[PATCH] mark qlogicfc broken

[PATCH] mark qlogicfc broken

[Christoph Hellwig]

The hardware is supported by qla2xxx which is much better (and
maintained) driver.  Keeping the driver around confuses people to
actually use it with severe results, e.g.

	http://marc.theaimsgroup.com/?l=linux-kernel&m=110656624623547&w=2


--- 1.95/drivers/scsi/Kconfig	2005-01-28 18:14:29 +01:00
+++ edited/drivers/scsi/Kconfig	2005-02-10 14:59:56 +01:00
@@ -1270,7 +1248,7 @@
 
 config SCSI_QLOGIC_FC
 	tristate "Qlogic ISP FC SCSI support"
-	depends on PCI && SCSI
+	depends on PCI && SCSI && BROKEN
 	help
 	  This is a driver for the QLogic ISP2100 SCSI-FCP host adapter.
 

Re: [PATCH] mark qlogicfc broken

[Ricky Beam]

On Thu, 10 Feb 2005, Christoph Hellwig wrote:
>The hardware is supported by qla2xxx which is much better (and
>maintained) driver.  Keeping the driver around confuses people to
>actually use it with severe results, e.g.

That's a nice theory, but not so true in practice.  qla2xxx will not
correctly drive a 2100.  I've complained about this for several years
now and it still doesn't work.  QLogic doesn't test the 2100 anymore.
(Maybe not the 2200's either.)

It's a command queuing problem.  qlogicfc used to have the same problem,
but over the last decade I've managed to stop it from killing itself.
Command queuing to that card cannot be explained as anything but a
cluster f***.  The queue size is dynamic -- dependent on the exact
commands queued.

qlogicfc was removed from the tree once before only to be put back
about 30sec later because people still use it.  It's certainly not
BROKEN.  Not recommended, but not broken either.  BROKEN is a tag for
things that are _broken_ -- they don't compile and don't work when they
do.  qlogicfc both compiles and works, so BROKEN is the wrong thing to
use.

--Ricky

Re: [PATCH] mark qlogicfc broken

[Christoph Hellwig]

On Fri, Feb 11, 2005 at 09:06:14AM -0500, Ricky Beam wrote:
> On Thu, 10 Feb 2005, Christoph Hellwig wrote:
> >The hardware is supported by qla2xxx which is much better (and
> >maintained) driver.  Keeping the driver around confuses people to
> >actually use it with severe results, e.g.
> 
> That's a nice theory, but not so true in practice.  qla2xxx will not
> correctly drive a 2100.  I've complained about this for several years
> now and it still doesn't work.  QLogic doesn't test the 2100 anymore.
> (Maybe not the 2200's either.)

Arjan tested it on the 2100 a while ago and it worked just fine.  Have
you tested it recently?

> qlogicfc was removed from the tree once before only to be put back
> about 30sec later because people still use it.  It's certainly not
> BROKEN.  Not recommended, but not broken either.  BROKEN is a tag for
> things that are _broken_ -- they don't compile and don't work when they
> do.  qlogicfc both compiles and works, so BROKEN is the wrong thing to
> use.

It _is_ broken.  See the url in my initial post.

Re: [PATCH] mark qlogicfc broken

[Ricky Beam]

On Fri, 11 Feb 2005, Christoph Hellwig wrote:
>Arjan tested it on the 2100 a while ago and it worked just fine.  Have
>you tested it recently?

Yes.  If you sling enough traffic at it, it hangs, times out, and eventually
restarts the card goes on about it's business for a few more seconds before
hanging again.

>It _is_ broken.  See the url in my initial post.

No, it is not.  I'm using it right now -- and have been for ~5 years.
Tridge's message doesn't say anything -- 4000 4 can happen with either
driver.

--Ricky

RE: [PATCH] mark qlogicfc broken

[Duane Grigsby]

If you could run the qla2xxx driver with debug enabled and send me the message file after the failure then maybe I can determine whats wrong with it and fix for good.
-- Duane

> -----Original Message-----
> From: linux-scsi-owner@vger.kernel.org
> [mailto:linux-scsi-owner@vger.kernel.org]On Behalf Of Ricky Beam
> Sent: Friday, February 11, 2005 6:30 AM
> To: Christoph Hellwig
> Cc: Linux SCSI List
> Subject: Re: [PATCH] mark qlogicfc broken
> 
> 
> On Fri, 11 Feb 2005, Christoph Hellwig wrote:
> >Arjan tested it on the 2100 a while ago and it worked just 
> fine.  Have
> >you tested it recently?
> 
> Yes.  If you sling enough traffic at it, it hangs, times out, 
> and eventually
> restarts the card goes on about it's business for a few more 
> seconds before
> hanging again.
> 
> >It _is_ broken.  See the url in my initial post.
> 
> No, it is not.  I'm using it right now -- and have been for ~5 years.
> Tridge's message doesn't say anything -- 4000 4 can happen with either
> driver.
> 
> --Ricky
> 
> 
> -
> 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
>