[OpenAFS] Re: multiplt kernel call traces
Sat, 20 Mar 2010 16:42:33 -0400
On Sat, Mar 20, 2010 at 4:37 PM, Buhrmaster, Gary <email@example.com> w=
>> From: firstname.lastname@example.org [mailto:openafs-info-admin@openafs.=
org] On Behalf Of Marc Dionne
>> On Mon, Mar 15, 2010 at 11:05 AM, Andrew Deason <email@example.com>=
>> > On Mon, 15 Mar 2010 09:16:48 -0500
>> > Andrew Deason <firstname.lastname@example.org> wrote:
>> >> On Mon, 15 Mar 2010 07:08:48 -0700 (PDT)
>> >> email@example.com wrote:
>> >> > As a follow up to your response- will this problem "go away" for
>> >> > kernel version 2.6.33? =A0If so, perhaps I can upgrade to that vers=
>> >> >
>> >> > As an alternative- can I fix this by downgrading the kernel to some
>> >> > lesser version (say 2.6.30)?
>> >> "The IMA problem" effectively exists in 2.6.30-2.6.32.
>> > ...however, Simon tells me that the fedora 2.6.30 packages may not hav=
>> > enabled IMA, so he's probably right that 2.6.30 would be fine.
>> Yeah, pretty sure there was no Fedora 2.6.30 kernel with IMA enabled,
>> that came with 2.6.31.
>> The problem with the IMA fix as it was done in 2.6.33 is that it makes
>> substantial changes to how/where the IMA accounting is done - probably
>> not something you'd want as-is for 2.6.32-stable.
>> What I hope is that Fedora 12 will get an update to 2.6.33 in the not
>> too distant future. =A0If we take 2.6.32 as an example, it was added to
>> F12 roughly a month after its release, and 2.6.33 was released on Feb.
>> 24. =A0I don't know what the plan is, and looking in Koji I haven't seen
>> any 2.6.33 builds targeted for F12 yet.
> 2.6.33 may have some badness ahead too. =A0Somewhere in
> the release cycle the INIT_WORK definition changed to
> call the (new) __init_work function, which is GPL only.
> This breaks the released OpenAFS kernel modules.
Only if CONFIG_DEBUG_OBJECTS_WORK is set, otherwise __init_work is
just an empty inline function. This shouldn't be set for release