[PATCH 0/2] gitlab: Add config for Loongarch64 custom runner

[PATCH 0/2] gitlab: Add config for Loongarch64 custom runner

[Philippe Mathieu-Daudé]

Loongson Technology Corporation Limited offered to sponsor QEMU
with a Loongarch64 custom runner.

Philippe Mathieu-Daudé (2):
  gitlab: Introduce Loongarch64 runner
  gitlab: Add Loongarch64 KVM-only build

 docs/devel/ci-jobs.rst.inc                    |  6 +++
 .gitlab-ci.d/custom-runners.yml               |  1 +
 .../openeuler-22.03-loongarch64.yml           | 43 +++++++++++++++++++
 3 files changed, 50 insertions(+)
 create mode 100644 .gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml

-- 
2.41.0

[PATCH 1/2] gitlab: Introduce Loongarch64 runner

[Philippe Mathieu-Daudé]

Full build config to run CI tests on a Loongarch64 host.

Forks might enable this by setting LOONGARCH64_RUNNER_AVAILABLE
in their CI namespace settings, see:
https://www.qemu.org/docs/master/devel/ci.html#maintainer-controlled-job-variables

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
 docs/devel/ci-jobs.rst.inc                    |  6 ++++++
 .gitlab-ci.d/custom-runners.yml               |  1 +
 .../openeuler-22.03-loongarch64.yml           | 21 +++++++++++++++++++
 3 files changed, 28 insertions(+)
 create mode 100644 .gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml

diff --git a/docs/devel/ci-jobs.rst.inc b/docs/devel/ci-jobs.rst.inc
index 4c39cdb2d9..b821a33112 100644
--- a/docs/devel/ci-jobs.rst.inc
+++ b/docs/devel/ci-jobs.rst.inc
@@ -189,6 +189,12 @@ used as a gitlab-CI runner, you can set this variable to enable the
 tests that require this kind of host. The runner should be tagged with
 both "centos_stream_8" and "x86_64".
 
+LOONGARCH64_RUNNER_AVAILABLE
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+If you've got access to a Loongarch64 host that can be used as a gitlab-CI
+runner, you can set this variable to enable the tests that require this
+kind of host. The runner should be tagged with "loongarch64".
+
 CCACHE_DISABLE
 ~~~~~~~~~~~~~~
 The jobs are configured to use "ccache" by default since this typically
diff --git a/.gitlab-ci.d/custom-runners.yml b/.gitlab-ci.d/custom-runners.yml
index 8e5b9500f4..152ace4492 100644
--- a/.gitlab-ci.d/custom-runners.yml
+++ b/.gitlab-ci.d/custom-runners.yml
@@ -32,3 +32,4 @@ include:
   - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch64.yml'
   - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch32.yml'
   - local: '/.gitlab-ci.d/custom-runners/centos-stream-8-x86_64.yml'
+  - local: '/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml'
diff --git a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
new file mode 100644
index 0000000000..86d18f820e
--- /dev/null
+++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
@@ -0,0 +1,21 @@
+openeuler-22.03-loongarch64-all:
+ extends: .custom_runner_template
+ needs: []
+ stage: build
+ tags:
+ - oe2203
+ - loongarch64
+ rules:
+ - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && $CI_COMMIT_BRANCH =~ /^staging/'
+   when: manual
+   allow_failure: true
+ - if: "$LOONGARCH64_RUNNER_AVAILABLE"
+   when: manual
+   allow_failure: true
+ script:
+ - mkdir build
+ - cd build
+ - ../configure
+   || { cat config.log meson-logs/meson-log.txt; exit 1; }
+ - make --output-sync -j`nproc --ignore=40`
+ - make --output-sync -j`nproc --ignore=40` check
-- 
2.41.0

[NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[Philippe Mathieu-Daudé]

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
Used to test https://lore.kernel.org/qemu-devel/20231228084051.3235354-1-zhaotianrui@loongson.cn/
---
 .../openeuler-22.03-loongarch64.yml           | 22 +++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
index 86d18f820e..60674b8d0f 100644
--- a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
+++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
@@ -19,3 +19,25 @@ openeuler-22.03-loongarch64-all:
    || { cat config.log meson-logs/meson-log.txt; exit 1; }
  - make --output-sync -j`nproc --ignore=40`
  - make --output-sync -j`nproc --ignore=40` check
+
+openeuler-22.03-loongarch64-kvm:
+ extends: .custom_runner_template
+ needs: []
+ stage: build
+ tags:
+ - oe2203
+ - loongarch64
+ rules:
+ - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && $CI_COMMIT_BRANCH =~ /^staging/'
+   when: manual
+   allow_failure: true
+ - if: "$LOONGARCH64_RUNNER_AVAILABLE"
+   when: manual
+   allow_failure: true
+ script:
+ - mkdir build
+ - cd build
+ - ../configure --enable-kvm --disable-tcg
+   || { cat config.log meson-logs/meson-log.txt; exit 1; }
+ - make --output-sync -j`nproc --ignore=40`
+ - make --output-sync -j`nproc --ignore=40` check
-- 
2.41.0

Re: [PATCH 1/2] gitlab: Introduce Loongarch64 runner

[Thomas Huth]

On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
> Full build config to run CI tests on a Loongarch64 host.
> 
> Forks might enable this by setting LOONGARCH64_RUNNER_AVAILABLE
> in their CI namespace settings, see:
> https://www.qemu.org/docs/master/devel/ci.html#maintainer-controlled-job-variables
> 
> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> ---
>   docs/devel/ci-jobs.rst.inc                    |  6 ++++++
>   .gitlab-ci.d/custom-runners.yml               |  1 +
>   .../openeuler-22.03-loongarch64.yml           | 21 +++++++++++++++++++
>   3 files changed, 28 insertions(+)
>   create mode 100644 .gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
> 
...
