Lite5200 NFS mount issue

Lite5200 NFS mount issue

[Björn Östby]

Hi all,
I'm using u-boot 1.1.4 (CVS downloaded 2006-05-04) and ELDK 4.0. I've =
encountered=20
problems when trying to mount a NFS root filesystem when using the Lite =
MPC5200
development board. The system mounts ok when the board is connected to =
office wide=20
network switches but if the board is connected directly through a small =
100/10Mbit switch
to the NFS server, the mount halts as shown in the log below (The =
100/10Mbit switch
do work as other boards can mount their filesystem through these). Does =
anyone
recognize this problem and know a suitable work-around/solution?

=3D>run flash_nfs
 ## Booting image at ff0a0000 ...
   Image Name:   Linux-2.4.25
   Created:      2006-06-16  13:11:56 UTC
   Image Type:   PowerPC Linux Kernel Image (gzip compressed)
   Data Size:    945550 Bytes =3D 923.4 kB
   Load Address: 00000000
   Entry Point:  00000000
   Verifying Checksum ... OK
   Uncompressing Kernel Image ... OK
Memory BAT mapping: BAT2=3D64Mb, BAT3=3D0Mb, residual: 0Mb
Linux version 2.4.25 (tord@toshiba) (gcc version 4.0.0 (DENX ELDK 4.0 =
4.0.0)) #1 fre jun 16 14:46:35 CEST 2006
On node 0 totalpages: 16384
zone(0): 16384 pages.
zone(1): 0 pages.
zone(2): 0 pages.
Kernel command line: root=3D/dev/nfs rw =
nfsroot=3D10.3.67.98:/opt/eldk_4_0/ppc_6xx =
ip=3D10.3.67.99:10.3.67.98:10.3.67.1:::eth0:off panic=3D1
Calibrating delay loop... 263.78 BogoMIPS
Memory: 62080k available (1608k kernel code, 484k data, 80k init, 0k =
highmem)
Dentry cache hash table entries: 8192 (order: 4, 65536 bytes)
Inode cache hash table entries: 4096 (order: 3, 32768 bytes)
Mount cache hash table entries: 512 (order: 0, 4096 bytes)
Buffer cache hash table entries: 4096 (order: 2, 16384 bytes)
Page-cache hash table entries: 16384 (order: 4, 65536 bytes)
POSIX conformance testing by UNIFIX
PCI: Probing PCI hardware
Linux NET4.0 for Linux 2.4
Based upon Swansea University Computer Society NET3.039
Initializing RT netlink socket
Starting kswapd
Journalled Block Device driver loaded
JFFS2 version 2.2. (NAND) (C) 2001-2003 Red Hat, Inc.
i2c-core.o: i2c core module version 2.6.1 (20010830)
i2c-proc.o version 2.6.1 (20010830)
pty: 256 Unix98 ptys configured
ttyS0 on PSC1
ttyS1 on PSC2
ttyS2 on PSC3
RAMDISK driver initialized: 16 RAM disks of 16384K size 1024 blocksize
loop: loaded (max 8 devices)
Uniform Multi-Platform E-IDE driver Revision: 7.00beta4-2.4
ide: Assuming 33MHz system bus speed for PIO modes; override with =
idebus=3Dxx
Port Config is: 0x11050004
ipb=3D66MHz, set clock period to 15
GPIO config: 11050004
ATA invalid: 00800000
ATA hostcnf: 03000000
ATA pio1   : 100a0a00
ATA pio2   : 02040600
XLB Arb cnf: 0000a366
mpc5xxx_ide: Setting up IDE interface ide0...
ATA DMA task: 5
Probing IDE interface ide0...
SCSI subsystem driver Revision: 1.00
Icecube-0: Found 1 x8 devices at 0x0 in 8-bit bank
Icecube-0: Found 1 x8 devices at 0x800000 in 8-bit bank
 Amd/Fujitsu Extended Query Table at 0x0040
Icecube-0: CFI does not contain boot bank location. Assuming top.
number of CFI chips: 2
cfi_cmdset_0002: Disabling erase-suspend-program due to code brokenness.
Icecube flash bank 0: Using static image partition definition
Creating 5 MTD partitions on "Icecube-0":
0x00000000-0x00800000 : "Spare"
0x00800000-0x00900000 : "kernel"
0x00900000-0x00c00000 : "initrd"
0x00c00000-0x00f00000 : "jffs"
0x00f00000-0x01000000 : "Firmware"
usb.c: registered new driver hub
host/usb-ohci.c: USB OHCI at membase 0xf0001000, IRQ 44
host/usb-ohci.c: usb-0, Built-In ohci
usb.c: new USB bus registered, assigned bus number 1
hub.c: USB hub found
hub.c: 1 port detected
bluetooth.c: USB Bluetooth support registered
usb.c: registered new driver bluetty
bluetooth.c: USB Bluetooth tty driver v0.13
Initializing USB Mass Storage driver...
usb.c: registered new driver usb-storage
USB Mass Storage support registered.
i2c-icecube.o: I2C module #2 installed
NET4: Linux TCP/IP 1.0 for NET4.0
IP Protocols: ICMP, UDP, TCP, IGMP
IP: routing cache hash table of 512 buckets, 4Kbytes
TCP: Hash tables configured (established 4096 bind 8192)
eth0: Phy @ 0x0, type LXT971 (0x001378e2)
IP-Config: Guessing netmask 255.0.0.0
IP-Config: Complete:
      device=3Deth0, addr=3D10.3.67.99, mask=3D255.0.0.0, =
gw=3D10.3.67.1,
     host=3D10.3.67.99, domain=3D, nis-domain=3D(none),
     bootserver=3D10.3.67.98, rootserver=3D10.3.67.98, rootpath=3D
NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
Looking up port of RPC 100003/2 on 10.3.67.98
Looking up port of RPC 100005/1 on 10.3.67.98
VFS: Mounted root (nfs filesystem).
Freeing unused kernel memory: 80k init
nfs: server 10.3.67.98 not responding, still trying
nfs: server 10.3.67.98 OK
nfs: server 10.3.67.98 not responding, still trying
nfs: server 10.3.67.98 OK
[time to this about 5 minutes]
INIT: version 2.85 booting
nfs: server 10.3.67.98 not responding, still trying
nfs: server 10.3.67.98 OK
nfs: server 10.3.67.98 not responding, still trying
nfs: server 10.3.67.98 OK

