Date: Tue, 11 Nov 1997 09:33:38 -0700 (MST) From: Nate Williams <nate@mt.sri.com> To: Guido van Rooij <guido@gvr.org> Cc: freebsd-mobile@freebsd.org Subject: Re: sio: probe test 3 failed Message-ID: <199711111633.JAA16494@rocky.mt.sri.com> In-Reply-To: <199711110940.KAA28845@gvr.gvr.org> References: <199711110940.KAA28845@gvr.gvr.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> Sometimes, after suspending my system, my modem card is not properly > reallocated. What happens is that the kernel prints: > Card inserted, slot 1 > sio1: probe test 3 failed > > Removing and resinserting the card does not help anymore. > Taking a look at the sio driver, one can read: > if (idev->id_irq != 0) > failures[3] = isa_irq_pending(idev) ? 0 : 1; > Inserting another modem card (same irq) gives: > Card inserted, slot 1 > Nov 10 23:34:22 mouse pccardd[41]: No free configuration for card Intelligent It may be related to 'interrupts' not being properly let go in the system. The interrupt masks and such are assumed to always be 'added' in update_intr_mask(), so it's possible that the interrupt is lost. > So somehow something seems still resident in the sio driver preventing this > new modem card to be properly initialised. The same thing also makes > the third test fail. Would you have a clue on what can be wrong? The sio driver *is* resident, due to hardware race bugs, but marked as unused. However, as you are already aware, it's possible that something is still using it. :( > Bruce suggested that the irq might have changed preventing test 3 from > succeeding. This would also explain that sio1 is kept resident depsite the > removal of the modem card which leads to pccardd logging that no free > driver is available. How could the irq have changed? The pccard sources *always* use the same interrupt (especially if it's hard-coded), and assuming that they recognize the card's removal and insertion (and nothing else has occurred in between) they will always use the same interrupt as well, since they keep track of their own interrupt mask for consistancy (much to Bruce's chargrin). Nate
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199711111633.JAA16494>