> diff --git a/.gitlab-ci.d/custom-runners.yml b/.gitlab-ci.d/custom-runners.yml
> index 8e5b9500f4..152ace4492 100644
> --- a/.gitlab-ci.d/custom-runners.yml
> +++ b/.gitlab-ci.d/custom-runners.yml
> @@ -32,3 +32,4 @@ include:
>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch64.yml'
>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch32.yml'
>     - local: '/.gitlab-ci.d/custom-runners/centos-stream-8-x86_64.yml'
> +  - local: '/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml'
> diff --git a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
> new file mode 100644
> index 0000000000..86d18f820e
> --- /dev/null
> +++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
> @@ -0,0 +1,21 @@
> +openeuler-22.03-loongarch64-all:
> + extends: .custom_runner_template
> + needs: []
> + stage: build
> + tags:
> + - oe2203
> + - loongarch64
> + rules:
> + - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && $CI_COMMIT_BRANCH =~ /^staging/'
> +   when: manual
> +   allow_failure: true
> + - if: "$LOONGARCH64_RUNNER_AVAILABLE"
> +   when: manual
> +   allow_failure: true
> + script:
> + - mkdir build
> + - cd build
> + - ../configure
> +   || { cat config.log meson-logs/meson-log.txt; exit 1; }
> + - make --output-sync -j`nproc --ignore=40`
> + - make --output-sync -j`nproc --ignore=40` check

Does this system really have more than 40 CPU threads? Or is this a 
copy-n-past from one of the other scripts? In the latter case, I'd suggest 
to adjust the --ignore=40 to a more reasonable value.

  Thomas

Re: [NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[Thomas Huth]

On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> ---
> Used to test https://lore.kernel.org/qemu-devel/20231228084051.3235354-1-zhaotianrui@loongson.cn/

So why is it NOTFORMERGE ? Don't we want to test KVM-only builds for 
loongarch in the long run?

  Thomas


> ---
>   .../openeuler-22.03-loongarch64.yml           | 22 +++++++++++++++++++
>   1 file changed, 22 insertions(+)
> 
> diff --git a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
> index 86d18f820e..60674b8d0f 100644
> --- a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
> +++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
> @@ -19,3 +19,25 @@ openeuler-22.03-loongarch64-all:
>      || { cat config.log meson-logs/meson-log.txt; exit 1; }
>    - make --output-sync -j`nproc --ignore=40`
>    - make --output-sync -j`nproc --ignore=40` check
> +
> +openeuler-22.03-loongarch64-kvm:
> + extends: .custom_runner_template
> + needs: []
> + stage: build
> + tags:
> + - oe2203
> + - loongarch64
> + rules:
> + - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && $CI_COMMIT_BRANCH =~ /^staging/'
> +   when: manual
> +   allow_failure: true
> + - if: "$LOONGARCH64_RUNNER_AVAILABLE"
> +   when: manual
> +   allow_failure: true
> + script:
> + - mkdir build
> + - cd build
> + - ../configure --enable-kvm --disable-tcg
> +   || { cat config.log meson-logs/meson-log.txt; exit 1; }
> + - make --output-sync -j`nproc --ignore=40`
> + - make --output-sync -j`nproc --ignore=40` check

Re: [PATCH 1/2] gitlab: Introduce Loongarch64 runner

[gaosong]

Hi,

在 2024/1/11 下午3:08, Thomas Huth 写道:
> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>> Full build config to run CI tests on a Loongarch64 host.
>>
>> Forks might enable this by setting LOONGARCH64_RUNNER_AVAILABLE
>> in their CI namespace settings, see:
>> https://www.qemu.org/docs/master/devel/ci.html#maintainer-controlled-job-variables 
>>
>>
>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>> ---
>>   docs/devel/ci-jobs.rst.inc                    |  6 ++++++
>>   .gitlab-ci.d/custom-runners.yml               |  1 +
>>   .../openeuler-22.03-loongarch64.yml           | 21 +++++++++++++++++++
>>   3 files changed, 28 insertions(+)
>>   create mode 100644 
>> .gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>
> ...
>> diff --git a/.gitlab-ci.d/custom-runners.yml 
>> b/.gitlab-ci.d/custom-runners.yml
>> index 8e5b9500f4..152ace4492 100644
>> --- a/.gitlab-ci.d/custom-runners.yml
>> +++ b/.gitlab-ci.d/custom-runners.yml
>> @@ -32,3 +32,4 @@ include:
>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch64.yml'
>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch32.yml'
>>     - local: '/.gitlab-ci.d/custom-runners/centos-stream-8-x86_64.yml'
>> +  - local: 
>> '/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml'
>> diff --git 
>> a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml 
>> b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>> new file mode 100644
>> index 0000000000..86d18f820e
>> --- /dev/null
>> +++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>> @@ -0,0 +1,21 @@
>> +openeuler-22.03-loongarch64-all:
>> + extends: .custom_runner_template :-)
>> + needs: []
>> + stage: build
>> + tags:
>> + - oe2203
>> + - loongarch64
>> + rules:
>> + - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && $CI_COMMIT_BRANCH 
>> =~ /^staging/'
>> +   when: manual
>> +   allow_failure: true
>> + - if: "$LOONGARCH64_RUNNER_AVAILABLE"
>> +   when: manual
>> +   allow_failure: true
>> + script:
>> + - mkdir build
>> + - cd build
>> + - ../configure
>> +   || { cat config.log meson-logs/meson-log.txt; exit 1; }
>> + - make --output-sync -j`nproc --ignore=40`
>> + - make --output-sync -j`nproc --ignore=40` check
>
> Does this system really have more than 40 CPU threads? Or is this a 
> copy-n-past from one of the other scripts? In the latter case, I'd 
> suggest to adjust the --ignore=40 to a more reasonable value.
>
>  Thomas
No,  only 32.   I think it should be --ignore=32 or 16.

I create a same runner on this machine,  and I  find  some check error.
but I am not sure how to fix it. :-)

See:

https://gitlab.com/gaosong/qemu/-/jobs/5906269934

Thanks.
Song Gao

Re: [NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[gaosong]

Hi,

在 2024/1/11 下午3:10, Thomas Huth 写道:
> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>> ---
>> Used to test 
>> https://lore.kernel.org/qemu-devel/20231228084051.3235354-1-zhaotianrui@loongson.cn/
>
> So why is it NOTFORMERGE ? Don't we want to test KVM-only builds for 
> loongarch in the long run?
>
>  Thomas
>
I think we can drop this title.

I tested this job by the latest loongarch kvm patches.  buf I find a 
test-hmp check error.

See:
https://gitlab.com/gaosong/qemu/-/jobs/5906385234

If you want to log in to this machine, we can create an account for you.

Thanks.
