public inbox for linux-mtd@lists.infradead.org
 help / color / mirror / Atom feed
* DOC2000 and Grub
@ 2002-06-11 16:13 Matt Vestal
  2002-06-11 16:27 ` Dzuy Nguyen
  2002-06-11 17:41 ` Ilguiz Latypov
  0 siblings, 2 replies; 11+ messages in thread
From: Matt Vestal @ 2002-06-11 16:13 UTC (permalink / raw)
  To: linux-mtd

I 'm having the same problem as others who have posted, in getting the
"Selected disk does not exist" error.  I'm using a 96MB DOC with a
VersaLogic VSBC-8 motherboard.

-using Grub 0.92
-patched with grub-2002-02-19-doc.patch
-ran "aclocal && automake && autoconf"   (needed automake-1.5-8)
-./configure --enable-diskonchip-2000 --enable-ext2fs
-../util/doc_loadbios /dev/mtd0 grub_firmware
-./nftl_format /dev/mtd0 98304
-fdisk /dev/nftla into multiple portions.  also tried leaving it full
-mke2fs
-copy over necessary files (kernel and such)
-reboot

The DOC loads Grub with no problem.  However, Grub has never been able to
see the DOC filesystem(s).  I've tried root (dc0), root (dc0,0), root
(dc0,1), etc, but nothing works.  I always get the "Selected disk does not
exist, error"

Any help would be greatly appreciated, as I've been pouring over this for
days, now.

Thanks,

Matt

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: DOC2000 and Grub
  2002-06-11 16:13 Matt Vestal
@ 2002-06-11 16:27 ` Dzuy Nguyen
  2002-06-11 17:46   ` Matt Vestal
  2002-06-11 17:41 ` Ilguiz Latypov
  1 sibling, 1 reply; 11+ messages in thread
From: Dzuy Nguyen @ 2002-06-11 16:27 UTC (permalink / raw)
  To: linux-mtd

It's possible that the grub_firmware and the nftl are overlapping.  When 
run doc_loadbios,
observe the number of bytes written into DoC.  It should be the size of 
the grub_firmware.
I'd add a couple of blocks (erase size) to that as the offet in nftl_format.

Matt Vestal wrote:

>I 'm having the same problem as others who have posted, in getting the
>"Selected disk does not exist" error.  I'm using a 96MB DOC with a
>VersaLogic VSBC-8 motherboard.
>
>-using Grub 0.92
>-patched with grub-2002-02-19-doc.patch
>-ran "aclocal && automake && autoconf"   (needed automake-1.5-8)
>-./configure --enable-diskonchip-2000 --enable-ext2fs
>-../util/doc_loadbios /dev/mtd0 grub_firmware
>-./nftl_format /dev/mtd0 98304
>-fdisk /dev/nftla into multiple portions.  also tried leaving it full
>-mke2fs
>-copy over necessary files (kernel and such)
>-reboot
>
>The DOC loads Grub with no problem.  However, Grub has never been able to
>see the DOC filesystem(s).  I've tried root (dc0), root (dc0,0), root
>(dc0,1), etc, but nothing works.  I always get the "Selected disk does not
>exist, error"
>
>Any help would be greatly appreciated, as I've been pouring over this for
>days, now.
>
>Thanks,
>
>Matt
>
>
>
>______________________________________________________
>Linux MTD discussion mailing list
>http://lists.infradead.org/mailman/listinfo/linux-mtd/
>
>

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: DOC2000 and Grub
  2002-06-11 16:13 Matt Vestal
  2002-06-11 16:27 ` Dzuy Nguyen
@ 2002-06-11 17:41 ` Ilguiz Latypov
  2002-06-11 18:04   ` Matt Vestal
  1 sibling, 1 reply; 11+ messages in thread
From: Ilguiz Latypov @ 2002-06-11 17:41 UTC (permalink / raw)
  To: Matt Vestal; +Cc: linux-mtd

On Tue, 11 Jun 2002, Matt Vestal wrote:

> The DOC loads Grub with no problem.  However, Grub has never been able to
> see the DOC filesystem(s).  I've tried root (dc0), root (dc0,0), root
> (dc0,1), etc, but nothing works.  I always get the "Selected disk does not
> exist, error"

Any debug output on the screen when GRUB starts?

