MVME 2306 almost up and running

MVME 2306 almost up and running

[David De Ridder]

Hi,

just reporting some progress for those who are interested...

The kernel, v2.2.1, correctly boots, but was unable to mount its
NFS root fs. I have tracked the problem down to BOOTP. I do not
know whether this a flaw in the Linux kernel BOOTP method or an
error made by me (if you can clear this up, please do. My ego
can take a couple more hits ;-).

Because I wanted to separate the root fs from the tftpboot directories,
I have to use both the 'hd' tag as the 'rp' (remote path) tag in BOOTP.

The problem was located at the 'rp' tag. I used a full NFS-style name,
'nfshost:/path', while Linux will want to mount 'nfshost:/path' on 
host 'nfshost', instead of '/path' on 'nfshost'. After changing the
'rp' value to '/path', it works fine. My question : then how do I
set an other NFS server than the BOOTP server for the client ?
(I know : nfsaddr can do the job - for Linux clients).

I always thought it was possible to use different hosts.

BTW, the Linux BOOTP kernel sequence doesn't seem to recognize
the domain name, "domain=(none)", any ideas on that ? 

My final question : our board now finally boots, it can mount an
NFS partition as root fs, but does anybody know where on the Internet
I can find a file system tree with the binaries for LinuxPPC ? I mean,
like Slackware, but for LinuxPPC. (We have no other [spare] PPC based
machines, so I need a full binary file tree to set up on the NFS fs).

Kind regards,

+-----------------------------------------------------------------+
|   David 'Septimus' De Ridder   | "Never underestimate the       |  
|                                |  power of the soul unleashed." |
| <david.de.ridder@bitsmart.com> |   - 'Essence Flare', M:TG      |
+-----------------------------------------------------------------+

--
"- His high exaltedness the great Jabba the Hutt has decreed that you are
   to be terminated immediately.
 - Good. I hate long waits."
 Han Solo, 'Star Wars : Return of the Jedi'




[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to  Cc linuxppc-dev  if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]

Re: MVME 2306 almost up and running

[Gerd Knorr]

> Hi,
> 
> My final question : our board now finally boots, it can mount an
> NFS partition as root fs, but does anybody know where on the Internet
> I can find a file system tree with the binaries for LinuxPPC ? I mean,
> like Slackware, but for LinuxPPC. (We have no other [spare] PPC based
> machines, so I need a full binary file tree to set up on the NFS fs).

Just download the basic rpm's and install them on the NFS Server
with "rpm --root=/export/nfsroot4ppc -i *.rpm".

  Gerd


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Michael Meissner]

On Thu, Mar 04, 1999 at 08:51:55PM +0100, Gerd Knorr wrote:
> 
> > Hi,
> > 
> > My final question : our board now finally boots, it can mount an
> > NFS partition as root fs, but does anybody know where on the Internet
> > I can find a file system tree with the binaries for LinuxPPC ? I mean,
> > like Slackware, but for LinuxPPC. (We have no other [spare] PPC based
> > machines, so I need a full binary file tree to set up on the NFS fs).
> 
> Just download the basic rpm's and install them on the NFS Server
> with "rpm --root=/export/nfsroot4ppc -i *.rpm".

I think the point is that the normal rpm's (ie, ftp.linuxppc.{org,com}) won't
work too well on an ppc860, since that machine does not have a FPU, and the
kernel support for emulating fp might not be present or even if present, too
slow.

One thought is to build a cross compiler on some other system, and manually
build the compiler and enough tools to run native and then bootstrap the old
fashioned way (you probably want to use the --with-cpu=860 switch when
configuring, so that binaries by default don't reference the fpu).  However,
setting up a cross compiler is somewhat tricky if you've never done it before.

-- 
Michael Meissner, Cygnus Solutions
PO Box 98, Ayer Massachusetts, USA 01432-0098
meissner@cygnus.com

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Gabriel Paubert]

On Thu, 4 Mar 1999, Michael Meissner wrote:

> I think the point is that the normal rpm's (ie, ftp.linuxppc.{org,com}) won't
> work too well on an ppc860, since that machine does not have a FPU, and the
> kernel support for emulating fp might not be present or even if present, too
> slow.

