On Thu, 1 Oct 2009, Tom Chen wrote:
> Randy,
>
> On Ultra45, "uadmin 3 8" has the same result as "uadmin 3 22" and
> does not crash.
Figures, but I will see what I can find with th 7 fcn (but feel free
to file a bug).
>
> Instead of testing whole system, how I can test Suspend/Resume
> capability of a specific driver? I tried the "uadmin 3 25 224" to
> test NIC device "bge" driver whose major number is 224. But the
> 'uadmin 3 25 224" results in a hang state. Is "25" only good for x86
> server?
As this fcn is driven by the same bug that allows 22, it will
execute on Sparc. However, as it was initially intended for x86, it
is possible that there are bugs on Sparc. The cpr module must be
loaded for s/r to even consider working, so 'cpr_debug/W0xff' should
at least display the last place the code was at before hanging (but
only on the console).
You should also take note that this is unsupported for testing, and
drivers *might* depend on other system state w.r.t. s/r, so it is
possible that normal behavior is a hang.
>
> bash-3.2# cat /etc/name_to_major | grep bge
> bge 224
>
> Also, on Ultra45, mdb -K does not work.
> bash-3.2# mdb -K
> mdb: -F must also be supplied to start kmdb from non-console tty
You might want to be on the console for these tests, as much more
usable output will be there (as well as better debuggability).
Possibly use command-line login or a serial console. Once you
know that things are working well, then running tests on non-console
tty's will be more practical.
I also don't know if networking-discuss or driver-discuss is
interested in all this detail, and maybe we should reduce the cc line,
unless someone on either alias indicate that they are listening and
would like it to continue there.
Cheers!
---- Randy
>
>
> Tom
> --
> This message posted from opensolaris.org
> _______________________________________________
> networking-discuss mailing list
> [email protected]
>
_______________________________________________
networking-discuss mailing list
[email protected]
