Adjust 'libgomp.c++/target-exceptions-pr118794-1.C' for 'targetm.arm_eabi_unwinder' [PR118794] (was: [Linaro-TCWG-CI] gcc-15-9463-gaa3e72f9430: 2 regressions on arm)
Hi!
On 2025-04-17T18:15:50+0000, ci_notify--- via Gcc-regression gcc-regression@gcc.gnu.org wrote:
Our automatic CI has detected problems related to your patch(es). Please find some details below.
In bootstrap_check master-arm-check_bootstrap, after: | commit gcc-15-9463-gaa3e72f9430 | Author: Thomas Schwinge tschwinge@baylibre.com | Date: Thu Mar 27 23:06:37 2025 +0100 | | Add test cases for exception handling constructs in dead code for GCN, nvptx target and OpenMP 'target' offloading [PR118794] | | PR target/118794 | gcc/testsuite/ | * g++.target/gcn/exceptions-pr118794-1.C: New. | ... 7 lines of the commit log omitted.
Produces 2 regressions: | | regressions.sum: | Running libgomp:libgomp.c++/c++.exp ... | FAIL: libgomp.c++/target-exceptions-pr118794-1.C scan-tree-dump-times optimized "gimple_call <__builtin_eh_pointer, " 1 | FAIL: libgomp.c++/target-exceptions-pr118794-1.C scan-tree-dump-times optimized "gimple_call <__builtin_unwind_resume, " 1
Ah, sorry for that. This is due to 'targetm.arm_eabi_unwinder', as per:
gcc/config/arm/arm.cc:#define TARGET_ARM_EABI_UNWINDER true gcc/config/c6x/c6x.cc:#define TARGET_ARM_EABI_UNWINDER true
..., which for ARM is conditional to '#if ARM_UNWIND_INFO' (defined in 'gcc/config/arm/bpabi.h', used for various GCC configurations), and for C6x unconditional.
This gets us:
--- target-exceptions-pr118794-1.C.269t.optimized +++ target-exceptions-pr118794-1.C.270t.optimized [...] __attribute__((omp declare target)) void f () [...] gimple_call <__dt_comp , NULL, &c> - gimple_call <__builtin_eh_pointer, _7, 2> - gimple_call <__builtin_unwind_resume, NULL, _7> + gimple_call <__builtin_cxa_end_cleanup, NULL>
} [...]
There doesn't appear to be an effective-target keyword for 'targetm.arm_eabi_unwinder' specifically, do we need to add one? Or, other test cases appear to use effective-target 'arm_eabi' to disambiguate the two variants; is that the right thing to use here, too? (..., plus 'tic6x-*-*' in this case?) OK to push the attached "Adjust 'libgomp.c++/target-exceptions-pr118794-1.C' for 'targetm.arm_eabi_unwinder' [PR118794]"? (Could Arm/C6x maintainers please test this for me?)
Grüße Thomas
Used configuration : *CI config* tcwg_bootstrap_check master-arm-check_bootstrap *configure and test flags:* none, autodetected on armv8l-unknown-linux-gnueabihf
We track this bug report under https://linaro.atlassian.net/browse/GNU-1562. Please let us know if you have a fix.
If you have any questions regarding this report, please ask on linaro-toolchain@lists.linaro.org mailing list.
-----------------8<--------------------------8<--------------------------8<--------------------------
The information below contains the details of the failures, and the ways to reproduce a debug environment:
You can find the failure logs in *.log.1.xz files in
The full lists of regressions and improvements as well as configure and make commands are in
The list of [ignored] baseline and flaky failures are in
Current build : https://ci.linaro.org/job/tcwg_bootstrap_check--master-arm-check_bootstrap-b... Reference build : https://ci.linaro.org/job/tcwg_bootstrap_check--master-arm-check_bootstrap-b...
Instruction to reproduce the build : https://git-us.linaro.org/toolchain/ci/interesting-commits.git/plain/gcc/sha...
Full commit : https://gcc.gnu.org/git/?p=gcc.git%3Ba=commitdiff%3Bh=aa3e72f943032e5f074b2b...
Hi!
On Tue, 22 Apr 2025 at 13:55, Thomas Schwinge tschwinge@baylibre.com wrote:
Hi!