Does this diagnostic message show up?

   Spare Media Header for NFTL %d found in flash sector %d

Ilguiz

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: DOC2000 and Grub
  2002-06-11 16:27 ` Dzuy Nguyen
@ 2002-06-11 17:46   ` Matt Vestal
  2002-06-11 17:56     ` Ilguiz Latypov
  0 siblings, 1 reply; 11+ messages in thread
From: Matt Vestal @ 2002-06-11 17:46 UTC (permalink / raw)
  To: Dzuy Nguyen; +Cc: linux-mtd

Thanks for the reply.  I had read that in previous posts, and made sure to
leave enough space.

../util/doc_loadbios /dev/mtd0 grub_firmware 

Performing Flash Erase of length 16384 at offset 0
Performing Flash Erase of length 16384 at offset 16384
Performing Flash Erase of length 16384 at offset 32768
Performing Flash Erase of length 16384 at offset 49152
Performing Flash Erase of length 16384 at offset 65536
Performing Flash Erase of length 16384 at offset 81920
Writing the firmware of length 88544 at 0... Done.

So I did an nftl_format /dev/mtd0 98304.  I've also tried using 114688,
just to make sure.

Also, the partition table gets easily corrupted.  Here's what I got on a
recent reboot.  I had started at cylinder 2 in case it was overlapping
something in the firmware.

[~]$ fdisk /dev/nftla
Device contains neither a valid DOS partition table, nor Sun, SGI or OSF
disklabel
Building a new DOS disklabel. Changes will remain in memory only,
until you decide to write them. After that, of course, the previous
content won't be recoverable.


Command (m for help): p

Disk /dev/nftla: 16 heads, 12 sectors, 1022 cylinders
Units = cylinders of 192 * 512 bytes

     Device Boot    Start       End    Blocks   Id  System
/dev/nftla1      12453227  21715648 889192448   ff  BBT
Partition 1 has different physical/logical beginnings (non-Linux?):
     phys=(1023, 232, 60) logical=(12453226, 9, 11)
Partition 1 has different physical/logical endings:
     phys=(389, 255, 24) logical=(21715647, 15, 2)
Partition 1 does not end on cylinder boundary:
     phys=(389, 255, 24) should be (389, 15, 12)
/dev/nftla3   ?   6313488   4245801 1948985768   48  Unknown
Partition 3 has different physical/logical beginnings (non-Linux?):
     phys=(592, 88, 11) logical=(6313487, 2, 1)
Partition 3 has different physical/logical endings:
     phys=(583, 82, 11) logical=(4245800, 13, 12)
Partition 3 does not end on cylinder boundary:
     phys=(583, 82, 11) should be (583, 15, 12)
/dev/nftla4   ?  11218423  11218573     14375+  83  Linux
Partition 4 has different physical/logical beginnings (non-Linux?):
     phys=(1023, 255, 63) logical=(11218422, 5, 5)
Partition 4 has different physical/logical endings:
     phys=(133, 196, 32) logical=(11218572, 1, 3)
Partition 4 does not end on cylinder boundary:
     phys=(133, 196, 32) should be (133, 15, 12)

Partition table entries are not in disk order

Command (m for help): q

[~]$ e2fsck /dev/nftla1
e2fsck 1.19, 13-Jul-2000 for EXT2 FS 0.5b, 95/08/09
/dev/nftla1: clean, 2372/24576 files, 31540/98016 blocks
[~]$ fdisk /dev/nftla

Command (m for help): p

Disk /dev/nftla: 16 heads, 12 sectors, 1022 cylinders
Units = cylinders of 192 * 512 bytes

     Device Boot    Start       End    Blocks   Id  System
/dev/nftla1             2      1022     98016   83  Linux

