Somnath kotur wrote:
Hi,
In order to get my driver to work on reboot, i copied it to
/kernel/drv/amd64. The driver on reboot seems to load fine ,up until a point
where it tries to plumb(?) ...ie it enters my driver's xxx_start() entry point
and spits out this error msg below:
Problem 1:
ct 14 12:39:57 opensolaris be2net: [ID 498891 kern.warning] WARNING:
benet_start linkstatus=1
Oct 14 12:39:57 opensolaris ip: [ID 926743 kern.error] be2net0: DL_BIND_REQ
failed: DL_SYSERR (errno 6)
Oct 14 12:39:57 opensolaris ip: [ID 738341 kern.error] be2net0: DL_UNBIND_REQ
failed: DL_OUTSTATE
where benet_start is my <driver_start> entry point and be2net is my driver name
Sounds like maybe the instance isn't available? DL_BIND_REQ *should*
succeed. The fact that it didn't is slightly worrisome.
Soon after this the driver's detach entry pt gets called and driver gets
unloaded!! (No apparent reason why detach was triggered), Any idea why this
could be so? -------- Problem 2
If the device isn't held busy, then it can be detached as a result of
modunload -i 0. The debug kernel does this periodically, but even a
production kernel does it *once* during boot.
After a gap of say 10 mins , my driver gets loaded again(automatically) ,albeit this time succesfully and i am able to manually plumb the driver -------------------------- Problem/Situation 3:
Is this expected behaviour?
No, but perhaps NWAM is doing this?
Problem Scenario without reboot:
This is when i copy my driver to /tmp and create a symbolic link to it (as is
the suggested practice during initial driver development) . On add_drv of my
driver , around 75% of the time the command ' ifconfig be2net0 plumb' fails the
1st time around with the same DL_BIND_REQ failure as above.
I've never used this symbolic link approach. I usually directly
modload the driver, and then use devfsadm to plumb it, during development.
However it invariably succeeds on my 2nd / 3rd/.... attempt
Could somebody pls tell me what could be going wrong here?
I'm not sure, but the above might help.
- Garrett
Thanks
Som
_______________________________________________
driver-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/driver-discuss
