SATA CD/DVD I/O error render system unresponsive, 100% cpu load
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
linux (Ubuntu) |
Invalid
|
Undecided
|
Unassigned |
Bug Description
I have a broken CDROM (one that has read errors, a toaster). Using this CD-ROM renders my system unresponsive - it will have a high CPU load or will sometimes even freeze from 10 to 60 seconds.
Kernel log says things like:
Nov 15 17:05:51 stout kernel: [17182964.856000] Info fld=0xe
Nov 15 17:05:51 stout kernel: [17182964.856000] end_request: I/O error, dev sr0, sector 56
Nov 15 17:05:51 stout kernel: [17182964.856000] Buffer I/O error on device sr0, logical block 7
Nov 15 17:06:09 stout kernel: [17182982.868000] sr 0:0:1:0: SCSI error: return code = 0x8000002
Nov 15 17:06:09 stout kernel: [17182982.868000] sr0: Current: sense key: Medium Error
Nov 15 17:06:09 stout kernel: [17182982.868000] Additional sense: No seek complete
Nov 15 17:06:09 stout kernel: [17182982.868000] Info fld=0x0
Nov 15 17:06:09 stout kernel: [17182982.868000] end_request: I/O error, dev sr0, sector 0
Nov 15 17:06:09 stout kernel: [17182982.868000] Buffer I/O error on device sr0, logical block 0
Nov 15 17:06:27 stout kernel: [17183000.892000] sr 0:0:1:0: SCSI error: return code = 0x8000002
Nov 15 17:06:27 stout kernel: [17183000.892000] sr0: Current: sense key: Medium Error
Nov 15 17:06:27 stout kernel: [17183000.892000] Additional sense: No seek complete
... which is correct, as the CD I'm using is broken. But this should not freeze the system to the extent where you even think it's unusable (namely: 60 seconds or more).
My feeling says that this is some sort of non-reentrancy in the SATA layer, where a CD-ROM reading problem will block requests to the HDD, but this is just speculation.
Core2 duo, 82801GB/GR/GH (ICH7 Family) Serial ATA Storage Controller, driver=ata_piix
ST3320620AS hdd and Optiarc DVD RW AD-7170S 1.00 Oct27,2006
2.6.15-28-686 #1 SMP PREEMPT Wed Jul 18 22:57:30 UTC 2007
Ubuntu 6.06.1 LTS
This bug seems to relate to #75295 but seems different.
Bug #84603 is another version of this bug - this time related to an optical drive that times out when there's no disk in it.
Then there's #64587, that also shows a freezing machine due to SATA hangs.
So it seems that no SATA drive can be reached, once there's at least one SATA drive busy. Is this a ata_piix driver issue, a chipset issue (i.e. the chipset can only handle one timeout at a time) or anything else?