Date: Sat, 22 Jan 2005 17:54:18 -0700 From: "Justin T. Gibbs" <gibbs@scsiguy.com> To: Anton Berezin <tobez@tobez.org>, freebsd-scsi@freebsd.org Subject: Re: 29320A: tons of "unexpected busfree while idle" Message-ID: <92B2C5A0199CF60FA976BBDC@aslan.scsiguy.com> In-Reply-To: <20050121135814.GA35988@heechee.tobez.org> References: <20050121135814.GA35988@heechee.tobez.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> Hi,
>
> I am having a lot of "unexpected busfree while idle" events with Adaptec
> 29320A Ultra320 SCSI adapter.
>
> While trying to eliminate various possibilities, it turned out that the
> problem disappears if I downgrade sys/dev/aic7xxx directory to
>
> -D '01 Sep 2004' -rRELENG_5
>
> So it looks like the problem was introduced sometime after that.
I know which change is causing your problem, but not why. I'm going
to run some experiments here to try replicate the failure here.
> Jan 18 10:15:56 seifert kernel: ahd0: Unexpected busfree while idle, 0 SCBs aborted, PRGMCNT == 0xb9
> Jan 18 10:15:56 seifert kernel: >>>>>>>>>>>>>>>>>> Dump Card State Begins <<<<<<<<<<<<<<<<<
> Jan 18 10:15:56 seifert kernel: ahd0: Dumping Card State at program address 0xb7 Mode 0x33
> Jan 18 10:15:56 seifert kernel: Card was paused
> Jan 18 10:15:56 seifert kernel: INTSTAT[0x8] SELOID[0x4] SELID[0x0] HS_MAILBOX[0x0]
In the future, please compile your kernels with:
options AHD_REG_PRETTY_PRINT # Print register bitfields in debug
# output. Adds ~215k to driver.
This adds the string table necessary to auto-decode the card state dump.
It saves me having to dig through register listings to figure out the
source of a problem. There is no run-time penalty for enabling this
option.
> Another strangety is how the thing recognizes the disks (any version of
> the driver):
>
> Jan 18 10:15:56 seifert kernel: da0 at ahd0 bus 0 target 0 lun 0
> Jan 18 10:15:56 seifert kernel: da0: <HITACHI HUS157373EL3600 A5A5> Fixed Direct Access SCSI-3 device
> Jan 18 10:15:56 seifert kernel: da0: 40.000MB/s transfers (20.000MHz, offset 81, 16bit), Tagged Queueing Enabled
This is caused by running your disks on the Single-Ended segment of the
29320A. This will limit operation to Ultra SCSI speeds. If you move your
string of disks to the other internal port, you should see faster speeds
if your disks support Low-Voltage-Differential operating modes.
--
Justin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?92B2C5A0199CF60FA976BBDC>