- From: Michael Borland <michael_borland_at_lynceantech.com>
- Date: Mon, 08 Sep 2003 12:21:07 -0700
Frank, Thanks!! That fixed it. I did 500 iterations and there were no problems. Prior to the fix, I typically got 15 iterations before a crash. Sorry it was so annoying. Your efforts are really appreciated. --Michael Frank Mori Hess wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > On Sunday 07 September 2003 02:00 pm, Michael Borland wrote: > >>Following the suggestion from Frank I was able to get a look at the >>kernel messages on the console. (Although I have to switch to the >>console before the machine crashes, since after the crash it is >>completely unresponsive.) >> >>When the machine crashes, I see this message shortly before the crash: >> plx9080: cancel() timed out waiting for dma 1 done set >> >>The machine crashes shortly after that, and not always in the same >>place. I added calls to syslog() to the mmap demo, so I could see what >>it was asking comedi to do. Usually, the machine crashes before mmap is >>able to open the comedi device again, but sometimes it crashes sooner. >> >>--Michael > > > That was an incredibly annoying bug. It should be fixed in cvs now. The > plx9080.h file is really the only thing you need to update. Apparently, > if you disable the dma channel a microsecond before setting the dma abort > bit (instead of doing both in the same writeb()) there is a small chance > the board will lock up. After that happens, any attempt to do a readw() > of one of the board's daq registers will hang forever. > > - -- > Frank > > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.0.6 (GNU/Linux) > Comment: For info see http://www.gnupg.org > > iD8DBQE/XM1G5vihyNWuA4URAq5ZAJ4nNJCAf+Cd2qaUNt4eWy/q2Qgx5QCfUEzs > Sm3Gnd0pp+oF9meLuHw/kvM= > =Nc5u > -----END PGP SIGNATURE----- > >
Received on 2003-09-08Z18:21:07