2.9.29.2: ide-tape: panic when probing device at boot

2.9.29.2: ide-tape: panic when probing device at boot

[Jonathan Woithe]

Hi

I am experiencing a reproducable kernel panic during (I think) IDE probing
of an ide-connected tape unit in 2.6.29.2.  Under 2.6.24.2 things seem to
work ok, although some strange behaviour in recent weeks has me thinking
that the tape drive might have a problem.  In any case though, having
2.6.29.2 panic while 2.6.24.2 doesn't seems a bit strange.

For reference, the "strange behaviour" amounts to a whole stack of
the following messages appearing in syslog.
 WARNING: at include/asm/dma-mapping_32.h:45 dma_map_sg()
 Pid: 0, comm: swapper Not tainted 2.6.24.2 #1
  [<c02f38fe>] ide_build_sglist+0x85/0xd8
  [<c02f398b>] ide_build_dmatable+0x3a/0x158
  [<c0126d3d>] autoremove_wake_function+0x13/0x33
  [<c02f3cb4>] ide_dma_setup+0x26/0x8d
  [<c02fbbb5>] idetape_issue_packet_command+0x143/0x21d
  [<c016ff58>] mpage_end_io_read+0x0/0x50
  [<c02fc4b1>] idetape_do_request+0x2aa/0x2b2
  [<c02ef0a9>] ide_wait_stat+0x46/0x73
  [<c02ee332>] start_request+0x126/0x142
  [<c02ee5f3>] ide_do_request+0x287/0x2be
  [<c02eea27>] ide_intr+0xe0/0x101
  [<c02f37d0>] ide_dma_intr+0x0/0x9c
  [<c01345a1>] handle_IRQ_event+0x1a/0x3f
  [<c013512b>] handle_level_irq+0x50/0x85
  [<c0105325>] do_IRQ+0x74/0x8b
  [<c0103d0b>] common_interrupt+0x23/0x28
  [<c0102018>] default_idle+0x0/0x39
  [<c010203f>] default_idle+0x27/0x39
  [<c0102098>] cpu_idle+0x44/0x60
  [<c04ea769>] start_kernel+0x195/0x199
  [<c04ea32b>] unknown_bootoption+0x0/0x139
Eventually DMA gets disabled on all IDE devices, seemingly as a side effect
of this.

The tape drive concerned is a Sony "AIT-turbo" drive connected as
/dev/hdb.  The IDE interface is a VIA VT82C586A/B/VT82C686/A/B/VT823x/A/C.
The system hard drive is /dev/hda.  Unfortunately my resources are limited
here at present and the best I could get in terms of a panic report is the
following snippet.  Please excuse errors - I had to copy the details to
paper the old fashioned way and then type them in since I didn't have a
digital camera handy.  The extent of the dump is limited at the top by the
number of lines on the console (60).  The call trace stops prematurely
because I couldn't afford to keep the machine down any longer than I did (it
is a production server).

I guess at this point I'm interested to know whether this is indicative of a
kernel problem or whether there really is something potentially wrong with
this drive.  I'm happy to run additional tests if it will help (although a
full bisect is probably not practical given the age of the machine and the
fact that it is a production machine).  In particular, if the last half of
the call trace is deemed interesting I could get that in due course.

Please CC me to ensure I don't miss any replies.

Regards
  jonathan

Kernel oops under 2.6.29.2:

Process swapper (pid: 1 ti=df83e000 task=df822cc0 task.ti=df83e000)

Stack:                                             v-- or 7
  dfaa9000 dfb87000 df83fb7c c02f3ebe 00000000 fda29be0 dfb87800 dfa29be0
  000004ec df8e3de0 c02ebcec ffffffff ffffffff 00000000 dfb87800 dfa29be0
  dfb87000 c02e6e2a df8ae128 df8ae128 dfa29be0 dfa29be0 df8ae128 00000000

Call trace:
[<c02f3e6e>] idetape_do_request+0x1e5/0x1ec
[<c02e6cec>] start_request+0xfd/0x120
[<c02e6e2a>] do_ide_request+0xfa/0x13c
[<c0259ec8>] blk_start_queueing+0x15/0x1e
[<c0258c08>] elv_insert+0x68/0x14f
[<c025c7c0>] blk_execute_rq_nowait+0x4d/0x65
[<c025c84b>] blk_execute_rq+0x73/0x94
[<c025c750>] blk_end_sync_rq+0x0/0x23
[<c02ebc4e>] ide_queue_pc_tail+0x6b/0x7d
[<c02f54ce>] ide_tape_get_inquiry_results+0x28/0xc5
[<c03af803>] schedule_timeout+0x13/0x84
[<c0264870>] kobject_get+0xf/0x13
[<c02f57c9>] ide_tape_setup+0x99/0x1e6
:

Code: b6 53 89 cb 83 7c 24 08 00 74 07 e8 88 fb ff ff eb 05 08 f3 fa ff ff  
89 d8 5b c3 57 89 c7 56 53 86 70 18 90 d3 8b 80 9c 01 00 00<80>38 03 75 10
80 3a 03 75 0b 68 a3 0d 4a c0 e8 9c 4e e2 ff 58

EIP: [<c02f43a527>] idetape_issue_pc+0x10/0de SS:ESP 0068:df83fa20

---[ end trace b172b6d3ec773d8b ]---

Kernel panic - not syncing: Attempted to kill init!

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Bartlomiej Zolnierkiewicz]

[ cc:ing linux-ide and knowledgeable people ]

Borislav/Tejun:

Is the 2.6.29 problem the same thing that was fixed recently?

Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
or there is more needed to bring ide-tape to the world of living?

