Messages in this thread |  | | | Date | Mon, 26 May 2008 10:58:24 +0200 | | From | Ingo Molnar <> | | Subject | Re: [patch, -git] pcie hotplug bootup crash fix | |
* Andrew Morton <akpm@linux-foundation.org> wrote:
> On Mon, 26 May 2008 10:47:09 +0200 Ingo Molnar <mingo@elte.hu> wrote:
>
> >
> > * Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> wrote:
> >
> > > I updated Ingo's patch. If it's ok, I'll send it to Jess Barnes with
> > > some other patches for the other pciehp regression problems.
> >
> > looks good to me, thanks Kenji.
> >
>
> It's a bit sad to add a large workaround like this. I'm surprised
> that fixing it properly is considered unviable for 2.6.26. Normally
> these fixes are pretty simple - just request the IRQ a bit later?
hm, will that solve the problem? These irqs can be shared and there can
be multiple of them, so making the handlers robust against
half-constructed global state seems inevitable (the other option would
be extra locking). But i've only looked briefly ...
Ingo
|  |