mesa_jenkins issueshttps://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues2018-06-22T05:19:44Zhttps://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/30jenkins: bisect_mesa description mismatch2018-06-22T05:19:44Zclayton craftjenkins: bisect_mesa description mismatch*Created by: dcbaker*
The page for bisect mesa failures says:
`sha of the mesa build which did not have the failures. Just pass the sha, do not include the "mesa="`
However, the script clearly expects to be passed '<project>=<sha>'.
*Created by: dcbaker*
The page for bisect mesa failures says:
`sha of the mesa build which did not have the failures. Just pass the sha, do not include the "mesa="`
However, the script clearly expects to be passed '<project>=<sha>'.
https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/25debian_installer: support for mmcblk drives2018-06-22T05:19:44Zclayton craftdebian_installer: support for mmcblk drives*Created by: dcbaker*
currently if the debian installer needs to be pointed at non /dev/sda (such as /dev/mmcblk0) the installs script must be modified and a new image created. It would be ideal if autodetection could happen to set this...*Created by: dcbaker*
currently if the debian installer needs to be pointed at non /dev/sda (such as /dev/mmcblk0) the installs script must be modified and a new image created. It would be ideal if autodetection could happen to set this correctly
https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/21debian_installer: select working ethernet rather than automatically selecting...2018-06-22T05:19:44Zclayton craftdebian_installer: select working ethernet rather than automatically selecting eth0*Created by: dcbaker*
if eth0 doesn't work the installer will get stuck. If eth1 does work, it won't try to select it. This should be fixable
*Created by: dcbaker*
if eth0 doesn't work the installer will get stuck. If eth1 does work, it won't try to select it. This should be fixable
https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/19debian_installer: sometimes finalize.sh doesn't complete2018-06-22T05:19:44Zclayton craftdebian_installer: sometimes finalize.sh doesn't complete*Created by: dcbaker*
Sometimes machines come out without dist-upgrade completing, leaving the system half updated and with many uninstall packages.
*Created by: dcbaker*
Sometimes machines come out without dist-upgrade completing, leaving the system half updated and with many uninstall packages.
https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/17debian_installer: dont' ask to load firmware2018-06-22T05:19:44Zclayton craftdebian_installer: dont' ask to load firmware*Created by: dcbaker*
It would be ideal if the system didn't ask to load non-free firmware, since we have never needed it.
*Created by: dcbaker*
It would be ideal if the system didn't ask to load non-free firmware, since we have never needed it.
https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/15debian installer raises error about /mnt/media2018-06-22T05:19:44Zclayton craftdebian installer raises error about /mnt/media*Created by: dcbaker*
I'm leaving this as a reminder (apparently I can't file bugs against my clone of your repo).
Basically this bug boils down to the line in debian_installer/jenkins.cfg:
``` sh
d-i partman/early_command string umou...*Created by: dcbaker*
I'm leaving this as a reminder (apparently I can't file bugs against my clone of your repo).
Basically this bug boils down to the line in debian_installer/jenkins.cfg:
``` sh
d-i partman/early_command string umount /media
```
which always tries to unmount /media, really this needs to check to see if /media is mounted, and only try to unmount if /media is actually mounted.
https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/31Allow to build mesa with different options2018-10-15T20:27:11ZSamuel PitoisetAllow to build mesa with different optionsCurrently, the script that builds Mesa is only for Intel. It would be nice to have some options, like for selecting which Vulkan drivers to build etc.Currently, the script that builds Mesa is only for Intel. It would be nice to have some options, like for selecting which Vulkan drivers to build etc.https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/32Add a module for building LLVM2018-10-15T20:28:29ZSamuel PitoisetAdd a module for building LLVMRADV and RadeonSI require LLVM. Should be easy to add llvm/build.py.RADV and RadeonSI require LLVM. Should be easy to add llvm/build.py.https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/33-Werror=missing-prototypes issues are not caught2019-01-25T17:29:37ZEmil Velikov-Werror=missing-prototypes issues are not caughtIn light of https://gitlab.freedesktop.org/mesa/mesa/merge_requests/152
The patch/series which causes the issue ran through the Intel CI and returned a clean bill of health[1]. Additionally local testing and Travis did not flag the issu...In light of https://gitlab.freedesktop.org/mesa/mesa/merge_requests/152
The patch/series which causes the issue ran through the Intel CI and returned a clean bill of health[1]. Additionally local testing and Travis did not flag the issue either.
