ATA timeout with TRIM flag set

Sascha Klauder sklauder at
Thu Aug 2 12:52:57 UTC 2012

Hi all,

 I've bought a 60 GB Intel 330 series SSD, connected to a
Promise TX4 (FastTrak 4310) SATA300 controller (PDC40719).
Running 9.0-STABLE dated mid-January 2012.

 Writing to any filesystem that has TRIM enabled will make
the disk timeout and eventually panic the system.

atapci0: <Promise PDC40719 SATA300 controller> port 0xe000-0xe07f,0xd000-0xd0ff mem 0xf6042000-0xf6042fff,0xf6000000-0xf601ffff irq 11 at device 8.0 on pci0
ata5: SIGNATURE: 00000101
ada0 at ata5 bus 0 scbus5 target 0 lun 0
ada0: <INTEL SSDSC2CT060A3 300i> ATA-9 SATA 2.x device
ada0: 300.000MB/s transfers (SATA 2.x, UDMA5, PIO 8192bytes)
ada0: 57241MB (117231408 512 byte sectors: 16H 63S/T 16383C)

 Writing to a TRIM-enabled UFS filesystem will almost 
immediately stall whatever process running in ufs or
wdrain, with the disk seemingly dropping off the bus
after about 15s:

ata5: SIGNATURE: 00000101
ata5: SIGNATURE: ffffffff
ata5: timeout waiting to issue command
ata5: error issuing ATA_IDENTIFY command
ata5: SIGNATURE: ffffffff
(ada0:ata5:0:0:0): lost device
ata5: timeout waiting to issue command
ata5: error issuing unknown CMD (0x06) command
ata5: SIGNATURE: ffffffff
g_vfs_done():ada0s1a[WRITE(offset=40711716864, length=32768)]error = 6
/dev: got error 6 while accessing filesystem
panic: softdep_deallocate_dependencies: unrecovered I/O error
Uptime: 14m36s
ata5: timeout waiting to issue command
ata5: error issuing FLUSHCACHE48 command
ata5: SIGNATURE: ffffffff
(ada0:ata5:0:0:0): Synchronize cache failed

 The drive works fine without TRIM on this controller.  I
also had no issues whatsoever with two SATA harddisks pre-
viously running on the same card for several months.

 I believe there's an issue with the controller since I've
verified that the drive works fine with TRIM enabled on two
different SATA controllers (nForce3 and nForce MCP77 AHCI
on Gigabyte mainboards).

 Any use opening a PR or should I get another controller?
I'm aware there are "issues" at least with PDC40718-based


More information about the freebsd-stable mailing list