Re: Bug in gpt_probe()

[rahul dev]

Yeah, its very simple. Just zero out the primary gpt header (ie lba=1) using dd 
command.
Try to read the gpt partition table. libparted will fail to read it although 
the secondary partition table at the end of disk is intact and still valid.

thanks,
rahul

--- On Mon, 1/8/11, Brian C. Lane <address@hidden> wrote:

> From: Brian C. Lane <address@hidden>
> Subject: Re: Bug in gpt_probe()
> To: "rahul dev" <address@hidden>
> Cc: address@hidden
> Date: Monday, 1 August, 2011, 10:29 PM
> On Mon, Aug 01, 2011 at 10:00:58AM
> +0530, rahul dev wrote:
> > Guys,
> > 
> >   Looks like there is a bug in
> gpt_probe(). In case it reads the primary header
> successfully but doesn't find a gpt signature, it returns 0
> and doesn't check for the secondary gpt table.
> > 
> > This seems to be a serious bug. Can anyone take a look
> ?
> 
> Can you describe a reliable way to reproduce it? Preferably
> using a
> disk image file as a test.
> 
> -- 
> Brian C. Lane | Anaconda Team | IRC: bcl #anaconda | Port
> Orchard, WA (PST8PDT)
>