Song Gao
>
>> ---
>>   .../openeuler-22.03-loongarch64.yml           | 22 +++++++++++++++++++
>>   1 file changed, 22 insertions(+)
>>
>> diff --git 
>> a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml 
>> b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>> index 86d18f820e..60674b8d0f 100644
>> --- a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>> +++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>> @@ -19,3 +19,25 @@ openeuler-22.03-loongarch64-all:
>>      || { cat config.log meson-logs/meson-log.txt; exit 1; }
>>    - make --output-sync -j`nproc --ignore=40`
>>    - make --output-sync -j`nproc --ignore=40` check
>> +
>> +openeuler-22.03-loongarch64-kvm:
>> + extends: .custom_runner_template
>> + needs: []
>> + stage: build
>> + tags:
>> + - oe2203
>> + - loongarch64
>> + rules:
>> + - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && $CI_COMMIT_BRANCH 
>> =~ /^staging/'
>> +   when: manual
>> +   allow_failure: true
>> + - if: "$LOONGARCH64_RUNNER_AVAILABLE"
>> +   when: manual
>> +   allow_failure: true
>> + script:
>> + - mkdir build
>> + - cd build
>> + - ../configure --enable-kvm --disable-tcg
>> +   || { cat config.log meson-logs/meson-log.txt; exit 1; }
>> + - make --output-sync -j`nproc --ignore=40`
>> + - make --output-sync -j`nproc --ignore=40` check

Re: [NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[Thomas Huth]

On 11/01/2024 08.37, gaosong wrote:
> Hi,
> 
> 在 2024/1/11 下午3:10, Thomas Huth 写道:
>> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>> ---
>>> Used to test 
>>> https://lore.kernel.org/qemu-devel/20231228084051.3235354-1-zhaotianrui@loongson.cn/
>>
>> So why is it NOTFORMERGE ? Don't we want to test KVM-only builds for 
>> loongarch in the long run?
>>
>>  Thomas
>>
> I think we can drop this title.
> 
> I tested this job by the latest loongarch kvm patches.  buf I find a 
> test-hmp check error.
Can you recreate the error manually? i.e. compile with configure 
--disable-tcg and then run:

  V=2 QTEST_QEMU_BINARY=./qemu-system-loongarch64 tests/qtest/test-hmp

That should likely provide you with a hint where it is crashing

  Thomas

Re: [PATCH 1/2] gitlab: Introduce Loongarch64 runner

[Thomas Huth]

On 11/01/2024 08.25, gaosong wrote:
> Hi,
> 
> 在 2024/1/11 下午3:08, Thomas Huth 写道:
>> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>>> Full build config to run CI tests on a Loongarch64 host.
>>>
>>> Forks might enable this by setting LOONGARCH64_RUNNER_AVAILABLE
>>> in their CI namespace settings, see:
>>> https://www.qemu.org/docs/master/devel/ci.html#maintainer-controlled-job-variables
>>>
>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>> ---
>>>   docs/devel/ci-jobs.rst.inc                    |  6 ++++++
>>>   .gitlab-ci.d/custom-runners.yml               |  1 +
>>>   .../openeuler-22.03-loongarch64.yml           | 21 +++++++++++++++++++
>>>   3 files changed, 28 insertions(+)
>>>   create mode 100644 
>>> .gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>
>> ...
>>> diff --git a/.gitlab-ci.d/custom-runners.yml 
>>> b/.gitlab-ci.d/custom-runners.yml
>>> index 8e5b9500f4..152ace4492 100644
>>> --- a/.gitlab-ci.d/custom-runners.yml
>>> +++ b/.gitlab-ci.d/custom-runners.yml
>>> @@ -32,3 +32,4 @@ include:
>>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch64.yml'
>>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch32.yml'
>>>     - local: '/.gitlab-ci.d/custom-runners/centos-stream-8-x86_64.yml'
>>> +  - local: '/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml'
>>> diff --git a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml 
>>> b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>> new file mode 100644
>>> index 0000000000..86d18f820e
>>> --- /dev/null
>>> +++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>> @@ -0,0 +1,21 @@
>>> +openeuler-22.03-loongarch64-all:
>>> + extends: .custom_runner_template :-)
>>> + needs: []
>>> + stage: build
>>> + tags:
>>> + - oe2203
>>> + - loongarch64
>>> + rules:
>>> + - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && $CI_COMMIT_BRANCH =~ 
>>> /^staging/'
>>> +   when: manual
>>> +   allow_failure: true
>>> + - if: "$LOONGARCH64_RUNNER_AVAILABLE"
>>> +   when: manual
>>> +   allow_failure: true
>>> + script:
>>> + - mkdir build
>>> + - cd build
>>> + - ../configure
>>> +   || { cat config.log meson-logs/meson-log.txt; exit 1; }
>>> + - make --output-sync -j`nproc --ignore=40`
>>> + - make --output-sync -j`nproc --ignore=40` check
>>
>> Does this system really have more than 40 CPU threads? Or is this a 
>> copy-n-past from one of the other scripts? In the latter case, I'd suggest 
>> to adjust the --ignore=40 to a more reasonable value.
>>
>>  Thomas
> No,  only 32.   I think it should be --ignore=32 or 16.

--ignore=32 then also does not make much sense, that would still be the same 
as simply omitting the -j parameter. I guess --ignore=16 should be fine.

> I create a same runner on this machine,  and I  find  some check error.
> but I am not sure how to fix it. :-)
> 
> See:
> 
> https://gitlab.com/gaosong/qemu/-/jobs/5906269934

Seems to be related to RAM backing... for example, the erst-test is failing, 
which is doing something like:

     setup_vm_cmd(&state,
         "-object memory-backend-file,"
             "mem-path=acpi-erst.XXXXXX,"
             "size=64K,"
             "share=on,"
             "id=nvram "
         "-device acpi-erst,"
             "memdev=nvram");

So it seems like -object memory-backend-file" is not correctly working in 
your gitlab runner? Is there some setup missing?

  Thomas

Re: [NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[gaosong]

在 2024/1/11 下午4:20, Thomas Huth 写道:
> On 11/01/2024 08.37, gaosong wrote:
>> Hi,
>>
>> 在 2024/1/11 下午3:10, Thomas Huth 写道:
>>> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>>> ---
>>>> Used to test 
>>>> https://lore.kernel.org/qemu-devel/20231228084051.3235354-1-zhaotianrui@loongson.cn/
>>>
>>> So why is it NOTFORMERGE ? Don't we want to test KVM-only builds for 
>>> loongarch in the long run?
>>>
>>>  Thomas
>>>
>> I think we can drop this title.
>>
>> I tested this job by the latest loongarch kvm patches.  buf I find a 
>> test-hmp check error.
