mutex(void*)’:
/n/16/grosser/daily_git_builds/tmp_src/2010_12_18_00_03_09/libstdc++-v3/src/debug.cc:45:3:
internal compiler error: in scan_tree_for_params, at
graphite-sese-to-poly.c:851
Please submit a full bug report,
with preprocessed source if appropriate.
See for instructions.
[Bug tree-optimization/47021] graphite branch fails to bootstrap
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47021 --- Comment #1 from Sebastian Pop 2010-12-21 22:46:19 UTC --- Author: spop Date: Tue Dec 21 22:46:05 2010 New Revision: 168138 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168138 Log: Fix PR47021: ADDR_EXPRs don't contain SCoP parameters. 2010-12-21 Sebastian Pop PR tree-optimization/47021 * graphite-sese-to-poly.c (scan_tree_for_params): Handle ADDR_EXPR. Modified: branches/graphite/gcc/ChangeLog.graphite branches/graphite/gcc/graphite-sese-to-poly.c
[Bug tree-optimization/47021] graphite branch fails to bootstrap
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47021 --- Comment #2 from Sebastian Pop 2010-12-21 22:48:13 UTC --- Fixed in the graphite branch. I will close the PR when the patch will be committed to trunk.
[Bug tree-optimization/47002] [4.6 Regression] segmentation fault in find_uses_to_rename_use
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47002 Sebastian Pop changed: What|Removed |Added Status|NEW |WAITING --- Comment #4 from Sebastian Pop 2010-12-22 06:23:54 UTC --- Patch fixing the memory leaks related to loop distribution: http://gcc.gnu.org/ml/gcc-patches/2010-12/msg01670.html I still cannot reproduce the error reported in this PR. HJ, it seems like you were able to see the ICE. Could you please specify which flags and architecture you used? Thanks.
[Bug bootstrap/45146] Bootstrap broken at -O3
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45146 Sebastian Pop changed: What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution||FIXED --- Comment #1 from Sebastian Pop 2010-12-22 16:15:28 UTC --- Trunk at r168000 passes bootstrap at -O3.
[Bug middle-end/47019] [4.6 Regression] ICE: in rename_uses, at sese.c:535 with -O -ftree-pre -fgraphite-identity -fno-tree-copy-prop
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47019 Sebastian Pop changed: What|Removed |Added Status|UNCONFIRMED |ASSIGNED Last reconfirmed||2010.12.22 16:55:09 AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | Ever Confirmed|0 |1 --- Comment #1 from Sebastian Pop 2010-12-22 16:55:09 UTC --- This is due to the fact that the SSA representation is not in loop-closed form: we are defining and using an SSA_NAME in two different loops without a loop-close-phi node in between: loop_1 pretmp.15_60 = offset.3_18 + pretmp.15_59; end_1 loop_2 D.1564_26 = pretmp.15_60; end_2
[Bug middle-end/47019] [4.6 Regression] ICE: in rename_uses, at sese.c:535 with -O -ftree-pre -fgraphite-identity -fno-tree-copy-prop
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47019 --- Comment #2 from Sebastian Pop 2010-12-22 17:07:27 UTC --- The SSA is in loop-closed form, the two loops are nested: loop_1 D_60 = ... loop_2 ... = D_60 end_2 end_1 The problem is elsewhere. I'm still investigating.
[Bug middle-end/47019] [4.6 Regression] ICE: in rename_uses, at sese.c:535 with -O -ftree-pre -fgraphite-identity -fno-tree-copy-prop
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47019 --- Comment #3 from Sebastian Pop 2010-12-22 20:00:00 UTC --- Patch: http://gcc.gnu.org/ml/gcc-patches/2010-12/msg01725.html
[Bug middle-end/47019] [4.6 Regression] ICE: in rename_uses, at sese.c:535 with -O -ftree-pre -fgraphite-identity -fno-tree-copy-prop
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47019 --- Comment #4 from Sebastian Pop 2010-12-22 21:19:38 UTC --- Author: spop Date: Wed Dec 22 21:19:34 2010 New Revision: 168182 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168182 Log: Fix PR47019: parameters of a SCoP are SCEV analyzable. 2010-12-22 Sebastian Pop PR tree-optimization/47019 * sese.h (scev_analyzable_p): Parameters of a SCoP are SCEV analyzable. * gfortran.dg/graphite/pr47019.f: New. Added: trunk/gcc/testsuite/gfortran.dg/graphite/pr47019.f Modified: trunk/gcc/ChangeLog trunk/gcc/sese.h trunk/gcc/testsuite/ChangeLog
[Bug middle-end/47019] [4.6 Regression] ICE: in rename_uses, at sese.c:535 with -O -ftree-pre -fgraphite-identity -fno-tree-copy-prop
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47019 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #5 from Sebastian Pop 2010-12-22 21:21:01 UTC --- Fixed.
