headers
Trevor Batley | 2 Dec 2008 10:54
Picon
Gravatar

Re: Unable to see HDD in /dev

On Mon, 2008-12-01 at 07:22 -0800, Dan Burns wrote:
> I've checked on Knoppix and it is trying to use the ahci driver, looking in
> Thinstation I've got this enabled so I'm guessing I need to get a more up to
> date version of the driver ?
> 
Probably. It may also be a very old version of udev (but more likely the
driver).

I was trying to get out an updated kernel for 2.2, but I ran into a few
difficulties (if I add in supermount I seem to get a corrupted kernel,
so am assuming that it is a bad patch...).

If I manage to get it out, it may help.

I haven't answered much on this as I'm still a little confused.
Originally I thought that our debugging showed that the kernel didn't
kick of any hotplug events for these card readers, hence I stopped
looking at udev (if it doesn't get called it's unlikely to be able to
respond correctly). I guess that diagnosis was incorrect.

We'll see what we get with a newer kernel (and possibly updated udev).
Unfortunately the only multi-card reader I have doesn't seem to do
anything....

Trevor B

> Regards
> 
> Dan
>
(Continue reading)

Permalink | Reply |
headers
Dan Burns | 2 Dec 2008 13:25
Picon
Favicon

Re: Unable to see HDD in /dev


Hi Trevor,

Just seen this reply (I think it's appeared under a different post !).  I
probably wasn't clear when I tried to explain my original diagnosis so
apologies for that.  UDev does detect the actual card reader being plugged
in (and creates each relevant block device in /dev - i.e. a four slot reader
creates sdb, sdc, sdd and sde).  The problem is UDev failed to then detect a
card being plugged into any of the slots and there by creating the relevant
device under /dev and run the usb.sh script.

As previously mentioned I found that I could force UDev to rescan the USB
bus that the card reader was on by attempting to mount each block device
originally created for the card reader, for example mount /dev/sdb /mnt/tmp. 
Although this will obviously fail it triggered a scan of UDev which then
detected the newly inserted card and there by created the device etc.  After
a lot of work (and complaints from users :o) ) I created the script I placed
in the other post which seems to work quite nicely, I created UDev rules so
that I knew where each card type would mount, this way I could script the
mapping of client drives for our Citrix users.

I think you're absoloutely right that it's something the newer kernel will
probably tackle a lot better (since my Ubuntu workstation works quite
happily with cards being hot plugged into the card reader), in the meantime
the script is doing the job.  I really appreciate the time you've spent
looking into this though, I thought I'd share my script etc in case anyone
else needed a workaround (or could suggest how to do it even smarter !!).

Thanks again !
(Continue reading)

Permalink | Reply |
headers
Trevor Batley | 5 Dec 2008 06:42
Picon
Gravatar

Re: Unable to see HDD in /dev

Quoting "Dan Burns" <daninleics@...>:

>
> Hi Trevor,
>
> Just seen this reply (I think it's appeared under a different post !).  I
> probably wasn't clear when I tried to explain my original diagnosis so
> apologies for that.  UDev does detect the actual card reader being plugged
> in (and creates each relevant block device in /dev - i.e. a four slot reader
> creates sdb, sdc, sdd and sde).  The problem is UDev failed to then detect a
> card being plugged into any of the slots and there by creating the relevant
> device under /dev and run the usb.sh script.

Just remembered.

You DID apply the udev patch for Thinstation 2.2 from sourceforge.net?

Trevor B
>
> As previously mentioned I found that I could force UDev to rescan the USB
> bus that the card reader was on by attempting to mount each block device
> originally created for the card reader, for example mount /dev/sdb /mnt/tmp.
> Although this will obviously fail it triggered a scan of UDev which then
> detected the newly inserted card and there by created the device etc.  After
> a lot of work (and complaints from users :o) ) I created the script I placed
> in the other post which seems to work quite nicely, I created UDev rules so
> that I knew where each card type would mount, this way I could script the
> mapping of client drives for our Citrix users.
>
> I think you're absoloutely right that it's something the newer kernel will
(Continue reading)

Permalink | Reply |
headers
Trevor Batley | 5 Dec 2008 06:51
Picon
Gravatar

Re: Unable to see HDD in /dev

Please disregard, as I fired off before remembering that you are  
running 2.2.2 which already has this patch.

