Date: Sun, 18 Jun 2000 20:29:44 -0400 From: "Sean O'Connell" <sean@stat.Duke.EDU> To: FreeBSD mobile <freebsd-mobile@FreeBSD.ORG> Cc: imp@village.org Subject: Re: 4.0-RELEASE, Netgear FA410TXC, device timeout Message-ID: <20000618202944.F9791@stat.Duke.EDU> In-Reply-To: <20000618164044.A9791@stat.Duke.EDU>; from sean@stat.Duke.EDU on Sun, Jun 18, 2000 at 04:40:44PM -0400 References: <20000617225842.A36405@vobiscum.styx.org> <20000617170915.A1296@vobiscum.styx.org> <200006171835.LAA00874@mass.osd.bsdi.com> <20000617225842.A36405@vobiscum.styx.org> <200006180622.AAA47458@harmony.village.org> <20000618164044.A9791@stat.Duke.EDU>
next in thread | previous in thread | raw e-mail | index | archive | help
Sean O'Connell stated: > Warner Losh stated: > > In message <20000617225842.A36405@vobiscum.styx.org> Marc Fonvieille writes: > > : and 'ed0: device timeout' message in the kernel log. > > > > That's almost always (99%) a case of interrupts not being routed > > correctly. > > > Warner- > > In the case of the Netgear card, if_ed is broken. I have attempted > to help two people off-line trying to use these cards, and in both > cases, they _never_ worked. These were on truly free IRQs (usually > irq 3 with COM2 disabled in the BIOS and not even in the kernel config > file). In both cases, replacing "device ed0 at isa?.." with just > "device ed" really only was a cosmetic fix to change ed1 timeouts to > ed0 timeouts. These cards just plain do not work as is. > > if_ed is also badly broken for the Linksys PCMP100. I can only get > it to probe with the correct hardware ethernet address after adding > in the hack to make the Linksys test always succeed (of course, this > breaks the detection for my older Linksys E2CT card). However, the > performance is abyssmal, not akin to a duplex mismatch (either that > or if_ed is dropping beacoup packets). As a for instance, scp'ing > the netscape tarball usually takes about 20+ seconds on a 10Mbit > hub usinge either the EC2T or a 3Com 3c589D, but with 10/100 Linksys > card I am seeing 2hours being reported by scp (CTRL-C is deployed > long before this). Hi- As a follow-up, I applied the patch to if_ed.c (oops, hadn't looked at cvs-all before earlier email). This does allow both my Linksys EC2T card and the Linksys 10/100 (not mine) to be probed/attached by the same kernel; however, the same performance problems were experienced. scp of a 17168KB file took 00:46 (accrding to ouput) with my old card, but it was claiming an ETA of 34:36 ... eek. this is all on a 10Mb hub. S -- 1--------01---------01--------01--------01--------01--------01--------0 Sean O'Connell sean@stat.Duke.EDU To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-mobile" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000618202944.F9791>