On Thu, 2007-08-02 at 15:06 -0700, David Miller wrote:
quoted text
> From: "Michael Chan" <mchan@broadcom.com>
> Date: Thu, 02 Aug 2007 12:10:29 -0700
>
> > Alternatively, we can also fix it by calling pci_enable_device() again
> > in tg3_open().  But I think it is better to just always save and restore
> > in suspend/resume.  bnx2.c will also require the same fix.
> 
> We could do it that way.  But don't you think it's more reliable to
> save and restore around the event we know will be what clobbers the
> PCI config space on us? :-)
> 

Yes for sure when netif state is running and we were already doing that.

quoted text
> Other things might happen between ->resume() and ->open() that could
> modify PCI config space, and we could overwrite such changes if we do
> the PCI restore in ->open().

I suggested calling pci_enable_device() in ->open(), not calling
pci_restore_state() in ->open().  I ultimately decided against it
because some devices do not enable memory as a workaround and it would
be messy to deal with it again in tg3_open().

I definitely agree that calling PCI restore in ->open() is a bad idea.
We used to save PCI state in ->probe() once and restore PCI state after
every chip reset.  This sequence caused many subtle problems.

-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/