The MVME23xx is 603 or 604 based. 

	Gabriel.




[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[David De Ridder]

On Thu, 4 Mar 1999, Michael Meissner wrote:

> On Thu, Mar 04, 1999 at 08:51:55PM +0100, Gerd Knorr wrote:
> > 
> > > My final question : our board now finally boots, it can mount an
> > > NFS partition as root fs, but does anybody know where on the Internet
> > > I can find a file system tree with the binaries for LinuxPPC ? I mean,
> > > like Slackware, but for LinuxPPC. (We have no other [spare] PPC based
> > > machines, so I need a full binary file tree to set up on the NFS fs).
> > 
> > Just download the basic rpm's and install them on the NFS Server
> > with "rpm --root=/export/nfsroot4ppc -i *.rpm".
> 
> I think the point is that the normal rpm's (ie, ftp.linuxppc.{org,com}) won't
> work too well on an ppc860, since that machine does not have a FPU, and the
> kernel support for emulating fp might not be present or even if present, too
> slow.

 Mmmh, our board, an MVME2306 has a PPC604, so I don't see a problem
 there.

> One thought is to build a cross compiler on some other system, and manually
> build the compiler and enough tools to run native and then bootstrap the old
> fashioned way (you probably want to use the --with-cpu=860 switch when
> configuring, so that binaries by default don't reference the fpu).  However,
> setting up a cross compiler is somewhat tricky if you've never done it before.

 I did build a cross-gcc and cross-binutils, and yes, this is *extremely*
 tricky. The docs are little bit too sparse on how to _correctly_ build
 the cross-tools. Anyway, my first attempts to build a PPC kernel from
 the standard Linux source tree failed. So I really want a native gcc
 on the board...

 Regards,

+-----------------------------------------------------------------+
|   David 'Septimus' De Ridder   | "Never underestimate the       |  
|                                |  power of the soul unleashed." |
| <david.de.ridder@bitsmart.com> |   - 'Essence Flare', M:TG      |
+-----------------------------------------------------------------+

--
"Orcs are not exactly known for their valor - although
 most Orcs have seen countless battles, only a handful 
 have actually fought in them."
 - 'Orcish Veteran' (M:TG)


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[David De Ridder]

On Thu, 4 Mar 1999, Gerd Knorr wrote:

> Just download the basic rpm's and install them on the NFS Server
> with "rpm --root=/export/nfsroot4ppc -i *.rpm".

 Thanks ! I'll try this ASAP (and if it fails, I'll come back
 whining ;-)

 Regards,

+-----------------------------------------------------------------+
|   David 'Septimus' De Ridder   | "Never underestimate the       |  
|                                |  power of the soul unleashed." |
| <david.de.ridder@bitsmart.com> |   - 'Essence Flare', M:TG      |
+-----------------------------------------------------------------+

--
"Orcs are not exactly known for their valor - although
 most Orcs have seen countless battles, only a handful 
 have actually fought in them."
 - 'Orcish Veteran' (M:TG)


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Gabriel Paubert]

On Thu, 4 Mar 1999, David De Ridder wrote:

> 
> Hi,
> 
> just reporting some progress for those who are interested...
> 
> The kernel, v2.2.1, correctly boots, but was unable to mount its
> NFS root fs. I have tracked the problem down to BOOTP. I do not
> know whether this a flaw in the Linux kernel BOOTP method or an
> error made by me (if you can clear this up, please do. My ego
> can take a couple more hits ;-).
> 
> Because I wanted to separate the root fs from the tftpboot directories,
> I have to use both the 'hd' tag as the 'rp' (remote path) tag in BOOTP.
> 
> The problem was located at the 'rp' tag. I used a full NFS-style name,
> 'nfshost:/path', while Linux will want to mount 'nfshost:/path' on 
> host 'nfshost', instead of '/path' on 'nfshost'. After changing the
> 'rp' value to '/path', it works fine. My question : then how do I
> set an other NFS server than the BOOTP server for the client ?
> (I know : nfsaddr can do the job - for Linux clients).

AFAICT, there is no provision to use a different host for the loaded
file (the kernel image) and the root path. You may use the sa tag in your
bootptab file to tell which TFTP server to use and which NFS host to mount
the root file system. 