On 2025-04-17T18:15:50+0000, ci_notify--- via Gcc-regression gcc-regression@gcc.gnu.org wrote:
Our automatic CI has detected problems related to your patch(es). Please find some details below.
In bootstrap_check master-arm-check_bootstrap, after: | commit gcc-15-9463-gaa3e72f9430 | Author: Thomas Schwinge tschwinge@baylibre.com | Date: Thu Mar 27 23:06:37 2025 +0100 | | Add test cases for exception handling constructs in dead code for GCN, nvptx target and OpenMP 'target' offloading [PR118794] | | PR target/118794 | gcc/testsuite/ | * g++.target/gcn/exceptions-pr118794-1.C: New. | ... 7 lines of the commit log omitted.
Produces 2 regressions: | | regressions.sum: | Running libgomp:libgomp.c++/c++.exp ... | FAIL: libgomp.c++/target-exceptions-pr118794-1.C scan-tree-dump-times optimized "gimple_call <__builtin_eh_pointer, " 1 | FAIL: libgomp.c++/target-exceptions-pr118794-1.C scan-tree-dump-times optimized "gimple_call <__builtin_unwind_resume, " 1
Ah, sorry for that. This is due to 'targetm.arm_eabi_unwinder', as per:
gcc/config/arm/arm.cc:#define TARGET_ARM_EABI_UNWINDER true gcc/config/c6x/c6x.cc:#define TARGET_ARM_EABI_UNWINDER true..., which for ARM is conditional to '#if ARM_UNWIND_INFO' (defined in 'gcc/config/arm/bpabi.h', used for various GCC configurations), and for C6x unconditional.
This gets us:
--- target-exceptions-pr118794-1.C.269t.optimized +++ target-exceptions-pr118794-1.C.270t.optimized [...] __attribute__((omp declare target)) void f () [...] gimple_call <__dt_comp , NULL, &c> - gimple_call <__builtin_eh_pointer, _7, 2> - gimple_call <__builtin_unwind_resume, NULL, _7> + gimple_call <__builtin_cxa_end_cleanup, NULL> } [...]There doesn't appear to be an effective-target keyword for 'targetm.arm_eabi_unwinder' specifically, do we need to add one? Or, other test cases appear to use effective-target 'arm_eabi' to disambiguate the two variants; is that the right thing to use here, too? (..., plus 'tic6x-*-*' in this case?) OK to push the attached "Adjust 'libgomp.c++/target-exceptions-pr118794-1.C' for 'targetm.arm_eabi_unwinder' [PR118794]"? (Could Arm/C6x maintainers please test this for me?)
I confirm that on arm-linux-gnueabihf, with your patch: PASS: libgomp.c++/target-exceptions-pr118794-1.C (test for excess errors) PASS: libgomp.c++/target-exceptions-pr118794-1.C execution test PASS: libgomp.c++/target-exceptions-pr118794-1.C scan-tree-dump-times optimized "gimple_call <__builtin_cxa_end_cleanup, " 1 and the two FAILs have disappeared.
Thanks,
Christophe
Grüße Thomas
Used configuration : *CI config* tcwg_bootstrap_check master-arm-check_bootstrap *configure and test flags:* none, autodetected on armv8l-unknown-linux-gnueabihf
We track this bug report under https://linaro.atlassian.net/browse/GNU-1562. Please let us know if you have a fix.
If you have any questions regarding this report, please ask on linaro-toolchain@lists.linaro.org mailing list.
-----------------8<--------------------------8<--------------------------8<--------------------------
The information below contains the details of the failures, and the ways to reproduce a debug environment:
You can find the failure logs in *.log.1.xz files in
The full lists of regressions and improvements as well as configure and make commands are in
The list of [ignored] baseline and flaky failures are in
Current build : https://ci.linaro.org/job/tcwg_bootstrap_check--master-arm-check_bootstrap-b... Reference build : https://ci.linaro.org/job/tcwg_bootstrap_check--master-arm-check_bootstrap-b...
Instruction to reproduce the build : https://git-us.linaro.org/toolchain/ci/interesting-commits.git/plain/gcc/sha...
Full commit : https://gcc.gnu.org/git/?p=gcc.git%3Ba=commitdiff%3Bh=aa3e72f943032e5f074b2b...
linaro-toolchain@lists.linaro.org
-
Christophe Lyon -
Thomas Schwinge