> Can you recreate the error manually? i.e. compile with configure 
> --disable-tcg and then run:
>
>  V=2 QTEST_QEMU_BINARY=./qemu-system-loongarch64 tests/qtest/test-hmp
>
> That should likely provide you with a hint where it is crashing
>
>  Thomas
Thank you,

LoongArch no support these cmds  or some problems .
-    "gva2gpa 0",
-    "memsave 0 4096 \"/dev/null\"",
-    "x /8i 0x100",
-    "xp /16x 0",

Could we disable these 4 cmds or the test_temp check?
After we fix the cmds problems, we can enable them.

Thanks.
Song gao

Re: [NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[Thomas Huth]

On 11/01/2024 09.50, gaosong wrote:
> 在 2024/1/11 下午4:20, Thomas Huth 写道:
>> On 11/01/2024 08.37, gaosong wrote:
>>> Hi,
>>>
>>> 在 2024/1/11 下午3:10, Thomas Huth 写道:
>>>> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>>>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>>>> ---
>>>>> Used to test 
>>>>> https://lore.kernel.org/qemu-devel/20231228084051.3235354-1-zhaotianrui@loongson.cn/
>>>>
>>>> So why is it NOTFORMERGE ? Don't we want to test KVM-only builds for 
>>>> loongarch in the long run?
>>>>
>>>>  Thomas
>>>>
>>> I think we can drop this title.
>>>
>>> I tested this job by the latest loongarch kvm patches.  buf I find a 
>>> test-hmp check error.
>> Can you recreate the error manually? i.e. compile with configure 
>> --disable-tcg and then run:
>>
>>  V=2 QTEST_QEMU_BINARY=./qemu-system-loongarch64 tests/qtest/test-hmp
>>
>> That should likely provide you with a hint where it is crashing
>>
>>  Thomas
> Thank you,
> 
> LoongArch no support these cmds  or some problems .
> -    "gva2gpa 0",
> -    "memsave 0 4096 \"/dev/null\"",
> -    "x /8i 0x100",
> -    "xp /16x 0",
> 
> Could we disable these 4 cmds or the test_temp check?
> After we fix the cmds problems, we can enable them.

Even if loongarch does not support one of these commands, it should not 
crash QEMU. So please fix the crashes first before considering to enable the 
KVM-only test in the CI.

  Thanks
   Thomas

Re: [NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[Philippe Mathieu-Daudé]

On 11/1/24 08:10, Thomas Huth wrote:
> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>> ---
>> Used to test 
>> https://lore.kernel.org/qemu-devel/20231228084051.3235354-1-zhaotianrui@loongson.cn/
> 
> So why is it NOTFORMERGE ? Don't we want to test KVM-only builds for 
> loongarch in the long run?

We do. I was just waiting the loongarch64/KVM support was first.

Re: [NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[gaosong]

在 2024/1/11 下午5:04, Thomas Huth 写道:
> On 11/01/2024 09.50, gaosong wrote:
>> 在 2024/1/11 下午4:20, Thomas Huth 写道:
>>> On 11/01/2024 08.37, gaosong wrote:
>>>> Hi,
>>>>
>>>> 在 2024/1/11 下午3:10, Thomas Huth 写道:
>>>>> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>>>>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>>>>> ---
>>>>>> Used to test 
>>>>>> https://lore.kernel.org/qemu-devel/20231228084051.3235354-1-zhaotianrui@loongson.cn/
>>>>>
>>>>> So why is it NOTFORMERGE ? Don't we want to test KVM-only builds 
>>>>> for loongarch in the long run?
>>>>>
>>>>>  Thomas
>>>>>
>>>> I think we can drop this title.
>>>>
>>>> I tested this job by the latest loongarch kvm patches.  buf I find 
>>>> a test-hmp check error.
>>> Can you recreate the error manually? i.e. compile with configure 
>>> --disable-tcg and then run:
>>>
>>>  V=2 QTEST_QEMU_BINARY=./qemu-system-loongarch64 tests/qtest/test-hmp
>>>
>>> That should likely provide you with a hint where it is crashing
>>>
>>>  Thomas
>> Thank you,
>>
>> LoongArch no support these cmds  or some problems .
>> -    "gva2gpa 0",
>> -    "memsave 0 4096 \"/dev/null\"",
>> -    "x /8i 0x100",
>> -    "xp /16x 0",
>>
>> Could we disable these 4 cmds or the test_temp check?
>> After we fix the cmds problems, we can enable them.
>
> Even if loongarch does not support one of these commands, it should 
> not crash QEMU. So please fix the crashes first before considering to 
> enable the KVM-only test in the CI.
>

Sure,  we will fix the cmds problems first.

Thanks.
Song Gao

Re: [NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[Philippe Mathieu-Daudé]

On 11/1/24 10:51, gaosong wrote:
> 在 2024/1/11 下午5:04, Thomas Huth 写道:
>> On 11/01/2024 09.50, gaosong wrote:
>>> 在 2024/1/11 下午4:20, Thomas Huth 写道:
>>>> On 11/01/2024 08.37, gaosong wrote:

>>> LoongArch no support these cmds  or some problems .
>>> -    "gva2gpa 0",
>>> -    "memsave 0 4096 \"/dev/null\"",
>>> -    "x /8i 0x100",
>>> -    "xp /16x 0",
>>>
>>> Could we disable these 4 cmds or the test_temp check?
>>> After we fix the cmds problems, we can enable them.
>>
>> Even if loongarch does not support one of these commands, it should 
>> not crash QEMU. So please fix the crashes first before considering to 
>> enable the KVM-only test in the CI.
>>
> 
> Sure,  we will fix the cmds problems first.

The issue might be missing get_phys_page_attrs_debug() implementation.

Re: [NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[gaosong]

在 2024/1/11 下午7:26, Philippe Mathieu-Daudé 写道:
> On 11/1/24 10:51, gaosong wrote:
>> 在 2024/1/11 下午5:04, Thomas Huth 写道:
>>> On 11/01/2024 09.50, gaosong wrote:
>>>> 在 2024/1/11 下午4:20, Thomas Huth 写道:
>>>>> On 11/01/2024 08.37, gaosong wrote:
>
>>>> LoongArch no support these cmds  or some problems .
>>>> -    "gva2gpa 0",
>>>> -    "memsave 0 4096 \"/dev/null\"",
>>>> -    "x /8i 0x100",
>>>> -    "xp /16x 0",
>>>>
>>>> Could we disable these 4 cmds or the test_temp check?
>>>> After we fix the cmds problems, we can enable them.
