Bug 14033

Summary:	Data acq process gets stuck in vmopar
Product:	Base System	Reporter:	Daniel O'Connor <doconnor>
Component:	kern	Assignee:	freebsd-bugs (Nobody) <bugs>
Status:	Closed FIXED
Severity:	Affects Only Me
Priority:	Normal
Version:	3.3-RELEASE
Hardware:	Any
OS:	Any

Description Daniel O'Connor 1999-09-29 07:40:01 UTC

We have a PCI data acquisition card which effectivly maps a FIFO into memory
and the kernel reads it during interrupt, and passes the data off to a userland
process in read().

Occasionally (happens much more often when the system is loaded, ie buildworld,
and doing networking) the process reading from the card gets stuck in vmopar. 

I looked in the archives and there was a suggestion that it was NFS related so
I umount'd the NFS partitions but no joy.

This problem does not occur in 2.2.8, and the driver code is almost identical,
(except poll) so I suspect the OS, but I do not know where to start looking.

Fix: 

Go back to 2.2.8+CAM (blerg)
How-To-Repeat: 
Rather complicated :(

Comment 1 iedowse freebsd_committer

2001-06-05 21:40:59 UTC

State Changed
From-To: open->feedback


Is this still a problem with recent releases? The most common case 
of processes getting stuck in 'vmopar' was fixed before 4.2-RELEASE 
(nfs_subs.c rev 1.96 + related changes).

Comment 2 iedowse freebsd_committer

2001-06-06 01:15:26 UTC

State Changed
From-To: feedback->closed


Submitter confirms that this is no longer a problem.