> I always thought it was possible to use different hosts.

2 hosts, yes, but not 3 (bootp server+ TFTP server for image to download +
server to mount root on NFS). The last 2 must be the same if I understand
correctly, or at least from the code in the current linux implementation. 
(Although the RFC are quite nebulous in this respect, see RFC1533 and tell
me if you understand it differently). 

> BTW, the Linux BOOTP kernel sequence doesn't seem to recognize
> the domain name, "domain=(none)", any ideas on that ? 

Don't mix up DNS domain names and YP domain names. Yes, I know it's
confusing and I'm not yet sure to have understood all the subtleties
(I'm not a network guru by any stretch of the imagination):
hostname 
hostname -d <=> dnsdomainname
hostname -f <=> hostname.dnsdomainname
hostname -y <=> ypdomainname

which is/are used it depends on your /etc/nsswitch.conf file.

> NFS partition as root fs, but does anybody know where on the Internet
> I can find a file system tree with the binaries for LinuxPPC ? I mean,
> like Slackware, but for LinuxPPC. (We have no other [spare] PPC based
> machines, so I need a full binary file tree to set up on the NFS fs).

I would recommend for now Linux/PPC R4, but with some compiler/libraries
upgrade. Find a well stocked mirror nearby, for me it's Spain or
Switzerland (I never got to upload even a minimal R5 release since it 
seems RPM formats have changed, but I did not try very hard). 

	Regards,
	Gabriel.


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[David De Ridder]

On Fri, 5 Mar 1999, Gabriel Paubert wrote:

> On Thu, 4 Mar 1999, David De Ridder wrote:
> > 
> > The problem was located at the 'rp' tag. I used a full NFS-style name,
> > 'nfshost:/path', while Linux will want to mount 'nfshost:/path' on 
> > host 'nfshost', instead of '/path' on 'nfshost'. After changing the
> > 'rp' value to '/path', it works fine. My question : then how do I
> > set an other NFS server than the BOOTP server for the client ?
> > (I know : nfsaddr can do the job - for Linux clients).
> 
> AFAICT, there is no provision to use a different host for the loaded
> file (the kernel image) and the root path. You may use the sa tag in your
> bootptab file to tell which TFTP server to use and which NFS host to mount
> the root file system. 

 I see.

> > I always thought it was possible to use different hosts.
> 
> 2 hosts, yes, but not 3 (bootp server+ TFTP server for image to download +
> server to mount root on NFS). The last 2 must be the same if I understand
> correctly, or at least from the code in the current linux implementation. 

 You might be right : I may have mixed up those two. It has been a couple
 of months since I last read the RFCs.

> (Although the RFC are quite nebulous in this respect, see RFC1533 and tell
> me if you understand it differently). 

 I'll reread the concerning RFCs, and try to find a conclusive answer.

> > BTW, the Linux BOOTP kernel sequence doesn't seem to recognize
> > the domain name, "domain=(none)", any ideas on that ? 
> 
> Don't mix up DNS domain names and YP domain names. Yes, I know it's
> confusing and I'm not yet sure to have understood all the subtleties
> (I'm not a network guru by any stretch of the imagination):
> hostname 
> hostname -d <=> dnsdomainname
> hostname -f <=> hostname.dnsdomainname
> hostname -y <=> ypdomainname
> 
> which is/are used it depends on your /etc/nsswitch.conf file.

 Thanks for the explanation. Aside from that, it showed an error
 on my server : seems like I did not set the DNS domainname correctly.
 (The bootp server does have a tag for it, but the server doesn't know
  its own domain - stupid me).

> > NFS partition as root fs, but does anybody know where on the Internet
> > I can find a file system tree with the binaries for LinuxPPC ? I mean,
> > like Slackware, but for LinuxPPC. (We have no other [spare] PPC based
> > machines, so I need a full binary file tree to set up on the NFS fs).
> 
> I would recommend for now Linux/PPC R4, but with some compiler/libraries
> upgrade. Find a well stocked mirror nearby, for me it's Spain or
> Switzerland (I never got to upload even a minimal R5 release since it 
> seems RPM formats have changed, but I did not try very hard). 

 OK, I'll start with an R4, and then upgrade when necessary.

