scsi bus-reset in qlogic causes system stall

scsi bus-reset in qlogic causes system stall

[devnetfs]

Hello,

I am running RedHat 2.4.9-13 with qlogic qla2x00 driver (ver 4.31.7b).
Trying to do a bus-reset (in an multi-initiator, shared-scsi scenario)
stalls one (or more) nodes in the system for a while. dmesg shows the
message: "Waiting for LIP to complete...", on all nodes. Everything
freezes for a while (around 10-40secs). It happens on UP/SMP both.

Is this normal? or a bug in qlogic driver? 

Thanks in advance for any insight,
A.


__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com

Re: scsi bus-reset in qlogic causes system stall

[Steven Dake]

bug in qlogic driver I think this has been fixed to a schedule_timeout 
in the 6.X drivers (from a spin on i/o)

I have done this same thing with the 6.x driver and the system is still 
usable.  Keep in mind that the I/O's will retry until the FC LIP has 
completed.

Thanks
-steve

devnetfs wrote:

>Hello,
>
>I am running RedHat 2.4.9-13 with qlogic qla2x00 driver (ver 4.31.7b).
>Trying to do a bus-reset (in an multi-initiator, shared-scsi scenario)
>stalls one (or more) nodes in the system for a while. dmesg shows the
>message: "Waiting for LIP to complete...", on all nodes. Everything
>freezes for a while (around 10-40secs). It happens on UP/SMP both.
>
>Is this normal? or a bug in qlogic driver? 
>
>Thanks in advance for any insight,
>A.
>
>
>__________________________________________________
>Do you Yahoo!?
>Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
>http://mailplus.yahoo.com
>-
>To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>
>
>  
>

RE: scsi bus-reset in qlogic causes system stall

[Ravi Anand]

Please use the following link to download the latest driver-v6.04.00b8

http://www.qlogic.com/support/os_detail.asp?productid=125&osid=26

There has been some other fixes too in the bus reset code when issued through the backdoor.

Regards
Ravi

-----Original Message-----
From: Steven Dake [mailto:sdake@mvista.com]
Sent: Friday, January 17, 2003 10:33 AM
To: devnetfs
Cc: linux-scsi@vger.kernel.org
Subject: Re: scsi bus-reset in qlogic causes system stall


bug in qlogic driver I think this has been fixed to a schedule_timeout 
in the 6.X drivers (from a spin on i/o)

I have done this same thing with the 6.x driver and the system is still 
usable.  Keep in mind that the I/O's will retry until the FC LIP has 
completed.

Thanks
-steve

devnetfs wrote:

>Hello,
>
>I am running RedHat 2.4.9-13 with qlogic qla2x00 driver (ver 4.31.7b).
>Trying to do a bus-reset (in an multi-initiator, shared-scsi scenario)
>stalls one (or more) nodes in the system for a while. dmesg shows the
>message: "Waiting for LIP to complete...", on all nodes. Everything
>freezes for a while (around 10-40secs). It happens on UP/SMP both.
>
>Is this normal? or a bug in qlogic driver? 
>
>Thanks in advance for any insight,
>A.
>
>
>__________________________________________________
>Do you Yahoo!?
>Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
>http://mailplus.yahoo.com
>-
>To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>
>
>  
>

-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

RE: scsi bus-reset in qlogic causes system stall

[devnetfs]

Thanks Ravi and Steven for the information.
I will try the new driver you pointed out and update back.

I had couple of questions though:

1. What do you mean by "issued thru the backdoor"? I am using the
standard(?) SG_SCSI_RESET ioctl. Is there any other way of doing it?

2. Can this "Waiting for FC LIP..." be disabled (still bus resets 
work)? [sorry I am a newbie in the fiberchannel/scsi world]. The reason
I am asking this is, while doing a BUS RESET from a node all other
nodes also *wait* for FC LIP to complete -- and that makes all
the nodes on the shared scsi vulnerable to this hang/stall thing.

The way I've connected shared scsi is: Multiple nodes (each having one
qlogic 2200 FC HBA) connect to a SAN switch(brocade) to which a JBOD is
also connected.

Thanks once again,
A.

