> > Everything seems functional under SMP.......at 2.1.123/124/125 there
> > appears to be a hang and DSC timeout. I am hunting through the changes
The first potential problem that I have found is as follows ::
Oct 23 00:49:58 Zosma kernel: Command Packet DRQ Type: <6>Interrupt DRQ
Oct 23 00:49:58 Zosma kernel: Command Packet Size: <6>12 bytes
Oct 23 00:49:58 Zosma kernel: Model: HP COLORADO 5GB
Oct 23 00:49:58 Zosma kernel: Firmware Revision: 2.010000
Oct 23 00:49:58 Zosma kernel: Serial Number: US04123977
Oct 23 00:49:58 Zosma kernel: Write buffer size: 0 bytes
Oct 23 00:49:58 Zosma kernel: DMA: Yes
Oct 23 00:49:58 Zosma kernel: LBA: Yes
Oct 23 00:49:58 Zosma kernel: IORDY can be disabled: Yes
Oct 23 00:49:58 Zosma kernel: IORDY supported: Yes
Oct 23 00:49:58 Zosma kernel: ATAPI overlap supported: No
Oct 23 00:49:58 Zosma kernel: PIO Cycle Timing Category: 2
Oct 23 00:49:58 Zosma kernel: DMA Cycle Timing Category: 2
HP COLORADO 5GB
ht0
name value min max mode
---- ----- --- --- ----
buffer 512 0 32767 r
dsc_overlap 1 0 1 rw
io_32bit 0 0 3 rw
keepsettings 0 0 1 rw
The tape driver defaults to "ATAPI overlap supported" and is not flag
tested and updated. Since the "HP COLORADO 5GB" reports that this is not
supported hardware via the logs and the driver thinks that it is support,
here is the first suspect in causing timing faults.
If I understand the concept of "dsc_overlap" and how it was implimented by
Gadi, IMHO the hardware and driver speak different dialects (sp) of the
same language.
I am going to try next disabling this setting with a hack and see the
results. If the solves the problem, then another function or option
setting may be needed to correctly configure the two to the same level.
Cheers,
Andre Hedrick
The IDE-FNG for Linux
The APC UPS Specialist for Linux
http://www.dyer.vanderbilt.edu/server/udma/
http://www.dyer.vanderbilt.edu/server/apcupsd/ SUSPENDED
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/