 Kind regards,

+-----------------------------------------------------------------+
|   David 'Septimus' De Ridder   | "Never underestimate the       |  
|                                |  power of the soul unleashed." |
| <david.de.ridder@bitsmart.com> |   - 'Essence Flare', M:TG      |
+-----------------------------------------------------------------+

--
"I want another cup of coffffeeee !"
 - drunk in 'Some like it hot' (1959)


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Michael Meissner]

On Fri, Mar 05, 1999 at 11:21:53AM +0100, Gabriel Paubert wrote:
> 
> 
> On Thu, 4 Mar 1999, Michael Meissner wrote:
> 
> > I think the point is that the normal rpm's (ie, ftp.linuxppc.{org,com}) won't
> > work too well on an ppc860, since that machine does not have a FPU, and the
> > kernel support for emulating fp might not be present or even if present, too
> > slow.
> 
> The MVME23xx is 603 or 604 based. 

Right you are, sorry about that.

-- 
Michael Meissner, Cygnus Solutions
PO Box 98, Ayer Massachusetts, USA 01432-0098
meissner@cygnus.com

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Michael Lundkvist]

David De Ridder <david.de.ridder@bitsmart.com> writes:

> Michael Lundkvist, with bold words breaking the horrid silence thus began :
> 
> >Do you have any instructions on how to do it? How difficult is it to
> >set up a crosscompile environment on an x86 Linux box?
> 
>  Well, I tried setting up a cross-gcc with cross-binutils on my Linux PC.
>  It was a nightmare. I finally could determine the correct procedure, but the
>  cross-compilation of the kernel itself failed. So I started looking for
>  a precompiled kernel (less headaches).
> 

To bad. It is kind of hard to debug when you can't even compile your
own kernel. 


>  3) Again, using the ENV command, locate the setting 
>     "VME3PCI Master Master Enable" and set it to N[o]. This one causes the
> kernel
>     to hang during boot.

Aaaaah! This one helped a lot!

But I'm not quite there yet. Now the kernel panics when it tries to
configure the ethernet devices.

It looks to me like it dies in net/ipv4/ipconfig.c:ic_open_devs since
i see the line "IP-Config: Entered." but I never see the 
"IP-Config: Opened %s (able=%d)\n", dev->name, able).

I'm attaching the boot log and I'm cc-ing to linuxppc-dev to see if
anyone there has any help to give me.

I'm using Gabriel Pauberts zImage.nfsroot-2.2.1 with the preploader.

My board is a MVME 2307SC-1 with 64MB RAM and a DEC 21143-TD etherne chip.

Is there anymore info I can provide. I feel like I know absolutely
nothing about this hardware.

/Micke

--

Michael Lundkvist
Software Designer
Epact Technology AB
ml@epact.se

Bootup info:
----------------------------------------------

 
Residual-Data Located at: $03F88000
 
Model:                            (e2)
Serial: MOT0333B9B  
Processor/Bus frequencies (Hz): 300026368/66668960
Time Base Divisor: 4000
Original MSR: 3040
Original HID0: 82
Original R31: 0
 
PCI: Probing PCI hardware
 
Linux/PPC load: console=ttyS0 root=/dev/nfs nfsroot=150.132.80.47:/export/home/proj_y7/ nfsaddrs=150.132.80.68:150.132.80.47:150.132.80.1:255.255.255.0:tss25:eth0:off
 