--- Ravi Anand <ravi.anand@qlogic.com> wrote:
> Please use the following link to download the latest
> driver-v6.04.00b8
> 
> http://www.qlogic.com/support/os_detail.asp?productid=125&osid=26
> 
> There has been some other fixes too in the bus reset code when issued
> through the backdoor.
> 
> Regards
> Ravi
> 
> -----Original Message-----
> From: Steven Dake [mailto:sdake@mvista.com]
> Sent: Friday, January 17, 2003 10:33 AM
> To: devnetfs
> Cc: linux-scsi@vger.kernel.org
> Subject: Re: scsi bus-reset in qlogic causes system stall
> 
> 
> bug in qlogic driver I think this has been fixed to a
> schedule_timeout 
> in the 6.X drivers (from a spin on i/o)
> 
> I have done this same thing with the 6.x driver and the system is
> still 
> usable.  Keep in mind that the I/O's will retry until the FC LIP has 
> completed.
> 
> Thanks
> -steve
> 
> devnetfs wrote:
> 
> >Hello,
> >
> >I am running RedHat 2.4.9-13 with qlogic qla2x00 driver (ver
> 4.31.7b).
> >Trying to do a bus-reset (in an multi-initiator, shared-scsi
> scenario)
> >stalls one (or more) nodes in the system for a while. dmesg shows
> the
> >message: "Waiting for LIP to complete...", on all nodes. Everything
> >freezes for a while (around 10-40secs). It happens on UP/SMP both.
> >
> >Is this normal? or a bug in qlogic driver? 
> >
> >Thanks in advance for any insight,
> >A.
> >
> >
> >__________________________________________________
> >Do you Yahoo!?
> >Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
> >http://mailplus.yahoo.com
> >-
> >To unsubscribe from this list: send the line "unsubscribe
> linux-scsi" in
> >the body of a message to majordomo@vger.kernel.org
> >More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >
> >
> >
> >  
> >
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-scsi"
> in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html


__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com

RE: scsi bus-reset in qlogic causes system stall

[Ravi Anand]

>>>1. What do you mean by "issued thru the backdoor"? I am using the
standard(?) SG_SCSI_RESET ioctl. Is there any other way of doing it?

By backdoor I meant using the sg interface.In general bus_reset hook of the lld is called during error recovery scenario if needed.

>>>Can this "Waiting for FC LIP..." be disabled (still bus resets 
work)? [sorry I am a newbie in the fiberchannel/scsi world]. The reason
I am asking this is, while doing a BUS RESET from a node all other
nodes also *wait* for FC LIP to complete -- and that makes all
the nodes on the shared scsi vulnerable to this hang/stall thing.

why do u want to reset the bus?whats the purpose of doing it?.

ravi
--- Ravi Anand <ravi.anand@qlogic.com> wrote:
> Please use the following link to download the latest
> driver-v6.04.00b8
> 
> http://www.qlogic.com/support/os_detail.asp?productid=125&osid=26
> 
> There has been some other fixes too in the bus reset code when issued
> through the backdoor.
> 
> Regards
> Ravi
> 
> -----Original Message-----
> From: Steven Dake [mailto:sdake@mvista.com]
> Sent: Friday, January 17, 2003 10:33 AM
> To: devnetfs
> Cc: linux-scsi@vger.kernel.org
> Subject: Re: scsi bus-reset in qlogic causes system stall
> 
> 
> bug in qlogic driver I think this has been fixed to a
> schedule_timeout 
> in the 6.X drivers (from a spin on i/o)
> 
> I have done this same thing with the 6.x driver and the system is
> still 
> usable.  Keep in mind that the I/O's will retry until the FC LIP has 
> completed.
> 
> Thanks
> -steve
> 
> devnetfs wrote:
> 
> >Hello,
> >
> >I am running RedHat 2.4.9-13 with qlogic qla2x00 driver (ver
> 4.31.7b).
> >Trying to do a bus-reset (in an multi-initiator, shared-scsi
> scenario)
> >stalls one (or more) nodes in the system for a while. dmesg shows
> the
> >message: "Waiting for LIP to complete...", on all nodes. Everything
> >freezes for a while (around 10-40secs). It happens on UP/SMP both.
> >
> >Is this normal? or a bug in qlogic driver? 
> >
> >Thanks in advance for any insight,
> >A.
> >
> >
> >__________________________________________________
> >Do you Yahoo!?
> >Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
> >http://mailplus.yahoo.com
> >-
> >To unsubscribe from this list: send the line "unsubscribe
> linux-scsi" in
> >the body of a message to majordomo@vger.kernel.org
> >More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >
> >
> >
> >  
> >
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-scsi"
> in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html


