Regression with grub-probefs

Regression with grub-probefs

[Robert Millan]

Looks like grub-probefs is broken in current CVS:

$ sudo grub-probefs -v /
grub-probefs: info: changing current directory to /dev
[...]
grub-probefs: info: /dev/hda1 starts from 63
grub-probefs: info: opening the device hd0
grub-probefs: info: the size of hd0 is 156301488
grub-probefs: info: opening the device `/dev/hda'
grub-probefs: info: DOS partition 0 starts from 63
grub-probefs: info: opening hd0,0
grub-probefs: info: the size of hd0 is 156301488
grub-probefs: error: unknown device

I've performed a binary search and found that it stopped working after:

2006-06-05  Yoshinori K. Okuji  <okuji@enbug.org>

	Count partitions from 1 instead of 0 in the string representation
	of partitions. Still use 0-based internally.

I'm not sure if this applies to device.map.  Note that grub-mkdevicemap both
from before and after this commit generates the same output for me.

My (hand-editted) file is:

$ cat /boot/grub/device.map
(hd0)	/dev/hda

I also tried with (hd1) and even (hd-1), with same result.

-- 
Robert Millan

My spam trap is honeypot@aybabtu.com.  Note: this address is only intended for
spam harvesters.  Writing to it will get you added to my black list.

Re: Regression with grub-probefs

[Yoshinori K. Okuji]

On Friday 22 September 2006 00:08, Robert Millan wrote:
> Looks like grub-probefs is broken in current CVS:

Thanks. I will take a look tomorrow.

Okuji

Re: Regression with grub-probefs

[Robert Millan]

On Fri, Sep 22, 2006 at 01:57:41PM +0200, Yoshinori K. Okuji wrote:
> On Friday 22 September 2006 00:08, Robert Millan wrote:
> > Looks like grub-probefs is broken in current CVS:
> 
> Thanks. I will take a look tomorrow.

Hi Okuji,

Any news on that?  In debian we have to maintain a patch that reverts this
commit, but this might have undesired effects.  We'd like to stay close to
upstream if possible.

-- 
Robert Millan

My spam trap is honeypot@aybabtu.com.  Note: this address is only intended for
spam harvesters.  Writing to it will get you added to my black list.

Re: Regression with grub-probefs

[Yoshinori K. Okuji]

On Saturday 14 October 2006 21:18, Robert Millan wrote:
> Any news on that?  In debian we have to maintain a patch that reverts this
> commit, but this might have undesired effects.  We'd like to stay close to
> upstream if possible.

Done.

Okuji

Re: Regression with grub-probefs

[Robert Millan]

On Sat, Oct 14, 2006 at 11:08:17PM +0200, Yoshinori K. Okuji wrote:
> On Saturday 14 October 2006 21:18, Robert Millan wrote:
> > Any news on that?  In debian we have to maintain a patch that reverts this
> > commit, but this might have undesired effects.  We'd like to stay close to
> > upstream if possible.
> 
> Done.

Oops.  I noticed this other regression after my commit, and was working on it
but looks like you beat me to it :)

	* util/i386/pc/grub-probe.c (probe): Print DEVICE_NAME instead of
	DRIVE_NAME when grub_util_biosdisk_get_grub_dev fails. Open
	DRIVE_NAME instead of DEVICE_NAME. Make sure that DEVICE_NAME and
	DRIVE_NAME are always freed.

Anyways, glad to see CVS is usable again on i386!

-- 
Robert Millan

My spam trap is honeypot@aybabtu.com.  Note: this address is only intended for
spam harvesters.  Writing to it will get you added to my black list.