-
Hi Dave,
I'm sorry for the long delay, much trouble in work on other projects.
I attached an archive with the files created from your debug commands and the kernel config.
According to citrix website, the kernel, along with lots of other sources can be downloaded here: http://www.citrix.com/lang/English/lp/lp_1688623.asp
while testing, dmesg said this:
e1000: eth1...
2009-11-20 11:50:45 UTC by seggi
-
Both fixes described here are lined up for version e1000e-1.1.2 (or higher) , and on their way upstream.
2009-11-18 18:51:19 UTC by davegraham
-
Please let me know the result if & when you do retest with the platform updates from Supermicro.
2009-11-18 17:37:24 UTC by davegraham
-
Here's another patch, e1000-8.0.16.dbg2.sf289538.patch to the e1000-8.0.16 base driver that will allow us to see what transmit desciptors the driver is reading, if any. Lets run this driver, and a very precise "script" of actions to figure out where things are going wrong.
1) Disable network scripts
First, please disable any network start stcripts, because if we have full manual control of...
2009-11-18 01:34:37 UTC by davegraham
-
John, can you direct me to the sources and kernel config used for this Xen kernel [2.6.18-128.1.6.el5.xs5.5.0.496.1012xen] please. I have an 82541PI. I want to get this same build running here.
2009-11-17 22:43:45 UTC by davegraham
-
OK thanks. That driver should have worked even if the NIC was improperly posting the descriptors. I like your idea of getting a tcpdump from wither a mirrored switch port or any machine that is a target of the first message.
I'll work on some other debug instrumentation for tomorrow.
2009-11-16 22:59:40 UTC by davegraham
-
Hi Erik, I have checked that the 2.6.30 kernel you are running, with e1000e version 0.3.3.4-k4, has the latest driver changes that were made for this issue (the e1000e commit that contained the fixes was 75eb0fad , which first appeared in driver e1000e-0.3.3.3-k6 in kernel 2.6.29).
One these platforms, an FW layer exists, and its function includes to abstract the BMC from being concerned...
2009-11-16 22:24:56 UTC by davegraham
-
Hi Dave!
Thank you for your patience and time!
I applied your patch and tried again, this time I didn't get any errors in the logfiles or in dmesg and the interface didn't get resetted as soon as the box tried to send data. Nevertheless, I was not able to establish any network connection (not even ping) because the ARP lookups didn't succeed. But I'm not sure whether this is an other problem...
2009-11-16 21:33:21 UTC by seggi
-
Because (from your debug output) I see that the NIC thinks it is transmitting (or more precisely that it is ready to receive more TX data to send), I want to know if it is sending data to the wire. I've attached a patch e1000-8.0.16.tdh.sf2897538.patch which may allow us to see if the NIC is really transmitting. If the patched driver allows you to obtain an IP address and perform limited...
2009-11-16 20:32:41 UTC by davegraham
-
Hi John,
The HW thinks it has transmitted something, but the driver doesn't think it has. It's possible its a bad card, and I'd like to collect some more data about the failure. I'm preparing instructions & debug code that will allow us to capture that additional detail, and will attach it shortly.
2009-11-16 18:49:13 UTC by davegraham
