Help with booting with very large initrd

Help with booting with very large initrd

[Howard, Marc]

Hi,

I'm developing a PPC440GX based board that uses U-Boot to boot a
multi-file boot image composed of the kernel and a very large (> 96MB
uncompressed) initrd file.  The board has 512MB of RAM of which the
upper 16MB is reserved for dedicated hardware.  The 16MB block is
reserved via "mem=3D496M" and U-Boot is told to stay out of that area by
setting "initrd_high=3D1f000000".

Before anyone asks there are several reasons for doing things this way.
NFS is not an option in the target environment.

I can tftp the combined boot image to my board.  I checked the crc with
the crc32 command and it agrees exactly with the result obtained on the
host machine using the boot file.  Therefore there is not a "TFTP >32MB"
problem here.

If I boot I get the following:

=3D> boot
Waiting for PHY auto negotiation to complete... done
ENET Speed is 1000 Mbps - FULL duplex connection
Using ppc_4xx_eth2 device
TFTP from server 192.168.168.108; our IP address is 192.168.168.111
Filename 'pMulti-ramdisk'.
Load address: 0x400000
Loading: *
done
Bytes transferred =3D 38825407 (2506dbf hex)
Automatic boot of image at addr 0x00400000 ...
## Booting image at 00400000 ...
   Image Name:   Linux-2.6.10_mvl401-440gx_eval-I
   Created:      2006-08-25   1:01:29 UTC
   Image Type:   PowerPC Linux Multi-File Image (gzip compressed)
   Data Size:    38825343 Bytes =3D 37 MB
   Load Address: 00000000
   Entry Point:  00000000
   Contents:
   Image 0:  1137986 Bytes =3D  1.1 MB
   Image 1: 37687343 Bytes =3D 35.9 MB
   Verifying Checksum ... OK
   Uncompressing Multi-File Image ... OK
   Loading Ramdisk to 1cc0e000, end 1efff02f ... OK
Linux version 2.6.10_mvl401-440gx_eval (cram@toaster.kla-tencor.com)
(gcc version 3.4.3 (MontaVista 3.4.3-25.0.107.0601076 2006-07-21)) #46
Thu Aug 24 17:28:09 PDT 2006
IBM Ocotea port (MontaVista Software, Inc. <source@mvista.com>)
Built 1 zonelists
Kernel command line: ramdisk_size=3D262144 root=3D/dev/ram rw
console=3DttyS0,115200
ip=3D192.168.168.111:192.168.168.108::255.255.255.0:scpu2:eth0: off
mem=3D496M
PID hash table entries: 2048 (order: 11, 32768 bytes)

......stuff deleted......

RAMDISK driver initialized: 8 RAM disks of 262144K size 1024 blocksize
loop: loaded (max 8 devices)

......more stuff deleted......

eth0: link is down
eth0: link is up, 1000 FDX, pause enabled
IP-Config: Complete:
      device=3Deth0, addr=3D192.168.168.111, mask=3D255.255.255.0,
gw=3D255.255.255.255,     host=3Dscpu2, domain=3D, nis-domain=3D(none),
     bootserver=3D192.168.168.108, rootserver=3D192.168.168.108, =
rootpath=3D
RAMDISK: Compressed image found at block 0
crc error (orig 0x9a278d64, CRC_VALUE 0xa7bcd2e3 -- ignoring!
length error (orig =3D 0x0c000000, bytes_out =3D 0x0c000015 -- ignored
VFS: Mounted root (ext2 filesystem).
Freeing unused kernel memory: 120k init
Warning: unable to open an initial console.
Kernel panic - not syncing: No init found.  Try passing init=3D option =
to
kernel.
 <0>Rebooting in 180 seconds..

(I modified lib/inflate.c so that the crc and length checks would
provide more info).

Since the overall file CRC is good and U-Boot checksums are also okay
this looks like some sort of size limitation with the inflate routine.
BTW, The kerenel was made assuming a 256MB ramdisk; likewise the command
line specs one as well.  The initrd image would fit easily in that
space.

Have any of you worked on this problem before and come up with a
solution?

Thanks,

Marc Howard

Re: Help with booting with very large initrd

[Reeve Yang]

[-- Attachment #1: Type: text/plain, Size: 3938 bytes --]

I had a similiar problem long time ago. I remember it was related to the
size of ramdisk. Double check if your ramdisk size big enough.

