Re: HEADS UP! ATAng committed

On Tue, 26 Aug 2003, Wiktor Niesiobedzki wrote:

> On Sun, Aug 24, 2003 at 11:27:05AM +0200, Soren Schmidt wrote:
> >
> > ATAng has just been committed. You need to make world after this update
> > as atacontrol etc needs to pick up the changes.
> >
> After updating to ATAng my DVD drive isn't detected. I get following message:
> ata1-slave: FAILURE - ATA_IDENTIFY status=51<READY,DSC,ERROR> error=4<ABORTED>

Try backing out rev.1.5 of ata-lowlevel.c.  It clobbers the signature of
all 3 atapi drives on one of my systems here.  This results the atapi
drives being further probed as ata drives and the probe soon fails with
the above error.

Output from ATAog for these atapi drives:

Aug 27 00:16:50 gamplex kernel: afd0: 96MB <IOMEGA ZIP 100 ATAPI> [96/64/32] at ata0-slave PIO0
Aug 27 00:16:50 gamplex kernel: acd0: CD-RW <RICOH CD-RW MP7320A> at ata1-slave UDMA33
Aug 27 00:16:50 gamplex kernel: acd1: CDROM <ATAPI 44X CDROM> at ata2-slave PIO4
Aug 27 00:16:50 gamplex kernel: cd1 at ata2 bus 0 target 1 lun 0
Aug 27 00:16:50 gamplex kernel: cd1: <ATAPI 44X CDROM 3.40> Removable CD-ROM SCSI-0 device
Aug 27 00:16:50 gamplex kernel: cd1: 16.000MB/s transfers
Aug 27 00:16:50 gamplex kernel: cd1: cd present [317714 x 2048 byte records]
Aug 27 00:16:50 gamplex kernel: cd0 at ata1 bus 0 target 1 lun 0
Aug 27 00:16:50 gamplex kernel: cd0: <RICOH CD-RW MP7320A bp13> Removable CD-ROM SCSI-0 device
Aug 27 00:16:50 gamplex kernel: cd0: 33.000MB/s transfers
Aug 27 00:16:50 gamplex kernel: cd0: cd present [320930 x 2048 byte records]
Aug 27 00:16:50 gamplex kernel: da0 at ata0 bus 0 target 1 lun 0
Aug 27 00:16:50 gamplex kernel: da0: <IOMEGA ZIP 100 14.A> Removable Direct Access SCSI-0 device
Aug 27 00:16:50 gamplex kernel: da0: 3.300MB/s transfers
Aug 27 00:16:50 gamplex kernel: da0: 96MB (196608 512 byte sectors: 64H 32S/T 96C)

ata0 and ata1 are on a BX chipset, and ata2 and ata3 are on an old highpoint
chipset.  ata0 has a working ata drive on the master.  ata1 has a broken
ata drive on the master (I keep it for debugging ata; it probes correctly
but recently became unreadable on block 0).  ata2 has no master.  ata3 has
no drives.

Breaking the probe of the atapi drives avoids some other bugs:
- a panic for atapicam (already reported).
- spurious interrupts on ata2.  -current somehow recovers after reporting
  about 10 such interrupts, but my -mycurrent spins endlessly on them.  I
  suspect Giant locking problems.

Bruce