mbox series

[BUGFIX,0/6] block, bfq: series of fixes, and not only, for some recently reported issues

Message ID 20200131092409.10867-1-paolo.valente@linaro.org (mailing list archive)
Headers show
Series block, bfq: series of fixes, and not only, for some recently reported issues | expand

Message

Paolo Valente Jan. 31, 2020, 9:24 a.m. UTC
Hi Jens,
these patches are mostly fixes for the issues reported in [1, 2]. All
patches have been publicly tested in the dev version of BFQ.

Thanks,
Paolo

[1] https://bugzilla.redhat.com/show_bug.cgi?id=1767539
[2] https://bugzilla.kernel.org/show_bug.cgi?id=205447

Paolo Valente (6):
  block, bfq: do not plug I/O for bfq_queues with no proc refs
  block, bfq: do not insert oom queue into position tree
  block, bfq: get extra ref to prevent a queue from being freed during a
    group move
  block, bfq: extend incomplete name of field on_st
  block, bfq: get a ref to a group when adding it to a service tree
  block, bfq: clarify the goal of bfq_split_bfqq()

 block/bfq-cgroup.c  | 12 ++++++++++--
 block/bfq-iosched.c | 20 +++++++++++++++++++-
 block/bfq-iosched.h |  3 ++-
 block/bfq-wf2q.c    | 27 ++++++++++++++++++++++-----
 4 files changed, 53 insertions(+), 9 deletions(-)

--
2.20.1

Comments

Jens Axboe Feb. 1, 2020, 4:48 a.m. UTC | #1
On 1/31/20 2:24 AM, Paolo Valente wrote:
> Hi Jens,
> these patches are mostly fixes for the issues reported in [1, 2]. All
> patches have been publicly tested in the dev version of BFQ.
> 
> Thanks,
> Paolo
> 
> [1] https://bugzilla.redhat.com/show_bug.cgi?id=1767539
> [2] https://bugzilla.kernel.org/show_bug.cgi?id=205447
> 
> Paolo Valente (6):
>   block, bfq: do not plug I/O for bfq_queues with no proc refs
>   block, bfq: do not insert oom queue into position tree
>   block, bfq: get extra ref to prevent a queue from being freed during a
>     group move
>   block, bfq: extend incomplete name of field on_st
>   block, bfq: get a ref to a group when adding it to a service tree
>   block, bfq: clarify the goal of bfq_split_bfqq()
> 
>  block/bfq-cgroup.c  | 12 ++++++++++--
>  block/bfq-iosched.c | 20 +++++++++++++++++++-
>  block/bfq-iosched.h |  3 ++-
>  block/bfq-wf2q.c    | 27 ++++++++++++++++++++++-----
>  4 files changed, 53 insertions(+), 9 deletions(-)

I wish some of these had been sent sooner, they really should have been
sent in a few weeks ago. Just took a quick look at the bug reports, and
at least one of the bugs mentions looks like it had a fix available 2
months ago. Have they been in -next? They are all marked as bug fixes,
should they have stable tags? All of them, some of them?
Paolo Valente Feb. 3, 2020, 8:50 a.m. UTC | #2
> Il giorno 1 feb 2020, alle ore 05:48, Jens Axboe <axboe@kernel.dk> ha scritto:
> 
> On 1/31/20 2:24 AM, Paolo Valente wrote:
>> Hi Jens,
>> these patches are mostly fixes for the issues reported in [1, 2]. All
>> patches have been publicly tested in the dev version of BFQ.
>> 
>> Thanks,
>> Paolo
>> 
>> [1] https://bugzilla.redhat.com/show_bug.cgi?id=1767539
>> [2] https://bugzilla.kernel.org/show_bug.cgi?id=205447
>> 
>> Paolo Valente (6):
>>  block, bfq: do not plug I/O for bfq_queues with no proc refs
>>  block, bfq: do not insert oom queue into position tree
>>  block, bfq: get extra ref to prevent a queue from being freed during a
>>    group move
>>  block, bfq: extend incomplete name of field on_st
>>  block, bfq: get a ref to a group when adding it to a service tree
>>  block, bfq: clarify the goal of bfq_split_bfqq()
>> 
>> block/bfq-cgroup.c  | 12 ++++++++++--
>> block/bfq-iosched.c | 20 +++++++++++++++++++-
>> block/bfq-iosched.h |  3 ++-
>> block/bfq-wf2q.c    | 27 ++++++++++++++++++++++-----
>> 4 files changed, 53 insertions(+), 9 deletions(-)
> 
> I wish some of these had been sent sooner, they really should have been
> sent in a few weeks ago. Just took a quick look at the bug reports, and
> at least one of the bugs mentions looks like it had a fix available 2
> months ago.

The first fix(es) didn't work with the issue reported in [2], which
was in turn very similar to that in [1].  Since I didn't know why, I
couldn't be sure that the first fix was correct and did not introduce
other issues.

> Have they been in -next?

Nope. I proposed the full series in this thread, the day after the
full fix was confirmed to work.  I didn't propose any partial series
patch before, for the above reason.

> They are all marked as bug fixes,
> should they have stable tags?

I guess they should, as fixes to bugs that may cause crashes.  If
there are other rules for these tags, I'm sorry but I'm not aware of
them.

> All of them, some of them?

The only two non-fix patches are non-functional, trivial code
improvements made along the way.

Submitting a V2.

Thanks,
Paolo

> 
> -- 
> Jens Axboe