freebsd stable - Dec 2009 - problems with SATA controller after recent RELENG_8 upgrade

Hi,

I've just upgraded my 8 from around the 6th of december 2 days ago. Now
the system won't boot up. When it is going to mount the rootfs, it
receives some ICRC error and the harddisk gets accessed massivly. The the
error shown on the screenshot is repeating and repeating. Apart from my
custom kernel I also compiled and tried the GENERIC kernel with some
additional modules (ipfw, dummynet, smb, intpm, pcfclock - nothing which
should interfear)
The filesystem is labeled with glabel/tunefs.

Could you advise me what to do next? Right now I'm using the old kernel...

Screenshot (where I tried to reach at least single user):

   http://pics.pofo.de/gallery/v/misc/P1090111.JPG.html


The controller and harddisk in question:

atapci1: <Promise PDC40775 SATA300 controller> port
0xec00-0xec7f,0xe800-0xe8ff mem 0xffaff000-0xffafffff,0xffac0000-0xffadffff irq
11 at device 17.0 on pci0
atapci1: [ITHREAD]
atapci1: [ITHREAD]
ata2: <ATA channel 0> on atapci1
ata2: SIGNATURE: 00000101
ata2: [ITHREAD]
ata3: <ATA channel 1> on atapci1
ata3: [ITHREAD]
ata4: <ATA channel 2> on atapci1
ata4: [ITHREAD]

ad4: 238475MB <WDC WD2500KS-00MJB0 02.01C03> at ata2-master SATA300
GEOM: ad4s1: geometry does not match label (255h,63s != 16h,63s).
Trying to mount root from ufs:/dev/ufs/root


-- 
 Oliver Lehmann
  http://www.pofo.de/
  http://wishlist.ans-netz.de/

Oliver Lehmann wrote:
> I've just upgraded my 8 from around the 6th of december 2 days ago. Now
> the system won't boot up. When it is going to mount the rootfs, it
> receives some ICRC error and the harddisk gets accessed massivly. The the
> error shown on the screenshot is repeating and repeating. Apart from my
> custom kernel I also compiled and tried the GENERIC kernel with some
> additional modules (ipfw, dummynet, smb, intpm, pcfclock - nothing which
> should interfear)
> The filesystem is labeled with glabel/tunefs.
> 
> Could you advise me what to do next? Right now I'm using the old
kernel...
> 
> Screenshot (where I tried to reach at least single user):
> 
>    http://pics.pofo.de/gallery/v/misc/P1090111.JPG.html
Looks like it was working first, until something happened. I've reread
all Promise related changes and don't see problem there. The only idea I
have is that it could be larger transfer, which was not used before. Try
to apply this patch to get limitation back for these controllers:

--- ata-promise.c.prev  2009-12-15 21:35:43.000000000 +0200
+++ ata-promise.c       2009-12-15 21:35:24.000000000 +0200
@@ -957,6 +957,7 @@ ata_promise_mio_dmainit(device_t dev)
     ata_dmainit(dev);
     /* note start and stop are not used here */
     ch->dma.setprd = ata_promise_mio_setprd;
+    ch->dma.max_iosize = 65536;
 }


-- 
Alexander Motin