Description of problem: There is aproblem with the Fedora kernel updates of 3.8.x: the driver for the Atheros LAN chip (atl1c). After a while (from 30 minutes to a few hours) the kernel starts logging errors like "DMA: Out of SW-IOMMU space for XXXX bytes" on the device, the performance of the network degrates and finally stops. You have to reboot the machine to get it working again. It seems that there is a fix for this in the kernel, that will appear in 3.8.3 (according to my understaning). But what will be done in between with the fedora updates, where 3.8.1 got installed in my system? Version-Release number of selected component (if applicable): Fedora 18 kernel 3.8.1. How reproducible: Use the network for a while in a pc with the Atheros lan chip. Steps to Reproduce: 1. 2. 3. Actual results: Network stops Expected results: Network does not stop Additional info:
Same problem here
(In reply to comment #0) > Description of problem: > There is aproblem with the Fedora kernel updates of 3.8.x: the driver for > the Atheros LAN chip (atl1c). After a while (from 30 minutes to a few hours) > the kernel starts logging errors like "DMA: Out of SW-IOMMU space for XXXX > bytes" on the device, the performance of the network degrates and finally > stops. > You have to reboot the machine to get it working again. > > It seems that there is a fix for this in the kernel, that will appear in > 3.8.3 (according to my understaning). Do you have any more information about this patch? I took a quick look at the web interface for 3.8-stable and did not see any patches that seemed to modify the atl1c driver. I'm cloning the git tree now in case I missed it, but more information about this specific patch would be helpful.
Unfortunately no, I haven't stored the pages I have read about it. Does this e-mail provide any useful information? http://www.spinics.net/lists/stable/msg01076.html commit 7cb08d7f3a5ea6131f4f243c2080530ac41cb293 Author: Huang, Xiong <xiong@xxxxxxxxxxxxxxxx> Date: Tue Feb 19 07:23:09 2013 +0000 atl1c: restore buffer state in the previous commit : f1f220ea1dda078, the BUSY state of buffer is wrongly deleted. this patch just restore it. Signed-off-by: xiong <xiong@xxxxxxxxxxxxxxxx> Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>
And I think I have read about the patch not applied in kernel 3.8.2 here: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1132477 Comment 44, 48 & 49. But I know nothing about kernel bugs, so I cannot verify any of these comments.
(In reply to comment #3) > Unfortunately no, I haven't stored the pages I have read about it. > Does this e-mail provide any useful information? > > http://www.spinics.net/lists/stable/msg01076.html > > commit 7cb08d7f3a5ea6131f4f243c2080530ac41cb293 > Author: Huang, Xiong <xiong@xxxxxxxxxxxxxxxx> > Date: Tue Feb 19 07:23:09 2013 +0000 > > atl1c: restore buffer state > > in the previous commit : f1f220ea1dda078, the BUSY state of buffer is > wrongly > deleted. this patch just restore it. > > Signed-off-by: xiong <xiong@xxxxxxxxxxxxxxxx> > Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx> That is helpful. Thank you!
Problem persists with 3.8.3 :(
I tried a 3.9 kernel from f19. It seems, that fixed the issue, though that kernel has debugging enabled and thus feels slooow.
Hello, it looks like that patch for this issue finally made it into 3.8.4 kernel. I have tested now with 3.8.4-201.fc18.x86_64 and the issue did not appear any more for me. (after 2,5hrs of testing) Cheers, Michael
Yes, I also think that it has been fixed in recent kernels. The bug should be closed.
Laptop is in the shop currently so I cannot confirm one way or the other... So I will trust your experience :)
Well, the pc is network-stable for many days now, with kernel 3.8.5. I cannot complain... :-)