Date: Mon, 28 Feb 2000 12:52:22 +0100 From: Ben Gras <ben@euro.net> To: freebsd-hackers@freebsd.org Subject: Question: Problem with PCI network cards (e.g. ed0 and vx0) - no interrupts? Message-ID: <20000228125222.A41924@euronet.nl>
next in thread | raw e-mail | index | archive | help
Greetings -hackers, Here's a question I sent to -questions a long time ago. The matter is quite kludged over (by buying an ISA card), but I'm kinda curious regardless. Is this a known thing? =Ben ----- Forwarded message from Ben Gras <ben@scum.org> ----- Subject: Problem with PCI network cards (e.g. ed0 and vx0) - no interrupts? To: freebsd-questions@freebsd.org Date: Fri, 25 Jun 1999 20:43:53 +0200 (CEST) From: Ben Gras <ben@scum.org> Greetings -questions denizens, Any information available on the following problem would be welcome; the symptom is that various PCI network cards will be recognized but Don't Work. vx0 and ed0 will be seen (in booting) but produce "ed%d: device timeout" style errors; reading the code leads me to believe the drivers are expecting interrupts but aren't generated by the card. Searching the archives reveals this is usually caused (with the NE2000 clones at least) by irq conflicts, or the BNC cable not being attached; however, I've removed all other cards (except for the videocard) from the system and tried all possible IRQs (using the BIOS setup), which the system indeed finds on probing it (it uses and configures the irq set in the BIOS, whether it's 5, 9, 10, 11 or .. and makes no difference). The machine is an AMD 486 (80 or 100 MHz), the BIOS is an Award modular BIOS V4.50G, V4.26GN2 (12/07/94; don't ask me about the meanings of those two version numbers). The board calls itself a "486-VIP-IO" mainboard, and it's "Based on the VIA GMC chipset" (this is my bluffing mode) according to the manual. (ISA/VL+PCI). It mentions a VIA VT82C505 as a bus bridge, if that's useful. It claims PCI2.0 compliance. I have more details of this thing if useful. Other things I've tried include other (pci) slots, settings in BIOS (other than changing the irq). The system doesn't say anything about spurious interrupts, so I presume the card isn't generating the wrong ones, but none at all. vmstat -i reports nothing out of the ordinary, except for no interrupts from any network cards of course. Also if I set up nothing in the BIOS, the IRQ will show up as 0 or 255 -- hmm. The manual mentions, a jumper I can twiddle for IRQ 11 if a PCI card is inserted; under "IRQ pull up/pull down", where up is default, and down says, `pull down if pci card is inserted'. I twiddled it without knowing what I was doing; it had no effect though. I presumed this would be a bizarreness of the el-cheapo NE2000 clone I first used, but even the rather sophisticated 3C905B had no luck; same symptom: probe OK, then no interfacing. The videocard, incidentally, is a PCI card too and is OK, although as a clever friend mentioned, it probably doesn't need interrupts to do something useful. Other BIOS PCI features settings I twiddled are "Int <A/B/C/D> using IRQ: " which are set to N/A. IRQ Mode can be Transparent, Transparent invert, Converse and Converse & monitor EOI. The manual reckons the choice depends on the IDE controller card. I can set the latency timer too, per irq, which I've had on 0, 32 and 255 with no difference. "CPU to PCI write buffer", "PCI Master Write Buffer" and "Pci Master Pre-fch [fetch I suppose] buffer" I've twiddled with no effect either. The on-board NCR BIOS is disabled, which has little effect I suppose. The CPU on 80 or 100MHz has no effect. As you probably can tell, I have little experience in dealing with this kind of low-level system problem.. I hope I haven't forgotten any things I've tried or information obtained, I've been plodding on with this thing for months now :/ So far I'm presuming I'm a victim of obsolete or bogus hardware, but I'd really appreciate information folks :) or even information on how to get more information? I'd include dmesg and-so-on output if it as easier, I'm not particularly networked at the moment you see :) cheers, =Ben ----- End forwarded message ----- To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000228125222.A41924>