On 8/25/06, Howard, Marc <Marc.Howard@kla-tencor.com> wrote:
>
> Hi,
>
> I'm developing a PPC440GX based board that uses U-Boot to boot a
> multi-file boot image composed of the kernel and a very large (> 96MB
> uncompressed) initrd file.  The board has 512MB of RAM of which the
> upper 16MB is reserved for dedicated hardware.  The 16MB block is
> reserved via "mem=496M" and U-Boot is told to stay out of that area by
> setting "initrd_high=1f000000".
>
> Before anyone asks there are several reasons for doing things this way.
> NFS is not an option in the target environment.
>
> I can tftp the combined boot image to my board.  I checked the crc with
> the crc32 command and it agrees exactly with the result obtained on the
> host machine using the boot file.  Therefore there is not a "TFTP >32MB"
> problem here.
>
> If I boot I get the following:
>
> => boot
> Waiting for PHY auto negotiation to complete... done
> ENET Speed is 1000 Mbps - FULL duplex connection
> Using ppc_4xx_eth2 device
> TFTP from server 192.168.168.108; our IP address is 192.168.168.111
> Filename 'pMulti-ramdisk'.
> Load address: 0x400000
> Loading: *
> done
> Bytes transferred = 38825407 (2506dbf hex)
> Automatic boot of image at addr 0x00400000 ...
> ## Booting image at 00400000 ...
>    Image Name:   Linux-2.6.10_mvl401-440gx_eval-I
>    Created:      2006-08-25   1:01:29 UTC
>    Image Type:   PowerPC Linux Multi-File Image (gzip compressed)
>    Data Size:    38825343 Bytes = 37 MB
>    Load Address: 00000000
>    Entry Point:  00000000
>    Contents:
>    Image 0:  1137986 Bytes =  1.1 MB
>    Image 1: 37687343 Bytes = 35.9 MB
>    Verifying Checksum ... OK
>    Uncompressing Multi-File Image ... OK
>    Loading Ramdisk to 1cc0e000, end 1efff02f ... OK
> Linux version 2.6.10_mvl401-440gx_eval (cram@toaster.kla-tencor.com)
> (gcc version 3.4.3 (MontaVista 3.4.3-25.0.107.0601076 2006-07-21)) #46
> Thu Aug 24 17:28:09 PDT 2006
> IBM Ocotea port (MontaVista Software, Inc. <source@mvista.com>)
> Built 1 zonelists
> Kernel command line: ramdisk_size=262144 root=/dev/ram rw
> console=ttyS0,115200
> ip=192.168.168.111:192.168.168.108::255.255.255.0:scpu2:eth0: off
> mem=496M
> PID hash table entries: 2048 (order: 11, 32768 bytes)
>
> ......stuff deleted......
>
> RAMDISK driver initialized: 8 RAM disks of 262144K size 1024 blocksize
> loop: loaded (max 8 devices)
>
> ......more stuff deleted......
>
> eth0: link is down
> eth0: link is up, 1000 FDX, pause enabled
> IP-Config: Complete:
>       device=eth0, addr=192.168.168.111, mask=255.255.255.0,
> gw=255.255.255.255,     host=scpu2, domain=, nis-domain=(none),
>      bootserver=192.168.168.108, rootserver=192.168.168.108, rootpath=
> RAMDISK: Compressed image found at block 0
> crc error (orig 0x9a278d64, CRC_VALUE 0xa7bcd2e3 -- ignoring!
> length error (orig = 0x0c000000, bytes_out = 0x0c000015 -- ignored
> VFS: Mounted root (ext2 filesystem).
> Freeing unused kernel memory: 120k init
> Warning: unable to open an initial console.
> Kernel panic - not syncing: No init found.  Try passing init= option to
> kernel.
> <0>Rebooting in 180 seconds..
>
> (I modified lib/inflate.c so that the crc and length checks would
> provide more info).
>
> Since the overall file CRC is good and U-Boot checksums are also okay
> this looks like some sort of size limitation with the inflate routine.
> BTW, The kerenel was made assuming a 256MB ramdisk; likewise the command
> line specs one as well.  The initrd image would fit easily in that
> space.
>
> Have any of you worked on this problem before and come up with a
> solution?
>
> Thanks,
>
> Marc Howard
>
>
>
> _______________________________________________
> Linuxppc-embedded mailing list
> Linuxppc-embedded@ozlabs.org
> https://ozlabs.org/mailman/listinfo/linuxppc-embedded
>

[-- Attachment #2: Type: text/html, Size: 5268 bytes --]