On Friday 08 May 2009 09:00:04 Jonathan Woithe wrote:
> Hi
> 
> I am experiencing a reproducable kernel panic during (I think) IDE probing
> of an ide-connected tape unit in 2.6.29.2.  Under 2.6.24.2 things seem to
> work ok, although some strange behaviour in recent weeks has me thinking
> that the tape drive might have a problem.  In any case though, having
> 2.6.29.2 panic while 2.6.24.2 doesn't seems a bit strange.
> 
> For reference, the "strange behaviour" amounts to a whole stack of
> the following messages appearing in syslog.
>  WARNING: at include/asm/dma-mapping_32.h:45 dma_map_sg()
>  Pid: 0, comm: swapper Not tainted 2.6.24.2 #1
>   [<c02f38fe>] ide_build_sglist+0x85/0xd8
>   [<c02f398b>] ide_build_dmatable+0x3a/0x158
>   [<c0126d3d>] autoremove_wake_function+0x13/0x33
>   [<c02f3cb4>] ide_dma_setup+0x26/0x8d
>   [<c02fbbb5>] idetape_issue_packet_command+0x143/0x21d
>   [<c016ff58>] mpage_end_io_read+0x0/0x50
>   [<c02fc4b1>] idetape_do_request+0x2aa/0x2b2
>   [<c02ef0a9>] ide_wait_stat+0x46/0x73
>   [<c02ee332>] start_request+0x126/0x142
>   [<c02ee5f3>] ide_do_request+0x287/0x2be
>   [<c02eea27>] ide_intr+0xe0/0x101
>   [<c02f37d0>] ide_dma_intr+0x0/0x9c
>   [<c01345a1>] handle_IRQ_event+0x1a/0x3f
>   [<c013512b>] handle_level_irq+0x50/0x85
>   [<c0105325>] do_IRQ+0x74/0x8b
>   [<c0103d0b>] common_interrupt+0x23/0x28
>   [<c0102018>] default_idle+0x0/0x39
>   [<c010203f>] default_idle+0x27/0x39
>   [<c0102098>] cpu_idle+0x44/0x60
>   [<c04ea769>] start_kernel+0x195/0x199
>   [<c04ea32b>] unknown_bootoption+0x0/0x139
> Eventually DMA gets disabled on all IDE devices, seemingly as a side effect
> of this.
> 
> The tape drive concerned is a Sony "AIT-turbo" drive connected as
> /dev/hdb.  The IDE interface is a VIA VT82C586A/B/VT82C686/A/B/VT823x/A/C.
> The system hard drive is /dev/hda.  Unfortunately my resources are limited
> here at present and the best I could get in terms of a panic report is the
> following snippet.  Please excuse errors - I had to copy the details to
> paper the old fashioned way and then type them in since I didn't have a
> digital camera handy.  The extent of the dump is limited at the top by the
> number of lines on the console (60).  The call trace stops prematurely
> because I couldn't afford to keep the machine down any longer than I did (it
> is a production server).
> 
> I guess at this point I'm interested to know whether this is indicative of a
> kernel problem or whether there really is something potentially wrong with
> this drive.  I'm happy to run additional tests if it will help (although a
> full bisect is probably not practical given the age of the machine and the
> fact that it is a production machine).  In particular, if the last half of
> the call trace is deemed interesting I could get that in due course.
> 
> Please CC me to ensure I don't miss any replies.
> 
> Regards
>   jonathan
> 
> Kernel oops under 2.6.29.2:
> 
> Process swapper (pid: 1 ti=df83e000 task=df822cc0 task.ti=df83e000)
> 
> Stack:                                             v-- or 7
>   dfaa9000 dfb87000 df83fb7c c02f3ebe 00000000 fda29be0 dfb87800 dfa29be0
>   000004ec df8e3de0 c02ebcec ffffffff ffffffff 00000000 dfb87800 dfa29be0
>   dfb87000 c02e6e2a df8ae128 df8ae128 dfa29be0 dfa29be0 df8ae128 00000000
> 
> Call trace:
> [<c02f3e6e>] idetape_do_request+0x1e5/0x1ec
> [<c02e6cec>] start_request+0xfd/0x120
> [<c02e6e2a>] do_ide_request+0xfa/0x13c
> [<c0259ec8>] blk_start_queueing+0x15/0x1e
> [<c0258c08>] elv_insert+0x68/0x14f
> [<c025c7c0>] blk_execute_rq_nowait+0x4d/0x65
> [<c025c84b>] blk_execute_rq+0x73/0x94
> [<c025c750>] blk_end_sync_rq+0x0/0x23
> [<c02ebc4e>] ide_queue_pc_tail+0x6b/0x7d
> [<c02f54ce>] ide_tape_get_inquiry_results+0x28/0xc5
> [<c03af803>] schedule_timeout+0x13/0x84
> [<c0264870>] kobject_get+0xf/0x13
> [<c02f57c9>] ide_tape_setup+0x99/0x1e6
> :
> 
> Code: b6 53 89 cb 83 7c 24 08 00 74 07 e8 88 fb ff ff eb 05 08 f3 fa ff ff  
> 89 d8 5b c3 57 89 c7 56 53 86 70 18 90 d3 8b 80 9c 01 00 00<80>38 03 75 10
> 80 3a 03 75 0b 68 a3 0d 4a c0 e8 9c 4e e2 ff 58
> 
> EIP: [<c02f43a527>] idetape_issue_pc+0x10/0de SS:ESP 0068:df83fa20
> 
> ---[ end trace b172b6d3ec773d8b ]---
> 
> Kernel panic - not syncing: Attempted to kill init!
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Borislav Petkov]

Hi,

On Sat, May 09, 2009 at 05:57:49PM +0200, Bartlomiej Zolnierkiewicz wrote:
> 
> [ cc:ing linux-ide and knowledgeable people ]
> 
> Borislav/Tejun:
> 
> Is the 2.6.29 problem the same thing that was fixed recently?

