aic79xx

aic79xx

[Stephan Rattai]

Hi list,

I would like to bother you with a problem that might be a little discrepancy 
in the aic79xx driver. I'm not quite sure if this is a aic79xx problem at 
all. I have an Adaptec 29320 controller with a LTO (U320) drive connected. I 
needed to turn down the speed to 160 which I did via the adapter bios. After 
having booted I found the following output in /proc/scsi/aic79xx/1:

Target 1 Negotiation Settings
        User: 160.000MB/s transfers (80.000MHz DT, 16bit)
        Goal: 320.000MB/s transfers (160.000MHz DT|IU|RTI, 16bit)
        Curr: 320.000MB/s transfers (160.000MHz DT|IU|RTI, 16bit)
     
So the speed for the device is set to 320 despite my bios setting (at least 
that's what I'm being told through the proc interface). 

With kernel 2.6.13 I get:

Target 1 Negotiation Settings
        User: 160.000MB/s transfers (80.000MHz RDSTRM|DT, 16bit)
        Goal: 160.000MB/s transfers (80.000MHz DT, 16bit)
        Curr: 160.000MB/s transfers (80.000MHz DT, 16bit)

Is this a bug or a feature ? 

Regards
Stephan

Re: aic79xx

[James Bottomley]

On Wed, 2006-02-22 at 16:18 +0100, Stephan Rattai wrote:
> Hi list,
> 
> I would like to bother you with a problem that might be a little discrepancy 
> in the aic79xx driver. I'm not quite sure if this is a aic79xx problem at 
> all. I have an Adaptec 29320 controller with a LTO (U320) drive connected. I 
> needed to turn down the speed to 160 which I did via the adapter bios. After 
> having booted I found the following output in /proc/scsi/aic79xx/1:
> 
> Target 1 Negotiation Settings
>         User: 160.000MB/s transfers (80.000MHz DT, 16bit)
>         Goal: 320.000MB/s transfers (160.000MHz DT|IU|RTI, 16bit)
>         Curr: 320.000MB/s transfers (160.000MHz DT|IU|RTI, 16bit)
>      
> So the speed for the device is set to 320 despite my bios setting (at least 
> that's what I'm being told through the proc interface). 
> 
> With kernel 2.6.13 I get:
> 
> Target 1 Negotiation Settings
>         User: 160.000MB/s transfers (80.000MHz RDSTRM|DT, 16bit)
>         Goal: 160.000MB/s transfers (80.000MHz DT, 16bit)
>         Curr: 160.000MB/s transfers (80.000MHz DT, 16bit)
> 
> Is this a bug or a feature ? 

It's definitely a bug.  What's supposed to happen in 2.6.16-rc3 is that
aic79xx picks the parameters out of the bios and places them into the
min_period/max_offset variables.

could you look at

cat /sys/class/spi_transport/target<whatever>/min_period

and tell me what it says.

Thanks,

James
 

Re: aic79xx

[Stephan Rattai]

Am Mittwoch, 22. Februar 2006 18:14 schrieben Sie:
> On Wed, 2006-02-22 at 16:18 +0100, Stephan Rattai wrote:
> > Hi list,
> >
> > I would like to bother you with a problem that might be a little
> > discrepancy in the aic79xx driver. I'm not quite sure if this is a
> > aic79xx problem at all. I have an Adaptec 29320 controller with a LTO
> > (U320) drive connected. I needed to turn down the speed to 160 which I
> > did via the adapter bios. After having booted I found the following
> > output in /proc/scsi/aic79xx/1:
> >
> > Target 1 Negotiation Settings
> >         User: 160.000MB/s transfers (80.000MHz DT, 16bit)
> >         Goal: 320.000MB/s transfers (160.000MHz DT|IU|RTI, 16bit)
> >         Curr: 320.000MB/s transfers (160.000MHz DT|IU|RTI, 16bit)
> >
> > So the speed for the device is set to 320 despite my bios setting (at
> > least that's what I'm being told through the proc interface).
> >
> > With kernel 2.6.13 I get:
> >
> > Target 1 Negotiation Settings
> >         User: 160.000MB/s transfers (80.000MHz RDSTRM|DT, 16bit)
> >         Goal: 160.000MB/s transfers (80.000MHz DT, 16bit)
> >         Curr: 160.000MB/s transfers (80.000MHz DT, 16bit)
> >
> > Is this a bug or a feature ?
>
> It's definitely a bug.  What's supposed to happen in 2.6.16-rc3 is that
> aic79xx picks the parameters out of the bios and places them into the
> min_period/max_offset variables.
>
> could you look at
>
> cat /sys/class/spi_transport/target<whatever>/min_period
>
> and tell me what it says.
>
> Thanks,
>
> James