>>>
>>> Even if loongarch does not support one of these commands, it should 
>>> not crash QEMU. So please fix the crashes first before considering 
>>> to enable the KVM-only test in the CI.
>>>
>>
>> Sure,  we will fix the cmds problems first.
>
> The issue might be missing get_phys_page_attrs_debug() implementation.

yes, I see,  from  hmp_gva2gpa().  I think we need implement it.

Thanks.
Song Gao

Re: [PATCH 1/2] gitlab: Introduce Loongarch64 runner

[gaosong]

在 2024/1/11 下午4:26, Thomas Huth 写道:
> On 11/01/2024 08.25, gaosong wrote:
>> Hi,
>>
>> 在 2024/1/11 下午3:08, Thomas Huth 写道:
>>> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>>>> Full build config to run CI tests on a Loongarch64 host.
>>>>
>>>> Forks might enable this by setting LOONGARCH64_RUNNER_AVAILABLE
>>>> in their CI namespace settings, see:
>>>> https://www.qemu.org/docs/master/devel/ci.html#maintainer-controlled-job-variables 
>>>>
>>>>
>>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>>> ---
>>>>   docs/devel/ci-jobs.rst.inc                    |  6 ++++++
>>>>   .gitlab-ci.d/custom-runners.yml               |  1 +
>>>>   .../openeuler-22.03-loongarch64.yml           | 21 
>>>> +++++++++++++++++++
>>>>   3 files changed, 28 insertions(+)
>>>>   create mode 100644 
>>>> .gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>>
>>> ...
>>>> diff --git a/.gitlab-ci.d/custom-runners.yml 
>>>> b/.gitlab-ci.d/custom-runners.yml
>>>> index 8e5b9500f4..152ace4492 100644
>>>> --- a/.gitlab-ci.d/custom-runners.yml
>>>> +++ b/.gitlab-ci.d/custom-runners.yml
>>>> @@ -32,3 +32,4 @@ include:
>>>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch64.yml'
>>>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch32.yml'
>>>>     - local: '/.gitlab-ci.d/custom-runners/centos-stream-8-x86_64.yml'
>>>> +  - local: 
>>>> '/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml'
>>>> diff --git 
>>>> a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml 
>>>> b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>> new file mode 100644
>>>> index 0000000000..86d18f820e
>>>> --- /dev/null
>>>> +++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>> @@ -0,0 +1,21 @@
>>>> +openeuler-22.03-loongarch64-all:
>>>> + extends: .custom_runner_template :-)
>>>> + needs: []
>>>> + stage: build
>>>> + tags:
>>>> + - oe2203
>>>> + - loongarch64
>>>> + rules:
>>>> + - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && 
>>>> $CI_COMMIT_BRANCH =~ /^staging/'
>>>> +   when: manual
>>>> +   allow_failure: true
>>>> + - if: "$LOONGARCH64_RUNNER_AVAILABLE"
>>>> +   when: manual
>>>> +   allow_failure: true
>>>> + script:
>>>> + - mkdir build
>>>> + - cd build
>>>> + - ../configure
>>>> +   || { cat config.log meson-logs/meson-log.txt; exit 1; }
>>>> + - make --output-sync -j`nproc --ignore=40`
>>>> + - make --output-sync -j`nproc --ignore=40` check
>>>
>>> Does this system really have more than 40 CPU threads? Or is this a 
>>> copy-n-past from one of the other scripts? In the latter case, I'd 
>>> suggest to adjust the --ignore=40 to a more reasonable value.
>>>
>>>  Thomas
>> No,  only 32.   I think it should be --ignore=32 or 16.
>
> --ignore=32 then also does not make much sense, that would still be 
> the same as simply omitting the -j parameter. I guess --ignore=16 
> should be fine.
>
>> I create a same runner on this machine, and I  find  some check error.
>> but I am not sure how to fix it. :-)
>>
>> See:
>>
>> https://gitlab.com/gaosong/qemu/-/jobs/5906269934
>
> Seems to be related to RAM backing... for example, the erst-test is 
> failing, which is doing something like:
>
>     setup_vm_cmd(&state,
>         "-object memory-backend-file,"
>             "mem-path=acpi-erst.XXXXXX,"
>             "size=64K,"
>             "share=on,"
>             "id=nvram "
>         "-device acpi-erst,"
>             "memdev=nvram");
>
> So it seems like -object memory-backend-file" is not correctly working 
> in your gitlab runner? Is there some setup missing?
>
>  Thomas
>
>
This is my runner config.

     concurrent = 32
     check_interval = 0
     shutdown_timeout = 0

     [session_server]
       session_timeout = 1800

     ...

     [[runners]]
       name = "loongarch64"
       request_concurrency = 24
       url = "https://gitlab.com"
       id = 31426483
       token = "glrt-bGugocYrR2yqcu3ma7ka"
       token_obtained_at = 2024-01-10T08:31:29Z
       token_expires_at = 0001-01-01T00:00:00Z
       executor = "shell"
       builds_dir = "/data/gitlab-runner/builds"
       cache_dir = "/data/gitlab-runner/cache"
       [runners.cache]
         MaxUploadedArchiveSize = 0

I create a project runner 'loongarch64' for my branch ci-master.
Do we need some special configuration?

and  I just './configure  --target-list="x86_64-softmmu" 
--enable-debug'  on LoongArch host.   make , make check.
I got the same error.

[7/361] 🌗 qemu:qtest+qtest-x86_64 / 
qtest-x86_64/qos-test                                    2928s   60 
subtests passed

^CWARNING: Received SIGTERM, exiting
   7/361 qemu:qtest+qtest-x86_64 / 
qtest-x86_64/qos-test                        INTERRUPT 4671.32s   killed 
by signal 11 SIGSEGV
 >>> PYTHON=/home/gs/gitlab/qemu/build/pyvenv/bin/python3 
G_TEST_DBUS_DAEMON=/home/gs/gitlab/qemu/tests/dbus-vmstate-daemon.sh 
MALLOC_PERTURB_=198 QTEST_QEMU_IMG=./qemu-img 
QTEST_QEMU_BINARY=./qemu-system-x86_64 
QTEST_QEMU_STORAGE_DAEMON_BINARY=./storage-daemon/qemu-storage-daemon 
/home/gs/gitlab/qemu/build/tests/qtest/qos-test --tap -k
――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――― ✀ 
―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
stderr:
qemu-system-x86_64: unable to map backing store for guest RAM: Invalid 
argument
Broken pipe
../tests/qtest/libqtest.c:195: kill_qemu() tried to terminate QEMU 
process but encountered exit status 1 (expected 0)

TAP parsing error: Too few tests run (expected 119, got 60)


I also use V=2  to check.