I'll let Tejun answer that one since I don't get an oops. However, now
that I have an ide-tape hardware here, I do get something DMA-related
failing during boot:

[    1.677012] hdb: Seagate STT8000A, ATAPI TAPE drive

..

[    1.701097] hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4
[    1.708289] hdb: bad DMA info in identify block
[    1.712872] hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4

..

[    3.450381] Uniform CD-ROM driver Revision: 3.20
[    5.467347] ide-tape: hdb <-> ht0: Seagate STT8000A rev 5.44
[    5.484852] ide-tape: hdb <-> ht0: 88KBps, 14*26kB buffer, 400ms tDS

and more specifically the bad DMA info in identify block thing above. I
haven't looked into it yet but its next on my TODO.

> Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
> ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
> or there is more needed to bring ide-tape to the world of living?

... and I'm afraid we'll have to do some serious bugfixing since this
driver is behaving really funny :). Stay tuned, I'm on it.

-- 
Regards/Gruss,
    Boris.

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Tejun Heo]

Bartlomiej Zolnierkiewicz wrote:
> [ cc:ing linux-ide and knowledgeable people ]
> 
> Borislav/Tejun:
> 
> Is the 2.6.29 problem the same thing that was fixed recently?
> 
> Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
> ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
> or there is more needed to bring ide-tape to the world of living?

The oops will probably fixed the above commit.  I don't have much idea
about the DMA problem on 2.6.24 tho.  But even with the above commit
fixed, I doubt it would work.  The buffer allocation code is broken
and reliably triggered OOM on my test machine.  While trying to locate
the bug, I realized the complex code didn't do much good to begin with
and just stripped it down, so I didn't actually tracked down the
actual bug && the patch to simplify buffer management is way too large
for 2.6.29, so unless someone is willing to hunt down the bug just for
2.6.29, we're kind of stuck.  :-(

Thanks.

-- 
tejun

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Tejun Heo]

Hello,

Borislav Petkov wrote:
> Hi,
> 
> On Sat, May 09, 2009 at 05:57:49PM +0200, Bartlomiej Zolnierkiewicz wrote:
>> [ cc:ing linux-ide and knowledgeable people ]
>>
>> Borislav/Tejun:
>>
>> Is the 2.6.29 problem the same thing that was fixed recently?
> 
> I'll let Tejun answer that one since I don't get an oops. However, now
> that I have an ide-tape hardware here, I do get something DMA-related
> failing during boot:
> 
> [    1.677012] hdb: Seagate STT8000A, ATAPI TAPE drive
> 
> ..
> 
> [    1.701097] hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> [    1.708289] hdb: bad DMA info in identify block
> [    1.712872] hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> 
> ..
> 
> [    3.450381] Uniform CD-ROM driver Revision: 3.20
> [    5.467347] ide-tape: hdb <-> ht0: Seagate STT8000A rev 5.44
> [    5.484852] ide-tape: hdb <-> ht0: 88KBps, 14*26kB buffer, 400ms tDS
> 
> and more specifically the bad DMA info in identify block thing above. I
> haven't looked into it yet but its next on my TODO.

Yeah, I've been seeing that too (before and after my changes).
Haven't investigated it yet.  Good that it's on your todo list.  :-)

>> Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
>> ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
>> or there is more needed to bring ide-tape to the world of living?
> 
> ... and I'm afraid we'll have to do some serious bugfixing since this
> driver is behaving really funny :). Stay tuned, I'm on it.

Cool.

Thanks.

-- 
tejun

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Tejun Heo]

Hello,

Borislav Petkov wrote:
> Hi,
> 
> On Sat, May 09, 2009 at 05:57:49PM +0200, Bartlomiej Zolnierkiewicz wrote:
>> [ cc:ing linux-ide and knowledgeable people ]
>>
>> Borislav/Tejun:
>>
>> Is the 2.6.29 problem the same thing that was fixed recently?
> 
> I'll let Tejun answer that one since I don't get an oops. However, now
> that I have an ide-tape hardware here, I do get something DMA-related
> failing during boot:
> 
> [    1.677012] hdb: Seagate STT8000A, ATAPI TAPE drive
> 
> ..
> 
> [    1.701097] hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> [    1.708289] hdb: bad DMA info in identify block
> [    1.712872] hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> 
> ..
> 
> [    3.450381] Uniform CD-ROM driver Revision: 3.20
> [    5.467347] ide-tape: hdb <-> ht0: Seagate STT8000A rev 5.44
> [    5.484852] ide-tape: hdb <-> ht0: 88KBps, 14*26kB buffer, 400ms tDS
> 
> and more specifically the bad DMA info in identify block thing above. I
> haven't looked into it yet but its next on my TODO.

Yeah, I've been seeing that too (before and after my changes).
Haven't investigated it yet.  Good that it's on your todo list.  :-)

>> Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
>> ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
>> or there is more needed to bring ide-tape to the world of living?
> 
> ... and I'm afraid we'll have to do some serious bugfixing since this
> driver is behaving really funny :). Stay tuned, I'm on it.

Cool.

Thanks.

-- 
tejun

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Jonathan Woithe]

> On Sat, May 09, 2009 at 05:57:49PM +0200, Bartlomiej Zolnierkiewicz wrote:
> > 
> > [ cc:ing linux-ide and knowledgeable people ]
> > 
> > Borislav/Tejun:
> > Is the 2.6.29 problem the same thing that was fixed recently?
> 
> I'll let Tejun answer that one since I don't get an oops. However, now
> that I have an ide-tape hardware here, I do get something DMA-related
> failing during boot ...
> :
> and more specifically the bad DMA info in identify block thing above. I
> haven't looked into it yet but its next on my TODO.

