SEO

#1 (**permalink**) 01-18-2008, 06:48 AM

Hi group,

I am not sure if this is a bug in sd.c or setup problem or even
something else but on my system (2.6.8 - Fedora based distro) with
AHA-2940 SCSI controller and a removable syquest drive (SQ3270S) I
encounter the problem as per subject line. On boot-up, whenever the
relevant modules are loaded, it tries to spin-up the SyQuest drive and
this fails unless there is medium inserted in the drive. The problem is
that it tries to do it at all but even worse that it doesn't give up!
The solution is to insert the medium and allow it to spin-up. The
booting can then continue but whenever I remove the medium - the eternal
"spinning up drive" message and the 1s period of activity LED blinking
come back :-( In the /proc entries the drive is properly recognised as
"removable".

Any clues on how to get rid of that?

TNX.

#2 (**permalink**) 01-18-2008, 06:48 AM

Silver Dream ! wrote:

> Hi group,
>
> I am not sure if this is a bug in sd.c or setup problem or even
> something else but on my system (2.6.8 - Fedora based distro) with
> AHA-2940 SCSI controller and a removable syquest drive (SQ3270S) I
> encounter the problem as per subject line. On boot-up, whenever the
> relevant modules are loaded, it tries to spin-up the SyQuest drive and
> this fails unless there is medium inserted in the drive. The problem is
> that it tries to do it at all but even worse that it doesn't give up!
> The solution is to insert the medium and allow it to spin-up. The
> booting can then continue but whenever I remove the medium - the eternal
> "spinning up drive" message and the 1s period of activity LED blinking
> come back :-( In the /proc entries the drive is properly recognised as
> "removable".
>

this seems to be bad intercation between hardware and driver. Your drive
does not properly indicates "no media" condition (or even does not declare
itself removable drive at all) and driver attempts forever to start it
(well, that it does not stop is a bug too).

I suggest you post lkml or (better) linux-scsi with description of this
problem.

> Any clues on how to get rid of that?
>

As a quick workaround - add dev_flags parameter with flag BLIST_NOSTARTONADD
(normally 0x1000, check inlcude/scsi/scsi_devinfo.h):

{pts/2}% modinfo scsi-mod
filename: /lib/modules/2.6.10-1mdk/kernel/drivers/scsi/scsi_mod.ko.gz
....
parm: dev_flags:Given scsi_dev_flags=vendor:model:flags[,v:m:f]
add black/white list entries for vendor and model with an integer value of
flags to the scsi device info list

=arvi=

#3 (**permalink**) 01-18-2008, 06:49 AM

kermit wrote:

>>I am not sure if this is a bug in sd.c or setup problem or even
>>something else but on my system (2.6.8 - Fedora based distro) with
>>AHA-2940 SCSI controller and a removable syquest drive (SQ3270S) I
>>encounter the problem as per subject line. On boot-up, whenever the
>>relevant modules are loaded, it tries to spin-up the SyQuest drive and
>>this fails unless there is medium inserted in the drive. The problem is
>>that it tries to do it at all but even worse that it doesn't give up!
>>The solution is to insert the medium and allow it to spin-up. The
>>booting can then continue but whenever I remove the medium - the eternal
>>"spinning up drive" message and the 1s period of activity LED blinking
>>come back :-( In the /proc entries the drive is properly recognised as
>>"removable".
>>
>
>
> this seems to be bad intercation between hardware and driver. Your drive
> does not properly indicates "no media" condition

I guess it does as I have never had any problem with it for years, but
of course am not 100% sure here.

> (or even does not declare
> itself removable drive at all)

This I have checked in /proc, as I have written above - it is recognised
as "removable" drive.

> and driver attempts forever to start it
> (well, that it does not stop is a bug too).

Right. Looks so.

>
> I suggest you post lkml or (better) linux-scsi with description of this
> problem.
>

OK.

