Bug#524876: sata_mv: frozen/hard reset on 4-port 5041 chip
Hi Dave,
Dave Alitz wrote:
> After upgrading a SuperMicro SuperServer 5013-MT server from etch to
> lenny I started getting numerous hard resets on all of the sata
> ports. The 4-port sata controller is a Marvell MV88SX5041. Looking
> around a bit it seems that the error reported is a timeout error. A
> very similar bug was filed under number 514155 for the 508x/6081
> 8-port controller chips.
>
> I am using LVM2 on MD raid 1 and raid 5. Disabling write caching
> significantly reduced the number of resets; but didn't eliminate
> them.
>
> I'm using all four ports. Two of each of the following drives:
[...]
> [577365.394906] EXT3-fs: mounted filesystem with ordered data mode.
> [593245.683756] ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 frozen
> [593245.683793] ata2.00: cmd ca/00:08:9b:d6:3b/00:00:00:00:00/e0 tag 0 dma 4096 out
> [593245.683797] res 40/00:00:00:00:00/00:00:00:00:00/40 Emask 0x4 (timeout)
> [593245.683862] ata2.00: status: { DRDY }
> [593245.683893] ata2: hard resetting link
Thanks for reporting it, and sorry for the slow response. Basic
questions:
- Does downgrading the kernel again to the version from etch help?
It should be possible to test this by testing in the installer
environments for etch and lenny separately, for example.
- Do more current kernels behave better? (I doubt they would, but
it's always worth a try.)
- Could you attach full dmesg output from bootup of the last working
and first non-working kernel you have tried? http://snapshot.debian.org/
has many kernels if you'd like to narrow the regression range.
- Any other weird symptoms? Do you have any ideas about what could
be causing this?
Reply to: