John Austin wrote:
Hi
I have been fighting my problem with system freezes
described in the previous thread
I have added a 2nd NIC to the machine
and tested across an NFS mount using md2sum as before
The machine now has
An on board Marvell 8380 GB/s nic that uses the sky2 driver
A D-Link 530T GB/s nic that uses the skge driver
I am using a Netgear GS108T GB/s switch to a Centos 5.4
nfs4 server.
The machine is booted with only one NIC connected at a time
D-link Marvell
F11 OK OK
F12 OK Fails
Fails = Hardware button press required after "random" time during
read/checksum calc of 8GB file set
My conclusion was that it must be the F12 sky2 driver not NFS
I then down loaded the latest "Working" F11 kernel source rpm and
extracted sky2.c and sky2.h
and compiled these against the F12 kernel
I thought this must fix the problem - but no!
The machine still freezes during the md5sum test
I am rapidly running out of ideas !!
Any help appreciated
I have no help, but I do have information. There have been reports of problems
in the sky2 driver in the 2.6.32-rc tests. These may have nothing to do with
your problem, or may lead to a patch which will appear in the Fedora kernels.
If you want to compare the sources from the two kernels it might tell you
something, but the issue may be in interrupt handling, not the driver itself. I
saw something on a chat suggesting that selecting a different hardware source
for the clock may solve network hang issues. I neither know what issues or why
that would help, unless it's some time jumps backward issue still with us.
You now have all the relevant info I do, whether it qualifies as "help" is up to
you.
--
Bill Davidsen <davidsen@xxxxxxx>
"We have more to fear from the bungling of the incompetent than from
the machinations of the wicked." - from Slashdot
--
fedora-list mailing list
fedora-list@xxxxxxxxxx
To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
Guidelines: http://fedoraproject.org/wiki/Communicate/MailingListGuidelines