>
>>Any clues on how to get rid of that?
>>
>
>
>
> As a quick workaround - add dev_flags parameter with flag BLIST_NOSTARTONADD
> (normally 0x1000, check inlcude/scsi/scsi_devinfo.h):

OK. 0x1000 is there. But could you tell me where should I add this to
make it effective on boot? Maybe an example? Thank you in advance. It
might be a newbie-style question but this is a different distro and
different kernel. I am quite fluent with 2.4 and Debian based distros -
yet this one gives me a lot of PITA ...

#4 (**permalink**) 01-18-2008, 06:49 AM

Silver Dream ! wrote:

> kermit wrote:
>
>>>I am not sure if this is a bug in sd.c or setup problem or even
>>>something else but on my system (2.6.8 - Fedora based distro) with
>>>AHA-2940 SCSI controller and a removable syquest drive (SQ3270S) I
>>>encounter the problem as per subject line. On boot-up, whenever the
>>>relevant modules are loaded, it tries to spin-up the SyQuest drive and
>>>this fails unless there is medium inserted in the drive. The problem is
>>>that it tries to do it at all but even worse that it doesn't give up!
>>>The solution is to insert the medium and allow it to spin-up. The
>>>booting can then continue but whenever I remove the medium - the eternal
>>>"spinning up drive" message and the 1s period of activity LED blinking
>>>come back :-( In the /proc entries the drive is properly recognised as
>>>"removable".
>>>
>>
>>
>> this seems to be bad intercation between hardware and driver. Your drive
>> does not properly indicates "no media" condition
>
> I guess it does as I have never had any problem with it for years, but
> of course am not 100% sure here.
>

OK, it does not return the information that driver interprets as "medium not
present"

>> (or even does not declare
>> itself removable drive at all)
>
> This I have checked in /proc, as I have written above - it is recognised
> as "removable" drive.
>
>> and driver attempts forever to start it
>> (well, that it does not stop is a bug too).
>
> Right. Looks so.
>
>>
>> I suggest you post lkml or (better) linux-scsi with description of this
>> problem.
>>
>
> OK.
>
>>
>>>Any clues on how to get rid of that?
>>>
>>
>>
>>
>> As a quick workaround - add dev_flags parameter with flag
>> BLIST_NOSTARTONADD (normally 0x1000, check inlcude/scsi/scsi_devinfo.h):
>
> OK. 0x1000 is there. But could you tell me where should I add this to
> make it effective on boot?

if you have scsi_mod as module, just add to /etc/modprobe.conf

options scsi_mod dev_flags=VENDOR:MODEL:0x1000

replace VENDOR and MODEL with correct values for your drive. If scsi_mod is
loaded in initrd, you may need to regenerate it.

regards

=arvi=

> Maybe an example? Thank you in advance. It
> might be a newbie-style question but this is a different distro and
> different kernel. I am quite fluent with 2.4 and Debian based distros -
> yet this one gives me a lot of PITA ...

#5 (**permalink**) 01-18-2008, 06:50 AM

kermit wrote:

[...]

>>>As a quick workaround - add dev_flags parameter with flag
>>>BLIST_NOSTARTONADD (normally 0x1000, check inlcude/scsi/scsi_devinfo.h):
>>
>>OK. 0x1000 is there. But could you tell me where should I add this to
>>make it effective on boot?
>
>
> if you have scsi_mod as module, just add to /etc/modprobe.conf
>
> options scsi_mod dev_flags=VENDOR:MODEL:0x1000
>
> replace VENDOR and MODEL with correct values for your drive.

Hm, this might be the problem. My regular Debian based 2.4 setup without
initrd recognises the Vendor: SyQuest, Model: SQ3270S, Rev: 1_27 etc.

The 2.6.9 Fedoraalike gets only the Rev: 1_27 field while having both
VENDOR:MODEL empty! I have a gut feeling that this might be tightly
connected to the problem I experience.

> If scsi_mod is
> loaded in initrd, you may need to regenerate it.

Yes, that might be the case. I immediately found and tried modprobe.conf
but with no apparent results. Then i tried to blame the "hotplug" stuff
but blacklisting the module(s) didn't help either. OK. Going to check
what I can do with the initrd. I may probably need to change the "aic"
driver module too... Thank you for the good pointers.

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

SEO

sd_mod spinning up empty removable drive... forever!