* RE: e1000 close (NAPI)
@ 2003-10-08 5:58 Feldman, Scott
2003-10-08 9:16 ` Robert Olsson
0 siblings, 1 reply; 5+ messages in thread
From: Feldman, Scott @ 2003-10-08 5:58 UTC (permalink / raw)
To: Robert Olsson; +Cc: Jeff Garzik, netdev
> - if(work_done < work_to_do) {
> + if(work_done < work_to_do || !netif_running(netdev)) {
> netif_rx_complete(netdev);
> e1000_irq_enable(adapter);
> }
Is e1000 unique in this regard, or do the other NAPI-enabled drivers
have the same issue?
-scott
^ permalink raw reply [flat|nested] 5+ messages in thread* RE: e1000 close (NAPI)
2003-10-08 5:58 e1000 close (NAPI) Feldman, Scott
@ 2003-10-08 9:16 ` Robert Olsson
2003-10-08 12:07 ` David S. Miller
0 siblings, 1 reply; 5+ messages in thread
From: Robert Olsson @ 2003-10-08 9:16 UTC (permalink / raw)
To: Feldman, Scott; +Cc: Robert Olsson, Jeff Garzik, netdev
Feldman, Scott writes:
> > - if(work_done < work_to_do) {
> > + if(work_done < work_to_do || !netif_running(netdev)) {
> > netif_rx_complete(netdev);
> > e1000_irq_enable(adapter);
> > }
>
> Is e1000 unique in this regard, or do the other NAPI-enabled drivers
> have the same issue?
Hello!
Other drivers may have the same issue. This was spotted while stress-testing
tulip. Patch was sent here some days ago. It's only seen if trying to close
device under extreme loads. See the syncing code in dev_close()
You may try the e100 as well. We update the docs when see that this is OK.
Cheers.
--ro
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: e1000 close (NAPI)
2003-10-08 9:16 ` Robert Olsson
@ 2003-10-08 12:07 ` David S. Miller
0 siblings, 0 replies; 5+ messages in thread
From: David S. Miller @ 2003-10-08 12:07 UTC (permalink / raw)
To: Robert Olsson; +Cc: scott.feldman, Robert.Olsson, jgarzik, netdev
On Wed, 8 Oct 2003 11:16:02 +0200
Robert Olsson <Robert.Olsson@data.slu.se> wrote:
> Feldman, Scott writes:
> > > - if(work_done < work_to_do) {
> > > + if(work_done < work_to_do || !netif_running(netdev)) {
> > > netif_rx_complete(netdev);
> > > e1000_irq_enable(adapter);
> > > }
> >
> > Is e1000 unique in this regard, or do the other NAPI-enabled drivers
> > have the same issue?
>
> Other drivers may have the same issue. This was spotted while stress-testing
> tulip. Patch was sent here some days ago. It's only seen if trying to close
> device under extreme loads. See the syncing code in dev_close()
>
> You may try the e100 as well. We update the docs when see that this is OK.
We should just fix all the NAPI drivers in this regard at one
time. Undoubtedly tg3 needs this change too.
^ permalink raw reply [flat|nested] 5+ messages in thread
* RE: e1000 close (NAPI)
@ 2003-10-10 18:45 Feldman, Scott
0 siblings, 0 replies; 5+ messages in thread
From: Feldman, Scott @ 2003-10-10 18:45 UTC (permalink / raw)
To: Robert Olsson; +Cc: Jeff Garzik, netdev
Robert Olsson wrote:
> Other drivers may have the same issue. This was spotted
> while stress-testing tulip. Patch was sent here some days
> ago. It's only seen if trying to close device under extreme
> loads. See the syncing code in dev_close()
Got it, thanks.
^ permalink raw reply [flat|nested] 5+ messages in thread
* e1000 close (NAPI)
@ 2003-10-07 12:34 Robert Olsson
0 siblings, 0 replies; 5+ messages in thread
From: Robert Olsson @ 2003-10-07 12:34 UTC (permalink / raw)
To: scott.feldman; +Cc: Jeff Garzik, Robert Olsson, netdev
Hello!
Here is an patch to check netif_running in dev->poll so we don't have to hang in
dev_close until all polls are finished.
Cheers.
--ro
--- drivers/net/e1000/e1000_main.c.031007 2003-05-08 16:29:08.000000000 +0200
+++ drivers/net/e1000/e1000_main.c 2003-10-07 14:26:12.000000000 +0200
@@ -2040,7 +2040,7 @@
*budget -= work_done;
netdev->quota -= work_done;
- if(work_done < work_to_do) {
+ if(work_done < work_to_do || !netif_running(netdev)) {
netif_rx_complete(netdev);
e1000_irq_enable(adapter);
}
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2003-10-10 18:45 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-10-08 5:58 e1000 close (NAPI) Feldman, Scott
2003-10-08 9:16 ` Robert Olsson
2003-10-08 12:07 ` David S. Miller
-- strict thread matches above, loose matches on Subject: below --
2003-10-10 18:45 Feldman, Scott
2003-10-07 12:34 Robert Olsson
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).