Trevor B

Quoting "Trevor Batley" <trevorbatley@...>:

> Quoting "Dan Burns" <daninleics@...>:
>
>>
>> Hi Trevor,
>>
>> Just seen this reply (I think it's appeared under a different post !).  I
>> probably wasn't clear when I tried to explain my original diagnosis so
>> apologies for that.  UDev does detect the actual card reader being plugged
>> in (and creates each relevant block device in /dev - i.e. a four slot reader
>> creates sdb, sdc, sdd and sde).  The problem is UDev failed to then detect a
>> card being plugged into any of the slots and there by creating the relevant
>> device under /dev and run the usb.sh script.
>
> Just remembered.
>
> You DID apply the udev patch for Thinstation 2.2 from sourceforge.net?
>
> Trevor B
>>
>> As previously mentioned I found that I could force UDev to rescan the USB
>> bus that the card reader was on by attempting to mount each block device
>> originally created for the card reader, for example mount /dev/sdb /mnt/tmp.
>> Although this will obviously fail it triggered a scan of UDev which then
(Continue reading)

Permalink | Reply |
headers
Trevor Batley | 2 Dec 2008 22:45
Picon
Gravatar

Re: Unable to see HDD in /dev

Quoting "Dan Burns" <daninleics@...>:

>
> Hi Trevor,
>
> Just seen this reply (I think it's appeared under a different post !).  I
> probably wasn't clear when I tried to explain my original diagnosis so
> apologies for that.  UDev does detect the actual card reader being plugged
> in (and creates each relevant block device in /dev - i.e. a four slot reader
> creates sdb, sdc, sdd and sde).  The problem is UDev failed to then detect a
> card being plugged into any of the slots and there by creating the relevant
> device under /dev and run the usb.sh script.

OK, I had probably forgotten that part.. :-) It sounds familiar, I had  
just remembered the no trigger, but it applies to the card, not the  
device. In my case the device doesn't show either.
>
> As previously mentioned I found that I could force UDev to rescan the USB
> bus that the card reader was on by attempting to mount each block device
> originally created for the card reader, for example mount /dev/sdb /mnt/tmp.
> Although this will obviously fail it triggered a scan of UDev which then
> detected the newly inserted card and there by created the device etc.  After
> a lot of work (and complaints from users :o) ) I created the script I placed
> in the other post which seems to work quite nicely, I created UDev rules so
> that I knew where each card type would mount, this way I could script the
> mapping of client drives for our Citrix users.
>
> I think you're absoloutely right that it's something the newer kernel will
> probably tackle a lot better (since my Ubuntu workstation works quite
> happily with cards being hot plugged into the card reader), in the meantime
(Continue reading)

Permalink | Reply |
headers
Trevor Batley | 2 Dec 2008 11:06
Picon
Gravatar

Re: Unable to see HDD in /dev

If you are confused by my answer, so am I....

It was really against the USB issue, not the NVRAM one.

In the case of NVRAM, I think the achi driver is a helper driver and
there was probably some other driver that specifically supports that
device.

Trevor B

On Tue, 2008-12-02 at 20:54 +1100, Trevor Batley wrote:
> On Mon, 2008-12-01 at 07:22 -0800, Dan Burns wrote:
> > I've checked on Knoppix and it is trying to use the ahci driver, looking in
> > Thinstation I've got this enabled so I'm guessing I need to get a more up to
> > date version of the driver ?
> > 
> Probably. It may also be a very old version of udev (but more likely the
> driver).
> 
> I was trying to get out an updated kernel for 2.2, but I ran into a few
> difficulties (if I add in supermount I seem to get a corrupted kernel,
> so am assuming that it is a bad patch...).
> 
> If I manage to get it out, it may help.
> 
> I haven't answered much on this as I'm still a little confused.
> Originally I thought that our debugging showed that the kernel didn't
> kick of any hotplug events for these card readers, hence I stopped
> looking at udev (if it doesn't get called it's unlikely to be able to
> respond correctly). I guess that diagnosis was incorrect.
(Continue reading)

Permalink | Reply |

Gmane