Hi,
unfortunately I do not have arm system at hand. I tried to build this commit on QEMU (qemu-system-arm) running Debian (armhf architecture) and I cannot reproduce any of the regressions.
I'd appreciate some information on how to build and run suitable armv8l system on QEMU (or, if that's easier, on Odroid-N2) - I could not find much.
Thanks! Jan
On Sun, 2025-11-30 at 10:46 +0000, ci_notify@linaro.org wrote:
Dear contributor,
Our automatic CI has detected problems related to your patch(es). Please find some details below.
In gdb_check master-arm, after: | commit gdb-17-branchpoint-981-gcc1fc6af415 | Author: Jan Vrany jan.vrany@labware.com | Date: Fri Nov 28 13:47:02 2025 +0000 | | gdb: change blockvector::contains() to handle blockvectors with "holes" | | This commit slightly changes the logic in blockvector::contains() | to handle a case where the blockvector contains blocks with disjoint | regions (see the comment in blockvector::contains for details). | ... 18 lines of the commit log omitted.
Produces 77 regressions: | | regressions.sum: | Running gdb:gdb.base/annota1.exp ... | FAIL: gdb.base/annota1.exp: send SIGUSR1 (timeout) | Running gdb:gdb.base/annota3.exp ... | FAIL: gdb.base/annota3.exp: send SIGUSR1 (pattern 8) | Running gdb:gdb.base/sigstep.exp ... | ... and 87 more
Used configuration : *CI config* tcwg_gdb_check master-arm *configure and test flags:* none, autodetected on armv8l-unknown-linux-gnueabihf
We track this bug report under https://linaro.atlassian.net/browse/GNU-1767. 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_gdb_check--master-arm-build/3149/artifact/art... Reference build : https://ci.linaro.org/job/tcwg_gdb_check--master-arm-build/3148/artifact/art...
Instruction to reproduce the build : https://gitlab.com/LinaroLtd/tcwg/ci/interesting-commits/-/raw/master/gdb/sh...
Full commit : https://sourceware.org/git/?p=binutils-gdb.git%3Ba=commitdiff%3Bh=cc1fc6af41...
Hi Thiago,
Would you please help Jan investigate these regressions?
Thanks!
-- Maxim Kuvyrkov https://www.linaro.org
On Dec 4, 2025, at 03:23, Jan Vraný via linaro-toolchain linaro-toolchain@lists.linaro.org wrote:
Hi,
unfortunately I do not have arm system at hand. I tried to build this commit on QEMU (qemu-system-arm) running Debian (armhf architecture) and I cannot reproduce any of the regressions.
I'd appreciate some information on how to build and run suitable armv8l system on QEMU (or, if that's easier, on Odroid-N2) - I could not find much.
Thanks! Jan
On Sun, 2025-11-30 at 10:46 +0000, ci_notify@linaro.org wrote:
Dear contributor,
Our automatic CI has detected problems related to your patch(es). Please find some details below.
In gdb_check master-arm, after: | commit gdb-17-branchpoint-981-gcc1fc6af415 | Author: Jan Vrany jan.vrany@labware.com | Date: Fri Nov 28 13:47:02 2025 +0000 | | gdb: change blockvector::contains() to handle blockvectors with "holes" | | This commit slightly changes the logic in blockvector::contains() | to handle a case where the blockvector contains blocks with disjoint | regions (see the comment in blockvector::contains for details). | ... 18 lines of the commit log omitted.
Produces 77 regressions: | | regressions.sum: | Running gdb:gdb.base/annota1.exp ... | FAIL: gdb.base/annota1.exp: send SIGUSR1 (timeout) | Running gdb:gdb.base/annota3.exp ... | FAIL: gdb.base/annota3.exp: send SIGUSR1 (pattern 8) | Running gdb:gdb.base/sigstep.exp ... | ... and 87 more
Used configuration : *CI config* tcwg_gdb_check master-arm *configure and test flags:* none, autodetected on armv8l-unknown-linux-gnueabihf
We track this bug report under https://linaro.atlassian.net/browse/GNU-1767. 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_gdb_check--master-arm-build/3149/artifact/art... Reference build : https://ci.linaro.org/job/tcwg_gdb_check--master-arm-build/3148/artifact/art...
