| To: | maximilian attems <janitor@xxxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: sis900 oops on resume |
| From: | Daniele Venzano <webvenza@xxxxxxxxx> |
| Date: | Fri, 19 Nov 2004 11:01:46 +0100 |
| Cc: | NetDev <netdev@xxxxxxxxxxx> |
| In-reply-to: | <20041118094602.GB2287@stro.at> |
| Mail-followup-to: | maximilian attems <janitor@xxxxxxxxxxxxxx>, NetDev <netdev@xxxxxxxxxxx> |
| References: | <20041117182403.GC2252@stro.at> <20041117201008.GA11439@elf.ucw.cz> <20041118021425.GA2128@stro.at> <20041118084431.GA13870@gateway.milesteg.arr> <20041118094602.GB2287@stro.at> |
| Sender: | netdev-bounce@xxxxxxxxxxx |
On Thu, Nov 18, 2004 at 10:46:02AM +0100, maximilian attems wrote: > On Thu, 18 Nov 2004, Daniele Venzano wrote: > > I'm aware of this problem, sis900 looses the first interrupt after > > resume, I have to check the code. > > On my tests the card, after that lost interrupt, worked normally. Is > > that your case, or do you have a non working sis900 after resume ? > > well it works, as long as i do not use guessnet(8), > as described in the bug report, but the lost interrupt is reported > on every resume. Unfortunately I'm no longer able to reproduce this problem. I just tried to suspend/resume with a running ping -f and I can't see any lost or abnormal interrupts. There is a noticeable delay after resume and before the NIC is on line and able to handle traffic, but that is caused by the link detection/autonegotiation and I can do little to speed things up. Have you tried delaying the guessnet invocation ? -- ----------------------------- Daniele Venzano Web: http://teg.homeunix.org |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Fix ipv6_ifa_notify race (was: IPv6 "badness" in recent releases), Herbert Xu |
|---|---|
| Next by Date: | Re: Fw: [Bugme-new] [Bug 3765] New: Network link down, when writting to sata disk AND network in use, Surakshan Mendis |
| Previous by Thread: | Re: sis900 oops on resume, maximilian attems |
| Next by Thread: | [patch 2.6.10-rc2] tulip: make tulip_stop_rxtx() wait for DMA to fully stop, John W. Linville |
| Indexes: | [Date] [Thread] [Top] [All Lists] |