- From: Omri Schwarz <ocschwar_at_MIT.EDU>
- Date: Mon, 28 Jun 2004 11:46:32 -0400
I'm afraid the crash is still there, although it takes a little longer now, and comes with a different call trace. (Comedi off CVS, linux 2.6.7, compiled with spinlock debugging and GDB symbols). It's time I moved the machine to somewhere it can chat with a neighbor over the serial line so there could be more useful debugging. I'll have more information once I have done so. > I put a fix into cvs that should prevent the driver's interrupt handler > from getting run before the initialization is done. Would you give that a > try? > > On Thursday 24 June 2004 09:31 pm, Omri Schwarz wrote: > > > Is your board sharing an interrupt? It's possible an interrupt that > > > occurs > > > > /sbin/lspci -vv says yes. I should kick myself for not > > doing this sooner. > > > > > after the driver registers its interrupt handler but before > > > comedi_device_attach() completely finishes could cause a crash, but > > > you said the crash happens after comedi_device_attach() returns? > > > > It's hard to gauge. The oops() message scrolls the terminal > > all the way through, giving me only a split second to see the > > printk()s and other output, and then locks up. > > > > - -- > Frank > > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.2.4 (GNU/Linux) > > iD8DBQFA328v5vihyNWuA4URAktOAKCs2ay4Qh6+PyXfn7ehFRO+PNtSMwCgw5tl > cI70YxlXGvaenK+VdmVRVdQ= > =1jO2 > -----END PGP SIGNATURE----- > > _______________________________________________ > comedi mailing list > comedi_at_comedi.org > https://cvs.comedi.org/cgi-bin/mailman/listinfo/comedi
Received on 2004-06-28Z14:46:32