Command (m for help): 


On Tue, 11 Jun 2002, Dzuy Nguyen wrote:

> It's possible that the grub_firmware and the nftl are overlapping.  When 
> run doc_loadbios,
> observe the number of bytes written into DoC.  It should be the size of 
> the grub_firmware.
> I'd add a couple of blocks (erase size) to that as the offet in nftl_format.
> 
> Matt Vestal wrote:
> 
> >I 'm having the same problem as others who have posted, in getting the
> >"Selected disk does not exist" error.  I'm using a 96MB DOC with a
> >VersaLogic VSBC-8 motherboard.
> >
> >-using Grub 0.92
> >-patched with grub-2002-02-19-doc.patch
> >-ran "aclocal && automake && autoconf"   (needed automake-1.5-8)
> >-./configure --enable-diskonchip-2000 --enable-ext2fs
> >-../util/doc_loadbios /dev/mtd0 grub_firmware
> >-./nftl_format /dev/mtd0 98304
> >-fdisk /dev/nftla into multiple portions.  also tried leaving it full
> >-mke2fs
> >-copy over necessary files (kernel and such)
> >-reboot
> >
> >The DOC loads Grub with no problem.  However, Grub has never been able to
> >see the DOC filesystem(s).  I've tried root (dc0), root (dc0,0), root
> >(dc0,1), etc, but nothing works.  I always get the "Selected disk does not
> >exist, error"
> >
> >Any help would be greatly appreciated, as I've been pouring over this for
> >days, now.
> >
> >Thanks,
> >
> >Matt
> >
> >
> >
> >______________________________________________________
> >Linux MTD discussion mailing list
> >http://lists.infradead.org/mailman/listinfo/linux-mtd/
> >
> >
> 
> 
> 
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/
> 

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: DOC2000 and Grub
  2002-06-11 17:46   ` Matt Vestal
@ 2002-06-11 17:56     ` Ilguiz Latypov
  0 siblings, 0 replies; 11+ messages in thread
From: Ilguiz Latypov @ 2002-06-11 17:56 UTC (permalink / raw)
  To: Matt Vestal; +Cc: Linux MTD mailing list

On Tue, 11 Jun 2002, Matt Vestal wrote:

> Also, the partition table gets easily corrupted.

You might have noticed the today's posting by David Woodhouse on 
importance of NFTL driver being inactive when re-formatting the chip.  I 
reboot few times to achieve that:

    erase_all /dev/mtd0
    doc_loadbios /dev/mtd0 grub_firmware
    reboot

    # Now it is unlikely the NFTL driver will find an NFTL signature on 
    # the chip.
    erase_all /dev/mtd0   # erase again since the previous erase might
                          # be spoiled by the NFTL driver
    doc_loadbios /dev/mtd0 grub_firmware
    nftl_format /dev/mtd0 98304
    fdisk /dev/nftla
    mke2fs /dev/nftlaX
    ...

Ilguiz

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: DOC2000 and Grub
  2002-06-11 17:41 ` Ilguiz Latypov
@ 2002-06-11 18:04   ` Matt Vestal
  2002-06-11 18:34     ` Ilguiz Latypov
  0 siblings, 1 reply; 11+ messages in thread
From: Matt Vestal @ 2002-06-11 18:04 UTC (permalink / raw)
  To: Ilguiz Latypov; +Cc: linux-mtd

Now, with DOC_DEBUG defined -

[~/mtd/grub]$ ../util/doc_loadbios /dev/mtd0 grub_firmware 
Performing Flash Erase of length 16384 at offset 0
Performing Flash Erase of length 16384 at offset 16384
Performing Flash Erase of length 16384 at offset 32768
Performing Flash Erase of length 16384 at offset 49152
Performing Flash Erase of length 16384 at offset 65536
Performing Flash Erase of length 16384 at offset 81920
Writing the firmware of length 90112 at 0... Done.

