On 10/29/2011 2:01 PM, Nix wrote: > On 29 Oct 2011, Kyle Brantley told this: > >> On 10/28/2011 3:41 PM, Nix wrote: >>> setpci -s 03:00.0 CAP_EXP+10.b=40 >>> >>> (where 03:00.0 is the PCI device address). >>> >>> I really must get back to diagnosing it, but my 82574Ls are on an >>> always-on server with my home directory on it, so debugging is *really* >>> annoying. >> How did you get the OS installed? I take it that you didn't use PXE. > Oh, this isn't on a net6501, it's a server, hardware RAID, 24Gb RAM and > all. OS installation was via CD-ROM, then via network, but that was OK > because this was back in 2.6.30 days, and the kernel only started > turning ASPM on for *anything* around 2.6.35. This problem immediately > appeared and has not left since :( > >> This bug made it outright impossible for me to install CentOS 6 on the >> 6501. I wound up installing Fedora in the interim, where interestingly >> enough, it is not present... > The e1000e guys have given up trying to track this: it's not an e1000e > driver bug but something in the PCI layer. Nobody has brought this to > the attention of the PCI guys yet (and I blame myself for this, I had a > kernel bugzilla bug open with lots and lots of painstakingly-collected > data on this problem, and I didn't tell the PCI guys and then kernel.org > was penetrated, and now kernel.org bugzilla is gone, possibly never to > return.) >
Surely you can recreate the bug from the emails though, right? I mean, that sucks don't get me wrong, but technically nothing should be *lost.* And now that we have a device that can easily reproduce the issue (that doesn't take out your homedir with it!)... If I can help here, let me know how. --Kyle _______________________________________________ Soekris-tech mailing list [email protected] http://lists.soekris.com/mailman/listinfo/soekris-tech