Ok, 

cat /sys/class/spi_transport/target1\:0\:1/min_period
6.25

2.6.16-rc3 ? Whatever causes this behaviour must have been introduced in 
2.6.14. The cat above is taken from 2.6.15.4. The output from 2.6.14 is the 
same (as is the output from /proc/scsi/aic79xx/1), 2.6.16-rc4 dito. 2.6.13 
was working as expected.

Thanks
Stephan

Re: aic79xx

[James Bottomley]

On Wed, 2006-02-22 at 18:39 +0100, Stephan Rattai wrote:
> cat /sys/class/spi_transport/target1\:0\:1/min_period
> 6.25
> 
> 2.6.16-rc3 ? Whatever causes this behaviour must have been introduced in 
> 2.6.14. The cat above is taken from 2.6.15.4. The output from 2.6.14 is the 
> same (as is the output from /proc/scsi/aic79xx/1), 2.6.16-rc4 dito. 2.6.13 
> was working as expected.

OK, I'm afraid this is a bit more serious than I thought.  Apparently
the aic7xxx code that pulls the settings out of the seeprom didn't get
ported over to the aic79xx driver.

Temporarily, to achive what you want, you can just do

echo 12.5 > /sys/class/spi_transport/target1\:0\:1/min_period
echo 1 > /sys/class/spi_transport/target1\:0\:1/revalidate

and that should bring the speed down to U160.

I'll see what it will take to get the seeprom functions ported, but they
probably won't make 2.6.16.

James

Re: aic79xx

[Stephan Rattai]

> OK, I'm afraid this is a bit more serious than I thought.  Apparently
> the aic7xxx code that pulls the settings out of the seeprom didn't get
> ported over to the aic79xx driver.
>
> Temporarily, to achive what you want, you can just do
>
> echo 12.5 > /sys/class/spi_transport/target1\:0\:1/min_period
> echo 1 > /sys/class/spi_transport/target1\:0\:1/revalidate
>
> and that should bring the speed down to U160.
>
> I'll see what it will take to get the seeprom functions ported, but they
> probably won't make 2.6.16.
>
> James

Well, if I try this, the 
echo 1 > /sys/class/spi_transport/target1\:0\:1/revalidate
never returns. 

cat /proc/scsi/aic79xx/1 says:
Target 1 Negotiation Settings
        User: 160.000MB/s transfers (80.000MHz RDSTRM|DT|IU|RTI|QAS, 16bit)
        Goal: 3.300MB/s transfers
        Curr: 3.300MB/s transfers

Thinks like mt -f /dev/nst0 never return. 

I think I'll just use 2.6.13, the target is working fine there.

Thanks,
Stephan

Re: aic79xx

[Stephen Degler]

Hello,

I've observed this problem as well.  I'd be happy to test a patch, or 
derive one if the patch isn't too involved.

skd

James Bottomley wrote:

>On Wed, 2006-02-22 at 18:39 +0100, Stephan Rattai wrote:
>  
>
>>cat /sys/class/spi_transport/target1\:0\:1/min_period
>>6.25
>>
>>2.6.16-rc3 ? Whatever causes this behaviour must have been introduced in 
>>2.6.14. The cat above is taken from 2.6.15.4. The output from 2.6.14 is the 
>>same (as is the output from /proc/scsi/aic79xx/1), 2.6.16-rc4 dito. 2.6.13 
>>was working as expected.
>>    
>>
>
>OK, I'm afraid this is a bit more serious than I thought.  Apparently
>the aic7xxx code that pulls the settings out of the seeprom didn't get
>ported over to the aic79xx driver.
>
>Temporarily, to achive what you want, you can just do
>
>echo 12.5 > /sys/class/spi_transport/target1\:0\:1/min_period
>echo 1 > /sys/class/spi_transport/target1\:0\:1/revalidate
>
>and that should bring the speed down to U160.
>
>I'll see what it will take to get the seeprom functions ported, but they
>probably won't make 2.6.16.
>
>James
>
>
>-
>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
>
>  
>