linux-scsi.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* Hotplug does not work with Fusion MPT
@ 2006-05-15 12:08 Maik Hampel
  2006-05-15 13:44 ` Matthew Wilcox
  0 siblings, 1 reply; 5+ messages in thread
From: Maik Hampel @ 2006-05-15 12:08 UTC (permalink / raw)
  To: linux-scsi

Hello,

I'm using MPT-Fusion driver (linux-2.6.16.x) with LSI 53C1030 on SPI.

I recognized, that removing a scsi-drive (connected with SCA) did not
remove it from linux scsi subsystem. The physically removed drive was
still listed in /proc/scsi/scsi:

Host: scsi0 Channel: 00 Id: 04 Lun: 00
  Vendor: FUJITSU  Model: MAP3367NC        Rev: 0108
  Type:   Direct-Access                    ANSI SCSI revision: 03

On access via cfdisk /dev/sdd the following error messages occured in
syslog:

mptbase: ioc0: IOCStatus(0x0043): SCSI Device Not There
sd 0:0:4:0: SCSI error: return code = 0x10000
end_request: I/O error, dev sdd, sector 0
Buffer I/O error on device sdd, logical block 0
mptbase: ioc0: IOCStatus(0x0043): SCSI Device Not There
sd 0:0:4:0: SCSI error: return code = 0x10000
end_request: I/O error, dev sdd, sector 4
Buffer I/O error on device sdd, logical block 1
mptbase: ioc0: IOCStatus(0x0043): SCSI Device Not There
sd 0:0:4:0: SCSI error: return code = 0x10000
end_request: I/O error, dev sdd, sector 8
Buffer I/O error on device sdd, logical block 2
mptbase: ioc0: IOCStatus(0x0043): SCSI Device Not There
sd 0:0:4:0: SCSI error: return code = 0x10000
end_request: I/O error, dev sdd, sector 0
Buffer I/O error on device sdd, logical block 0
mptbase: ioc0: IOCStatus(0x0043): SCSI Device Not There
sd 0:0:4:0: SCSI error: return code = 0x10000
end_request: I/O error, dev sdd, sector 12
Buffer I/O error on device sdd, logical block 3
mptbase: ioc0: IOCStatus(0x0043): SCSI Device Not There
sd 0:0:4:0: SCSI error: return code = 0x10000
end_request: I/O error, dev sdd, sector 4
Buffer I/O error on device sdd, logical block 1
mptbase: ioc0: IOCStatus(0x0043): SCSI Device Not There
sd 0:0:4:0: SCSI error: return code = 0x10000
end_request: I/O error, dev sdd, sector 16
Buffer I/O error on device sdd, logical block 4
mptbase: ioc0: IOCStatus(0x0043): SCSI Device Not There
sd 0:0:4:0: SCSI error: return code = 0x10000
end_request: I/O error, dev sdd, sector 20
Buffer I/O error on device sdd, logical block 5
mptbase: ioc0: IOCStatus(0x0043): SCSI Device Not There
sd 0:0:4:0: SCSI error: return code = 0x10000
end_request: I/O error, dev sdd, sector 24

Any suggestions about that?


Maik Hampel



^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: Hotplug does not work with Fusion MPT
  2006-05-15 12:08 Hotplug does not work with Fusion MPT Maik Hampel
@ 2006-05-15 13:44 ` Matthew Wilcox
  0 siblings, 0 replies; 5+ messages in thread
From: Matthew Wilcox @ 2006-05-15 13:44 UTC (permalink / raw)
  To: Maik Hampel; +Cc: linux-scsi

On Mon, May 15, 2006 at 02:08:14PM +0200, Maik Hampel wrote:
> I recognized, that removing a scsi-drive (connected with SCA) did not
> remove it from linux scsi subsystem. The physically removed drive was
> still listed in /proc/scsi/scsi:
> 
> Host: scsi0 Channel: 00 Id: 04 Lun: 00
>   Vendor: FUJITSU  Model: MAP3367NC        Rev: 0108
>   Type:   Direct-Access                    ANSI SCSI revision: 03

Did you tell Linux you removed it?  Try:

echo scsi remove-single-device 0 0 4 0 > /proc/scsi/scsi

Parallel SCSI doesn't have a way to notify the OS that a device is gone,
so the user has to say.