Cool.

> > Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
> > ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
> > or there is more needed to bring ide-tape to the world of living?
>
> The oops will probably fixed the above commit.  I don't have much idea
> about the DMA problem on 2.6.24 tho.  But even with the above commit
> fixed, I doubt it would work.  The buffer allocation code is broken
> and reliably triggered OOM on my test machine.  While trying to locate
> the bug, I realized the complex code didn't do much good to begin with
> and just stripped it down, so I didn't actually tracked down the
> actual bug && the patch to simplify buffer management is way too large
> for 2.6.29, so unless someone is willing to hunt down the bug just for
> 2.6.29, we're kind of stuck.  :-(

So I take it that 2.6.30 (or one of the RCs) is worth trying (at least in so
far as preventing the oops on boot)?

> ... and I'm afraid we'll have to do some serious bugfixing since this
> driver is behaving really funny :). Stay tuned, I'm on it.

Ok, I'll keep an eye out for updates.  Presumedly we're talking 2.6.31 or
.32 here.

Regards
  jonathan

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Jonathan Woithe]

> On Sat, May 09, 2009 at 05:57:49PM +0200, Bartlomiej Zolnierkiewicz wrote:
> > 
> > [ cc:ing linux-ide and knowledgeable people ]
> > 
> > Borislav/Tejun:
> > Is the 2.6.29 problem the same thing that was fixed recently?
> 
> I'll let Tejun answer that one since I don't get an oops. However, now
> that I have an ide-tape hardware here, I do get something DMA-related
> failing during boot ...
> :
> and more specifically the bad DMA info in identify block thing above. I
> haven't looked into it yet but its next on my TODO.

Cool.

> > Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
> > ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
> > or there is more needed to bring ide-tape to the world of living?
>
> The oops will probably fixed the above commit.  I don't have much idea
> about the DMA problem on 2.6.24 tho.  But even with the above commit
> fixed, I doubt it would work.  The buffer allocation code is broken
> and reliably triggered OOM on my test machine.  While trying to locate
> the bug, I realized the complex code didn't do much good to begin with
> and just stripped it down, so I didn't actually tracked down the
> actual bug && the patch to simplify buffer management is way too large
> for 2.6.29, so unless someone is willing to hunt down the bug just for
> 2.6.29, we're kind of stuck.  :-(

So I take it that 2.6.30 (or one of the RCs) is worth trying (at least in so
far as preventing the oops on boot)?

> ... and I'm afraid we'll have to do some serious bugfixing since this
> driver is behaving really funny :). Stay tuned, I'm on it.

Ok, I'll keep an eye out for updates.  Presumedly we're talking 2.6.31 or
.32 here.

Regards
  jonathan

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Borislav Petkov]

Hi,

On Mon, May 18, 2009 at 03:02:59PM +0930, Jonathan Woithe wrote:
> > On Sat, May 09, 2009 at 05:57:49PM +0200, Bartlomiej Zolnierkiewicz wrote:
> > > 
> > > [ cc:ing linux-ide and knowledgeable people ]
> > > 
> > > Borislav/Tejun:
> > > Is the 2.6.29 problem the same thing that was fixed recently?
> > 
> > I'll let Tejun answer that one since I don't get an oops. However, now
> > that I have an ide-tape hardware here, I do get something DMA-related
> > failing during boot ...
> > :
> > and more specifically the bad DMA info in identify block thing above. I
> > haven't looked into it yet but its next on my TODO.
> 
> Cool.
> 
> > > Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
> > > ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
> > > or there is more needed to bring ide-tape to the world of living?
> >
> > The oops will probably fixed the above commit.  I don't have much idea
> > about the DMA problem on 2.6.24 tho.  But even with the above commit
> > fixed, I doubt it would work.  The buffer allocation code is broken
> > and reliably triggered OOM on my test machine.  While trying to locate
> > the bug, I realized the complex code didn't do much good to begin with
> > and just stripped it down, so I didn't actually tracked down the
> > actual bug && the patch to simplify buffer management is way too large
> > for 2.6.29, so unless someone is willing to hunt down the bug just for
> > 2.6.29, we're kind of stuck.  :-(
> 
> So I take it that 2.6.30 (or one of the RCs) is worth trying (at least in so
> far as preventing the oops on boot)?

You could try 30-rc6 which runs pretty stable on my machines. It has the
patch mentioned above in it. At least one user confirmed that it fixes
ide-tape in his case. In case you still catch any OOPSes/warnings etc
please send me the whole dmesg output along with how you trigger them.

> > ... and I'm afraid we'll have to do some serious bugfixing since this
> > driver is behaving really funny :). Stay tuned, I'm on it.
> 
> Ok, I'll keep an eye out for updates.  Presumedly we're talking 2.6.31 or
> .32 here.

Already in .31-rc1 there's will more patches cleaning up ide-tape so if you
could try that and send us bug reports will be simply swell :).

Thanks.

-- 
Regards/Gruss,
    Boris.