Uncompressing the kernel...Kernel at 0x00000000, size=0x1007bc
Initrd at 0x00000000, size=0x0
Residual data at 0x00101000
Command line at 0x00108000
done
Now booting...
Total memory = 64MB; using 256kB for hash table (at c0140000)
Linux version 2.2.1 (root@vcorr1) (gcc version egcs-2.91.60 19981201 (egcs-1.1.1 release)) #47 Fri Feb 5 11:51:47 WET 1999
OpenPIC found at d1000000.
Boot arguments: console=ttyS0 root=/dev/nfs nfsroot=150.132.80.47:/export/home/proj_y7/ nfsaddrs=150.132.80.68:150.132.80.47:150.132.80.1:255.255.255.0:tss25:eth0:off
OpenPIC Version ? (2 CPUs and 17 IRQ sources) at d1000000
OpenPIC Vendor 0 (Motorola), Device 0 (Raven), Stepping 2
OpenPIC timer frequency is 8333620 Hz
request_irq(): irq 16 handler c000bb00 name 82c59 cascade dev_id 0000
request_irq(): irq 2 handler c0005170 name cascade dev_id 0000
time_init: decrementer frequency = 999956280/60 (15MHz)
IP-Config: Parameter #0: `150.132.80.68'
IP-Config: Parameter #1: `150.132.80.47'
IP-Config: Parameter #2: `150.132.80.1'
IP-Config: Parameter #3: `255.255.255.0'
IP-Config: Parameter #4: `tss25'
IP-Config: Parameter #5: `eth0'
IP-Config: Parameter #6: `off'
Calibrating delay loop... 199.88 BogoMIPS
Memory: 63600k available (648k kernel code, 1228k data, 60k init) [c0000000,c4000000]
POSIX conformance testing by UNIFIX
PCI: Probing PCI hardware
Setting bridge 0
Interrupt mapping from 20 to 80
Linux NET4.0 for Linux 2.2
Based upon Swansea University Computer Society NET3.039
NET4: Linux TCP/IP 1.0 for NET4.0
IP Protocols: ICMP, UDP, TCP, IGMP
Starting kswapd v 1.5 
Serial driver version 4.27 with no serial options enabled
ttyS00 at 0x03f8 (irq = 4) is a 16550A
pty: 256 Unix98 ptys configured
eth0: DC21143 at 0x11000 (PCI bus 0, device 14), h/w address 08:00:3e:26:a4:52,
      and requires IRQ18 (provided by PCI BIOS).
de4x5.c:V0.543 1998/12/30 davies@maniac.ultranet.com
IP-Config: Entered.
NIP: C000FB7C XER: 00000000 LR: C000FB60 REGS: c0227cd0 TRAP: 0300
MSR: 00001032 EE: 0 PR: 0 FP: 0 ME: 1 IR/DR: 11
TASK = c0226000[1] 'swapper' mm->pgd c00ba000 Last syscall: 120 
last math 00000000
GPR00: 00001032 C0227DC0 C0226000 00000001 00008000 C01090D4 00000000 0001864F 
GPR08: 00000001 00007800 80000000 000001E0 0000001F 03F824C0 00000000 00000000 
GPR16: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 
GPR24: 00000000 00000000 C00E0000 C00E0000 00000000 00009032 C010976C C0227DC0 
Call backtrace: 
00000000 C009AB34 C00945E0 C0092F60 C0092C98 C0055724 C005678C 
C00D5828 C00D65DC C00D502C C00D6AE0 C00CB720 C0004024 C0008494 
00000000 00000000 
Instruction DUMP: 2c090000  912b0000  41820008 <91690004> 3b800001  48000008  3b800000  73aa8000  38000000 
Kernel panic: kernel access of bad area pc c000fb7c lr c000fb60 address 7804 tsk swapper/1
Rebooting in 180 seconds..


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Gabriel Paubert]

On 7 Mar 1999, Michael Lundkvist wrote:

> Aaaaah! This one helped a lot!
> 
> But I'm not quite there yet. Now the kernel panics when it tries to
> configure the ethernet devices.
> 
> It looks to me like it dies in net/ipv4/ipconfig.c:ic_open_devs since
> i see the line "IP-Config: Entered." but I never see the 
> "IP-Config: Opened %s (able=%d)\n", dev->name, able).
> 
> I'm attaching the boot log and I'm cc-ing to linuxppc-dev to see if
> anyone there has any help to give me.
> 

> My board is a MVME 2307SC-1 with 64MB RAM and a DEC 21143-TD etherne chip.

Matt Porter had a fix for the problems with the 21143, but I don't know
where they can be found. Perhaps it will be simpler now that vger is back
(although I won't switch back to vger for at least a few weeks). 

> Is there anymore info I can provide. I feel like I know absolutely
> nothing about this hardware.

BTW: my driver for the Universe (PCI<->VME bridge chip) is now close to
its final state. It still needs some cleanup, but I hope to be able to ask
for testers next week.  (Even DMA is working now, this was quite a
headache to get right and it took me several iterations to find a simple
and relatively easy to use interface for the callers). 