[Bug middle-end/46758] [4.5/4.6 Regression] -fgraphite-identity produces wrong code when using 64bit constants
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46758 Sebastian Pop changed: What|Removed |Added Status|NEW |ASSIGNED AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | --- Comment #6 from Sebastian Pop 2010-12-23 00:19:19 UTC --- Patch: http://gcc.gnu.org/ml/gcc-patches/2010-12/msg01743.html
[Bug tree-optimization/47002] [4.6 Regression] segmentation fault in find_uses_to_rename_use
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47002 --- Comment #5 from Sebastian Pop 2010-12-23 16:25:59 UTC --- Author: spop Date: Thu Dec 23 16:25:52 2010 New Revision: 168210 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168210 Log: Fix PR47002: memory leaks. 2010-12-23 Sebastian Pop PR tree-optimization/47002 * tree-data-ref.c (compute_data_dependences_for_loop): Pass in a pointer to the loop_nest. (analyze_all_data_dependences): Initialize and free the loop_nest. (free_dependence_relations): Do not free loop_nest. (build_rdg): Pass in the loop_nest, datarefs, and dependence_relations. (free_rdg): Also free the data on edges. * tree-data-ref.h (build_rdg): Update declaration. (compute_data_dependences_for_loop): Same. * tree-if-conv.c (if_convertible_loop_p_1): Pass in the loop_nest. (if_convertible_loop_p): Allocate and free loop_nest. * tree-loop-distribution.c (rdg_flag_loop_exits): Free conds. (free_rdg_components): VEC_free components. (distribute_loop): Update call to build_rdg. Allocate and free loop_nest, datarefs, and dependence_relations. * tree-loop-linear.c (linear_transform_loops): Allocate and free loop_nest. * tree-parloops.c (loop_parallel_p): Same. * tree-predcom.c (tree_predictive_commoning_loop): Same. * tree-vect-data-refs.c (vect_analyze_data_refs): Pass to compute_data_dependences_for_loop a pointer to LOOP_VINFO_LOOP_NEST. * tree-vect-loop.c (new_loop_vec_info): Initialize LOOP_VINFO_LOOP_NEST. (destroy_loop_vec_info): Free LOOP_VINFO_MAY_ALIAS_DDRS and LOOP_VINFO_LOOP_NEST. * tree-vect-slp.c (destroy_bb_vec_info): Call free_data_refs and free_dependence_relations. * tree-vectorizer.h (struct _loop_vec_info): Add a field loop_nest. (LOOP_VINFO_LOOP_NEST): New. Modified: trunk/gcc/ChangeLog trunk/gcc/tree-data-ref.c trunk/gcc/tree-data-ref.h trunk/gcc/tree-if-conv.c trunk/gcc/tree-loop-distribution.c trunk/gcc/tree-loop-linear.c trunk/gcc/tree-parloops.c trunk/gcc/tree-predcom.c trunk/gcc/tree-vect-data-refs.c trunk/gcc/tree-vect-loop.c trunk/gcc/tree-vect-slp.c trunk/gcc/tree-vectorizer.h
[Bug middle-end/46758] [4.5/4.6 Regression] -fgraphite-identity produces wrong code when using 64bit constants
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46758 --- Comment #7 from Sebastian Pop 2010-12-23 16:26:17 UTC --- Author: spop Date: Thu Dec 23 16:26:11 2010 New Revision: 168211 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168211 Log: Fix PR46758: Do not use int_cst_value. 2010-12-23 Sebastian Pop Richard Guenther PR tree-optimization/46758 * graphite-sese-to-poly.c (scan_tree_for_params_right_scev): Use tree_int_to_gmp instead of int_cst_value. (scan_tree_for_params_int): Same. (scan_tree_for_params): Same. (pdr_add_data_dimensions): Use ppl_set_inhomogeneous_tree. * gcc.dg/graphite/run-id-pr46758.c: New. Added: trunk/gcc/testsuite/gcc.dg/graphite/run-id-pr46758.c Modified: trunk/gcc/ChangeLog trunk/gcc/graphite-sese-to-poly.c trunk/gcc/testsuite/ChangeLog
[Bug middle-end/46758] [4.5/4.6 Regression] -fgraphite-identity produces wrong code when using 64bit constants
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46758 Sebastian Pop changed: What|Removed |Added Known to work||4.6.0 Version|4.6.0 |4.5.3 Known to fail|4.6.0 | --- Comment #8 from Sebastian Pop 2010-12-23 16:35:00 UTC --- Fixed on 4.6, I am backporting this to 4.5.
[Bug tree-optimization/46029] -ftree-loop-if-convert-stores causes FAIL: libstdc++-v3/testsuite/ext/pb_ds/example/tree_intervals.cc
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46029 Sebastian Pop changed: What|Removed |Added Status|UNCONFIRMED |ASSIGNED Last reconfirmed||2010.12.23 17:20:47 AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | Ever Confirmed|0 |1 --- Comment #3 from Sebastian Pop 2010-12-23 17:20:47 UTC --- This bug is fixed by the rewrite of the if-convert-stores patch: http://gcc.gnu.org/ml/gcc-patches/2010-11/msg00304.html
[Bug tree-optimization/43023] missing SSA_NAME def for -ftree-loop-distribution in 459.GemsFDTD
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43023 --- Comment #6 from Sebastian Pop 2010-12-23 18:51:55 UTC --- Author: spop Date: Thu Dec 23 18:51:51 2010 New Revision: 168212 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168212 Log: Fix PR43023: fuse_partitions_with_similar_memory_accesses. 2010-12-23 Sebastian Pop PR tree-optimization/43023 * tree-data-ref.c (mem_write_stride_of_same_size_as_unit_type_p): Removed. (stores_zero_from_loop): Call stmt_stores_zero. (stmt_with_adjacent_zero_store_dr_p): New. * tree-data-ref.h (stmt_with_adjacent_zero_store_dr_p): Declared. (stride_of_unit_type_p): New. * tree-loop-distribution.c (generate_memset_zero): Do not return a boolean. Call gcc_assert on stride_of_unit_type_p. (generate_builtin): Call stmt_stores_zero. (rdg_flag_all_uses): Removed. (rdg_flag_similar_memory_accesses): Removed. (build_rdg_partition_for_component): Removed parameter other_stores. Removed call to rdg_flag_similar_memory_accesses. (can_generate_builtin): New. (similar_memory_accesses): New. (fuse_partitions_with_similar_memory_accesses): New. (rdg_build_partitions): Call fuse_partitions_with_similar_memory_accesses. * gfortran.dg/ldist-1.f90: Adjust pattern. * gfortran.dg/ldist-pr43023.f90: New. Added: branches/gcc-4_5-branch/gcc/testsuite/gfortran.dg/ldist-pr43023.f90 Modified: branches/gcc-4_5-branch/gcc/ChangeLog branches/gcc-4_5-branch/gcc/testsuite/ChangeLog branches/gcc-4_5-branch/gcc/testsuite/gfortran.dg/ldist-1.f90 branches/gcc-4_5-branch/gcc/tree-data-ref.c branches/gcc-4_5-branch/gcc/tree-data-ref.h branches/gcc-4_5-branch/gcc/tree-loop-distribution.c
[Bug middle-end/46758] [4.5/4.6 Regression] -fgraphite-identity produces wrong code when using 64bit constants
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46758 --- Comment #9 from Sebastian Pop 2010-12-23 18:52:15 UTC --- Author: spop Date: Thu Dec 23 18:52:12 2010 New Revision: 168214 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168214 Log: Fix PR46758: Do not use int_cst_value. 2010-12-23 Sebastian Pop Richard Guenther PR tree-optimization/46758 * graphite-sese-to-poly.c (scan_tree_for_params_right_scev): Use tree_int_to_gmp instead of int_cst_value. (scan_tree_for_params_int): Same. (scan_tree_for_params): Same. (pdr_add_data_dimensions): Use ppl_set_inhomogeneous_tree. * gcc.dg/graphite/run-id-pr46758.c: New. Added: branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/graphite/run-id-pr46758.c Modified: branches/gcc-4_5-branch/gcc/ChangeLog branches/gcc-4_5-branch/gcc/graphite-sese-to-poly.c branches/gcc-4_5-branch/gcc/testsuite/ChangeLog
[Bug tree-optimization/45552] [graphite] ICE in sese_loop_depth, at sese.h:172
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45552 --- Comment #6 from Sebastian Pop 2010-12-23 18:52:08 UTC --- Author: spop Date: Thu Dec 23 18:52:04 2010 New Revision: 168213 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168213 Log: Fix PR45552: backport fix for PR45758 to 4.5 branch. 2010-12-23 Sebastian Pop Backport from mainline Fix PR45758: reset scevs before Graphite. 2010-09-24 Sebastian Pop PR tree-optimization/45552 * graphite.c (graphite_initialize): Call scev_reset. * gcc.dg/graphite/pr45552.c Added: branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/graphite/pr45552.c Modified: branches/gcc-4_5-branch/gcc/ChangeLog branches/gcc-4_5-branch/gcc/graphite.c branches/gcc-4_5-branch/gcc/testsuite/ChangeLog
[Bug tree-optimization/43023] missing SSA_NAME def for -ftree-loop-distribution in 459.GemsFDTD
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43023 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #7 from Sebastian Pop 2010-12-23 18:53:01 UTC --- Fixed.
[Bug middle-end/46758] [4.5/4.6 Regression] -fgraphite-identity produces wrong code when using 64bit constants
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46758 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #10 from Sebastian Pop 2010-12-23 18:53:48 UTC --- Fixed.
[Bug middle-end/47033] loop distribution has problems on sane testcases
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47033 --- Comment #2 from Sebastian Pop 2010-12-29 05:15:19 UTC --- Is the problem that the testcase is too fragile and we have to fix the pattern expected by the testcase? I suppose we can just disable the scan for memset and leave the testcase test for no ICE while compiling it.
[Bug tree-optimization/47021] graphite branch fails to bootstrap
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47021 --- Comment #3 from Sebastian Pop 2011-01-03 17:01:39 UTC --- Author: spop Date: Mon Jan 3 17:01:36 2011 New Revision: 168433 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168433 Log: Fix PR47021: ADDR_EXPRs don't contain SCoP parameters. 2011-01-03 Sebastian Pop PR tree-optimization/47021 * graphite-sese-to-poly.c (scan_tree_for_params): Handle ADDR_EXPR. Modified: trunk/gcc/ChangeLog trunk/gcc/ChangeLog.graphite trunk/gcc/graphite-sese-to-poly.c
[Bug tree-optimization/47021] graphite branch fails to bootstrap
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47021 Sebastian Pop changed: What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution||FIXED --- Comment #4 from Sebastian Pop 2011-01-03 17:02:15 UTC --- Fixed.
[Bug tree-optimization/45313] [graphite] interchange-7.c is miscompiled
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45313 Sebastian Pop changed: What|Removed |Added Status|NEW |ASSIGNED --- Comment #1 from Sebastian Pop 2011-01-08 05:24:31 UTC --- This still fails on trunk r168433. The miscompile seems to be linked to -floop-strip-mine: $ gcc -O2 -floop-strip-mine interchange-7.c $ ./a.out Aborted
[Bug tree-optimization/47271] [4.6 Regression] if-conversion removes a test (if), the function generates invalid outputs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47271 Sebastian Pop changed: What|Removed |Added Status|NEW |ASSIGNED AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | --- Comment #4 from Sebastian Pop 2011-01-12 22:40:37 UTC --- Mine
[Bug tree-optimization/47271] [4.6 Regression] if-conversion removes a test (if), the function generates invalid outputs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47271
--- Comment #5 from Sebastian Pop 2011-01-18 18:45:27
UTC ---
In this loop:
for (i=0, nops=0 ; i= 90) ? 3 : 1))
{
addrmap[i] = i - nops;
if (codestr[i] == 9)
nops++;
}
it looks like this part: i += ((codestr[i] >= 90) ? 3 : 1)
is miscompiled into:
# i_35 = PHI
iftmp.0_4 = [cond_expr] D.2702_12 != 9 ? 3 : 1;
i_19 = iftmp.0_4 + i_35;
the predicate for the increment is wrongly computed.
[Bug tree-optimization/43567] linear loop transform
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43567 --- Comment #5 from Sebastian Pop 2011-01-18 20:47:10 UTC --- *** Bug 43568 has been marked as a duplicate of this bug. ***
[Bug tree-optimization/43568] linear loop transform
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43568 Sebastian Pop changed: What|Removed |Added Status|UNCONFIRMED |RESOLVED CC||spop at gcc dot gnu.org Resolution||DUPLICATE --- Comment #2 from Sebastian Pop 2011-01-18 20:47:10 UTC --- Indeed this is a duplicate. *** This bug has been marked as a duplicate of bug 43567 ***
[Bug tree-optimization/43567] linear loop transform
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43567 --- Comment #6 from Sebastian Pop 2011-01-18 20:54:23 UTC --- Author: spop Date: Tue Jan 18 20:54:18 2011 New Revision: 168962 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168962 Log: Add testcase for PR43567. 2011-01-18 Sebastian Pop PR tree-optimization/43567 * gcc.dg/graphite/pr43567.c: New. Added: branches/graphite/gcc/testsuite/gcc.dg/graphite/pr43567.c Modified: branches/graphite/gcc/ChangeLog.graphite
[Bug tree-optimization/29832] -ftree-loop-linear miscompiles scalarize.f90
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29832 --- Comment #8 from Sebastian Pop 2011-01-18 20:54:30 UTC --- Author: spop Date: Tue Jan 18 20:54:26 2011 New Revision: 168963 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168963 Log: Add testcase for PR29832. 2011-01-18 Sebastian Pop PR tree-optimization/29832 * gfortran.dg/graphite/pr29832.f90: New. Added: branches/graphite/gcc/testsuite/gfortran.dg/graphite/pr29832.f90 Modified: branches/graphite/gcc/ChangeLog.graphite
[Bug tree-optimization/43657] [4.3/4.4/4.5/4.6 Regression] -ftree-loop-linear causes FAIL: gcc.dg/vect/vect-cond-5.c execution test
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43657 --- Comment #7 from Sebastian Pop 2011-01-18 20:54:38 UTC --- Author: spop Date: Tue Jan 18 20:54:34 2011 New Revision: 168964 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168964 Log: Add testcase for PR43657. 2011-01-18 Sebastian Pop PR tree-optimization/43657 * gcc.dg/graphite/pr43657.c: New. Added: branches/graphite/gcc/testsuite/gcc.dg/graphite/pr43657.c Modified: branches/graphite/gcc/ChangeLog.graphite
[Bug tree-optimization/46168] [4.3/4.4/4.5/4.6 Regression] ICE: verify_ssa failed: definition in block 6 does not dominate use in block 5 with -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46168 --- Comment #3 from Sebastian Pop 2011-01-18 20:54:43 UTC --- Author: spop Date: Tue Jan 18 20:54:40 2011 New Revision: 168965 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168965 Log: Add testcase for PR46168. 2011-01-18 Sebastian Pop PR tree-optimization/46168 * gcc.dg/graphite/pr46168.c: New. Added: branches/graphite/gcc/testsuite/gcc.dg/graphite/pr46168.c Modified: branches/graphite/gcc/ChangeLog.graphite
[Bug tree-optimization/46215] ICE: verify_stmts failed: type mismatch in comparison expression with -fstrict-overflow -fno-tree-ch -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46215 --- Comment #1 from Sebastian Pop 2011-01-18 20:54:50 UTC --- Author: spop Date: Tue Jan 18 20:54:46 2011 New Revision: 168966 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168966 Log: Add testcase for PR46215. 2011-01-18 Sebastian Pop PR tree-optimization/46215 * gcc.dg/graphite/pr46215.c: New. Added: branches/graphite/gcc/testsuite/gcc.dg/graphite/pr46215.c Modified: branches/graphite/gcc/ChangeLog.graphite
[Bug tree-optimization/46970] [4.3/4.4/4.5/4.6 Regression] wrong code with -Os -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46970 --- Comment #6 from Sebastian Pop 2011-01-18 20:54:55 UTC --- Author: spop Date: Tue Jan 18 20:54:52 2011 New Revision: 168967 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168967 Log: Add testcase for PR46970. 2011-01-18 Sebastian Pop PR tree-optimization/46970 * gcc.dg/graphite/pr46970.c: New. Added: branches/graphite/gcc/testsuite/gcc.dg/graphite/pr46970.c Modified: branches/graphite/gcc/ChangeLog.graphite
[Bug tree-optimization/43657] [4.3/4.4/4.5/4.6 Regression] -ftree-loop-linear causes FAIL: gcc.dg/vect/vect-cond-5.c execution test
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43657 Sebastian Pop changed: What|Removed |Added AssignedTo|aoliva at gcc dot gnu.org |spop at gcc dot gnu.org --- Comment #8 from Sebastian Pop 2011-01-18 20:55:48 UTC --- Fixed on the graphite branch.
[Bug tree-optimization/46970] [4.3/4.4/4.5/4.6 Regression] wrong code with -Os -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46970 Sebastian Pop changed: What|Removed |Added Status|NEW |ASSIGNED AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | --- Comment #7 from Sebastian Pop 2011-01-18 20:56:30 UTC --- Fixed on the graphite branch.
[Bug tree-optimization/43567] linear loop transform
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43567 Sebastian Pop changed: What|Removed |Added Status|UNCONFIRMED |ASSIGNED Last reconfirmed||2011.01.18 20:57:45 AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | Ever Confirmed|0 |1 --- Comment #7 from Sebastian Pop 2011-01-18 20:57:45 UTC --- Fixed on the graphite branch.
[Bug tree-optimization/46168] [4.3/4.4/4.5/4.6 Regression] ICE: verify_ssa failed: definition in block 6 does not dominate use in block 5 with -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46168 Sebastian Pop changed: What|Removed |Added Status|NEW |ASSIGNED AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | --- Comment #4 from Sebastian Pop 2011-01-18 20:58:04 UTC --- Fixed on the graphite branch.
[Bug tree-optimization/46215] ICE: verify_stmts failed: type mismatch in comparison expression with -fstrict-overflow -fno-tree-ch -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46215 Sebastian Pop changed: What|Removed |Added Status|UNCONFIRMED |ASSIGNED Last reconfirmed||2011.01.18 20:58:20 CC||spop at gcc dot gnu.org AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | Ever Confirmed|0 |1 --- Comment #2 from Sebastian Pop 2011-01-18 20:58:20 UTC --- Fixed on the graphite branch.
[Bug tree-optimization/29832] -ftree-loop-linear miscompiles scalarize.f90
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29832 Sebastian Pop changed: What|Removed |Added Known to fail|| --- Comment #9 from Sebastian Pop 2011-01-18 20:57:25 UTC --- Fixed on the graphite branch.
[Bug tree-optimization/47271] [4.6 Regression] if-conversion removes a test (if), the function generates invalid outputs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47271 --- Comment #6 from Sebastian Pop 2011-01-18 21:18:04 UTC --- This bug is similar to: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23115 PR23115 never has been fixed correctly... Sebastian
[Bug tree-optimization/47271] [4.6 Regression] if-conversion removes a test (if), the function generates invalid outputs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47271 --- Comment #8 from Sebastian Pop 2011-01-24 18:34:38 UTC --- The translation in predicate_all_scalar_phis assumes that the incoming edges OR up to true: it translates a phi node x = phi (y, z) into x = a ? y : z; In the testcase of this PR, the conditions of the two edges is not ORing up to true: we have something like this: x = a ? y : x; x = (!a and b) ? z : x; this is because we have a CFG containing diamond shape, and the write into x does not happen at every execution of the loop: (a and !b) should not modify the value of x. A possible solution is to filter out these cases in predicate_bbs.
[Bug tree-optimization/47271] [4.6 Regression] if-conversion removes a test (if), the function generates invalid outputs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47271 --- Comment #9 from Sebastian Pop 2011-01-24 18:38:13 UTC --- Sorry, (!a and !b) should not modify the value of x.
[Bug tree-optimization/47271] [4.6 Regression] if-conversion removes a test (if), the function generates invalid outputs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47271 --- Comment #10 from Sebastian Pop 2011-01-24 18:55:48 UTC --- The BB of the phi node to be rewritten should post-dominate the BBs in which the arguments of the phi node are defined. The following gcc_assert will ICE when the translation done in predicate_scalar_phi is unsafe: diff --git a/gcc/tree-if-conv.c b/gcc/tree-if-conv.c index cc7ad8a..9a3e2b5 100644 --- a/gcc/tree-if-conv.c +++ b/gcc/tree-if-conv.c @@ -992,6 +992,7 @@ if_convertible_loop_p_1 (struct loop *loop, return false; calculate_dominance_info (CDI_DOMINATORS); + calculate_dominance_info (CDI_POST_DOMINATORS); /* Allow statements that can be handled during if-conversion. */ ifc_bbs = get_loop_body_in_if_conv_order (loop); @@ -1244,6 +1245,15 @@ predicate_scalar_phi (gimple phi, tree cond, arg_1 = gimple_phi_arg_def (phi, 1); } + gcc_assert ((TREE_CODE (arg_0) != SSA_NAME + || dominated_by_p (CDI_POST_DOMINATORS, + gimple_bb (SSA_NAME_DEF_STMT (arg_0)), + bb)) + && (TREE_CODE (arg_1) != SSA_NAME + || dominated_by_p (CDI_POST_DOMINATORS, + gimple_bb (SSA_NAME_DEF_STMT (arg_1)), + bb))); + /* Build new RHS using selected condition and arguments. */ rhs = build3 (COND_EXPR, TREE_TYPE (res), unshare_expr (cond), arg_0, arg_1);
[Bug tree-optimization/47271] [4.6 Regression] if-conversion removes a test (if), the function generates invalid outputs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47271 --- Comment #12 from Sebastian Pop 2011-01-24 19:57:40 UTC --- Created attachment 23106 --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23106 proposed fix Thanks Jakub for thinking out a counter example, what about this fix?
[Bug tree-optimization/47271] [4.6 Regression] if-conversion removes a test (if), the function generates invalid outputs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47271 --- Comment #14 from Sebastian Pop 2011-01-24 20:14:55 UTC --- Right, thanks for catching that: I handled it only in the first place and not in the second: gcc_assert (bb == bb->loop_father->header || bb_postdominates_preds (bb)); I am seeing that this check is redundant, as just before calling predicate_scalar_phi, we have: if (bb == loop->header) continue; As this is in a gcc_assert I don't mind the extra check, just to make it more clear when somebody reads this code again. If so, I will send the amended patch to gcc-patches.
[Bug tree-optimization/47271] [4.6 Regression] if-conversion removes a test (if), the function generates invalid outputs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47271 --- Comment #16 from Sebastian Pop 2011-01-25 14:51:28 UTC --- Author: spop Date: Tue Jan 25 14:51:23 2011 New Revision: 169233 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169233 Log: Fix PR47271: only if-convert full writes. 2011-01-25 Sebastian Pop Jakub Jelinek PR tree-optimization/47271 * tree-if-conv.c (bb_postdominates_preds): New. (if_convertible_bb_p): Call bb_postdominates_preds. (if_convertible_loop_p_1): Compute CDI_POST_DOMINATORS. (predicate_scalar_phi): Call bb_postdominates_preds. * gcc.dg/tree-ssa/ifc-pr47271.c: New. Added: trunk/gcc/testsuite/gcc.dg/tree-ssa/ifc-pr47271.c Modified: trunk/gcc/ChangeLog trunk/gcc/testsuite/ChangeLog trunk/gcc/tree-if-conv.c
[Bug tree-optimization/47271] [4.6 Regression] if-conversion removes a test (if), the function generates invalid outputs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47271 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #17 from Sebastian Pop 2011-01-25 14:54:52 UTC --- Fixed.
[Bug tree-optimization/29832] -ftree-loop-linear miscompiles scalarize.f90
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29832 --- Comment #10 from Sebastian Pop 2011-01-25 21:24:51 UTC --- Author: spop Date: Tue Jan 25 21:24:44 2011 New Revision: 169253 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169253 Log: Add testcase for PR29832. 2011-01-25 Sebastian Pop PR tree-optimization/29832 * gfortran.dg/graphite/pr29832.f90: New. Added: trunk/gcc/testsuite/gfortran.dg/graphite/pr29832.f90 Modified: trunk/gcc/testsuite/ChangeLog
[Bug tree-optimization/46970] [4.3/4.4/4.5/4.6 Regression] wrong code with -Os -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46970 --- Comment #8 from Sebastian Pop 2011-01-25 21:25:35 UTC --- Author: spop Date: Tue Jan 25 21:25:24 2011 New Revision: 169257 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169257 Log: Add testcase for PR46970. 2011-01-18 Sebastian Pop PR tree-optimization/46970 * gcc.dg/graphite/pr46970.c: New. Added: trunk/gcc/testsuite/gcc.dg/graphite/pr46970.c Modified: trunk/gcc/testsuite/ChangeLog
[Bug tree-optimization/43567] linear loop transform
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43567 --- Comment #8 from Sebastian Pop 2011-01-25 21:24:41 UTC --- Author: spop Date: Tue Jan 25 21:24:35 2011 New Revision: 169252 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169252 Log: Add testcase for PR43567. 2011-01-25 Sebastian Pop PR tree-optimization/43567 * gcc.dg/graphite/pr43567.c: New. Added: trunk/gcc/testsuite/gcc.dg/graphite/pr43567.c Modified: trunk/gcc/ChangeLog.graphite trunk/gcc/testsuite/ChangeLog
[Bug tree-optimization/46215] ICE: verify_stmts failed: type mismatch in comparison expression with -fstrict-overflow -fno-tree-ch -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46215 --- Comment #3 from Sebastian Pop 2011-01-25 21:25:26 UTC --- Author: spop Date: Tue Jan 25 21:25:13 2011 New Revision: 169256 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169256 Log: Add testcase for PR46215. 2011-01-18 Sebastian Pop PR tree-optimization/46215 * gcc.dg/graphite/pr46215.c: New. Added: trunk/gcc/testsuite/gcc.dg/graphite/pr46215.c Modified: trunk/gcc/testsuite/ChangeLog
[Bug tree-optimization/43657] [4.3/4.4/4.5/4.6 Regression] -ftree-loop-linear causes FAIL: gcc.dg/vect/vect-cond-5.c execution test
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43657 --- Comment #9 from Sebastian Pop 2011-01-25 21:25:05 UTC --- Author: spop Date: Tue Jan 25 21:24:53 2011 New Revision: 169254 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169254 Log: Add testcase for PR43657. 2011-01-18 Sebastian Pop PR tree-optimization/43657 * gcc.dg/graphite/pr43657.c: New. Added: trunk/gcc/testsuite/gcc.dg/graphite/pr43657.c Modified: trunk/gcc/testsuite/ChangeLog
[Bug tree-optimization/46168] [4.3/4.4/4.5/4.6 Regression] ICE: verify_ssa failed: definition in block 6 does not dominate use in block 5 with -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46168 --- Comment #5 from Sebastian Pop 2011-01-25 21:25:13 UTC --- Author: spop Date: Tue Jan 25 21:25:02 2011 New Revision: 169255 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169255 Log: Add testcase for PR46168. 2011-01-18 Sebastian Pop PR tree-optimization/46168 * gcc.dg/graphite/pr46168.c: New. Added: trunk/gcc/testsuite/gcc.dg/graphite/pr46168.c Modified: trunk/gcc/testsuite/ChangeLog
[Bug tree-optimization/46215] ICE: verify_stmts failed: type mismatch in comparison expression with -fstrict-overflow -fno-tree-ch -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46215 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #4 from Sebastian Pop 2011-01-25 21:28:02 UTC --- Fixed.
[Bug tree-optimization/29832] -ftree-loop-linear miscompiles scalarize.f90
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29832 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #11 from Sebastian Pop 2011-01-25 21:28:32 UTC --- Fixed.
[Bug tree-optimization/43657] [4.3/4.4/4.5/4.6 Regression] -ftree-loop-linear causes FAIL: gcc.dg/vect/vect-cond-5.c execution test
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43657 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #10 from Sebastian Pop 2011-01-25 21:28:55 UTC --- Fixed.
[Bug tree-optimization/46168] [4.3/4.4/4.5/4.6 Regression] ICE: verify_ssa failed: definition in block 6 does not dominate use in block 5 with -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46168 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #6 from Sebastian Pop 2011-01-25 21:28:46 UTC --- Fixed.
[Bug tree-optimization/46970] [4.3/4.4/4.5/4.6 Regression] wrong code with -Os -ftree-loop-linear
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46970 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #9 from Sebastian Pop 2011-01-25 21:28:37 UTC --- Fixed.
[Bug tree-optimization/43567] linear loop transform
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43567 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #9 from Sebastian Pop 2011-01-25 21:28:23 UTC --- Fixed.
[Bug tree-optimization/44303] [graphite] Segmentation fault within CLooG
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44303 --- Comment #4 from Sebastian Pop 2011-01-27 23:17:30 UTC --- The testcase passes when GCC is configured with --enable-cloog-backend=isl
[Bug middle-end/40979] induct benchmark 60% slower when compiled with -fgraphite-identity
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40979 --- Comment #11 from Sebastian Pop 2011-01-31 18:12:38 UTC --- Here is a reduced testcase from induct.f90 for the first loop not vectorized with -fgraphite-identity: module mqc_m integer, parameter, private :: longreal = selected_real_kind(15,90) contains subroutine mutual_ind_quad_cir_coil (m, l12) real (kind = longreal), dimension(9), save :: w2gauss, w1gauss real (kind = longreal) :: l12_lower, numerator real (kind = longreal), dimension(3) :: current_vector, coil_current_vec w2gauss(1) = 16.0_longreal/81.0_longreal w1gauss(5) = 0.3302393550_longreal do i = 1, 2*m do j = 1, 9 do k = 1, 9 numerator = w1gauss(j) * w2gauss(k) * & dot_product(coil_current_vec,current_vector) l12_lower = l12_lower + numerator end do end do end do l12 = l12_lower end subroutine mutual_ind_quad_cir_coil end module mqc_m The problem seems to be that graphite introduces a Commutative_Associative_Reduction array that confuses the vectorizer. I looked at how to improve translate_scalar_reduction_to_array in order to avoid the creation of the temporary array, but it seems to be difficult as the result is written to memory under a different type than the reduction itself: l12_lower is a real whereas l12 is an integer: l12_lower_200 = some_computation; # l12_lower_9 = PHI D.1585_43 = (integer(kind=4)) l12_lower_9; # .MEM_48 = VDEF <.MEM_47> *l12_44(D) = D.1585_43; so we cannot use *l12_44(D) as a data reference in the loop to perform the reduction as it does not have the same precision as l12_lower: it seems to me that we cannot avoid creating the temporary array. The solution could be to clean up the temporary arrays after graphite.
[Bug tree-optimization/47561] Error message does not say to which option it refers
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47561 Sebastian Pop changed: What|Removed |Added Status|UNCONFIRMED |ASSIGNED Last reconfirmed||2011.02.01 04:52:02 CC||spop at gcc dot gnu.org AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | Ever Confirmed|0 |1 --- Comment #2 from Sebastian Pop 2011-02-01 04:52:02 UTC --- Patch here: http://gcc.gnu.org/ml/gcc-patches/2011-02/msg5.html
[Bug middle-end/40979] induct benchmark 60% slower when compiled with -fgraphite-identity
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40979 --- Comment #15 from Sebastian Pop 2011-02-01 16:46:54 UTC --- The vectorizer does not apply because it does not match the canonical form of a reduction: here is the reduction after graphite-identity: # l12__lsm.18_179 = PHI S1:l12_lower_188 = l12__lsm.18_179; l12_lower_184 = D.1589_34 + l12_lower_188; S2:l12__lsm.18_154 = l12_lower_184; Without S1 and S2, this would be recognized as a reduction by the vectorizer. Why we end up with the two extra copies? Here is the original code: # l12_lower_5 = PHI l12_lower_36 = D.1589_321 + l12_lower_5; Graphite does the following: l12_lower_5 = *l12_43(D); l12_lower_36 = D.1589_321 + l12_lower_5; *l12_43(D) = l12_lower_36; Note that at this point we cannot construct this code because we use data references and we are in Gimple form: *l12_43(D) = D.1589_321 + *l12_43(D); So I think that the code produced by Graphite is fine, and the problem is in the cleanups that we're doing after: for instance loop invariant motion could be improved to avoid the extra two statements S1 and S2: # l12__lsm.18_179 = PHI S1:l12_lower_188 = l12__lsm.18_179; l12_lower_184 = D.1589_34 + l12_lower_188; S2:l12__lsm.18_154 = l12_lower_184; I also have tried to run pass_rename_ssa_copies but that would just rename the base variable l12__lsm.18 into l12_lower and wait for the out-of-SSA to remove the extra copies. Constant propagation does not help either... any other suggestions?
[Bug tree-optimization/47561] Error message does not say to which option it refers
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47561 --- Comment #3 from Sebastian Pop 2011-02-01 18:38:34 UTC --- Author: spop Date: Tue Feb 1 18:38:31 2011 New Revision: 169496 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169496 Log: Fix PR47561: print the Graphite flags. 2011-02-01 Sebastian Pop PR tree-optimization/47561 * toplev.c (process_options): Print the Graphite flags. Add flag_loop_flatten to the list of options requiring Graphite. Modified: trunk/gcc/ChangeLog trunk/gcc/toplev.c
[Bug tree-optimization/47561] Error message does not say to which option it refers
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47561 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #4 from Sebastian Pop 2011-02-01 18:41:07 UTC --- Fixed.
[Bug middle-end/47576] [4.6 Regression] FAIL: gfortran.dg/graphite/pr42732.f
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47576 Sebastian Pop changed: What|Removed |Added Status|NEW |ASSIGNED CC||spop at gcc dot gnu.org AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | --- Comment #3 from Sebastian Pop 2011-02-01 19:52:23 UTC --- Patch here: http://gcc.gnu.org/ml/gcc-patches/2011-02/msg00064.html
[Bug middle-end/40979] induct benchmark 60% slower when compiled with -fgraphite-identity
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40979 --- Comment #21 from Sebastian Pop 2011-02-01 20:51:31 UTC --- Patch here: http://gcc.gnu.org/ml/gcc-patches/2011-02/msg00070.html
[Bug tree-optimization/47555] [4.4 Regression] Huge memory usage when optimizing
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47555 Sebastian Pop changed: What|Removed |Added Status|NEW |ASSIGNED AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | --- Comment #8 from Sebastian Pop 2011-02-01 19:56:46 UTC --- Patch here: http://gcc.gnu.org/ml/gcc-patches/2011-02/msg00064.html
[Bug middle-end/47048] misc vect.exp failures with -fgraphite-identity enabled at -O2.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47048
Sebastian Pop changed:
What|Removed |Added
Status|UNCONFIRMED |ASSIGNED
Last reconfirmed||2011.02.01 21:15:06
AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org
|gnu.org |
Ever Confirmed|0 |1
--- Comment #2 from Sebastian Pop 2011-02-01 21:15:06
UTC ---
Looks like the classic data dependence test can be improved:
A[0] does not overlap with A[1], and so there is no dependence:
Data ref a:
#(Data Ref:
# bb: 6
# stmt: points[D.2720_36][0] = start$0_8(D);
# ref: points[D.2720_36][0];
# base_object: points
# Access function 0: 0
# Access function 1: (int) {0, +, 1}_2;
#)
Data ref b:
#(Data Ref:
# bb: 6
# stmt: points[D.2720_36][1] = D.2723_40;
# ref: points[D.2720_36][1];
# base_object: points
# Access function 0: 1
# Access function 1: (int) {0, +, 1}_2;
#)
affine dependence test not usable: access function not affine or constant.
(dependence classified: scev_not_known)
)
/home/seb/gcc/trunk/gcc/testsuite/gcc.dg/vect/pr33866.c:13: note: not
vectorized: data ref analysis failed points[D.2720_36][0] = start$0_8(D);
[Bug middle-end/47048] misc vect.exp failures with -fgraphite-identity enabled at -O2.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47048 --- Comment #3 from Sebastian Pop 2011-02-01 23:28:33 UTC --- I think that the problem is (again) the type of the induction variables that we create in graphite: in pr33866.c we have D.2720_37 = (int) graphite_IV.5_34; and this (int) confuses the data reference analysis.
[Bug middle-end/47048] misc vect.exp failures with -fgraphite-identity enabled at -O2.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47048 --- Comment #4 from Sebastian Pop 2011-02-01 23:32:20 UTC --- More specifically, the code that is now confused by the (int) cast is: dr_analyze_innermost (dr); dr_analyze_indices (dr, nest, loop); in create_data_ref.
[Bug bootstrap/47044] bootstrap comparison failure when -O2 defaults on -fgraphite-identity and --enable-build-with-cxx
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47044 --- Comment #5 from Sebastian Pop 2011-02-02 15:52:27 UTC --- Author: spop Date: Wed Feb 2 15:52:21 2011 New Revision: 169531 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169531 Log: Fix PR40979 and PR47044: after LIM call copy_prop and DCE to clean up. 2011-02-02 Sebastian Pop Richard Guenther PR tree-optimization/40979 PR bootstrap/47044 * passes.c (init_optimization_passes): After LIM call copy_prop and DCE to clean up. * tree-ssa-loop.c (pass_graphite_transforms): Add TODO_dump_func. * gcc.dg/graphite/graphite.exp (DEFAULT_VECTCFLAGS): Add -ffast-math. * gcc.dg/graphite/pr35356-2.c: Adjust pattern. * gfortran.dg/graphite/graphite.exp: Run vect_files conditionally to check_vect_support_and_set_flags. * gfortran.dg/graphite/vect-pr40979.f90: New. Added: trunk/gcc/testsuite/gfortran.dg/graphite/vect-pr40979.f90 Modified: trunk/gcc/ChangeLog trunk/gcc/passes.c trunk/gcc/testsuite/ChangeLog trunk/gcc/testsuite/gcc.dg/graphite/graphite.exp trunk/gcc/testsuite/gcc.dg/graphite/pr35356-2.c trunk/gcc/testsuite/gfortran.dg/graphite/graphite.exp trunk/gcc/tree-ssa-loop.c
[Bug middle-end/40979] induct benchmark 60% slower when compiled with -fgraphite-identity
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40979 --- Comment #22 from Sebastian Pop 2011-02-02 15:52:26 UTC --- Author: spop Date: Wed Feb 2 15:52:21 2011 New Revision: 169531 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169531 Log: Fix PR40979 and PR47044: after LIM call copy_prop and DCE to clean up. 2011-02-02 Sebastian Pop Richard Guenther PR tree-optimization/40979 PR bootstrap/47044 * passes.c (init_optimization_passes): After LIM call copy_prop and DCE to clean up. * tree-ssa-loop.c (pass_graphite_transforms): Add TODO_dump_func. * gcc.dg/graphite/graphite.exp (DEFAULT_VECTCFLAGS): Add -ffast-math. * gcc.dg/graphite/pr35356-2.c: Adjust pattern. * gfortran.dg/graphite/graphite.exp: Run vect_files conditionally to check_vect_support_and_set_flags. * gfortran.dg/graphite/vect-pr40979.f90: New. Added: trunk/gcc/testsuite/gfortran.dg/graphite/vect-pr40979.f90 Modified: trunk/gcc/ChangeLog trunk/gcc/passes.c trunk/gcc/testsuite/ChangeLog trunk/gcc/testsuite/gcc.dg/graphite/graphite.exp trunk/gcc/testsuite/gcc.dg/graphite/pr35356-2.c trunk/gcc/testsuite/gfortran.dg/graphite/graphite.exp trunk/gcc/tree-ssa-loop.c
[Bug tree-optimization/47555] [4.4 Regression] Huge memory usage when optimizing
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47555 --- Comment #9 from Sebastian Pop 2011-02-02 15:52:15 UTC --- Author: spop Date: Wed Feb 2 15:52:08 2011 New Revision: 169530 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169530 Log: Fix PR47576 and PR47555: add PARAM_SCEV_MAX_EXPR_COMPLEXITY. 2011-02-02 Sebastian Pop PR tree-optimization/47576 PR tree-optimization/47555 * doc/invoke.texi (scev-max-expr-complexity): Documented. * params.def (PARAM_SCEV_MAX_EXPR_SIZE): Bump the value to 100. (PARAM_SCEV_MAX_EXPR_COMPLEXITY): Declared. * tree-scalar-evolution.c (follow_ssa_edge): Use PARAM_SCEV_MAX_EXPR_COMPLEXITY. Modified: trunk/gcc/ChangeLog trunk/gcc/doc/invoke.texi trunk/gcc/params.def trunk/gcc/tree-scalar-evolution.c
[Bug middle-end/47576] [4.6 Regression] FAIL: gfortran.dg/graphite/pr42732.f
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47576 --- Comment #4 from Sebastian Pop 2011-02-02 15:52:15 UTC --- Author: spop Date: Wed Feb 2 15:52:08 2011 New Revision: 169530 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169530 Log: Fix PR47576 and PR47555: add PARAM_SCEV_MAX_EXPR_COMPLEXITY. 2011-02-02 Sebastian Pop PR tree-optimization/47576 PR tree-optimization/47555 * doc/invoke.texi (scev-max-expr-complexity): Documented. * params.def (PARAM_SCEV_MAX_EXPR_SIZE): Bump the value to 100. (PARAM_SCEV_MAX_EXPR_COMPLEXITY): Declared. * tree-scalar-evolution.c (follow_ssa_edge): Use PARAM_SCEV_MAX_EXPR_COMPLEXITY. Modified: trunk/gcc/ChangeLog trunk/gcc/doc/invoke.texi trunk/gcc/params.def trunk/gcc/tree-scalar-evolution.c
[Bug bootstrap/47044] bootstrap comparison failure when -O2 defaults on -fgraphite-identity and --enable-build-with-cxx
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47044 Sebastian Pop changed: What|Removed |Added Status|NEW |RESOLVED Resolution||FIXED --- Comment #6 from Sebastian Pop 2011-02-02 15:54:17 UTC --- Fixed.
[Bug middle-end/47576] [4.6 Regression] FAIL: gfortran.dg/graphite/pr42732.f
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47576 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #5 from Sebastian Pop 2011-02-02 15:56:23 UTC --- Fixed.
[Bug middle-end/40979] induct benchmark 60% slower when compiled with -fgraphite-identity
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40979 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #23 from Sebastian Pop 2011-02-02 15:59:20 UTC --- Fixed.
[Bug tree-optimization/47555] [4.4 Regression] Huge memory usage when optimizing
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47555 --- Comment #10 from Sebastian Pop 2011-02-02 16:01:34 UTC --- What about backporting this patch to 4.4 and 4.5?
[Bug middle-end/42860] ICE in gcc-4.4.3 with graphite
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42860 Sebastian Pop changed: What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution||WONTFIX --- Comment #5 from Sebastian Pop 2011-02-02 16:23:27 UTC --- I won't fix graphite in 4.4, please use GCC 4.5.
[Bug tree-optimization/47555] [4.4 Regression] Huge memory usage when optimizing
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47555 Sebastian Pop changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #13 from Sebastian Pop 2011-02-02 18:44:29 UTC --- Alright, then let's mark this as fixed.
[Bug tree-optimization/46194] [4.5/4.6 Regression] gcc.dg/graphite/block-0.c FAILs with -ftree-parallelize-loops
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46194 Sebastian Pop changed: What|Removed |Added Status|NEW |ASSIGNED AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | --- Comment #6 from Sebastian Pop 2011-02-03 17:33:52 UTC --- Mine. I will have a look at this.
[Bug tree-optimization/46194] [4.5/4.6 Regression] gcc.dg/graphite/block-0.c FAILs with -ftree-parallelize-loops
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46194
--- Comment #8 from Sebastian Pop 2011-02-03 18:20:56
UTC ---
The problem seems to be in analyze_miv_subscript where I see this note:
/* FIXME: This is a MIV subscript, not yet handled.
Example: (A[{1, +, 1}_1] vs. A[{1, +, 1}_2]) that comes from
(A[i] vs. A[j]).
In the SIV test we had to solve a Diophantine equation with two
variables. In the MIV case we have to solve a Diophantine
equation with 2*n variables (if the subscript uses n IVs).
*/
[Bug tree-optimization/46194] [4.5/4.6 Regression] gcc.dg/graphite/block-0.c FAILs with -ftree-parallelize-loops
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46194 --- Comment #10 from Sebastian Pop 2011-02-04 04:00:22 UTC --- Patch: http://gcc.gnu.org/ml/gcc-patches/2011-02/msg00271.html
[Bug middle-end/45306] ICE (Segmentation fault) while building PyQt with -fgraphite-identity
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45306 Sebastian Pop changed: What|Removed |Added Status|UNCONFIRMED |RESOLVED CC||spop at gcc dot gnu.org Resolution||FIXED --- Comment #7 from Sebastian Pop 2011-02-04 05:58:01 UTC --- trunk@169530 works. I tested the original and reduced testcases with -fgraphite-identity and either -O2 or -O1.
[Bug middle-end/46834] [4.6 Regression] ICE: verify_ssa failed: missing definition with -fgraphite-identity -ffast-math -fno-tree-dce
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46834 Sebastian Pop changed: What|Removed |Added Status|NEW |ASSIGNED AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | --- Comment #2 from Sebastian Pop 2011-02-04 07:13:26 UTC --- Patch: http://gcc.gnu.org/ml/gcc-patches/2011-02/msg00277.html
[Bug tree-optimization/46994] [4.6 Regression] ICE: in check_loop_closed_ssa_use, at tree-ssa-loop-manip.c:422 with -fgraphite-identity -fno-tree-dce -ffast-math
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46994 Sebastian Pop changed: What|Removed |Added Status|NEW |ASSIGNED AssignedTo|unassigned at gcc dot |spop at gcc dot gnu.org |gnu.org | --- Comment #2 from Sebastian Pop 2011-02-04 07:55:56 UTC --- Patch: http://gcc.gnu.org/ml/gcc-patches/2011-02/msg00277.html
