On Mon, Jul 26, 2010 at 04:46:23AM +0200, Aurelien Jarno wrote:
> On Mon, Dec 29, 2008 at 04:44:06PM +0100, Giuseppe Sacco wrote:
> > Package: util-linux
> > Version: 2.13.1.1-1
> > Severity: grave
> > Tags: patch
> >
> > Yesterday I tried the d-i rc1 for lenny on a silicon graphics O2 (a
> > complete report is available as Debian bug #510060).
> >
> > The problem I found against fdisk is that when using large disks (i.e.,
> > disks with more than 4096 sectors per cylinder), the volhdr partition is
> > incorrectly sized because it is marked as starting and ending on
> > cylinder 0. When having this specific case, this is what happen:
> >
> > Command (m for help): p
> >
> > Disk /dev/sda (SGI disk label): 255 heads, 63 sectors, 8941 cylinders
> > Units = cylinders of 16065 * 512 bytes
> >
> > ----- partitions -----
> > Pt# Device Info Start End Sectors Id System
> > ----- Bootinfo -----
> > Bootfile: /unix
> > ----- Directory Entries -----
> > 0: arcboot sector 4 size 72492
> >
> > Command (m for help): n
> > Partition number (1-16): 1
> > Attempting to generate entire disk entry automatically.
> > First cylinder (1-8940, default 1):
> > Using default value 1
> > Last cylinder (1-8940, default 8940): 2000
> >
> > Command (m for help): p
> >
> > Disk /dev/sda (SGI disk label): 255 heads, 63 sectors, 8941 cylinders
> > Units = cylinders of 16065 * 512 bytes
> >
> > ----- partitions -----
> > Pt# Device Info Start End Sectors Id System
> > 1: /dev/sda1 boot 1 2000 32130000 83 Linux native
> > 9: /dev/sda2 0 0 4096 0 SGI volhdr
> > 11: /dev/sda3 0 8940 143637165 6 SGI volume
> > ----- Bootinfo -----
> > Bootfile: /unix
> > ----- Directory Entries -----
> > 0: arcboot sector 4 size 72492
> >
> >
> > Command (m for help): n
> > Partition number (1-16): 2
> > First cylinder (1-0): 1
> > Value out of range.
> > First cylinder (1-0): 0
> > Value out of range.
> > First cylinder (1-0): 2001
> > Value out of range.
> > First cylinder (1-0): ^C
> >
> > As you may see, once the first partition has been created, fdisk will
> > automatically create partitions 9 and 11, but it will wrongly set the
> > end boundary of volhdr. Starting with this error, it is not possible to
> > add any new partition, because the accepted cylinder range would be
> > "1-0".
> >
> > Please find attached a solution for this bug. This solution raise the
> > volhdr size of 4096 up to the first cylinder boundary. If the boundary
> > is less than 4096 then nothing is changed.
> >
>
> While I agree with the analysis, the patch looks more like a workaround
> than a real solution. The problem is that when free space doesn't start
> on a cylinder boundary (the case for the free space after the volume
> header), there might be less than one cylinder to create a partition.
> The code to ask for partition boundaries is using cylinders (probably
> because partition start should be cylinder aligned) and thus doesn't
> cope with this issue.
>
> The patch below fixes the issue by making sure free space always starts
> as a cylinder boundary, as anyway that's the only possible location to
> create a new partition.
>
> --- util-linux-2.13.1.1.orig/fdisk/fdisksgilabel.c
> +++ util-linux-2.13.1.1/fdisk/fdisksgilabel.c
> @@ -501,6 +501,9 @@
> }
> start = sgi_get_start_sector(Index[i])
> + sgi_get_num_sectors(Index[i]);
> + /* Align free space on cylinder boundary */
> + if (start % cylsize != 0)
> + start += cylsize - (start % cylsize);
> if (debug > 1) {
> if (verbose)
> printf("%2d:%12d\t%12d\t%12d\n", Index[i],
[Adding upstream to CC]
Karel,
what do you think about the patch proposed by Aurelien?
This is http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=510130
Cheers,
Moritz
--
To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