I'll investigate why my local test did not show the problem, but it will be nice if also addressed for the Jenkins runs.
[1] https://mesa-ci.01.org/evelikov/builds/33/group/63a9f0ea7bb98050796b649e85481845#subgroupshttps://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/36dEQP/CTS NotSupported tests are flagged as "failed"2019-02-06T02:41:40ZEmil VelikovdEQP/CTS NotSupported tests are flagged as "failed"Based on [1] and [2] seems that dEQP/CTS tests reporting `Not Supported` are flagged as `fail`.
Did I understand this correctly, what the reason behind this?
In piglit we use "skip" for similar cases and I think the CI has such a state....Based on [1] and [2] seems that dEQP/CTS tests reporting `Not Supported` are flagged as `fail`.
Did I understand this correctly, what the reason behind this?
In piglit we use "skip" for similar cases and I think the CI has such a state.
Thanks
[1] https://mesa-ci.01.org/wip_18_3/builds/23/results/2940440
[2] https://github.com/KhronosGroup/VK-GL-CTS/blob/master/external/vulkancts/modules/vulkan/wsi/vktWsiDisplayTests.cpp#L2089https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/35Missing information about generated tests2019-02-14T16:25:46ZEmil VelikovMissing information about generated testsAs mentioned in #34 there is no obvious link/reference to the actual code (be that generator or generated) for `generated.` tests.
For example [1] lists only the following, thus trying to understand what's happening or reproduce is impo...As mentioned in #34 there is no obvious link/reference to the actual code (be that generator or generated) for `generated.` tests.
For example [1] lists only the following, thus trying to understand what's happening or reproduce is impossible :-(
```
2019-01-29T12:22:23,110938-0800 i915_gem_userptr_mn_invalidate_range_start+0x167/0x190 [i915]
```
[1] https://mesa-ci.01.org/wip_18_3/builds/23/results/1601927https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/3418.3: passing invalid arguments to the meson generated test2019-03-22T15:30:44ZEmil Velikov18.3: passing invalid arguments to the meson generated testIn `master` branch the `pl111` driver got renamed to `kmsro` but not in `18.3`. Yet the generated tests use `kmsro`.
Causing a failure like the following:
https://mesa-ci.01.org/wip_18_3/builds/23/results/1534585
Mildly related: where'...In `master` branch the `pl111` driver got renamed to `kmsro` but not in `18.3`. Yet the generated tests use `kmsro`.
