Re: sata_sil problems (lockup at boottime)

[Gerald Hopf <gerald.hopf@nv-systems.net>]

Hello everyone,

i'm using this post to answer to multiple questions of more than one 
person, so bear in mind that the quotes are not all from the same 
person/post.

Oliver Bausinger wrote:

>>I have the same problem with this hardware. The harddisk is a SAMSUNG SP1614C
>>again and the controller is this one:
>>    
>>
Great ! I really like the idea that i'm not the only one with this 
problem :-)

Oliver Bausinger wrote:
>> I tried quite a lot to get rid of those problems, including:
>> - Flashing a new/different Bios (even with different versions of the
>> Silicon Image SATA Bios part: 4.2.47 and 4.2.50)

>Where did you get it? URL?

I got it from here: http://homepage.ntlworld.com/michael.mcclay/
It's the "A7N8X-Dlx Uber 1009-BETA Sata Enhanced with new SATA bios v4.2.50  <- new (20th nov 04)".

I read about this Bios in the http://www.nforcershq.com/ Forums.

Works fine, but didn't change much (well, actually, it changed nothing regarding this problem...).

Bartlomiej Zolnierkiewicz wrote:

>Looks like some other device is using the same IRQ
>(it is not handled correctly and gets disabled).
>
>  
>
On my system, according to the lspci command no IRQs are shared. Only 4 
IRQs are in use, and none of them shared (with all the onboard stuff 
disabled).

Oliver Bausinger wrote:

>>I think I can rule out hardware problems because a FreeBSD 5.3 install
>>worked flawlessly on this disk.
>>    
>>
Same here, this hardware combination has worked absolutely flawless in 
Windows XP for months!

I wrote and Bartlomiej Zolnierkiewicz commented/asked:

>>>This problem has been there in all of the latest 2.6.X kernel revisions.
>>>I'm currently using 2.6.10-rc3 (which came out today), but i've allready
>>>had this problem in 2.6.9 and 2.6.8 and probably even 2.6.7 if i
>>>remember correctly.
>>>      
>>>
>Was it working before 2.6.7?
>  
>
I only recently tried to make this work, so i was not sure about this.
Therefor I just did some (time-consuming) tests with all Kernels between 
2.6.5 and 2.6.10-rc3:

Kernel 2.6.5:
- Bootet once and seemed to work, however only ONCE! Unable to reproduce 
this even with many reboots.
- (also i am not 100% sure if it actually worked this ONE time, cfdisk 
claimed the disk had no partitioning information on it which i thought 
it should have had... partitioning and reiserfs-formating worked fine 
though)
- In all following boots, the same strange errors, but continued to boot.
- Pictures of the strange errors and continued booting:
- http://www.nv-systems.net/pics/lkml/sata_sil4.jpg (page 1)
- http://www.nv-systems.net/pics/lkml/sata_sil5.jpg (page 2)
- http://www.nv-systems.net/pics/lkml/sata_sil6.jpg (same as page 2 but 
slightly different error from another try)
- After booting, harddrive does not respond, commands which access it 
(like cfdisk or mount) hang forever.

Kernel 2.6.6:
- Same as 2.6.5, except it didn't even work once.

Kernel 2.6.7:
- Same as 2.6.5, except it didn't even work once.

Kernel 2.6.8.1:
- Same as 2.6.5, except it didn't even work once.

Kernel 2.6.9:
- Hangs during bootup, like described in my original Post here!

Kernel 2.6.10-rc3
- Same behaviour as in 2.6.9

I also tried the depreciated siimage driver (thanks to the 
CONFIG_BLK_DEV_IDE_SATA tip, i didn't notice this before and had the 
same problems) with kernel 2.6.9,
works flawless (as far as i can tell from such a short test, hdparm -t = 
60MB/s). No strange IRQ disabled errors during bootup. Flawless of 
course only if you ignore that it takes AGES until it finally gives up 
probing the second SATA port. I know someone who bought a second SATA 
harddrive just because booting took so long!

I guess i should switch to a more PROMISing SATA controller as soon as 
possible... ;-)

Yours sincerely,
                  Gerald Hopf