__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com

Re: scsi bus-reset in qlogic causes system stall

[Christoph Hellwig]

On Fri, Jan 17, 2003 at 02:23:07PM -0800, Ravi Anand wrote:
> Please use the following link to download the latest driver-v6.04.00b8
> 
> http://www.qlogic.com/support/os_detail.asp?productid=125&osid=26

Could you explain all the junk this release added under
#ifdef __VMWARE__?  Especially a pointer to the ill kernel this is supposed to
compile against would be quite helpfull.

RE: scsi bus-reset in qlogic causes system stall

[Ravi Anand]

Added to take care some of the changes made by the VMWARE guys in the kernel.Included at the request from our OEM's.This will compile against the modified kernel from VMWARE.

Hope this answer your question.

Ravi

-----Original Message-----
From: Christoph Hellwig [mailto:hch@infradead.org]
Sent: Monday, January 20, 2003 2:41 AM
To: Ravi Anand
Cc: Steven Dake; devnetfs; linux-scsi@vger.kernel.org
Subject: Re: scsi bus-reset in qlogic causes system stall


On Fri, Jan 17, 2003 at 02:23:07PM -0800, Ravi Anand wrote:
> Please use the following link to download the latest driver-v6.04.00b8
> 
> http://www.qlogic.com/support/os_detail.asp?productid=125&osid=26

Could you explain all the junk this release added under
#ifdef __VMWARE__?  Especially a pointer to the ill kernel this is supposed to
compile against would be quite helpfull.

Re: scsi bus-reset in qlogic causes system stall

[Christoph Hellwig]

On Mon, Jan 20, 2003 at 04:06:53PM -0800, Ravi Anand wrote:
> Added to take care some of the changes made by the VMWARE guys in the kernel.Included at the request from our OEM's.This will compile against the modified kernel from VMWARE.
> 
> Hope this answer your question.

OK.  Given the recent discussion about the qla2xxx driver in 2.5 I'd just
like to add that you shouldn't expect such ugly hacks for a non-public
kernel tree to ever get merged..

RE: scsi bus-reset in qlogic causes system stall

[Ravi Anand]

Our development is going on which will include the following changes:

		-v2.5.x changes
		-no hacks or changes for non public kernel tree
		-lot of other clean ups,
		-re-orginzation of sources 
		- suggestions and bug fixes from the linux scsi community etc...

before the driver is officially submitted to be  merged in the v2.5.x tree.

regards
ravi


-----Original Message-----
From: Christoph Hellwig [mailto:hch@infradead.org]
Sent: Monday, January 20, 2003 11:53 PM
To: Ravi Anand
Cc: Christoph Hellwig; Steven Dake; devnetfs; linux-scsi@vger.kernel.org
Subject: Re: scsi bus-reset in qlogic causes system stall


On Mon, Jan 20, 2003 at 04:06:53PM -0800, Ravi Anand wrote:
> Added to take care some of the changes made by the VMWARE guys in the kernel.Included at the request from our OEM's.This will compile against the modified kernel from VMWARE.
> 
> Hope this answer your question.

OK.  Given the recent discussion about the qla2xxx driver in 2.5 I'd just
like to add that you shouldn't expect such ugly hacks for a non-public
kernel tree to ever get merged..

Re: scsi bus-reset in qlogic causes system stall

[Christoph Hellwig]

On Wed, Jan 22, 2003 at 11:29:27AM -0800, Ravi Anand wrote:
> Our development is going on which will include the following changes:
> 
> 		-v2.5.x changes
> 		-no hacks or changes for non public kernel tree
> 		-lot of other clean ups,
> 		-re-orginzation of sources 
> 		- suggestions and bug fixes from the linux scsi community etc...
> 
> before the driver is officially submitted to be  merged in the v2.5.x tree.

Nice to hear that.  Feel free to announce important milestones here
on linux-scsi if you're interested in some feedback.