Patch 2/2
This adds support for struct pci_driver shutdown and negates the previous NAK'ed
reboot_notifier patch. It's much easier, wish I had know about this before. Thanks to
Christoph for pointing this out.
Signed-off-by: Mike Miller <[email protected]>
------------------------------------------------------------------------------------------
diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c
index 916aab0..1abf1f5 100644
--- a/drivers/block/cciss.c
+++ b/drivers/block/cciss.c
@@ -3428,9 +3428,10 @@ static void __devexit cciss_remove_one(s
memset(flush_buf, 0, 4);
return_code = sendcmd(CCISS_CACHE_FLUSH, i, flush_buf, 4, 0, 0, 0, NULL,
TYPE_CMD);
- if (return_code != IO_OK) {
- printk(KERN_WARNING "Error Flushing cache on controller %d\n",
- i);
+ if (return_code = IO_OK) {
+ printk(KERN_WARNING "Completed cache flush on controller %d\n", i);
+ } else {
+ printk(KERN_WARNING "Error Flushing cache on controller %d\n", i);
}
free_irq(hba[i]->intr[2], hba[i]);
@@ -3481,6 +3482,7 @@ static struct pci_driver cciss_pci_drive
.probe = cciss_init_one,
.remove = __devexit_p(cciss_remove_one),
.id_table = cciss_pci_device_id, /* id_table */
+ .shutdown = cciss_remove_one,
};
/*
On Thu, Feb 22, 2007 at 03:38:45PM -0600, Mike Miller (OS Dev) wrote:
> Patch 2/2
>
> This adds support for struct pci_driver shutdown and negates the previous NAK'ed
> reboot_notifier patch. It's much easier, wish I had know about this before. Thanks to
> Christoph for pointing this out.
Just sondering, where should we have put information about this so driver
writers/maintainers like you would have found it easily?
On Thu, 2007-02-22 at 15:38 -0600, Mike Miller (OS Dev) wrote:
> .remove = __devexit_p(cciss_remove_one),
> .id_table = cciss_pci_device_id, /* id_table */
> + .shutdown = cciss_remove_one,
You need a __devexit_p() wrapper for this one too.
James
On Thu, Feb 22, 2007 at 03:49:38PM -0600, James Bottomley wrote:
> On Thu, 2007-02-22 at 15:38 -0600, Mike Miller (OS Dev) wrote:
> > .remove = __devexit_p(cciss_remove_one),
> > .id_table = cciss_pci_device_id, /* id_table */
> > + .shutdown = cciss_remove_one,
>
> You need a __devexit_p() wrapper for this one too.
No. We want to call this even in the non-modular, non-hotplug
case. So we should remove __devexit from cciss_remove_one. Or
alternatively implement a separate cciss_shutdown that just does
the nessecary cache flushing, like most other drivers do.
On Thu, Feb 22, 2007 at 09:45:02PM +0000, Christoph Hellwig wrote:
> On Thu, Feb 22, 2007 at 03:38:45PM -0600, Mike Miller (OS Dev) wrote:
> > Patch 2/2
> >
> > This adds support for struct pci_driver shutdown and negates the previous NAK'ed
> > reboot_notifier patch. It's much easier, wish I had know about this before. Thanks to
> > Christoph for pointing this out.
>
> Just sondering, where should we have put information about this so driver
> writers/maintainers like you would have found it easily?
>
Excellent question. I wish I had a good answer. It's pretty difficult to monitor
everything going on, especially when I also have to consider HP's schedule and those
of our partners.
-- mikem
Mike Miller (OS Dev) wrote:
> + if (return_code = IO_OK) {
Shouldn't that be ==
> + printk(KERN_WARNING "Completed cache flush on controller %d\n", i);
> + } else {
> + printk(KERN_WARNING "Error Flushing cache on controller %d\n", i);
> }
> free_irq(hba[i]->intr[2], hba[i]);
On Thu, Feb 22, 2007 at 09:55:01PM +0000, Christoph Hellwig wrote:
> On Thu, Feb 22, 2007 at 03:49:38PM -0600, James Bottomley wrote:
> > On Thu, 2007-02-22 at 15:38 -0600, Mike Miller (OS Dev) wrote:
> > > .remove = __devexit_p(cciss_remove_one),
> > > .id_table = cciss_pci_device_id, /* id_table */
> > > + .shutdown = cciss_remove_one,
> >
> > You need a __devexit_p() wrapper for this one too.
>
> No. We want to call this even in the non-modular, non-hotplug
> case. So we should remove __devexit from cciss_remove_one. Or
> alternatively implement a separate cciss_shutdown that just does
> the nessecary cache flushing, like most other drivers do.
So if I remove __devexit from cciss_init_one do I also remove the __devexit_p wrapper
from the remove method?
-- mikem
On Thu, Feb 22, 2007 at 03:56:11PM -0600, Mike Miller (OS Dev) wrote:
> On Thu, Feb 22, 2007 at 09:45:02PM +0000, Christoph Hellwig wrote:
> > On Thu, Feb 22, 2007 at 03:38:45PM -0600, Mike Miller (OS Dev) wrote:
> > > Patch 2/2
> > >
> > > This adds support for struct pci_driver shutdown and negates the previous NAK'ed
> > > reboot_notifier patch. It's much easier, wish I had know about this before. Thanks to
> > > Christoph for pointing this out.
> >
> > Just sondering, where should we have put information about this so driver
> > writers/maintainers like you would have found it easily?
> >
> Excellent question. I wish I had a good answer. It's pretty difficult to monitor
> everything going on, especially when I also have to consider HP's schedule and those
> of our partners.
It is properly documented in Documentation/pci.txt as to how to do
this...
thanks,
greg k-h