  PYTHON=/home/gs/gitlab/qemu/build/pyvenv/bin/python3 
G_TEST_DBUS_DAEMON=/home/gs/gitlab/qemu/tests/dbus-vmstate-daemon.sh 
QTEST_QEMU_IMG=./qemu-img V=2 QTEST_QEMU_BINARY=./qemu-system-x86_64 
QTEST_QEMU_STORAGE_DAEMON_BINARY=./storage-daemon/qemu-storage-daemon 
MALLOC_PERTURB_=220 
/home/gs/gitlab/qemu/build/tests/qtest/bios-tables-test --tap -k

I got the error:

...
Looking for expected file 'tests/data/acpi/pc/FACS'
Using expected file 'tests/data/acpi/pc/FACS'
Looking for expected file 'tests/data/acpi/pc/DSDT.acpihmat'
Using expected file 'tests/data/acpi/pc/DSDT.acpihmat'
ok 15 /x86_64/acpi/piix4/acpihmat
# starting QEMU: exec ./qemu-system-x86_64 -qtest 
unix:/tmp/qtest-923048.sock -qtest-log /dev/null -chardev 
socket,path=/tmp/qtest-923048.qmp,id=char0 -mon 
chardev=char0,mode=control -display none -audio none -machine pc -accel 
kvm -accel tcg -net none  -object 
memory-backend-file,id=erstnvram,mem-path=/tmp/qemu-test-erst.N41SG2,size=0x10000,share=on 
-device acpi-erst,memdev=erstnvram -drive 
id=hd0,if=none,file=tests/acpi-test-disk-KIDwOM,format=raw -device 
ide-hd,drive=hd0  -accel qtest
qemu-system-x86_64: unable to map backing store for guest RAM: Invalid 
argument
Broken pipe
../tests/qtest/libqtest.c:195: kill_qemu() tried to terminate QEMU 
process but encountered exit status 1 (expected 0)
Aborted (core dumped)

is there need some  special  package for '-object memory-backend-file' ?

Thanks.
Song Gao

Re: [PATCH 1/2] gitlab: Introduce Loongarch64 runner

[Philippe Mathieu-Daudé]

On 11/1/24 13:05, gaosong wrote:
> 在 2024/1/11 下午4:26, Thomas Huth 写道:
>> On 11/01/2024 08.25, gaosong wrote:
>>> Hi,
>>>
>>> 在 2024/1/11 下午3:08, Thomas Huth 写道:
>>>> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>>>>> Full build config to run CI tests on a Loongarch64 host.
>>>>>
>>>>> Forks might enable this by setting LOONGARCH64_RUNNER_AVAILABLE
>>>>> in their CI namespace settings, see:
>>>>> https://www.qemu.org/docs/master/devel/ci.html#maintainer-controlled-job-variables
>>>>>
>>>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>>>> ---
>>>>>   docs/devel/ci-jobs.rst.inc                    |  6 ++++++
>>>>>   .gitlab-ci.d/custom-runners.yml               |  1 +
>>>>>   .../openeuler-22.03-loongarch64.yml           | 21 
>>>>> +++++++++++++++++++
>>>>>   3 files changed, 28 insertions(+)
>>>>>   create mode 100644 
>>>>> .gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>>>
>>>> ...
>>>>> diff --git a/.gitlab-ci.d/custom-runners.yml 
>>>>> b/.gitlab-ci.d/custom-runners.yml
>>>>> index 8e5b9500f4..152ace4492 100644
>>>>> --- a/.gitlab-ci.d/custom-runners.yml
>>>>> +++ b/.gitlab-ci.d/custom-runners.yml
>>>>> @@ -32,3 +32,4 @@ include:
>>>>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch64.yml'
>>>>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch32.yml'
>>>>>     - local: '/.gitlab-ci.d/custom-runners/centos-stream-8-x86_64.yml'
>>>>> +  - local: 
>>>>> '/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml'
>>>>> diff --git 
>>>>> a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml 
>>>>> b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>>> new file mode 100644
>>>>> index 0000000000..86d18f820e
>>>>> --- /dev/null
>>>>> +++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>>> @@ -0,0 +1,21 @@
>>>>> +openeuler-22.03-loongarch64-all:
>>>>> + extends: .custom_runner_template :-)
>>>>> + needs: []
>>>>> + stage: build
>>>>> + tags:
>>>>> + - oe2203
>>>>> + - loongarch64
>>>>> + rules:
>>>>> + - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && 
>>>>> $CI_COMMIT_BRANCH =~ /^staging/'
>>>>> +   when: manual
>>>>> +   allow_failure: true
>>>>> + - if: "$LOONGARCH64_RUNNER_AVAILABLE"
>>>>> +   when: manual
>>>>> +   allow_failure: true
>>>>> + script:
>>>>> + - mkdir build
>>>>> + - cd build
>>>>> + - ../configure
>>>>> +   || { cat config.log meson-logs/meson-log.txt; exit 1; }
>>>>> + - make --output-sync -j`nproc --ignore=40`
>>>>> + - make --output-sync -j`nproc --ignore=40` check
>>>>
>>>> Does this system really have more than 40 CPU threads? Or is this a 
>>>> copy-n-past from one of the other scripts? In the latter case, I'd 
>>>> suggest to adjust the --ignore=40 to a more reasonable value.
>>>>
>>>>  Thomas
>>> No,  only 32.   I think it should be --ignore=32 or 16.
>>
>> --ignore=32 then also does not make much sense, that would still be 
>> the same as simply omitting the -j parameter. I guess --ignore=16 
>> should be fine.
>>
>>> I create a same runner on this machine, and I  find  some check error.
>>> but I am not sure how to fix it. :-)
>>>
>>> See:
>>>
>>> https://gitlab.com/gaosong/qemu/-/jobs/5906269934
>>
>> Seems to be related to RAM backing... for example, the erst-test is 
>> failing, which is doing something like:
>>
>>     setup_vm_cmd(&state,
>>         "-object memory-backend-file,"
>>             "mem-path=acpi-erst.XXXXXX,"
>>             "size=64K,"
>>             "share=on,"
>>             "id=nvram "
>>         "-device acpi-erst,"
>>             "memdev=nvram");
>>
>> So it seems like -object memory-backend-file" is not correctly working 
>> in your gitlab runner? Is there some setup missing?