I get the following when grub loads:

setup_part(): filename /boot/grub/menu.lst
open_device(): errnum 0, current_drive 0, current_partition ffffff
same_partition: yes
open_device(): errnum 21
setup_part(): errnum 21

Then a root (dc0,0) yields:

set_device(): device (dc0,0), drive 0
same_partition: yes
open_device(): errnum 0, current_drive e0, current_partition ffffff
same_partition: yes
open_device(): errnum 21

Error 21:  Selected disk does not exist.


Matt

On Tue, 11 Jun 2002, Ilguiz Latypov wrote:

> 
> On Tue, 11 Jun 2002, Matt Vestal wrote:
> 
> > The DOC loads Grub with no problem.  However, Grub has never been able to
> > see the DOC filesystem(s).  I've tried root (dc0), root (dc0,0), root
> > (dc0,1), etc, but nothing works.  I always get the "Selected disk does not
> > exist, error"
> 
> Any debug output on the screen when GRUB starts?
> 
> Does this diagnostic message show up?
> 
>    Spare Media Header for NFTL %d found in flash sector %d
> 
> Ilguiz
> 
> 
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/
> 

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: DOC2000 and Grub
@ 2002-06-11 18:08 Mark Meade
  2002-06-11 18:15 ` Ilguiz Latypov
  0 siblings, 1 reply; 11+ messages in thread
From: Mark Meade @ 2002-06-11 18:08 UTC (permalink / raw)
  To: linux-mtd

On Tue, 11 Jun 2002, Ilguiz Latypov wrote:

> erase_all /dev/mtd0
> doc_loadbios /dev/mtd0 grub_firmware

Ilguiz,

Is the "erase_all" necessary?  Does this delete the Bad Block Table on the 
DoC 2000 or Millennuim?

I was able to get a Millennium booting without any "erase_all" commands.  I 
guess I'm not clear on why this is needed.

Thanks,

Mark

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: DOC2000 and Grub
  2002-06-11 18:08 DOC2000 and Grub Mark Meade
@ 2002-06-11 18:15 ` Ilguiz Latypov
  0 siblings, 0 replies; 11+ messages in thread
From: Ilguiz Latypov @ 2002-06-11 18:15 UTC (permalink / raw)
  To: Mark Meade; +Cc: linux-mtd

Mark,

On Tue, 11 Jun 2002, Mark Meade wrote:

> Is the "erase_all" necessary?  Does this delete the Bad Block Table on the 
> DoC 2000 or Millennuim?

I agree erase_all is dangerous because it will delete the bad block table.  
The smarter way of inactivating the NFTL driver would be obfuscating the 
NFTL headers or implementing the previously mentioned ioctl() interface to 
NFTL layer or relying on the M-Sys native MS-DOS based NFTL format 
utility.

> I was able to get a Millennium booting without any "erase_all" commands.  I 
> guess I'm not clear on why this is needed.

I was afraid that the NFTL driver would keep updating the /dev/mtd0 device 
even when nftl_format was running.

Best regards,

Ilguiz

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: DOC2000 and Grub
  2002-06-11 18:04   ` Matt Vestal
@ 2002-06-11 18:34     ` Ilguiz Latypov
  2002-06-11 18:46       ` Matt Vestal
  2002-06-11 20:44       ` Matt Vestal
  0 siblings, 2 replies; 11+ messages in thread
From: Ilguiz Latypov @ 2002-06-11 18:34 UTC (permalink / raw)
  To: Matt Vestal; +Cc: linux-mtd

Thanks, Matt.  It seems the stripped down version of MTD/DOC2000/NFTL
drivers bdev_diskonchip.c was unable to find an NFTL header in the flash
memory.  The diagnostics from bdev_diskonchip.c should appear earlier when
GRUB is starting.  Unfortunately, those messages are not currently routed
to the serial port.  Does your oard have a video chip?

If not, I think GRUB could be modified to switch to serial port earlier 
based on hard coded menu options.