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Borislav Petkov]

Small correction:

>> > > Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
>> > > ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
>> > > or there is more needed to bring ide-tape to the world of living?
>> >
>> > The oops will probably fixed the above commit.  I don't have much idea
>> > about the DMA problem on 2.6.24 tho.  But even with the above commit
>> > fixed, I doubt it would work.  The buffer allocation code is broken
>> > and reliably triggered OOM on my test machine.  While trying to locate
>> > the bug, I realized the complex code didn't do much good to begin with
>> > and just stripped it down, so I didn't actually tracked down the
>> > actual bug && the patch to simplify buffer management is way too large
>> > for 2.6.29, so unless someone is willing to hunt down the bug just for
>> > 2.6.29, we're kind of stuck.  :-(
>>
>> So I take it that 2.6.30 (or one of the RCs) is worth trying (at least in so
>> far as preventing the oops on boot)?
>
> You could try 30-rc6 which runs pretty stable on my machines. It has the
> patch mentioned above in it.

the patch mentioned above got actually merged after 30-rc6 so you could try
30-rc7 - if there will be one - or a current linux-next release.

Thanks.

-- 
Regards/Gruss,
Boris

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Jonathan Woithe]

Hi

> Borislav Petkov wrote:
> > Hi,
> > 
> > On Sat, May 09, 2009 at 05:57:49PM +0200, Bartlomiej Zolnierkiewicz wrote:
> >> [ cc:ing linux-ide and knowledgeable people ]
> >>
> >> Borislav/Tejun:
> >>
> >> Is the 2.6.29 problem the same thing that was fixed recently?
> > 
> > I'll let Tejun answer that one since I don't get an oops. However, now
> > that I have an ide-tape hardware here, I do get something DMA-related
> > failing during boot:
> > 
> > [    1.677012] hdb: Seagate STT8000A, ATAPI TAPE drive
> > 
> > ..
> > 
> > [    1.701097] hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> > [    1.708289] hdb: bad DMA info in identify block
> > [    1.712872] hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> > 
> > ..
> > 
> > [    3.450381] Uniform CD-ROM driver Revision: 3.20
> > [    5.467347] ide-tape: hdb <-> ht0: Seagate STT8000A rev 5.44
> > [    5.484852] ide-tape: hdb <-> ht0: 88KBps, 14*26kB buffer, 400ms tDS
> > 
> > and more specifically the bad DMA info in identify block thing above. I
> > haven't looked into it yet but its next on my TODO.
> 
> Yeah, I've been seeing that too (before and after my changes).
> Haven't investigated it yet.  Good that it's on your todo list.  :-)
> 
> >> Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
> >> ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
> >> or there is more needed to bring ide-tape to the world of living?
> > 
> > ... and I'm afraid we'll have to do some serious bugfixing since this
> > driver is behaving really funny :). Stay tuned, I'm on it.
> 
> Cool.

Has there been any movement on this in recent months?  I gave 2.6.30 a whirl
on the machine when that was released but it still gave major problems with
the ide-tape driver.  Is 2.6.32 likely to be improved?  If so I'll make
arrangements to test that version.

Presently I'm stuck with 2.6.24.x on this system since ide-tape holds
together enough on this version to work.  I really want to upgrade to a more
recent kernel but obviously I can't until ide-tape is functional again.

Regards
  jonathan

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Jonathan Woithe]

Hi

> Borislav Petkov wrote:
> > Hi,
> > 
> > On Sat, May 09, 2009 at 05:57:49PM +0200, Bartlomiej Zolnierkiewicz wrote:
> >> [ cc:ing linux-ide and knowledgeable people ]
> >>
> >> Borislav/Tejun:
> >>
> >> Is the 2.6.29 problem the same thing that was fixed recently?
> > 
> > I'll let Tejun answer that one since I don't get an oops. However, now
> > that I have an ide-tape hardware here, I do get something DMA-related
> > failing during boot:
> > 
> > [    1.677012] hdb: Seagate STT8000A, ATAPI TAPE drive
> > 
> > ..
> > 
> > [    1.701097] hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> > [    1.708289] hdb: bad DMA info in identify block
> > [    1.712872] hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> > 
> > ..
> > 
> > [    3.450381] Uniform CD-ROM driver Revision: 3.20
> > [    5.467347] ide-tape: hdb <-> ht0: Seagate STT8000A rev 5.44
> > [    5.484852] ide-tape: hdb <-> ht0: 88KBps, 14*26kB buffer, 400ms tDS
> > 
> > and more specifically the bad DMA info in identify block thing above. I
> > haven't looked into it yet but its next on my TODO.
> 
> Yeah, I've been seeing that too (before and after my changes).
> Haven't investigated it yet.  Good that it's on your todo list.  :-)
> 
> >> Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
> >> ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
> >> or there is more needed to bring ide-tape to the world of living?
> > 
> > ... and I'm afraid we'll have to do some serious bugfixing since this
> > driver is behaving really funny :). Stay tuned, I'm on it.
> 
> Cool.

Has there been any movement on this in recent months?  I gave 2.6.30 a whirl
on the machine when that was released but it still gave major problems with
the ide-tape driver.  Is 2.6.32 likely to be improved?  If so I'll make
arrangements to test that version.

Presently I'm stuck with 2.6.24.x on this system since ide-tape holds
together enough on this version to work.  I really want to upgrade to a more
recent kernel but obviously I can't until ide-tape is functional again.

Regards
  jonathan

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Borislav Petkov]

On Mon, Jan 04, 2010 at 11:45:12AM +1030, Jonathan Woithe wrote:
> Has there been any movement on this in recent months?  I gave 2.6.30 a whirl
> on the machine when that was released but it still gave major problems with
> the ide-tape driver.  Is 2.6.32 likely to be improved?  If so I'll make
> arrangements to test that version.

Yes, please do test 2.6.32 to see whether it fixes your problem. If you still
get that oops, please apply the following patch and send me the _whole_ dmesg
output of the machine.

Thanks.

--
diff --git a/drivers/ide/ide-tape.c b/drivers/ide/ide-tape.c
index 6a0e625..e4c3ffd 100644
--- a/drivers/ide/ide-tape.c
+++ b/drivers/ide/ide-tape.c
@@ -49,7 +49,7 @@
 #include <linux/mtio.h>
 
 /* define to see debug info */
-#undef IDETAPE_DEBUG_LOG
+#define IDETAPE_DEBUG_LOG
 
 #ifdef IDETAPE_DEBUG_LOG
 #define ide_debug_log(lvl, fmt, args...) __ide_debug_log(lvl, fmt, ## args)
@@ -1737,6 +1737,8 @@ static void idetape_setup(ide_drive_t *drive, idetape_tape_t *tape, int minor)
 	int buffer_size;
 	u16 *ctl = (u16 *)&tape->caps[12];
 
+	drive->debug_mask = 0xffffffff;
+
 	ide_debug_log(IDE_DBG_FUNC, "minor: %d", minor);
 
 	drive->pc_callback = ide_tape_callback;


-- 
Regards/Gruss,
    Boris.

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Robert Hancock]

On 01/03/2010 07:15 PM, Jonathan Woithe wrote:
>>>> Should I just push commit 1e75540ec5202cae63cd238c86bd880e3d496546
>>>> ("ide-tape: remove back-to-back REQUEST_SENSE detection") to Linus
>>>> or there is more needed to bring ide-tape to the world of living?
>>>
>>> ... and I'm afraid we'll have to do some serious bugfixing since this
>>> driver is behaving really funny :). Stay tuned, I'm on it.
>>
>> Cool.
>
> Has there been any movement on this in recent months?  I gave 2.6.30 a whirl
> on the machine when that was released but it still gave major problems with
> the ide-tape driver.  Is 2.6.32 likely to be improved?  If so I'll make
> arrangements to test that version.
>
> Presently I'm stuck with 2.6.24.x on this system since ide-tape holds
> together enough on this version to work.  I really want to upgrade to a more
> recent kernel but obviously I can't until ide-tape is functional again.

You could also try using libata instead of IDE drivers and see if they 
work any better for you..

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Jonathan Woithe]

> On Mon, Jan 04, 2010 at 11:45:12AM +1030, Jonathan Woithe wrote:
> > Has there been any movement on this in recent months?  I gave 2.6.30 a whirl
> > on the machine when that was released but it still gave major problems with
> > the ide-tape driver.  Is 2.6.32 likely to be improved?  If so I'll make
> > arrangements to test that version.
> 
> Yes, please do test 2.6.32 to see whether it fixes your problem. If you still
> get that oops, please apply the following patch and send me the _whole_ dmesg
> output of the machine.

Ok, will do.  Since it is a production machine and I've got linux.conf.au
coming up it may be a couple of weeks before I get to do the test.  I'll
advise the outcome once I have tested it.

Robert Hancock wrote:
> You could also try using libata instead of IDE drivers and see if they 
> work any better for you..

Yes, a good suggestion (I'm not sure why I neglected to try that initially). 
I'll test that too and report back.

Regards
  jonathan

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Jonathan Woithe]

> On Mon, Jan 04, 2010 at 11:45:12AM +1030, Jonathan Woithe wrote:
> > Has there been any movement on this in recent months?  I gave 2.6.30 a whirl
> > on the machine when that was released but it still gave major problems with
> > the ide-tape driver.  Is 2.6.32 likely to be improved?  If so I'll make
> > arrangements to test that version.
> 
> Yes, please do test 2.6.32 to see whether it fixes your problem. If you still
> get that oops, please apply the following patch and send me the _whole_ dmesg
> output of the machine.

Ok, will do.  Since it is a production machine and I've got linux.conf.au
coming up it may be a couple of weeks before I get to do the test.  I'll
advise the outcome once I have tested it.

Robert Hancock wrote:
> You could also try using libata instead of IDE drivers and see if they 
> work any better for you..

Yes, a good suggestion (I'm not sure why I neglected to try that initially). 
I'll test that too and report back.

Regards
  jonathan

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Jonathan Woithe]

Hi

Following up on my email from several months ago.  Sorry for the delay.

> > On Mon, Jan 04, 2010 at 11:45:12AM +1030, Jonathan Woithe wrote:
> > > Has there been any movement on this in recent months?  I gave 2.6.30 a whirl
> > > on the machine when that was released but it still gave major problems with
> > > the ide-tape driver.  Is 2.6.32 likely to be improved?  If so I'll make
> > > arrangements to test that version.
> > 
> > Yes, please do test 2.6.32 to see whether it fixes your problem. If you still
> > get that oops, please apply the following patch and send me the _whole_ dmesg
> > output of the machine.
> 
> Ok, will do.  Since it is a production machine and I've got linux.conf.au
> coming up it may be a couple of weeks before I get to do the test.  I'll
> advise the outcome once I have tested it.
> 
> Robert Hancock wrote:
> > You could also try using libata instead of IDE drivers and see if they 
> > work any better for you..
> 
> Yes, a good suggestion (I'm not sure why I neglected to try that initially). 
> I'll test that too and report back.

I've finally had an opportunity to test a few things.  I grabbed 2.6.33 and
configured libata support for the PATA controller (a VIA TV82CXXX) using the
"VIA PATA" option.  I disabled support for the legacy IDE driver.  Upon
booting 2.6.33's libata drivers correctly identified the two SATA HDDs in
the system, the PATA CDROM drive (secondary master) and the PATA tape unit
(primary master).  However, at this point things stopped - it seemed as if
the kernel lost the ability to communicate with either SATA HDD since it
reported timeouts on accesses to sda and sdb.

I haven't had a chance to test whether the inability to access the HDDs
comes as a result of the use of the VIA PATA libata driver or whether 2.6.33
has some inherant problem with VIA SATA controllers.  If I get a chance I'll
test that later today.  In any case, where do you suggest we go from here?

