[SRU][Groovy][PATCH 0/2] Revert tty changes to fix regression

[Krzysztof Kozlowski]

On 21/04/2021 19:47, Krzysztof Kozlowski wrote:
> On 21/04/2021 19:16, Kleber Sacilotto de Souza wrote:
>> BugLink: https://bugs.launchpad.net/bugs/1925290
>>
>> [Impact]
>> Testcase hangup01 from ubuntu_ltp_stable.pty is failing on Groovy 5.8.0-51.57.
>>
>> Test case failed with:
>>  hangup01 1 TFAIL : hangup01.c:132: unexpected message 3
>>  tag=hangup01 stime=1618856629 dur=1 exit=exited stat=1
>>
>> It was found that the regression was introduced by the following commits applied
>> via stable uddates:
>>
>> dd78b0c483e3 tty: implement read_iter
>> 3b830a9c34d5 tty: convert tty_ldisc_ops 'read()' function to take a kernel pointer
>>
>> These patches were applied via bug 1921960 ("Groovy update: upstream stable
>> patchset 2021-03-30").
>>
>> [Fix]
>> Revert the patches mentioned above until a fix can be found.
>>
>> [Test case]
>> Run the pty testsuite from ubuntu_ltp_stable.
>>
>> [Where problems could occur]
>> If reverting these two patches is not enough to bring the code to the behavior
>> of the tty driver before 5.8.0-51.57 then remaining issues could still be
>> present on the tty code. Those issues should be caught by the ltp testcases
>> though.
> 
> Did you notify it (or plan to) to Greg/upstream about these commits?
> Maybe they would have the answers for cause.

I'm not sure whether there is need because these should not get into
v5.8 in the first place. The 36e2c7421f02ch which is fixed by
dd78b0c483e3 got into v5.10. Not into v5.8.

The 3b830a9c34d5 does not fixes anything and was not backported to
stable. How did we get it?


Best regards,
Krzysztof