	Regards,
	Gabriel.



[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Matt Porter]

> Gabriel Paubert wrote:
> > I'm attaching the boot log and I'm cc-ing to linuxppc-dev to see if
> > anyone there has any help to give me.
> > 
> 
> > My board is a MVME 2307SC-1 with 64MB RAM and a DEC 21143-TD etherne chip.
> 
> Matt Porter had a fix for the problems with the 21143, but I don't know
> where they can be found. Perhaps it will be simpler now that vger is back
> (although I won't switch back to vger for at least a few weeks). 

Sorry for the big delay, lost Internet at home for a couple weeks. :-/

All 21143-based boards shipped to this point had a non-compliant DEC SROM
which caused overruns in the driver's private structure that leads to the
crash in del_timer.  I've attached a patch to handle this bad value and
worked with David Davies so it should make it's way into the kernel
eventually.  Newly manufactured boards should be leaving with corrected
SROMs...I'd like to give the person who misread the spec a swift kick in
the knees...turns out that the SROMs from all the PowerPlus boards ever
made are non-compliant only they don't crash the de4x5 driver. 

diff -ud linux-2.1-bak/drivers/net/de4x5.c linux-2.1/drivers/net/de4x5.c
--- linux-2.1-bak/drivers/net/de4x5.c	Tue Jan  5 09:37:38 1999
+++ linux-2.1/drivers/net/de4x5.c	Mon Mar  8 02:30:23 1999
@@ -4794,6 +4794,7 @@
     if (lp->state == INITIALISED) {
         lp->ibn = 3;
         lp->active = *p++;
+	if (MOTO_SROM_BUG) lp->active = 0;
 	lp->phy[lp->active].gep = (*p ? p : 0); p += (2 * (*p) + 1);
 	lp->phy[lp->active].rst = (*p ? p : 0); p += (2 * (*p) + 1);
 	lp->phy[lp->active].mc  = TWIDDLE(p); p += 2;
@@ -4805,6 +4806,7 @@
     } else if ((lp->media == INIT) && (lp->timeout < 0)) {
         lp->ibn = 3;
 	lp->active = *p;
+	if (MOTO_SROM_BUG) lp->active = 0;
 	lp->infoblock_csr6 = OMR_MII_100;
 	lp->useMII = TRUE;
 	lp->infoblock_media = ANS;
diff -ud linux-2.1-bak/drivers/net/de4x5.h linux-2.1/drivers/net/de4x5.h
--- linux-2.1-bak/drivers/net/de4x5.h	Tue Jan  5 09:37:40 1999
+++ linux-2.1/drivers/net/de4x5.h	Mon Mar  8 02:30:43 1999
@@ -599,6 +599,11 @@
 #define BLOCK0_MC      0x3f        /* Media Code */
 
 /*
+** SROM fixup
+*/
+#define MOTO_SROM_BUG	((lp->active == 8) &&
((le32_to_cpu(get_unaligned((u32 *)dev->dev_addr))&0x00ffffff)==0x3e0008))
+
+/*
 ** DC21040 Full Duplex Register (DE4X5_FDR)
 */
 #define FDR_FDACV  0x0000ffff      /* Full Duplex Auto Configuration
Value */

--
Matt Porter
mmporter@home.com
This is Linux Country. On a quiet night, you can hear Windows reboot.


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Michael Lundkvist]

Matt Porter <mmporter@home.com> writes:

> All 21143-based boards shipped to this point had a non-compliant DEC SROM
> which caused overruns in the driver's private structure that leads to the
> crash in del_timer.  I've attached a patch to handle this bad value and
> worked with David Davies so it should make it's way into the kernel
> eventually.  Newly manufactured boards should be leaving with corrected
> SROMs...I'd like to give the person who misread the spec a swift kick in
> the knees...turns out that the SROMs from all the PowerPlus boards ever
> made are non-compliant only they don't crash the de4x5 driver. 
> 

I just tried your fix on a MVME2307 with the DEC21143. I'm using
2.2.1+Gabriels patch. I still see the crash in del_timer as called
from disable_ast. The function type1_infoblock() isn't even called as
far as I can see.