Regards,
Bjorn Ostby

Re: Lite5200 NFS mount issue

[Kimmo Surakka]

Bjorn.Ostby@combitech.se wrote:

> Hi all,
> I'm using u-boot 1.1.4 (CVS downloaded 2006-05-04) and ELDK 4.0. I've 
encountered 
> problems when trying to mount a NFS root filesystem when using the Lite 
MPC5200
> development board. The system mounts ok when the board is connected to 
office wide 
> network switches but if the board is connected directly through a small 
100/10Mbit switch
> to the NFS server, the mount halts as shown in the log below (The 100/10Mbit 
switch
> do work as other boards can mount their filesystem through these). Does 
anyone
> recognize this problem and know a suitable work-around/solution?

Try adding the ",tcp" to the kernel's "nfsroot=..." parameter, i.e. use 
commandline "root=/dev/nfs rw nfsroot=10.3.67.98:/opt/eldk_4_0/ppc_6xx,tcp 
ip=10.3.67.99:10.3.67.98:10.3.67.1:::eth0:off panic=1".

I recall having similar problems when connecting through an 100/10Mbit switch 
and in our case, the ",tcp" switch did the trick.

Cheers,

Kimmo Surakka

This message has been scanned by F-Secure Anti-Virus

###  OXFORD INSTRUMENTS   http://www.oxford-instruments.com/  ###

Unless stated above to be non-confidential, this E-mail and any
attachments are private and confidential and are for the addressee
only and may not be used, copied or disclosed save to the addressee.
If you have received this E-mail in error please notify us upon receipt
and delete it from your records. Internet communications are not secure
and Oxford Instruments is not responsible for their abuse by third
parties nor for any alteration or corruption in transmission.

Re: Lite5200 NFS mount issue

[Hans Thielemans]

Hi there,

I remember similar problems. It all had to do with full or half duplex 
ethernet.

