[Fwd: Bug#503344: Grub-pc fails on IBM x3250 configured with RAID-1]

[Fwd: Bug#503344: Grub-pc fails on IBM x3250 configured with RAID-1]

[Felix Zielcke]

Is this an acceptable fix?
Though I have the feeling this isn't a proper one.

-------- Weitergeleitete Nachricht --------
> Von: Bob Gilligan <gilligan@vyatta.com>
> Reply-to: Bob Gilligan <gilligan@vyatta.com>, 503344@bugs.debian.org
> An: submit@bugs.debian.org
> Betreff: Bug#503344: Grub-pc fails on IBM x3250 configured with RAID-1
> Datum: Fri, 24 Oct 2008 16:19:59 -0700
> 
> Package: grub-pc
> Version: 1.96+20080724-10
> 
> After installing grub-pc on an IBM x3250 with two 160 GB drives, with 
> the root filesystem configured using mdadm for RAID-1, grub fails to 
> boot, printing:
> 
> Welcome to GRUB!
> 
> error: unknown device fd1
> Entering rescue mode ...
> grub rescue>
> 
> After troubleshooting, I found that the failure was occurring in 
> grub_file_open() when grub_dl_load_file() was attempting to load the 
> normal mode module.  The grub_file_open() function checks the return 
> status of its call to grub_file_get_device_name() by checking the value 
> of grub_errno.  If grub_errno is non-zero, grub_file_open() returns 
> failure.  But grub_file_get_device_name() can succeed, but the error 
> test fail, if grub_errno was set in some earlier operation.  In this 
> case, apparently an early attempt to open a non-existent device set 
> grub_errno.
> 
> The fix is to assign grub_errno before making the call to 
> grub_file_get_device_name():
> 
> diff --git a/kern/file.c b/kern/file.c
> index adf55da..5e605f9 100644
> --- a/kern/file.c
> +++ b/kern/file.c
> @@ -59,6 +59,7 @@ grub_file_open (const char *name)
>     char *device_name;
>     char *file_name;
> 
> +  grub_errno = GRUB_ERR_NONE;  /* Used as error flag */
>     device_name = grub_file_get_device_name (name);
>     if (grub_errno)
>       return 0;


-- 
Felix Zielcke
Proud Debian Maintainer

Re: [Fwd: Bug#503344: Grub-pc fails on IBM x3250 configured with RAID-1]

[Vladimir 'phcoder' Serbinenko]

On Sat, Jul 25, 2009 at 1:35 AM, Felix Zielcke<fzielcke@z-51.de> wrote:
> Is this an acceptable fix?
> Though I have the feeling this isn't a proper one.
>
>>
>> The fix is to assign grub_errno before making the call to
>> grub_file_get_device_name():
Just check whether returned device_name is 0 or not
>>
>> diff --git a/kern/file.c b/kern/file.c
>> index adf55da..5e605f9 100644
>> --- a/kern/file.c
>> +++ b/kern/file.c
>> @@ -59,6 +59,7 @@ grub_file_open (const char *name)
>>     char *device_name;
>>     char *file_name;
>>
>> +  grub_errno = GRUB_ERR_NONE;  /* Used as error flag */
>>     device_name = grub_file_get_device_name (name);
>>     if (grub_errno)
>>       return 0;
>
>
> --
> Felix Zielcke
> Proud Debian Maintainer
>
>
>
> _______________________________________________
> Grub-devel mailing list
> Grub-devel@gnu.org
> http://lists.gnu.org/mailman/listinfo/grub-devel
>



-- 
Regards
Vladimir 'phcoder' Serbinenko

Personal git repository: http://repo.or.cz/w/grub2/phcoder.git

Re: [Fwd: Bug#503344: Grub-pc fails on IBM x3250 configured with RAID-1]

[Felix Zielcke]

Am Samstag, den 25.07.2009, 10:37 +0200 schrieb Vladimir 'phcoder'
Serbinenko:
> On Sat, Jul 25, 2009 at 1:35 AM, Felix Zielcke<fzielcke@z-51.de> wrote:
> > Is this an acceptable fix?
> > Though I have the feeling this isn't a proper one.
> >
> >>
> >> The fix is to assign grub_errno before making the call to
> >> grub_file_get_device_name():
> Just check whether returned device_name is 0 or not

Ok, I should have checked the actual code. That was pretty clear.
Sorry for the noise.

> >>
> >> diff --git a/kern/file.c b/kern/file.c
> >> index adf55da..5e605f9 100644
> >> --- a/kern/file.c
> >> +++ b/kern/file.c
> >> @@ -59,6 +59,7 @@ grub_file_open (const char *name)
> >>     char *device_name;
> >>     char *file_name;
> >>
> >> +  grub_errno = GRUB_ERR_NONE;  /* Used as error flag */
> >>     device_name = grub_file_get_device_name (name);
> >>     if (grub_errno)
> >>       return 0;
> >
> >
> > --
> > Felix Zielcke
> > Proud Debian Maintainer
> >
> >
> >
> > _______________________________________________
> > Grub-devel mailing list
> > Grub-devel@gnu.org
> > http://lists.gnu.org/mailman/listinfo/grub-devel
> >
> 
> 
> 


-- 
Felix Zielcke
Proud Debian Maintainer

Re: [Fwd: Bug#503344: Grub-pc fails on IBM x3250 configured with RAID-1]

[Robert Millan]

On Sat, Jul 25, 2009 at 01:35:02AM +0200, Felix Zielcke wrote:
> Is this an acceptable fix?
> Though I have the feeling this isn't a proper one.
> 
> -------- Weitergeleitete Nachricht --------
> > Von: Bob Gilligan <gilligan@vyatta.com>
> > Reply-to: Bob Gilligan <gilligan@vyatta.com>, 503344@bugs.debian.org
> > An: submit@bugs.debian.org
> > Betreff: Bug#503344: Grub-pc fails on IBM x3250 configured with RAID-1
> > Datum: Fri, 24 Oct 2008 16:19:59 -0700
> > 
> > Package: grub-pc
> > Version: 1.96+20080724-10
> > 
> > After installing grub-pc on an IBM x3250 with two 160 GB drives, with 
> > the root filesystem configured using mdadm for RAID-1, grub fails to 
> > boot, printing:
> > 
> > Welcome to GRUB!
> > 
> > error: unknown device fd1
> > Entering rescue mode ...
> > grub rescue>

ISTR we disabled floppy support completely in debian in order to avoid this
kind of trouble (I would disable it in official GRUB as well, but I can
already feel there'd be some opposition :-)).

Anyway, my point is that floppies are too unreliable.  We detect them when
iterating even if they're not readable.  Perhaps we could actually probe them
for read during iteration to solve this.

> > +  grub_errno = GRUB_ERR_NONE;  /* Used as error flag */
> >     device_name = grub_file_get_device_name (name);
> >     if (grub_errno)
> >       return 0;

This looks like a hack.

-- 
Robert Millan

  The DRM opt-in fallacy: "Your data belongs to us. We will decide when (and
  how) you may access your data; but nobody's threatening your freedom: we
  still allow you to remove your data and not access it at all."

Re: [Fwd: Bug#503344: Grub-pc fails on IBM x3250 configured with RAID-1]

[Felix Zielcke]

Am Samstag, den 25.07.2009, 18:45 +0200 schrieb Robert Millan:
> On Sat, Jul 25, 2009 at 01:35:02AM +0200, Felix Zielcke wrote:
> > Is this an acceptable fix?
> > Though I have the feeling this isn't a proper one.
> > 
> > -------- Weitergeleitete Nachricht --------
> > > Von: Bob Gilligan <gilligan@vyatta.com>
> > > Reply-to: Bob Gilligan <gilligan@vyatta.com>, 503344@bugs.debian.org
> > > An: submit@bugs.debian.org
> > > Betreff: Bug#503344: Grub-pc fails on IBM x3250 configured with RAID-1
> > > Datum: Fri, 24 Oct 2008 16:19:59 -0700
> > > 
> > > Package: grub-pc
> > > Version: 1.96+20080724-10
> > > 
> > > After installing grub-pc on an IBM x3250 with two 160 GB drives, with 
> > > the root filesystem configured using mdadm for RAID-1, grub fails to 
> > > boot, printing:
> > > 
> > > Welcome to GRUB!
> > > 
> > > error: unknown device fd1
> > > Entering rescue mode ...
> > > grub rescue>
> 
> ISTR we disabled floppy support completely in debian in order to avoid this
> kind of trouble (I would disable it in official GRUB as well, but I can
> already feel there'd be some opposition :-)).

Only for util/hostdisk.c

> Anyway, my point is that floppies are too unreliable.  We detect them when
> iterating even if they're not readable.  Perhaps we could actually probe them
> for read during iteration to solve this.
> 
> > > +  grub_errno = GRUB_ERR_NONE;  /* Used as error flag */
> > >     device_name = grub_file_get_device_name (name);
> > >     if (grub_errno)
> > >       return 0;
> 
> This looks like a hack.
> 

r2444 should fix this. I just check now if it returns 0 and not for
grub_errno.

-- 
Felix Zielcke
Proud Debian Maintainer