On 9/23/22 23:12, Richard Purdie wrote:
> On Fri, 2022-09-23 at 11:18 +0800, He Zhe wrote:
>> lttng-tools ptest requires SYS_ppoll and SYS_pselect which are not
>> supported by qemuriscv32.
>>
>> Back port a not merged patch from upstream to be able to disable the test
>> build and disable the ptest build for qemuriscv32. This way the main
>> package can still be used or at least built.
>>
>> See the following link for more details.
>> https://github.com/lttng/lttng-tools/pull/162
> This commit message and patch are confused. It isn't a backport if it
> wasn't merged upstream first.
>
> The patch below has:
>
> "Upstream-Status: Inappropriate [oe-core specific]"
>
> which isn't really true either. The correct status may be Submitted or
> possibly Rejected but not Inappropriate. At the very least the link to
> the github issue should be in the patch header to explain the status
> and let people get more information.

Thanks for pointing out. I should have added the link here.

>
>> Signed-off-by: He Zhe <[email protected]>
>> ---
>>  meta/recipes-kernel/lttng/lttng-platforms.inc |  4 ++
>>  .../0001-configure.ac-add-disable-tests.patch | 38 +++++++++++++++++++
>>  .../lttng/lttng-tools_2.13.8.bb               |  6 ++-
>>  3 files changed, 46 insertions(+), 2 deletions(-)
>>  create mode 100644 
>> meta/recipes-kernel/lttng/lttng-tools/0001-configure.ac-add-disable-tests.patch
>>
>> diff --git a/meta/recipes-kernel/lttng/lttng-platforms.inc 
>> b/meta/recipes-kernel/lttng/lttng-platforms.inc
>> index 933c65d85d..2351df08b8 100644
>> --- a/meta/recipes-kernel/lttng/lttng-platforms.inc
>> +++ b/meta/recipes-kernel/lttng/lttng-platforms.inc
>> @@ -15,3 +15,7 @@ LTTNGUST:arc = ""
>>  
>>  COMPATIBLE_HOST:arc:pn-lttng-ust = "null"
>>  
>> +# Whether the platform supports lttng-tools-ptest
>> +# lttng-tools-ptest uses syscall SYS_ppoll and SYS_pselect6 which are not 
>> supported for some platforms.
>> +LTTNGTOOLSPTEST = "ptest"
>> +LTTNGTOOLSPTEST:qemuriscv32 = ""
> If the lttng-tools ptests don't work, will lttng itself work on the
> platform? Would we be better off simply disabling lttng on the platform
> until it is supported?

There is a comment in the issue saying that "disabling them is a quick-win to
make it possible to use a package on such an architecture." I thought it might
make some sense if we could at least build the main package, so I tended to just
disable the ptest. But, yes, now that riscv32 is not officially supported we
should disable lttng completely. I'll send a patch for that.

https://github.com/lttng/lttng-tools/pull/162


Thanks,
Zhe

>
> Cheers,
>
> Richard

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#171051): 
https://lists.openembedded.org/g/openembedded-core/message/171051
Mute This Topic: https://lists.openembedded.org/mt/93863613/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to