Regards

Hans

On Tue, 20 Jun 2006, Björn Östby wrote:

> Hi all,
> I'm using u-boot 1.1.4 (CVS downloaded 2006-05-04) and ELDK 4.0. I've encountered 
> problems when trying to mount a NFS root filesystem when using the Lite MPC5200
> development board. The system mounts ok when the board is connected to office wide 
> network switches but if the board is connected directly through a small 100/10Mbit switch
> to the NFS server, the mount halts as shown in the log below (The 100/10Mbit switch
> do work as other boards can mount their filesystem through these). Does anyone
> recognize this problem and know a suitable work-around/solution?
> 
> =>run flash_nfs
>  ## Booting image at ff0a0000 ...
>    Image Name:   Linux-2.4.25
>    Created:      2006-06-16  13:11:56 UTC
>    Image Type:   PowerPC Linux Kernel Image (gzip compressed)
>    Data Size:    945550 Bytes = 923.4 kB
>    Load Address: 00000000
>    Entry Point:  00000000
>    Verifying Checksum ... OK
>    Uncompressing Kernel Image ... OK
> Memory BAT mapping: BAT2=64Mb, BAT3=0Mb, residual: 0Mb
> Linux version 2.4.25 (tord@toshiba) (gcc version 4.0.0 (DENX ELDK 4.0 4.0.0)) #1 fre jun 16 14:46:35 CEST 2006
> On node 0 totalpages: 16384
> zone(0): 16384 pages.
> zone(1): 0 pages.
> zone(2): 0 pages.
> Kernel command line: root=/dev/nfs rw nfsroot=10.3.67.98:/opt/eldk_4_0/ppc_6xx ip=10.3.67.99:10.3.67.98:10.3.67.1:::eth0:off panic=1
> Calibrating delay loop... 263.78 BogoMIPS
> Memory: 62080k available (1608k kernel code, 484k data, 80k init, 0k highmem)
> Dentry cache hash table entries: 8192 (order: 4, 65536 bytes)
> Inode cache hash table entries: 4096 (order: 3, 32768 bytes)
> Mount cache hash table entries: 512 (order: 0, 4096 bytes)
> Buffer cache hash table entries: 4096 (order: 2, 16384 bytes)
> Page-cache hash table entries: 16384 (order: 4, 65536 bytes)
> POSIX conformance testing by UNIFIX
> PCI: Probing PCI hardware
> Linux NET4.0 for Linux 2.4
> Based upon Swansea University Computer Society NET3.039
> Initializing RT netlink socket
> Starting kswapd
> Journalled Block Device driver loaded
> JFFS2 version 2.2. (NAND) (C) 2001-2003 Red Hat, Inc.
> i2c-core.o: i2c core module version 2.6.1 (20010830)
> i2c-proc.o version 2.6.1 (20010830)
> pty: 256 Unix98 ptys configured
> ttyS0 on PSC1
> ttyS1 on PSC2
> ttyS2 on PSC3
> RAMDISK driver initialized: 16 RAM disks of 16384K size 1024 blocksize
> loop: loaded (max 8 devices)
> Uniform Multi-Platform E-IDE driver Revision: 7.00beta4-2.4
> ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
> Port Config is: 0x11050004
> ipb=66MHz, set clock period to 15
> GPIO config: 11050004
> ATA invalid: 00800000
> ATA hostcnf: 03000000
> ATA pio1   : 100a0a00
> ATA pio2   : 02040600
> XLB Arb cnf: 0000a366
> mpc5xxx_ide: Setting up IDE interface ide0...
> ATA DMA task: 5
> Probing IDE interface ide0...
> SCSI subsystem driver Revision: 1.00
> Icecube-0: Found 1 x8 devices at 0x0 in 8-bit bank
> Icecube-0: Found 1 x8 devices at 0x800000 in 8-bit bank
>  Amd/Fujitsu Extended Query Table at 0x0040
> Icecube-0: CFI does not contain boot bank location. Assuming top.
> number of CFI chips: 2
> cfi_cmdset_0002: Disabling erase-suspend-program due to code brokenness.
> Icecube flash bank 0: Using static image partition definition
> Creating 5 MTD partitions on "Icecube-0":
> 0x00000000-0x00800000 : "Spare"
> 0x00800000-0x00900000 : "kernel"
> 0x00900000-0x00c00000 : "initrd"
> 0x00c00000-0x00f00000 : "jffs"
> 0x00f00000-0x01000000 : "Firmware"
> usb.c: registered new driver hub
> host/usb-ohci.c: USB OHCI at membase 0xf0001000, IRQ 44
> host/usb-ohci.c: usb-0, Built-In ohci
> usb.c: new USB bus registered, assigned bus number 1
> hub.c: USB hub found
> hub.c: 1 port detected
> bluetooth.c: USB Bluetooth support registered
> usb.c: registered new driver bluetty
> bluetooth.c: USB Bluetooth tty driver v0.13
> Initializing USB Mass Storage driver...
> usb.c: registered new driver usb-storage
> USB Mass Storage support registered.
> i2c-icecube.o: I2C module #2 installed
> NET4: Linux TCP/IP 1.0 for NET4.0
> IP Protocols: ICMP, UDP, TCP, IGMP
> IP: routing cache hash table of 512 buckets, 4Kbytes
> TCP: Hash tables configured (established 4096 bind 8192)
> eth0: Phy @ 0x0, type LXT971 (0x001378e2)
> IP-Config: Guessing netmask 255.0.0.0
> IP-Config: Complete:
>       device=eth0, addr=10.3.67.99, mask=255.0.0.0, gw=10.3.67.1,
>      host=10.3.67.99, domain=, nis-domain=(none),
>      bootserver=10.3.67.98, rootserver=10.3.67.98, rootpath=
> NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
> Looking up port of RPC 100003/2 on 10.3.67.98
> Looking up port of RPC 100005/1 on 10.3.67.98
> VFS: Mounted root (nfs filesystem).
> Freeing unused kernel memory: 80k init
> nfs: server 10.3.67.98 not responding, still trying
> nfs: server 10.3.67.98 OK
> nfs: server 10.3.67.98 not responding, still trying
> nfs: server 10.3.67.98 OK
> [time to this about 5 minutes]
> INIT: version 2.85 booting
> nfs: server 10.3.67.98 not responding, still trying
> nfs: server 10.3.67.98 OK
> nfs: server 10.3.67.98 not responding, still trying
> nfs: server 10.3.67.98 OK
> 
> Regards,
> Bjorn Ostby
> _______________________________________________
> Linuxppc-embedded mailing list
> Linuxppc-embedded@ozlabs.org
> https://ozlabs.org/mailman/listinfo/linuxppc-embedded
> 

