Re: Problems with aic94xx (AIC9410W onboard)

[Tomas Ebenlendr <ebik@jyxo.com>]

Hello again.
I digged more into the log and also previous logs.
The fail differs from normal noise in following:

First, normal noise begins with:
aic94xx: escb_tasklet_complete: REQ_TASK_ABORT, reason=0x6
sas: command 0xd8c229c0, task 0xda2505c0, timed out: EH_NOT_HANDLED

Whereas there is no REQ TASK ABORT line when the problem occurs.
Then it continues with following. The last three lines
(task not in LU) are very suspicious.

sas: Enter sas_scsi_recover_host
sas: trying to find task 0xd7aae5c0
sas: sas_scsi_find_task: aborting task 0xd7aae5c0
aic94xx: tmf timed out
aic94xx: tmf came back
aic94xx: task 0xd7aae5c0 aborted, res: 0x5
sas: sas_scsi_find_task: querying task 0xd7aae5c0
aic94xx: tmf tasklet complete
sas: sas_scsi_find_task: task 0xd7aae5c0 not at LU
sas: task 0xd7aae5c0 is not at LU: I_T recover
sas: I_T nexus reset for dev 5000c5000704c79d

Few other lines, and then reseting continues:

asd_clear_nexus_tasklet_complete: here
asd_clear_nexus_tasklet_complete: opcode: 0x0
sending hard reset to phy-3:5
control_phy_tasklet_complete: phy5: sub_func:0x81
aic94xx: asd_clear_nexus_I_T: PRE
aic94xx: asd_clear_nexus_I_T: POST
aic94xx: asd_clear_nexus_I_T: clear nexus posted, waiting...
aic94xx: asd_clear_nexus_timedout: here
aic94xx: asd_clear_nexus_I_T: PRE
aic94xx: asd_clear_nexus_I_T: POST
aic94xx: asd_clear_nexus_I_T: clear nexus posted, waiting...
aic94xx: asd_clear_nexus_tasklet_complete: here
aic94xx: asd_clear_nexus_tasklet_complete: opcode: 0x13
sas: I_T 5000c5000704c79d recovered
sas: --- Exit sas_scsi_recover_host

Log says recovered, but device is inaccessible, and whole controller
is frozen for a while then. Maybe until all other commands/tasks
are aborted.

Any help appreciated.
Thanks. Tomas Ebenlendr, Jyxo s.r.o., Czech Republic.