Instruction to reproduce the build : https://gitlab.com/LinaroLtd/tcwg/ci/interesting-commits/-/raw/master/gdb/sh...
Full commit : https://sourceware.org/git/?p=binutils-gdb.git%3Ba=commitdiff%3Bh=cc1fc6af41...
linaro-toolchain mailing list -- linaro-toolchain@lists.linaro.org To unsubscribe send an email to linaro-toolchain-leave@lists.linaro.org
Hello Jan,
Jan Vraný via linaro-toolchain linaro-toolchain@lists.linaro.org writes:
unfortunately I do not have arm system at hand. I tried to build this commit on QEMU (qemu-system-arm) running Debian (armhf architecture) and I cannot reproduce any of the regressions.
I'd appreciate some information on how to build and run suitable armv8l system on QEMU (or, if that's easier, on Odroid-N2) - I could not find much.
Our CI runs on Ubuntu 24.04, so I was able to reproduce the regressions on real hardware running that OS.
I'll set up a QEMU VM with Debian to see if I can reproduce them there, but if you want you can also try with an Ubuntu 24.04 image. I think it's likely that it would show the problem.
You probably already know this, but for completeness: when running the GDB tests with QEMU emulation one normally needs to increase the timeout to account for the slowness, by putting something like "set gdb_test_timeout 300" in <build-dir>/gdb/testsuite/site.exp.
On Thu, 2025-12-04 at 01:15 -0300, Thiago Jung Bauermann wrote:
Hello Jan,
Jan Vraný via linaro-toolchain linaro-toolchain@lists.linaro.org writes:
unfortunately I do not have arm system at hand. I tried to build this commit on QEMU (qemu-system-arm) running Debian (armhf architecture) and I cannot reproduce any of the regressions.
I'd appreciate some information on how to build and run suitable armv8l system on QEMU (or, if that's easier, on Odroid-N2) - I could not find much.
Our CI runs on Ubuntu 24.04, so I was able to reproduce the regressions on real hardware running that OS.
OK, perfect. It turned out that this also happens on x86_64, see https://sourceware.org/bugzilla/show_bug.cgi?id=33679
I'll set up a QEMU VM with Debian to see if I can reproduce them there, but if you want you can also try with an Ubuntu 24.04 image. I think it's likely that it would show the problem.
I'll try that.
Thanks!
Jan
Jan Vraný Jan.Vrany@labware.com writes:
On Thu, 2025-12-04 at 01:15 -0300, Thiago Jung Bauermann wrote:
Hello Jan,
Jan Vraný via linaro-toolchain linaro-toolchain@lists.linaro.org writes:
unfortunately I do not have arm system at hand. I tried to build this commit on QEMU (qemu-system-arm) running Debian (armhf architecture) and I cannot reproduce any of the regressions.
I'd appreciate some information on how to build and run suitable armv8l system on QEMU (or, if that's easier, on Odroid-N2) - I could not find much.
Our CI runs on Ubuntu 24.04, so I was able to reproduce the regressions on real hardware running that OS.
OK, perfect. It turned out that this also happens on x86_64, see https://sourceware.org/bugzilla/show_bug.cgi?id=33679
Ah, that makes it easier then. If you want, when there's a patch fixing the problem on x86_64 I can test on armhf to see if it fixes the problem there too.
On Thu, 2025-12-04 at 15:25 -0300, Thiago Jung Bauermann wrote:
Jan Vraný Jan.Vrany@labware.com writes:
On Thu, 2025-12-04 at 01:15 -0300, Thiago Jung Bauermann wrote:
Hello Jan,
Jan Vraný via linaro-toolchain linaro-toolchain@lists.linaro.org writes:
unfortunately I do not have arm system at hand. I tried to build this commit on QEMU (qemu-system-arm) running Debian (armhf architecture) and I cannot reproduce any of the regressions.