Unfortunately I didn't have time to note the precise errors seeen under
2.6.33 since I couldn't keep the machine down for very long.  I'll see if I
can repeat the test in a less busy part of the day later to get this
information.

Some system details:
  00:0f.0 RAID bus controller: VIA Technologies, Inc. VIA VT6420 SATA RAID
    Controller (rev 80)
  00:0f.1 IDE interface: VIA Technologies, Inc.
    VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE (rev 06)

00:0f.0 has two WD2502ABYS SATA HDDs connected.  Both drives have the
"1.5Gbps" jumper set because this chipset only supports 1.5Gbps but the
drives by default won't auto-negotiate down to this speed.  00:0f.1 has
an IDE tape drive (SONY SDX-260V) attached to primary master and a CD writer
(LITE-ON LTR-32123S) attached to secondary master.

Interface notes during boot on 2.6.24.2:

  Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
  ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
  VP_IDE: IDE controller (0x1106:0x0571 rev 0x06) at  PCI slot 0000:00:0f.1
  ACPI: PCI Interrupt Link [LNKE] enabled at IRQ 5
  PCI: setting IRQ 5 as level-triggered
  ACPI: PCI Interrupt 0000:00:0f.1[A] -> Link [LNKE] -> GSI 5 (level, low) ->
    IRQ 5
  PCI: VIA VLink IRQ fixup for 0000:00:0f.1, from 255 to 5
  VP_IDE: not 100% native mode: will probe irqs later
  VP_IDE: VIA vt8237 (rev 00) IDE UDMA133 controller on pci0000:00:0f.1
      ide0: BM-DMA at 0xa000-0xa007, BIOS settings: hda:DMA, hdb:pio
      ide1: BM-DMA at 0xa008-0xa00f, BIOS settings: hdc:DMA, hdd:pio
  Probing IDE interface ide0...  
  hda: SONY SDX-260V, ATAPI TAPE drive
  hda: host max PIO5 wanted PIO255(auto-tune) selected PIO4
  hda: UDMA/100 mode selected 
  ide0 at 0x1f0-0x1f7,0x3f6 on irq 14 
  Probing IDE interface ide1...
  hdc: LITE-ON LTR-32123S, ATAPI CD/DVD-ROM drive
  hdc: host max PIO5 wanted PIO255(auto-tune) selected PIO4
  hdc: UDMA/33 mode selected
  ide1 at 0x170-0x177,0x376 on irq 15
  hdc: ATAPI 40X CD-ROM CD-R/RW drive, 2048kB Cache
  Uniform CD-ROM driver Revision: 3.20
  ide-tape: hda <-> ht0: SONY SDX-260V rev 0100
  ide-tape: hda: overriding capabilities->speed (assuming 650KB/sec)
  ide-tape: hda: overriding capabilities->max_speed (assuming 650KB/sec)
  ide-tape: decreasing stage size
  ide-tape: decreasing stage size
  ide-tape: decreasing stage size
  ide-tape: hda <-> ht0: 650KBps, 126*32kB buffer, 6336kB pipeline, 100ms
    tDSC, DMA
  st: Version 20070203, fixed bufsize 32768, s/g segs 256 
  :
  sata_via 0000:00:0f.0: version 2.3
  ACPI: PCI Interrupt Link [LNKF] BIOS reported IRQ 0, using IRQ 9
  ACPI: PCI Interrupt Link [LNKF] enabled at IRQ 9
  PCI: setting IRQ 9 as level-triggered
  ACPI: PCI Interrupt 0000:00:0f.0[B] -> Link [LNKF] -> GSI 9 (level, low) ->
    IRQ 9
  PCI: VIA VLink IRQ fixup for 0000:00:0f.0, from 5 to 9
  sata_via 0000:00:0f.0: routed to hard irq line 9
  scsi0 : sata_via
  scsi1 : sata_via
  ata1: SATA max UDMA/133 cmd 0xd000 ctl 0xb800 bmdma 0xa800 irq 9
  ata2: SATA max UDMA/133 cmd 0xb400 ctl 0xb000 bmdma 0xa808 irq 9
  ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
  ata1.00: ATA-8: WDC WD2502ABYS-01B7A0, 02.03B02, max UDMA/133
  ata1.00: 490350672 sectors, multi 16: LBA48 NCQ (depth 0/32) 
  ata1.00: configured for UDMA/133
  ata2: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
  ata2.00: ATA-8: WDC WD2502ABYS-01B7A0, 02.03B02, max UDMA/133
  ata2.00: 490350672 sectors, multi 16: LBA48 NCQ (depth 0/32) 
  ata2.00: configured for UDMA/133

Regards
  jonathan

Re: 2.9.29.2: ide-tape: panic when probing device at boot

[Jonathan Woithe]

Hi

Following up on my email from several months ago.  Sorry for the delay.

> > On Mon, Jan 04, 2010 at 11:45:12AM +1030, Jonathan Woithe wrote:
> > > Has there been any movement on this in recent months?  I gave 2.6.30 a whirl
> > > on the machine when that was released but it still gave major problems with
> > > the ide-tape driver.  Is 2.6.32 likely to be improved?  If so I'll make
> > > arrangements to test that version.
> > 
> > Yes, please do test 2.6.32 to see whether it fixes your problem. If you still
> > get that oops, please apply the following patch and send me the _whole_ dmesg
> > output of the machine.
> 
> Ok, will do.  Since it is a production machine and I've got linux.conf.au
> coming up it may be a couple of weeks before I get to do the test.  I'll
> advise the outcome once I have tested it.
> 
> Robert Hancock wrote:
> > You could also try using libata instead of IDE drivers and see if they 
> > work any better for you..
> 
> Yes, a good suggestion (I'm not sure why I neglected to try that initially). 
> I'll test that too and report back.