>>
>>  Thomas
>>
>>
> This is my runner config.
> 
>      concurrent = 32
>      check_interval = 0
>      shutdown_timeout = 0
> 
>      [session_server]
>        session_timeout = 1800
> 
>      ...
> 
>      [[runners]]
>        name = "loongarch64"
>        request_concurrency = 24
>        url = "https://gitlab.com"
>        id = 31426483
>        token = "glrt-bGugocYrR2yqcu3ma7ka"
>        token_obtained_at = 2024-01-10T08:31:29Z
>        token_expires_at = 0001-01-01T00:00:00Z
>        executor = "shell"
>        builds_dir = "/data/gitlab-runner/builds"
>        cache_dir = "/data/gitlab-runner/cache"
>        [runners.cache]
>          MaxUploadedArchiveSize = 0
> 
> I create a project runner 'loongarch64' for my branch ci-master.
> Do we need some special configuration?

We need to follow this process:
https://lore.kernel.org/qemu-devel/20240104160805.56856-1-philmd@linaro.org/

Re: [PATCH 1/2] gitlab: Introduce Loongarch64 runner

[gaosong]

在 2024/1/11 下午4:26, Thomas Huth 写道:
> On 11/01/2024 08.25, gaosong wrote:
>> Hi,
>>
>> 在 2024/1/11 下午3:08, Thomas Huth 写道:
>>> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>>>> Full build config to run CI tests on a Loongarch64 host.
>>>>
>>>> Forks might enable this by setting LOONGARCH64_RUNNER_AVAILABLE
>>>> in their CI namespace settings, see:
>>>> https://www.qemu.org/docs/master/devel/ci.html#maintainer-controlled-job-variables 
>>>>
>>>>
>>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>>> ---
>>>>   docs/devel/ci-jobs.rst.inc                    |  6 ++++++
>>>>   .gitlab-ci.d/custom-runners.yml               |  1 +
>>>>   .../openeuler-22.03-loongarch64.yml           | 21 
>>>> +++++++++++++++++++
>>>>   3 files changed, 28 insertions(+)
>>>>   create mode 100644 
>>>> .gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>>
>>> ...
>>>> diff --git a/.gitlab-ci.d/custom-runners.yml 
>>>> b/.gitlab-ci.d/custom-runners.yml
>>>> index 8e5b9500f4..152ace4492 100644
>>>> --- a/.gitlab-ci.d/custom-runners.yml
>>>> +++ b/.gitlab-ci.d/custom-runners.yml
>>>> @@ -32,3 +32,4 @@ include:
>>>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch64.yml'
>>>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch32.yml'
>>>>     - local: '/.gitlab-ci.d/custom-runners/centos-stream-8-x86_64.yml'
>>>> +  - local: 
>>>> '/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml'
>>>> diff --git 
>>>> a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml 
>>>> b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>> new file mode 100644
>>>> index 0000000000..86d18f820e
>>>> --- /dev/null
>>>> +++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>> @@ -0,0 +1,21 @@
>>>> +openeuler-22.03-loongarch64-all:
>>>> + extends: .custom_runner_template :-)
>>>> + needs: []
>>>> + stage: build
>>>> + tags:
>>>> + - oe2203
>>>> + - loongarch64
>>>> + rules:
>>>> + - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && 
>>>> $CI_COMMIT_BRANCH =~ /^staging/'
>>>> +   when: manual
>>>> +   allow_failure: true
>>>> + - if: "$LOONGARCH64_RUNNER_AVAILABLE"
>>>> +   when: manual
>>>> +   allow_failure: true
>>>> + script:
>>>> + - mkdir build
>>>> + - cd build
>>>> + - ../configure
>>>> +   || { cat config.log meson-logs/meson-log.txt; exit 1; }
>>>> + - make --output-sync -j`nproc --ignore=40`
>>>> + - make --output-sync -j`nproc --ignore=40` check
>>>
>>> Does this system really have more than 40 CPU threads? Or is this a 
>>> copy-n-past from one of the other scripts? In the latter case, I'd 
>>> suggest to adjust the --ignore=40 to a more reasonable value.
>>>
>>>  Thomas
>> No,  only 32.   I think it should be --ignore=32 or 16.
>
> --ignore=32 then also does not make much sense, that would still be 
> the same as simply omitting the -j parameter. I guess --ignore=16 
> should be fine.
>
>> I create a same runner on this machine, and I  find  some check error.
>> but I am not sure how to fix it. :-)
>>
>> See:
>>
>> https://gitlab.com/gaosong/qemu/-/jobs/5906269934
>
> Seems to be related to RAM backing... for example, the erst-test is 
> failing, which is doing something like:
>
>     setup_vm_cmd(&state,
>         "-object memory-backend-file,"
>             "mem-path=acpi-erst.XXXXXX,"
>             "size=64K,"
Hi,

We tested this  on
LoongArch host with the 5.10 kernel,  (openEuler 22.03),
x86_64 host with 5.10 kernel, (openEuler 22.03)
x86_64 host with 5.15kernel , (Ubuntu 20.04.3 LTS)

and didn't get any error.

but the CI machine use the  6.7_rc4 kernel.
we didn't update the x86_64 host kernel to tested this.

Is it possible that the new kernel is causing the problem?

> "share=on,"
>             "id=nvram "
>         "-device acpi-erst,"
>             "memdev=nvram");
>
> So it seems like -object memory-backend-file" is not correctly working 
> in your gitlab runner? Is there some setup missing?
>
>  Thomas
>
>

Re: [PATCH 1/2] gitlab: Introduce Loongarch64 runner

[gaosong]

在 2024/1/12 下午5:52, gaosong 写道:
> 在 2024/1/11 下午4:26, Thomas Huth 写道:
>> On 11/01/2024 08.25, gaosong wrote:
>>> Hi,
>>>
>>> 在 2024/1/11 下午3:08, Thomas Huth 写道:
>>>> On 02/01/2024 18.22, Philippe Mathieu-Daudé wrote:
>>>>> Full build config to run CI tests on a Loongarch64 host.
>>>>>
>>>>> Forks might enable this by setting LOONGARCH64_RUNNER_AVAILABLE
>>>>> in their CI namespace settings, see:
>>>>> https://www.qemu.org/docs/master/devel/ci.html#maintainer-controlled-job-variables 
>>>>>
>>>>>
>>>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>>>> ---
>>>>>   docs/devel/ci-jobs.rst.inc                    |  6 ++++++
>>>>>   .gitlab-ci.d/custom-runners.yml               |  1 +
>>>>>   .../openeuler-22.03-loongarch64.yml           | 21 
>>>>> +++++++++++++++++++
>>>>>   3 files changed, 28 insertions(+)
>>>>>   create mode 100644 
>>>>> .gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>>>
>>>> ...