Re: Lite5200 NFS mount issue

[Wolfgang Denk]

In message <BF8F9BE6CD17EB4B8C887AB03E6A9B0303F03C@CORPAPPL020.corp.saab.se> you wrote:
>
> I'm using u-boot 1.1.4 (CVS downloaded 2006-05-04) and ELDK 4.0. I've encountered 
> problems when trying to mount a NFS root filesystem when using the Lite MPC5200

Which version of the Linux kernel is this? Make sure to  use  top  of
tree in the git repository.

Alternatively, use ELDK 3.1 1 which is better suited for a 2.4 kernel
tree.

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@denx.de
core error - bus dumped

Re: Lite5200 NFS mount issue

[Wolfgang Denk]

In message <200606201503.27928.kimmo.surakka@oxinst.fi> you wrote:
> 
> Try adding the ",tcp" to the kernel's "nfsroot=..." parameter, i.e. use 
> commandline "root=/dev/nfs rw nfsroot=10.3.67.98:/opt/eldk_4_0/ppc_6xx,tcp 
> ip=10.3.67.99:10.3.67.98:10.3.67.1:::eth0:off panic=1".

This should not be necessary. If it does help,  this  means  you  are
just covering a bug.

Best regards,

Wolfgang Denk

-- 
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@denx.de
Those who hate and fight must stop themselves -- otherwise it is  not
stopped.
	-- Spock, "Day of the Dove", stardate unknown

Re: Lite5200 NFS mount issue

[Kimmo Surakka]

On Tuesday 20 June 2006 22:30, Wolfgang Denk wrote:
> In message <200606201503.27928.kimmo.surakka@oxinst.fi> you wrote:
> > Try adding the ",tcp" to the kernel's "nfsroot=..." parameter, i.e. use
> > commandline "root=/dev/nfs rw
> > nfsroot=10.3.67.98:/opt/eldk_4_0/ppc_6xx,tcp
> > ip=10.3.67.99:10.3.67.98:10.3.67.1:::eth0:off panic=1".
>
> This should not be necessary. If it does help,  this  means  you  are
> just covering a bug.