Causing a failure like the following:
https://mesa-ci.01.org/wip_18_3/builds/23/results/1534585
Mildly related: where's the code that generates those, can we have a link to that from the failing tests?https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/38Unblacklist more things on iris2019-06-25T17:46:05ZKenneth Graunkekenneth@whitecape.orgUnblacklist more things on irisLocally, [almost all CTS tests are passing on iris](http://whitecape.org/stuff/brw-iris-june/cts/problems.html), so hopefully we should be able to unblacklist the entire "flaky on iris" section. I fixed a whole bunch of transform feedba...Locally, [almost all CTS tests are passing on iris](http://whitecape.org/stuff/brw-iris-june/cts/problems.html), so hopefully we should be able to unblacklist the entire "flaky on iris" section. I fixed a whole bunch of transform feedback and copy image bugs recently. I'd say to use the i965 blacklist, but we actually pass `GTF-GL46.gtf32.GL3Tests.packed_pixels.packed_pixels*` as well, which i965 has blacklisted.https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/39Make build options case insensitive2019-06-25T18:41:25ZSagar Ghugesagar.ghuge@intel.comMake build options case insensitiveIn revision field, make tokens case insensitive. For e.g both MESA and mesa should work.In revision field, make tokens case insensitive. For e.g both MESA and mesa should work.https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/40waffle has moved to gitlab.freedesktop.org2019-08-22T21:47:46ZDylan Bakerwaffle has moved to gitlab.freedesktop.orgAnd the github site is not being updated currently.And the github site is not being updated currently.https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/41icelake-only developer builds2019-09-17T20:36:26ZKenneth Graunkekenneth@whitecape.orgicelake-only developer buildsI could use the ability to test Icelake on developer builds. Ideally this would be added to normal developer builds, but that may not be possible yet. In the meantime, Clayton gave me the ability to test Icelake *only* via a developer ...I could use the ability to test Icelake on developer builds. Ideally this would be added to normal developer builds, but that may not be possible yet. In the meantime, Clayton gave me the ability to test Icelake *only* via a developer build by doing a custom build:
> \<craftyguy\> Kayden: do custom build, for 'build support' use sha e69a3a83a457fe7fa705e5943e9efeb100cfa37b, select 'all-test' project in drop down. you can leave everything else as default
>
> \<craftyguy\> Kayden: also beware that this will only test ICL, so if there are regressions in other platforms.. they won't be found in this build...
However, this broke within a few days. Now, if I submit a custom job with that build support sha, it doesn't even register a new job in the mesa_custom build list. It doesn't fail, it just silently does absolutely nothing as far as I can tell.https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/42Re-enable Eric Anholt's Intel CI access2019-10-16T23:31:42ZKenneth Graunkekenneth@whitecape.orgRe-enable Eric Anholt's Intel CI accessPlease re-enable Eric's Intel CI access:
* Email: eric@anholt.net
* GitLab user: @anholt
* Repo: Please add him to https://gitlab.freedesktop.org/Mesa_CI/repos/mesa and give him a dev/anholt branch.
Please make sure his results appear ...Please re-enable Eric's Intel CI access:
* Email: eric@anholt.net
* GitLab user: @anholt
* Repo: Please add him to https://gitlab.freedesktop.org/Mesa_CI/repos/mesa and give him a dev/anholt branch.
Please make sure his results appear on the public website.
Thanks!https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/43Subgroup wise comparison on ci result2019-11-25T19:12:41ZSagar Ghugesagar.ghuge@intel.comSubgroup wise comparison on ci resultThere should be an option to compare subgroup from one build to another, that would make things faster and narrow down the area in which developer is interested.There should be an option to compare subgroup from one build to another, that would make things faster and narrow down the area in which developer is interested.https://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/18debian_installer: error out if there are not at least two drives connected2020-03-16T21:28:38Zclayton craftdebian_installer: error out if there are not at least two drives connected*Created by: dcbaker*
Don't allow the installer to install over itself.
*Created by: dcbaker*
Don't allow the installer to install over itself.
Installer: Make drive selection more robusthttps://gitlab.freedesktop.org/Mesa_CI/mesa_jenkins/-/issues/44intel-ci: icelake: pick kernel patch for gpu error state reporting2020-03-28T06:08:17ZD Scott Phillipsintel-ci: icelake: pick kernel patch for gpu error state reportingPlease pick this patch onto the current kernel used in the intel ci's icelake machines:
https://github.com/d-scott-phillips/linux/commit/3b512ab97758e3a11485012ed64645e2cbdbc46f
This will get request information back in gpu hang reports.Please pick this patch onto the current kernel used in the intel ci's icelake machines:
https://github.com/d-scott-phillips/linux/commit/3b512ab97758e3a11485012ed64645e2cbdbc46f
This will get request information back in gpu hang reports.