I'd appreciate some information on how to build and run suitable armv8l system on QEMU (or, if that's easier, on Odroid-N2) - I could not find much.
Our CI runs on Ubuntu 24.04, so I was able to reproduce the regressions on real hardware running that OS.
OK, perfect. It turned out that this also happens on x86_64, see https://sourceware.org/bugzilla/show_bug.cgi?id=33679
Ah, that makes it easier then. If you want, when there's a patch fixing the problem on x86_64 I can test on armhf to see if it fixes the problem there too.
Yes, that would be good.
On the bug reproducing note, I'm getting lost as I'm not familiar with ARM ecosystem (this is pretty much my first serious encounter).
I could not find Ubuntu 24.04 for 32bit "armhf", only for 64bit "arm64". IIUC the issue occurs on "armv8l" which is 32bit ARM code running on 64bit ARM CPU, right?
Running on armhf QEMU, compiling GDB with Debian-provided toolchain does not exhibit the issue for me. I could not find Linaro-provided toolchain for armhf-linux-gnueabi systems targeting armhf-linux-gnueabi, only cross-compilation toolchains.
Jan
Jan Vraný Jan.Vrany@labware.com writes:
On Thu, 2025-12-04 at 15:25 -0300, Thiago Jung Bauermann wrote:
Jan Vraný Jan.Vrany@labware.com writes:
On Thu, 2025-12-04 at 01:15 -0300, Thiago Jung Bauermann wrote:
Hello Jan,
Jan Vraný via linaro-toolchain linaro-toolchain@lists.linaro.org writes:
unfortunately I do not have arm system at hand. I tried to build this commit on QEMU (qemu-system-arm) running Debian (armhf architecture) and I cannot reproduce any of the regressions.
I'd appreciate some information on how to build and run suitable armv8l system on QEMU (or, if that's easier, on Odroid-N2) - I could not find much.
Our CI runs on Ubuntu 24.04, so I was able to reproduce the regressions on real hardware running that OS.
OK, perfect. It turned out that this also happens on x86_64, see https://sourceware.org/bugzilla/show_bug.cgi?id=33679
Ah, that makes it easier then. If you want, when there's a patch fixing the problem on x86_64 I can test on armhf to see if it fixes the problem there too.
Yes, that would be good.
Ok, I'll keep an eye on that bugzilla and test any patches about this issue.
On the bug reproducing note, I'm getting lost as I'm not familiar with ARM ecosystem (this is pretty much my first serious encounter).
I could not find Ubuntu 24.04 for 32bit "armhf", only for 64bit "arm64". IIUC the issue occurs on "armv8l" which is 32bit ARM code running on 64bit ARM CPU, right?
I have a QEMU VM with it installed so I'm pretty sure there was an image or installer available at some point, but I can't find it anymore.
Running on armhf QEMU, compiling GDB with Debian-provided toolchain does not exhibit the issue for me.
Do you have the libc6-dbg package installed?
I upgraded my old Debian image from oldstable to stable and GDB is currently building, so soon I'll be able to see if I can reproduce the problem there.
I could not find Linaro-provided toolchain for armhf-linux-gnueabi systems targeting armhf-linux-gnueabi, only cross-compilation toolchains.
We did provide toolchains in the past, but not nowadays.
Thiago Jung Bauermann thiago.bauermann@linaro.org writes:
Jan Vraný Jan.Vrany@labware.com writes:
IIUC the issue occurs on "armv8l" which is 32bit ARM code running on 64bit ARM CPU, right?
Sorry, I forgot to answer this question. That is correct. For userspace, there's no difference between 32-bit running on a 64-bit CPU or on an actual 32-bit CPU.
linaro-toolchain@lists.linaro.org
-
Jan Vraný -
Maxim Kuvyrkov -
Thiago Jung Bauermann