You can also use tools like scsiadd for this kind of task, which has a
more agreeable user interface.

^ permalink raw reply	[flat|nested] 5+ messages in thread

* RE: Hotplug does not work with Fusion MPT
@ 2006-05-15 15:04 Moore, Eric
  2006-05-15 15:51 ` Matthew Wilcox
  0 siblings, 1 reply; 5+ messages in thread
From: Moore, Eric @ 2006-05-15 15:04 UTC (permalink / raw)
  To: Matthew Wilcox, Maik Hampel; +Cc: linux-scsi

On Monday, May 15, 2006 7:44 AM, Matthew Wilcox wrote:
> Did you tell Linux you removed it?  Try:
> 
> echo scsi remove-single-device 0 0 4 0 > /proc/scsi/scsi
> 
> Parallel SCSI doesn't have a way to notify the OS that a 
> device is gone,
> so the user has to say.
> 

We have Hotplug device support in fustion drivers for SAS and Fibre, not
for SPI.

Eric Moore

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: Hotplug does not work with Fusion MPT
  2006-05-15 15:04 Moore, Eric
@ 2006-05-15 15:51 ` Matthew Wilcox
  0 siblings, 0 replies; 5+ messages in thread
From: Matthew Wilcox @ 2006-05-15 15:51 UTC (permalink / raw)
  To: Moore, Eric; +Cc: Maik Hampel, linux-scsi

On Mon, May 15, 2006 at 09:04:52AM -0600, Moore, Eric wrote:
> On Monday, May 15, 2006 7:44 AM, Matthew Wilcox wrote:
> > Did you tell Linux you removed it?  Try:
> > 
> > echo scsi remove-single-device 0 0 4 0 > /proc/scsi/scsi
> > 
> > Parallel SCSI doesn't have a way to notify the OS that a 
> > device is gone,
> > so the user has to say.
> 
> We have Hotplug device support in fustion drivers for SAS and Fibre, not
> for SPI.

Erm.  Are you saying it doesn't work?  Or you just don't test it?
If the user runs the above command, the host's ->slave_destroy is called.
The mptspi slave_destroy method looks like it works.  What's the problem?

^ permalink raw reply	[flat|nested] 5+ messages in thread

* RE: Hotplug does not work with Fusion MPT
@ 2006-05-15 16:04 Moore, Eric
  0 siblings, 0 replies; 5+ messages in thread
From: Moore, Eric @ 2006-05-15 16:04 UTC (permalink / raw)
  To: Matthew Wilcox; +Cc: Maik Hampel, linux-scsi

On Monday, May 15, 2006 9:52 AM,  Matthew Wilcox wrote:
> 
> On Mon, May 15, 2006 at 09:04:52AM -0600, Moore, Eric wrote:
> > On Monday, May 15, 2006 7:44 AM, Matthew Wilcox wrote:
> > > Did you tell Linux you removed it?  Try:
> > > 
> > > echo scsi remove-single-device 0 0 4 0 > /proc/scsi/scsi
> > > 
> > > Parallel SCSI doesn't have a way to notify the OS that a 
> > > device is gone,
> > > so the user has to say.
> > 
> > We have Hotplug device support in fustion drivers for SAS 
> and Fibre, not
> > for SPI.
> 
> Erm.  Are you saying it doesn't work?  Or you just don't test it?
> If the user runs the above command, the host's 
> ->slave_destroy is called.
> The mptspi slave_destroy method looks like it works.  What's 
> the problem?
>

Yes, what you suggested does work.  I wasn't disagreeing with that.

What I was saying as with SAS and FC, our firmware sends events to the
driver which tells when devices are pulled or pushed. When that event
occurs,
we call functions in the transport layers that handle hotpluging the
device
automatically, that doesn't require an enduser to echo something into
the /proc or /sysfs.

As for SPI, there are not events that notifiy the driver when a device
is pulled or pushed.  Therefore you will need to do what you suggested.

Eric 

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2006-05-15 16:04 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-05-15 12:08 Hotplug does not work with Fusion MPT Maik Hampel
2006-05-15 13:44 ` Matthew Wilcox
  -- strict thread matches above, loose matches on Subject: below --
2006-05-15 15:04 Moore, Eric
2006-05-15 15:51 ` Matthew Wilcox
2006-05-15 16:04 Moore, Eric

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).