Ilguiz

On Tue, 11 Jun 2002, Matt Vestal wrote:

> Now, with DOC_DEBUG defined -
> [...]
> open_device(): errnum 21

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: DOC2000 and Grub
  2002-06-11 18:34     ` Ilguiz Latypov
@ 2002-06-11 18:46       ` Matt Vestal
  2002-06-11 20:44       ` Matt Vestal
  1 sibling, 0 replies; 11+ messages in thread
From: Matt Vestal @ 2002-06-11 18:46 UTC (permalink / raw)
  To: Ilguiz Latypov; +Cc: linux-mtd

On Tue, 11 Jun 2002, Ilguiz Latypov wrote:

> 
> Thanks, Matt.  It seems the stripped down version of MTD/DOC2000/NFTL
> drivers bdev_diskonchip.c was unable to find an NFTL header in the flash
> memory.  The diagnostics from bdev_diskonchip.c should appear earlier when
> GRUB is starting.  Unfortunately, those messages are not currently routed
> to the serial port.  Does your oard have a video chip?

Yes, it does.  The VSBC-8 is an EBX form factor, with standard ports, so
I've been using a monitor with it.  I'm not using a serial console, if
that's what you mean by "messages are not currently routed to the serial
port".

So it the DOC stuff really enabled in my grub?  What might be the problem?

Thanks,

Matt

> 
> If not, I think GRUB could be modified to switch to serial port earlier 
> based on hard coded menu options.
> 
> Ilguiz
> 
> On Tue, 11 Jun 2002, Matt Vestal wrote:
> 
> > Now, with DOC_DEBUG defined -
> > [...]
> > open_device(): errnum 21
> 
> 

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: DOC2000 and Grub
  2002-06-11 18:34     ` Ilguiz Latypov
  2002-06-11 18:46       ` Matt Vestal
@ 2002-06-11 20:44       ` Matt Vestal
  1 sibling, 0 replies; 11+ messages in thread
From: Matt Vestal @ 2002-06-11 20:44 UTC (permalink / raw)
  To: Ilguiz Latypov; +Cc: linux-mtd

Hi Ilguiz, it turns out that BDEV_DISKONCHIP was never getting defined!  
I thought the DOC stuff was enabled because root (dc0) started returning
"Selected disk does not exist", rather than the previous string "Error
23: Error while parsing number".  Anyways, I added -DBDEV_DISKONCHIP to
stage2/shared.h and voila!

I think I'm all set now.  Thanks for the help,

Matt


On Tue, 11 Jun 2002, Ilguiz Latypov wrote:

> 
> Thanks, Matt.  It seems the stripped down version of MTD/DOC2000/NFTL
> drivers bdev_diskonchip.c was unable to find an NFTL header in the flash
> memory.  The diagnostics from bdev_diskonchip.c should appear earlier when
> GRUB is starting.  Unfortunately, those messages are not currently routed
> to the serial port.  Does your oard have a video chip?
> 
> If not, I think GRUB could be modified to switch to serial port earlier 
> based on hard coded menu options.
> 
> Ilguiz
> 
> On Tue, 11 Jun 2002, Matt Vestal wrote:
> 
> > Now, with DOC_DEBUG defined -
> > [...]
> > open_device(): errnum 21
> 
> 

^ permalink raw reply	[flat|nested] 11+ messages in thread

end of thread, other threads:[~2002-06-11 20:44 UTC | newest]

Thread overview: 11+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2002-06-11 18:08 DOC2000 and Grub Mark Meade
2002-06-11 18:15 ` Ilguiz Latypov
  -- strict thread matches above, loose matches on Subject: below --
2002-06-11 16:13 Matt Vestal
2002-06-11 16:27 ` Dzuy Nguyen
2002-06-11 17:46   ` Matt Vestal
2002-06-11 17:56     ` Ilguiz Latypov
2002-06-11 17:41 ` Ilguiz Latypov
2002-06-11 18:04   ` Matt Vestal
2002-06-11 18:34     ` Ilguiz Latypov
2002-06-11 18:46       ` Matt Vestal
2002-06-11 20:44       ` Matt Vestal

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox