[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 Bernd Edlinger changed: What|Removed |Added CC||bernd.edlinger at hotmail dot de --- Comment #11 from Bernd Edlinger --- (In reply to Jakub Jelinek from comment #9) > Is this fixed now? everything except the regression in arm/unaligned-memcpy-2/3.c the patch was considered too ugly: the last version was here (probably still too ugly): https://gcc.gnu.org/ml/gcc-patches/2019-09/msg00370.html but I did not consider this important enough to send any pings.
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 Christophe Lyon changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution|--- |FIXED --- Comment #10 from Christophe Lyon --- Yes, this is fixed.
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 Jakub Jelinek changed: What|Removed |Added CC||jakub at gcc dot gnu.org --- Comment #9 from Jakub Jelinek --- Is this fixed now?
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 --- Comment #8 from Bernd Edlinger --- Author: edlinger Date: Thu Sep 5 13:40:17 2019 New Revision: 275409 URL: https://gcc.gnu.org/viewcvs?rev=275409=gcc=rev Log: 2019-09-05 Bernd Edlinger PR middle-end/91615 * expr.c (expand_expr_real_1): Handle misaligned TARGET_MEM_REF without movmisalign optab. Modified: trunk/gcc/ChangeLog trunk/gcc/expr.c
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 --- Comment #7 from Christophe Lyon --- (In reply to Bernd Edlinger from comment #6) > Created attachment 46820 [details] > untested patch This patch fixes the armeb problems reported here, thanks! (in addition to the scan-assembler-times issues on unaligned-memcpy-[23].c)
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 Bernd Edlinger changed: What|Removed |Added Status|UNCONFIRMED |ASSIGNED Last reconfirmed||2019-09-04 Assignee|unassigned at gcc dot gnu.org |edlinger at gcc dot gnu.org Ever confirmed|0 |1 --- Comment #6 from Bernd Edlinger --- Created attachment 46820 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=46820=edit untested patch
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 --- Comment #5 from Christophe Lyon --- (In reply to Bernd Edlinger from comment #4) > Hi Christophe, > > many thanks for your invaluable help. > > I think except this one all regressions are fixed or > at least understood. > > Unfortunately I have a bit of trouble to reproduce this > could you please give the exact config options > and RUNTESTFLAGS, so I can build a cross to debug this? Sure, nothing fancy; I hoped the options I listed earlier were enough. Here is what I have at the end of gcc.log: Executing on host: /aci-gcc-fsf/builds/gcc-fsf-gccsrc/obj-armeb-none-linux-gnueabihf/gcc3/gcc/xgcc -v(timeout = 300) spawn -ignore SIGHUP /aci-gcc-fsf/builds/gcc-fsf-gccsrc/obj-armeb-none-linux-gnueabihf/gcc3/gcc/xgcc -v Using built-in specs. COLLECT_GCC=/aci-gcc-fsf/builds/gcc-fsf-gccsrc/obj-armeb-none-linux-gnueabihf/gcc3/gcc/xgcc COLLECT_LTO_WRAPPER=/aci-gcc-fsf/builds/gcc-fsf-gccsrc/tools/libexec/gcc/armeb-none-linux-gnueabihf/10.0.0/lto-wrapper Target: armeb-none-linux-gnueabihf Configured with: /configure --target=armeb-none-linux-gnueabihf --prefix=/aci-gcc-fsf/builds/gcc-fsf-gccsrc/tools --with-sysroot=/aci-gcc-fsf/builds/gcc-fsf-gccsrc/sysroot-armeb-none-linux-gnueabihf --disable-nls --disable-libgomp --disable-libmudflap --disable-libcilkrts --enable-checking --enable-languages=c,c++,fortran --with-float=hard --enable-build-with-cxx --with-mode=arm --with-cpu=cortex-a9 --with-fpu=neon-fp16 Thread model: posix Supported LTO compression algorithms: zlib gcc version 10.0.0 20190828 (experimental) [master revision 274985] (GCC) and nothing in RUNTESTFLAGS
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 --- Comment #4 from Bernd Edlinger --- Hi Christophe, many thanks for your invaluable help. I think except this one all regressions are fixed or at least understood. Unfortunately I have a bit of trouble to reproduce this could you please give the exact config options and RUNTESTFLAGS, so I can build a cross to debug this?
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 Richard Biener changed: What|Removed |Added Target Milestone|--- |10.0
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 --- Comment #3 from Bernd Edlinger --- yes that looks very likely. I was not able to reproduce this particular failure, but you can try out the patch I attached to pr91612 and see if it fixes you problem. I am currently short of test capability and cannot post the patch before I have done another bootstrap/regtest cycle.
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 David Binderman changed: What|Removed |Added CC||dcb314 at hotmail dot com --- Comment #2 from David Binderman --- I am not sure if this is related or not, but for a raspberry pi cross compiler (arm-linux-gnueabihf) and gcc/testsuite file gcc.dg/vect/pr57558-2.c, it seems to compile ok at -O2, but not at -O3: ./gcc.dg/vect/pr57558-2.c during RTL pass: expand ./gcc.dg/vect/pr57558-2.c: In function ‘foo’: ./gcc.dg/vect/pr57558-2.c:9:13: internal compiler error: in gen_movdi, at config/arm/arm.md:5257 9 | a[i] = a[i+1]; |~^ 0x768aef gen_movdi(rtx_def*, rtx_def*) /home/dcb/gcc/trunk/gcc/config/arm/arm.md:5257 0xa3a5b7 insn_gen_fn::operator()(rtx_def*, rtx_def*) const /home/dcb/gcc/trunk/gcc/recog.h:318 0xa3a5b7 emit_move_insn_1(rtx_def*, rtx_def*) /home/dcb/gcc/trunk/gcc/expr.c:3694 0xa3a98c emit_move_insn(rtx_def*, rtx_def*) /home/dcb/gcc/trunk/gcc/expr.c:3790
[Bug target/91615] [10 regression][armeb] ICEs since r274986
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91615 --- Comment #1 from Christophe Lyon --- There are also 2 regressions in gfortran --target armeb-none-linux-gnueabihf --with-cpu cortex-a9 --with-fpu neon-fp16 gfortran.dg/vect/no-vfa-pr32377.f90 -O (internal compiler error) gfortran.dg/vect/pr32377.f90 -O (internal compiler error) Same backtrace for both: during RTL pass: expand /gcc/testsuite/gfortran.dg/vect/no-vfa-pr32377.f90:10:0: internal compiler error: in gen_movv4sf, at config/arm/vec-common.md:30 0x12e016c gen_movv4sf(rtx_def*, rtx_def*) /gcc/config/arm/vec-common.md:30 0x90cfe2 insn_gen_fn::operator()(rtx_def*, rtx_def*) const /gcc/recog.h:318 0x90cfe2 emit_move_insn_1(rtx_def*, rtx_def*) /gcc/expr.c:3694 0x90d2d3 emit_move_insn(rtx_def*, rtx_def*) /gcc/expr.c:3790 0x8e80cc copy_to_mode_reg(machine_mode, rtx_def*) /gcc/explow.c:631 0xb81b01 maybe_legitimize_operand /gcc/optabs.c:7246 0xb81b01 maybe_legitimize_operands(insn_code, unsigned int, unsigned int, expand_operand*) /gcc/optabs.c:7378 0xb81e69 maybe_gen_insn(insn_code, unsigned int, expand_operand*) /gcc/optabs.c:7397 0xb88a6a expand_binop_directly /gcc/optabs.c:1122 0xb865b8 expand_binop(machine_mode, optab_tag, rtx_def*, rtx_def*, rtx_def*, int, optab_methods) /gcc/optabs.c:1210 0x8efa0d expand_mult(machine_mode, rtx_def*, rtx_def*, rtx_def*, int, bool) /gcc/expmed.c:3549 0x91a8ed expand_expr_real_2(separate_ops*, rtx_def*, machine_mode, expand_modifier) /gcc/expr.c:8937 0x909d42 expand_expr_real_1(tree_node*, rtx_def*, machine_mode, expand_modifier, rtx_def**, bool) /gcc/expr.c:9948 0x913b1e expand_expr /gcc/expr.h:281 0x913b1e expand_operands(tree_node*, tree_node*, rtx_def*, rtx_def**, rtx_def**, expand_modifier) /gcc/expr.c:7878 0x918c23 expand_expr_real_2(separate_ops*, rtx_def*, machine_mode, expand_modifier) /gcc/expr.c:8739 0x909d42 expand_expr_real_1(tree_node*, rtx_def*, machine_mode, expand_modifier, rtx_def**, bool) /gcc/expr.c:9948 0x914d32 store_expr(tree_node*, rtx_def*, int, bool, bool) /gcc/expr.c:5682 0x916a26 expand_assignment(tree_node*, tree_node*, bool) /gcc/expr.c:5441 0x7d7494 expand_gimple_stmt_1 /gcc/cfgexpand.c:3779