I've finally had an opportunity to test a few things.  I grabbed 2.6.33 and
configured libata support for the PATA controller (a VIA TV82CXXX) using the
"VIA PATA" option.  I disabled support for the legacy IDE driver.  Upon
booting 2.6.33's libata drivers correctly identified the two SATA HDDs in
the system, the PATA CDROM drive (secondary master) and the PATA tape unit
(primary master).  However, at this point things stopped - it seemed as if
the kernel lost the ability to communicate with either SATA HDD since it
reported timeouts on accesses to sda and sdb.

I haven't had a chance to test whether the inability to access the HDDs
comes as a result of the use of the VIA PATA libata driver or whether 2.6.33
has some inherant problem with VIA SATA controllers.  If I get a chance I'll
test that later today.  In any case, where do you suggest we go from here?

Unfortunately I didn't have time to note the precise errors seeen under
2.6.33 since I couldn't keep the machine down for very long.  I'll see if I
can repeat the test in a less busy part of the day later to get this
information.

Some system details:
  00:0f.0 RAID bus controller: VIA Technologies, Inc. VIA VT6420 SATA RAID
    Controller (rev 80)
  00:0f.1 IDE interface: VIA Technologies, Inc.
    VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE (rev 06)

00:0f.0 has two WD2502ABYS SATA HDDs connected.  Both drives have the
"1.5Gbps" jumper set because this chipset only supports 1.5Gbps but the
drives by default won't auto-negotiate down to this speed.  00:0f.1 has
an IDE tape drive (SONY SDX-260V) attached to primary master and a CD writer
(LITE-ON LTR-32123S) attached to secondary master.

Interface notes during boot on 2.6.24.2:

  Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
  ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
  VP_IDE: IDE controller (0x1106:0x0571 rev 0x06) at  PCI slot 0000:00:0f.1
  ACPI: PCI Interrupt Link [LNKE] enabled at IRQ 5
  PCI: setting IRQ 5 as level-triggered
  ACPI: PCI Interrupt 0000:00:0f.1[A] -> Link [LNKE] -> GSI 5 (level, low) ->
    IRQ 5
  PCI: VIA VLink IRQ fixup for 0000:00:0f.1, from 255 to 5
  VP_IDE: not 100% native mode: will probe irqs later
  VP_IDE: VIA vt8237 (rev 00) IDE UDMA133 controller on pci0000:00:0f.1
      ide0: BM-DMA at 0xa000-0xa007, BIOS settings: hda:DMA, hdb:pio
      ide1: BM-DMA at 0xa008-0xa00f, BIOS settings: hdc:DMA, hdd:pio
  Probing IDE interface ide0...  
  hda: SONY SDX-260V, ATAPI TAPE drive
  hda: host max PIO5 wanted PIO255(auto-tune) selected PIO4
  hda: UDMA/100 mode selected 
  ide0 at 0x1f0-0x1f7,0x3f6 on irq 14 
  Probing IDE interface ide1...
  hdc: LITE-ON LTR-32123S, ATAPI CD/DVD-ROM drive
  hdc: host max PIO5 wanted PIO255(auto-tune) selected PIO4
  hdc: UDMA/33 mode selected
  ide1 at 0x170-0x177,0x376 on irq 15
  hdc: ATAPI 40X CD-ROM CD-R/RW drive, 2048kB Cache
  Uniform CD-ROM driver Revision: 3.20
  ide-tape: hda <-> ht0: SONY SDX-260V rev 0100
  ide-tape: hda: overriding capabilities->speed (assuming 650KB/sec)
  ide-tape: hda: overriding capabilities->max_speed (assuming 650KB/sec)
  ide-tape: decreasing stage size
  ide-tape: decreasing stage size
  ide-tape: decreasing stage size
  ide-tape: hda <-> ht0: 650KBps, 126*32kB buffer, 6336kB pipeline, 100ms
    tDSC, DMA
  st: Version 20070203, fixed bufsize 32768, s/g segs 256 
  :
  sata_via 0000:00:0f.0: version 2.3
  ACPI: PCI Interrupt Link [LNKF] BIOS reported IRQ 0, using IRQ 9
  ACPI: PCI Interrupt Link [LNKF] enabled at IRQ 9
  PCI: setting IRQ 9 as level-triggered
  ACPI: PCI Interrupt 0000:00:0f.0[B] -> Link [LNKF] -> GSI 9 (level, low) ->
    IRQ 9
  PCI: VIA VLink IRQ fixup for 0000:00:0f.0, from 5 to 9
  sata_via 0000:00:0f.0: routed to hard irq line 9
  scsi0 : sata_via
  scsi1 : sata_via
  ata1: SATA max UDMA/133 cmd 0xd000 ctl 0xb800 bmdma 0xa800 irq 9
  ata2: SATA max UDMA/133 cmd 0xb400 ctl 0xb000 bmdma 0xa808 irq 9
  ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
  ata1.00: ATA-8: WDC WD2502ABYS-01B7A0, 02.03B02, max UDMA/133
  ata1.00: 490350672 sectors, multi 16: LBA48 NCQ (depth 0/32) 
  ata1.00: configured for UDMA/133
  ata2: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
  ata2.00: ATA-8: WDC WD2502ABYS-01B7A0, 02.03B02, max UDMA/133
  ata2.00: 490350672 sectors, multi 16: LBA48 NCQ (depth 0/32) 
  ata2.00: configured for UDMA/133

Regards
  jonathan