NAK: [PATCH Trusty SRU] Revert "ipmi: simplify locking"

[Stefan Bader]

On 19.11.2014 15:47, Chris J Arges wrote:
> On 11/19/2014 03:12 AM, Stefan Bader wrote:
>> That patch is in the kernel since 3.4 and it does not look like we
>> even waited for the feedback about 3.16 kernels. Before randomly
>> ripping out a change from the beginning of the delta between 3.2
>> and 3.13 and maybe risking other problems I would at least wait for
>> the feedback about 3.16.
>>
> Stefan,
> Fair enough, I suspect there are about 3-4 patches that may be needed
> to backport into 3.13. If that is too complex I still think reverting
> f60adf42a makes sense because its a reduction in locking vs removing a
> necessary feature.
> --chris
> 

My major worries would be that since there is several more patches since 3.4 and
3.14, that might subtly rely on the change to revert. And that a positive
feedback from one tester (iirc the same code covers multiple ways to expose the
impi interface as well as methods to access it, some need polling some have
interrupts) may not uncover all possibilities.

Having looked at the code again in more detail, I find the locking not very easy
to grasp. While comments for smi_event_handler say must be called with the lock
and interrupts disabled, it is not always obvious that that is done.

My hopes would be that its something on top like

commit 48e8ac2979920ffa39117e2d725afa3a749bfe8d
Author: Bodo Stroesser <bstroesser at ts.fujitsu.com>
Date:   Mon Apr 14 09:46:51 2014 -0500

    ipmi: Fix a race restarting the timer

That way we would at least go forward and not deviate from upstream in a way we
most certainly will forget for future releases. If its still broken upstream
maybe we could get it fixed...

Stefan

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <https://lists.ubuntu.com/archives/kernel-team/attachments/20141119/d8ac5991/attachment.sig>