Re: Sata 4726 Testing

Tejun,

I have completed some of the testing, things were working fine.. then
something died. It will not even recognize the 4726 anymore. During book
when the HBA does it scan nor in windows 2000 does anything work or be
recognized.

I will have a new eSATA cable this week sometime, maybe a pin got
shmucked if lucky. I've attached the logs of the good and the gone dead.

Here's the testing procedure:

Sata 4726 testing: 4726 on at boot:

1. Rebooted the computer, powered the sata on 4726 before booting.
    Notes: Everything works
   
    File: 1_Bootup.log
       
2. Turn 4726 off, wait 10 seconds, turn 4726 on
    Notes: Works
   
3. Repeat #2.
    Notes: Works
   
    File: 2_PowerOnOffTwice.log
   
4. Unplug  4726, wait 10 seconds, plug back in (leave 4726 powered up)
    Notes: Works
   
    File: 3_PlugUnplug.txt

5. Power 4726 off, wait 10 seconds, power back on 4726
    Notes: Something crapped out. (Timestamp 404-505). LED status lights
dark

6. Repeat step #5
    Notes: Repeating did not fix things. Same failures.
   
    File: 4_PowerOnOffTwice.log
   
   
7. Unplugged and re-plugged 4726 on same port
    Notes: Did not do a thing. Did not even generate a kernel message..
(Timestamp between 688 and 776)

8. Unplugged from port 1 and plugged 4726 into port 2
    Same results as #7. (Timestamp 776-836)
   
9. Powered unit off, unplugged cable from HBA, waited 10 seconds,
plugged cable
    back into HBA and powered up.
    Notes: Nothing. No kernel messages or activity.
   
10. Unloaded sata_sil24 module.

    File: 5_Broken.log

11. Reloaded sata_sil24
    Notes: Still unresponsive.
   
    File: 6_ModuleReload.log

12. Repeated steps 10 & 11 but removed sata_sil24 and libata
    Notes: Unresponsive. LED status lights are still dark also since the
initial failure.
   
    File 7_ModuleReload2.log




Tejun Heo wrote:

Andrew Ryder wrote:
  

quoted

Tejun,

No problem. I should be thanking you and your team for all the hard work
writing the drivers for the hardware..

I think this patch might have solved the issue. I did a quick test: warm
rebooted and turned the 4726 on and off 3 times and all disks were
successfully detected each time. I'll do a more thorough test tomorrow
or Sunday.
    

Great.  I forwarded the patch for #upstream-fixes as the patch is in the
right direction whether it fully fixes your problem or not.  Please
lemme know the test result.

Gaston, for the initial detection failure messages when connected to ICH
AHCI, I don't really have much idea what's going on or how to fix it.  I
suspect it's related to initial FISes from drives arriving
asynchronously after hard resets but can't tell much without bus
analyzer.  Anyways, as EH can handle the case quickly enough, I guess we
can leave it at that for now.  If you have ideas on what's going on or
how to fix it, please lemme know.

Thanks.