Yes, it may be just covering a bug. Still, it made the connection work, and 
that was enough for me. Also, the linux NFS faq at nfs.sourceforge.net 
suggests using TCP in some situations (namely when the network has a large 
number of IP fragmentation). So, be it a mere work-around or not, it still 
helped me. If there's a better and purer solution, I'm more than happy to use 
it. Meanwhile, I use what works.

Cheers,

Kimmo Surakka

This message has been scanned by F-Secure Anti-Virus

###  OXFORD INSTRUMENTS   http://www.oxford-instruments.com/  ###

Unless stated above to be non-confidential, this E-mail and any
attachments are private and confidential and are for the addressee
only and may not be used, copied or disclosed save to the addressee.
If you have received this E-mail in error please notify us upon receipt
and delete it from your records. Internet communications are not secure
and Oxford Instruments is not responsible for their abuse by third
parties nor for any alteration or corruption in transmission.

RE: Lite5200 NFS mount issue

[Andersson Tord]

Hi,

I have some additional information on the problems we had with NFS on =
the Lite5200 card and the=20
2.4.25 kernel (DENX linuxppc_2_4_devel 2006-04-06) .=20
When analyzing the IP traffic with Etherreal, it was seen that the =
problems were=20
caused by loss of fragmented IP-packets. When the IP fragmentation was =
removed by forcing NFS's=20
UDP size to less than the Ethernet MTU size, the problems disappeared.=20
We used the kernel NFS argument option to this effect;

	nfsroot=3D${serverip}:${rootpath},rsize=3D1024,wsize=3D1024

Still, the question remains, why do fragmented packets get lost? Do =
anyone know of any related FEC/BESTCOMM/driver/buffer issues which might =
cause this problem? The problem also exist with the 5200B version.=20

Regards,

Tord Andersson

=20

> -----Original Message-----
> From:
> linuxppc-embedded-bounces+tord.andersson=3Dcombitechsystems.com@
> ozlabs.org
> [mailto:linuxppc-embedded-bounces+tord.andersson=3Dcombitechsyst
> ems.com@ozlabs.org] On Behalf Of Bj=F6rn =D6stby
> Sent: den 20 juni 2006 10:15
> To: linuxppc-embedded@ozlabs.org
> Subject: Lite5200 NFS mount issue
>=20
>=20
> Hi all,
> I'm using u-boot 1.1.4 (CVS downloaded 2006-05-04) and ELDK
> 4.0. I've encountered=20
> problems when trying to mount a NFS root filesystem when=20
> using the Lite MPC5200 development board. The system mounts=20
> ok when the board is connected to office wide=20
> network switches but if the board is connected directly=20
> through a small 100/10Mbit switch to the NFS server, the=20
> mount halts as shown in the log below (The 100/10Mbit switch=20
> do work as other boards can mount their filesystem through=20
> these). Does anyone recognize this problem and know a=20
> suitable work-around/solution?
>=20
> =3D>run flash_nfs
>  ## Booting image at ff0a0000 ...
>    Image Name:   Linux-2.4.25
>    Created:      2006-06-16  13:11:56 UTC
>    Image Type:   PowerPC Linux Kernel Image (gzip compressed)
>    Data Size:    945550 Bytes =3D 923.4 kB
>    Load Address: 00000000
>    Entry Point:  00000000
>    Verifying Checksum ... OK
>    Uncompressing Kernel Image ... OK
> Memory BAT mapping: BAT2=3D64Mb, BAT3=3D0Mb, residual: 0Mb
> Linux version 2.4.25 (tord@toshiba) (gcc version 4.0.0 (DENX
> ELDK 4.0 4.0.0)) #1 fre jun 16 14:46:35 CEST 2006 On node 0=20
> totalpages: 16384
> zone(0): 16384 pages.
> zone(1): 0 pages.
> zone(2): 0 pages.
> Kernel command line: root=3D/dev/nfs rw=20
> nfsroot=3D10.3.67.98:/opt/eldk_4_0/ppc_6xx=20
> ip=3D10.3.67.99:10.3.67.98:10.3.67.1:::eth0:off panic=3D1=20
...
> Looking up port of RPC 100003/2 on 10.3.67.98
> Looking up port of RPC 100005/1 on 10.3.67.98
> VFS: Mounted root (nfs filesystem).
> Freeing unused kernel memory: 80k init
> nfs: server 10.3.67.98 not responding, still trying
> nfs: server 10.3.67.98 OK
> nfs: server 10.3.67.98 not responding, still trying
> nfs: server 10.3.67.98 OK
> [time to this about 5 minutes]
> INIT: version 2.85 booting
> nfs: server 10.3.67.98 not responding, still trying
> nfs: server 10.3.67.98 OK
> nfs: server 10.3.67.98 not responding, still trying
> nfs: server 10.3.67.98 OK
>=20
> Regards,
> Bjorn Ostby
> _______________________________________________
> Linuxppc-embedded mailing list
> Linuxppc-embedded@ozlabs.org
> https://ozlabs.org/mailman/listinfo/linuxppc-embedded
>=20

RE: Lite5200 NFS mount issue

[Edward Jubenville]

Tord Andersson wrote:
> When analyzing the IP traffic with Etherreal, it was seen that
> the problems were caused by loss of fragmented IP-packets. When
> the IP fragmentation was removed by forcing NFS's UDP size to
> less than the Ethernet MTU size, the problems disappeared.
> We used the kernel NFS argument option to this effect;
> 	nfsroot=${serverip}:${rootpath},rsize=1024,wsize=1024

You may not be having a board problem or a kernel problem.  I had exactly
the same problem with an almost identical configuration as you.  Using
Etherreal at various points in my network, I found that an Ethernet switch
in the communication path was dropping packets before the board ever saw
them.  I solved it the same way that you did (with rsize/wsize), and it has
worked fine ever since.

See my post on 7/26/2006, subject "Re: Slow boot with NFS, server not
responding".

Ed Jubenville

RE: Lite5200 NFS mount issue

[Andersson Tord]

Hi,

Thanks for your reply! However, I don't think we have a network or
switch related issue here,=20
as the problem is also found when using a crossed ethernet-cable between
the host and the target.

Best regards,

Tord


>=20
> You may not be having a board problem or a kernel problem.  I=20
> had exactly the same problem with an almost identical=20
> configuration as you.  Using Etherreal at various points in=20
> my network, I found that an Ethernet switch in the=20
> communication path was dropping packets before the board ever=20
> saw them.  I solved it the same way that you did (with=20
> rsize/wsize), and it has worked fine ever since.
>=20
> See my post on 7/26/2006, subject "Re: Slow boot with NFS,=20
> server not responding".
>=20
> Ed Jubenville
>=20
>=20
> _______________________________________________
> Linuxppc-embedded mailing list
> Linuxppc-embedded@ozlabs.org=20
> https://ozlabs.org/mailman/listinfo/linuxppc-embedded
>=20

Re: Lite5200 NFS mount issue

[Grant Likely]

On 9/21/06, Andersson Tord <Tord.Andersson@combitech.se> wrote:
> Hi,
>
> I have some additional information on the problems we had with NFS on the Lite5200 card and the
> 2.4.25 kernel (DENX linuxppc_2_4_devel 2006-04-06) .
> When analyzing the IP traffic with Etherreal, it was seen that the problems were
> caused by loss of fragmented IP-packets. When the IP fragmentation was removed by forcing NFS's
> UDP size to less than the Ethernet MTU size, the problems disappeared.
> We used the kernel NFS argument option to this effect;
>
>         nfsroot=${serverip}:${rootpath},rsize=1024,wsize=1024

After having similar problems, I've had good luck by forcing nfs to
use tcp by appending ',tcp' to the nfsroot= argument.

g.


-- 
Grant Likely, B.Sc. P.Eng.
Secret Lab Technologies Ltd.
grant.likely@secretlab.ca
(403) 399-0195