W dniu 30.10.2020 o 13:03, Marcin Juszkiewicz via Asa-dev pisze:
> I added new CI job: ldcg-sbsa-acs.
>
> It builds HEAD of qemu and edk2 and then use it to run Arm SBSA ACS
> testsuite for SBSA levels 3-6.
Some changes done:
1. As SBSA ACS gets new commits I changed CI job to build what's
in git HEAD instead of using releases.
2. sbsa-ref is now using Cortex-A72 instead of A57 during run.
3. We have two runs per SBSA level now. One is with default action
for watchdog (reset) while other ignores watchdog (action
none).
Shashi has some ideas why SBSA ACS are failing in watchdog tests and
will file an issue.
https://snapshots.linaro.org/ldcg/sbsa-acs/9/ should get populated in
few minutes with full logs.
I do wonder which hardware passes SBSA ACS test suite and is available
at Linaro lab.
Qualcomm machine I use in mlab is not one of them for sure. Testsuite
crashes badly before even starting any tests.
If someone has compliant hardware then please run SBSA ACS and provide
logs to compare.
On Fri, Oct 30, 2020 at 12:07 PM Marcin Juszkiewicz via Asa-dev <
asa-dev(a)op-lists.linaro.org> wrote:
> W dniu 30.10.2020 o 13:03, Marcin Juszkiewicz via Asa-dev pisze:
> > I added new CI job: ldcg-sbsa-acs.
> >
> > It builds HEAD of qemu and edk2 and then use it to run Arm SBSA ACS
> > testsuite for SBSA levels 3-6.
> >
> > https://snapshots.linaro.org/ldcg/sbsa-acs/3/ has logs from each run.
> >
> > This allows us to check which components are still missing for each
> level.
> >
> > I created some Jira cards yesterday to list missing parts:
> >
> > ASA-65 - add PMU counters
> > ASA-66 - add GIC ITS support
> > ASA-67 - add Virtual Host Extensions (required for level 4)
> > ASA-68 - add SMMU (it is not detected by SBSA ACS now)
> >
>
ASA-68 effectively requirers ASA-66 as far as I can tell. I did try an
instantiate a SMMUv3 from ACPI but
the kernel throws errors if there is not an ITS there. (I could however
have made a mistake in my tables).
The qemu patches for SMMUv3 are upstream now so at least its correct from
QEMU side.
Graeme
> > I plan to look at those logs and create additional cards so we can see
> > what is still missing (or not enabled).
> >
> > Next run will have larger timeout as 30s looks a bit too short.
>
> Current situation summary:
>
> SBSA level | tests run | passed | failed
> ----------------------------------------
> 3 | 45 | 32 | 6
> 4 | 87 | 44 | 13
> 5 | 92 | 44 | 20
> 6 | 102 | 44 | 29
>
>
> Lot of tests were skipped.
> --
> Asa-dev mailing list
> Asa-dev(a)op-lists.linaro.org
> https://op-lists.linaro.org/mailman/listinfo/asa-dev
>
W dniu 30.10.2020 o 13:07, Marcin Juszkiewicz via Asa-dev pisze:
> Current situation summary:
>
> SBSA level | tests run | passed | failed
> ----------------------------------------
> 3 | 45 | 32 | 6
> 4 | 87 | 44 | 13
> 5 | 92 | 44 | 20
> 6 | 102 | 44 | 29
As you see SBSA level 3 is near with just 6 failed tests:
12 : Check number of PMU counters
Failed on PE - 0 for Level= 3 : Result: --FAIL-- 1
102 : If PCIe, then GIC implements ITS
Failed on PE - 0 for Level= 3 : Result: --FAIL-- 1
104 : GIC Maintenance Interrupt
Failed on PE - 0 for Level= 3 : Result: --FAIL-- 1
206 : SYS Timer if PE Timer not ON
PE Timers are not always-on.
Failed on PE - 0 for Level= 3 : Result: --FAIL-- 1
301 : Check NS Watchdog Accessibility
No Watchdogs reported 0
Failed on PE - 0 for Level= 3 : Result: --FAIL-- 1
302 : Check Watchdog WS0 interrupt
No Watchdogs reported 0
Failed on PE - 0 for Level= 3 : Result: --FAIL-- 1
Cards for failed tests:
test 12 is ASA-65 (add PMU counters)
test 102 is ASA-66 (add ITS support)
test 104 is ASA-70 (add GIC Maintenance Interrupt)
test 206 is ASA-71 (fix always-on timers)
test 301/302 is ASA-57 (add Watchdog support)
From them ASA-57 is done in QEMU, approved for EDK2.
If you have comments/ideas etc then please comment in Jira cards or in
replies.
W dniu 30.10.2020 o 13:03, Marcin Juszkiewicz via Asa-dev pisze:
> I added new CI job: ldcg-sbsa-acs.
>
> It builds HEAD of qemu and edk2 and then use it to run Arm SBSA ACS
> testsuite for SBSA levels 3-6.
>
> https://snapshots.linaro.org/ldcg/sbsa-acs/3/ has logs from each run.
>
> This allows us to check which components are still missing for each level.
>
> I created some Jira cards yesterday to list missing parts:
>
> ASA-65 - add PMU counters
> ASA-66 - add GIC ITS support
> ASA-67 - add Virtual Host Extensions (required for level 4)
> ASA-68 - add SMMU (it is not detected by SBSA ACS now)
>
> I plan to look at those logs and create additional cards so we can see
> what is still missing (or not enabled).
>
> Next run will have larger timeout as 30s looks a bit too short.
Current situation summary:
SBSA level | tests run | passed | failed
----------------------------------------
3 | 45 | 32 | 6
4 | 87 | 44 | 13
5 | 92 | 44 | 20
6 | 102 | 44 | 29
Lot of tests were skipped.
I added new CI job: ldcg-sbsa-acs.
It builds HEAD of qemu and edk2 and then use it to run Arm SBSA ACS
testsuite for SBSA levels 3-6.
https://snapshots.linaro.org/ldcg/sbsa-acs/3/ has logs from each run.
This allows us to check which components are still missing for each level.
I created some Jira cards yesterday to list missing parts:
ASA-65 - add PMU counters
ASA-66 - add GIC ITS support
ASA-67 - add Virtual Host Extensions (required for level 4)
ASA-68 - add SMMU (it is not detected by SBSA ACS now)
I plan to look at those logs and create additional cards so we can see
what is still missing (or not enabled).
Next run will have larger timeout as 30s looks a bit too short.
See <https://ci.linaro.org/jenkins/job/ldcg-sbsa-acs/1/display/redirect>
Changes:
------------------------------------------
Started by user Marcin Juszkiewicz (marcin.juszkiewicz(a)linaro.org)
Running as SYSTEM
[EnvInject] - Loading node environment variables.
Building remotely on leg-aarch64-01-f1b55a93cb1d (docker-buster-arm64-leg) in workspace <https://ci.linaro.org/jenkins/job/ldcg-sbsa-acs/ws/>
[ldcg-sbsa-acs] $ /bin/bash /tmp/jenkins16867014569957094092.sh
+ source common-code.sh
/tmp/jenkins16867014569957094092.sh: line 7: common-code.sh: No such file or directory
Build step 'Execute shell' marked build as failure