>>>>> diff --git a/.gitlab-ci.d/custom-runners.yml 
>>>>> b/.gitlab-ci.d/custom-runners.yml
>>>>> index 8e5b9500f4..152ace4492 100644
>>>>> --- a/.gitlab-ci.d/custom-runners.yml
>>>>> +++ b/.gitlab-ci.d/custom-runners.yml
>>>>> @@ -32,3 +32,4 @@ include:
>>>>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch64.yml'
>>>>>     - local: '/.gitlab-ci.d/custom-runners/ubuntu-22.04-aarch32.yml'
>>>>>     - local: 
>>>>> '/.gitlab-ci.d/custom-runners/centos-stream-8-x86_64.yml'
>>>>> +  - local: 
>>>>> '/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml'
>>>>> diff --git 
>>>>> a/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml 
>>>>> b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>>> new file mode 100644
>>>>> index 0000000000..86d18f820e
>>>>> --- /dev/null
>>>>> +++ b/.gitlab-ci.d/custom-runners/openeuler-22.03-loongarch64.yml
>>>>> @@ -0,0 +1,21 @@
>>>>> +openeuler-22.03-loongarch64-all:
>>>>> + extends: .custom_runner_template :-)
>>>>> + needs: []
>>>>> + stage: build
>>>>> + tags:
>>>>> + - oe2203
>>>>> + - loongarch64
>>>>> + rules:
>>>>> + - if: '$CI_PROJECT_NAMESPACE == "qemu-project" && 
>>>>> $CI_COMMIT_BRANCH =~ /^staging/'
>>>>> +   when: manual
>>>>> +   allow_failure: true
>>>>> + - if: "$LOONGARCH64_RUNNER_AVAILABLE"
>>>>> +   when: manual
>>>>> +   allow_failure: true
>>>>> + script:
>>>>> + - mkdir build
>>>>> + - cd build
>>>>> + - ../configure
>>>>> +   || { cat config.log meson-logs/meson-log.txt; exit 1; }
>>>>> + - make --output-sync -j`nproc --ignore=40`
>>>>> + - make --output-sync -j`nproc --ignore=40` check
>>>>
>>>> Does this system really have more than 40 CPU threads? Or is this a 
>>>> copy-n-past from one of the other scripts? In the latter case, I'd 
>>>> suggest to adjust the --ignore=40 to a more reasonable value.
>>>>
>>>>  Thomas
>>> No,  only 32.   I think it should be --ignore=32 or 16.
>>
>> --ignore=32 then also does not make much sense, that would still be 
>> the same as simply omitting the -j parameter. I guess --ignore=16 
>> should be fine.
>>
>>> I create a same runner on this machine, and I  find  some check error.
>>> but I am not sure how to fix it. :-)
>>>
>>> See:
>>>
>>> https://gitlab.com/gaosong/qemu/-/jobs/5906269934
>>
>> Seems to be related to RAM backing... for example, the erst-test is 
>> failing, which is doing something like:
>>
>>     setup_vm_cmd(&state,
>>         "-object memory-backend-file,"
>>             "mem-path=acpi-erst.XXXXXX,"
>>             "size=64K,"
> Hi,
>
> We tested this  on
> LoongArch host with the 5.10 kernel,  (openEuler 22.03),
> x86_64 host with 5.10 kernel, (openEuler 22.03)
> x86_64 host with 5.15kernel , (Ubuntu 20.04.3 LTS)
>
> and didn't get any error.
>
> but the CI machine use the  6.7_rc4 kernel.
> we didn't update the x86_64 host kernel to tested this.
>
> Is it possible that the new kernel is causing the problem?
>
Hi,

The kernel adds the patch[1] can fix this problem.

[1] 
https://patchew.org/linux/20240106145501.3370364-1-chenhuacai@loongson.cn/

So
Tested-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Song Gao <gaosong@loongson.cn>

Thanks.
Song Gao

>> "share=on,"
>>             "id=nvram "
>>         "-device acpi-erst,"
>>             "memdev=nvram");
>>
>> So it seems like -object memory-backend-file" is not correctly 
>> working in your gitlab runner? Is there some setup missing?
>>
>>  Thomas
>>
>>
>

Re: [NOTFORMERGE PATCH 2/2] gitlab: Add Loongarch64 KVM-only build

[gaosong]

Hi,

在 2024/1/11 下午7:26, Philippe Mathieu-Daudé 写道:
> On 11/1/24 10:51, gaosong wrote:
>> 在 2024/1/11 下午5:04, Thomas Huth 写道:
>>> On 11/01/2024 09.50, gaosong wrote:
>>>> 在 2024/1/11 下午4:20, Thomas Huth 写道:
>>>>> On 11/01/2024 08.37, gaosong wrote:
>
>>>> LoongArch no support these cmds  or some problems .
>>>> -    "gva2gpa 0",
>>>> -    "memsave 0 4096 \"/dev/null\"",
>>>> -    "x /8i 0x100",
>>>> -    "xp /16x 0",
>>>>
>>>> Could we disable these 4 cmds or the test_temp check?
>>>> After we fix the cmds problems, we can enable them.
>>>
>>> Even if loongarch does not support one of these commands, it should 
>>> not crash QEMU. So please fix the crashes first before considering 
>>> to enable the KVM-only test in the CI.
>>>
>>
>> Sure,  we will fix the cmds problems first.
>
> The issue might be missing get_phys_page_attrs_debug() implementation.

We use  '--enable-kvm --disable-tcg ',
the get_phys_page_debug()  is NULL,  so the test-hmp failed.

target/loongarch/cpu.c
...
#ifndef CONFIG_USER_ONLY
#include "hw/core/sysemu-cpu-ops.h"

static const struct SysemuCPUOps loongarch_sysemu_ops = {
#ifdef CONFIG_TCG
     .get_phys_page_debug = loongarch_cpu_get_phys_page_debug,
#endif
};
...

I see the other architectures  only implement 
get_phys_page_attrs_debug() or  get_phys_page_debug()
and not only build these functions on tcg mode.

Should we need implement  get_phys_page_attrs_debug() ?  or just use 
get_phys_page_debug()

Thanks.
Song Gao