/Micke

--
Michael Lundkvist
Epact Technology AB
ml@epact.se

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Michael Lundkvist]

Just ignore my last mail. I messed up the patch. 

This works just great!!! Thanks! 

One more win for the LinuxPPC-developers. Now I'll put Linux on every
MVME2307 card I can find around here! :)

/Micke

--
Michael Lundkvist
Epact Technology AB
ml@epact.se


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[David De Ridder]

On 21 Mar 1999, Michael Lundkvist wrote:

> This works just great!!! Thanks! 
> 
> One more win for the LinuxPPC-developers. Now I'll put Linux on every
> MVME2307 card I can find around here! :)

Nice. I wish I could say the same for our MVME2306...
I now strongly suspect NFS of obstructing progress ;-)
I also believe I am doing something incredibly stupid, and being
even more stupid not to see what's wrong. So if anyone can help me,
go ahead.

OK, so here's how far I've come :

- The MVME2306 correctly finds its BOOTP and TFTP server (they are the
  same). This very server is a Linux/i386 with 2.0.36 installed, and
  also serves as NFS server for the board's root file system.
- After the BOOTP phase, the board fetches Gabriels kernel (2.2.1 with
  NFS root support) with TFTP.
- Thanks to prepboot, I am able to type ``root=/dev/nfs'' at the
  boot prompt.
- Now, the Linux kernel starts. It initializes eth0 etc., and then
  starts its own BOOTP sequence. The data shown in ``IP-Config:''
  makes sense, the last line being :
  ``IP-Config: host=143.169.5.29, domain=(none), path=/export/''

  And indeed, the board address should be this, and the path on the
  NFS server (which is also TFTP server) to mount as root, is /export/.
- The kernel now starts an RPC sequence, contacts portmap (I guess)
  on the NFS server, and then says :
  ``VFS: Mounted root (NFS filesystem) read-only''.
- In the server's /var/adm/messages, I also get (I stripped the dates) :
  ``elpc6 mountd[67]: NFS mount of /export attempted from 143.169.5.29
    elpc6 mountd[67]: /export has been mounted by 143.169.5.29''

   If my /etc/exports is incorrect, I get the expected messages in syslog.
   But this is not the case now, since the board is valid client.
 - However, then, after initializing the serial console (I think), the
   board says :
   ``Kernel panic: no init found. Try passing init= option to kernel.''

So, I'm stuck. FTR, I tried using the init= option, e.g. ``init=/bin/sh''.
The relevant line in my /etc/exports is :

``/export/        elvme1.uia.ac.be(rw,no_root_squash,insecure)''

The insecure option is AFAIK not necessary, and initially I left it out.
I've also tried ``nfsroot=/export'', but with no effect.

One other thing : I've tried exporting the same filesystem (/export/) to
DECstation 3100. Something strange goes on : I export it rw, when I mount
it, the DEC recognizes it as rw, but I only have ro access ! I can
browse directories, read files, but NOT create files or change their
timestamp.  Maybe this is a DEC problem (it's pretty old), or caused by
the reason I don't know. 

Any help is appreciated,

+-----------------------------------------------------------------+
|   David 'Septimus' De Ridder   | "Never underestimate the       |  
|                                |  power of the soul unleashed." |
| <david.de.ridder@bitsmart.com> |   - 'Essence Flare', M:TG      |
+-----------------------------------------------------------------+

--
"Every dog has its day, and today's mine."
 Terry Borer after winning the Masters finals at the New York M:TG Pro Tour 


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Michael Lundkvist]

David De Ridder <s970707@uia.ua.ac.be> writes:

> - In the server's /var/adm/messages, I also get (I stripped the dates) :
>   ``elpc6 mountd[67]: NFS mount of /export attempted from 143.169.5.29
>     elpc6 mountd[67]: /export has been mounted by 143.169.5.29''
> 
>    If my /etc/exports is incorrect, I get the expected messages in syslog.
>    But this is not the case now, since the board is valid client.
>  - However, then, after initializing the serial console (I think), the
>    board says :
>    ``Kernel panic: no init found. Try passing init= option to kernel.''
> 
> So, I'm stuck. FTR, I tried using the init= option, e.g. ``init=/bin/sh''.
> The relevant line in my /etc/exports is :
> 
> ``/export/        elvme1.uia.ac.be(rw,no_root_squash,insecure)''
> 
> The insecure option is AFAIK not necessary, and initially I left it out.
> I've also tried ``nfsroot=/export'', but with no effect.
> 
> One other thing : I've tried exporting the same filesystem (/export/) to
> DECstation 3100. Something strange goes on : I export it rw, when I mount
> it, the DEC recognizes it as rw, but I only have ro access ! I can
> browse directories, read files, but NOT create files or change their
> timestamp.  Maybe this is a DEC problem (it's pretty old), or caused by
> the reason I don't know. 
> 
> Any help is appreciated,
> 

Could it be a problem with library versions? Both /bin/sh and
/sbin/init are dynamically linked against libc and I remeber having
problems when I accidently used a libc from R5 with a binary from R4.

Here is the versions I run now:

[root@tss236 /root]# rpm -q SysVinit-2.76-2 glibc
SysVinit-2.76-2
glibc-2.1-1a



/Micke

--
Michael Lundkvist
Epact Technology AB
ml@epact.se

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[David De Ridder]

On 22 Mar 1999, Michael Lundkvist wrote:

> Could it be a problem with library versions? Both /bin/sh and
> /sbin/init are dynamically linked against libc and I remeber having
> problems when I accidently used a libc from R5 with a binary from R4.

 All files I downloaded are from an R4. However, I have some new info.
 I started rpc.nfsd with the '-l' option. And this is what shows up in
 /var/adm/messages :

 ``Mar 22 12:53:26 elpc6 nfsd[5351]: 143.169.5.29 < /export/sbin/init
   Mar 22 12:53:26 elpc6 nfsd[5351]: 143.169.5.29 < /export/bin/bash''

 So, the kernel seems to be requesting those files (and receiving them ?).
 You could be right about the dynamic libraries. But, before I reinstall
 the entire root fs, I want to make sure *that* is the problem. Does
 any of you kind people have a statically linked program I could use to
 test, e.g. a Hello World program ?

 BTW, the 'file' commando recognizes PPC binaries. But can it make a 
 distinction between statically and dynamically linked binaries (I know
 it can for i386 arch) ? E.g., when I do a 'file' on my PPC init :

 ``sbin/init: ELF 32-bit MSB executable, PowerPC or cisco 4500, version 1''

 and on my i386 /sbin/init :

 ``/sbin/init: ELF 32-bit LSB executable, Intel 80386, version 1,
               statically linked, stripped''

 Thank you,

+-----------------------------------------------------------------+
|   David 'Septimus' De Ridder   | "Never underestimate the       |  
|                                |  power of the soul unleashed." |
| <david.de.ridder@bitsmart.com> |   - 'Essence Flare', M:TG      |
+-----------------------------------------------------------------+

--
"Every dog has its day, and today's mine."
 Terry Borer after winning the Masters finals at the New York M:TG Pro Tour 


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Michael Lundkvist]

David De Ridder <s970707@uia.ua.ac.be> writes:

> 
>  So, the kernel seems to be requesting those files (and receiving them ?).
>  You could be right about the dynamic libraries. But, before I reinstall
>  the entire root fs, I want to make sure *that* is the problem. Does
>  any of you kind people have a statically linked program I could use to
>  test, e.g. a Hello World program ?
> 

Try /bin/rpm.

>  BTW, the 'file' commando recognizes PPC binaries. But can it make a 
>  distinction between statically and dynamically linked binaries (I know
>  it can for i386 arch) ? E.g., when I do a 'file' on my PPC init :
> 
>  ``sbin/init: ELF 32-bit MSB executable, PowerPC or cisco 4500, version 1''
> 
>  and on my i386 /sbin/init :
> 
>  ``/sbin/init: ELF 32-bit LSB executable, Intel 80386, version 1,
>                statically linked, stripped''
> 
>  Thank you,
> 

[root@tss236 /root]# file /sbin/init /bin/rpm
/sbin/init: ELF 32-bit MSB executable, PowerPC or cisco 4500, version 1, dynamically linked (uses shared libs), stripped
/bin/rpm:   ELF 32-bit MSB executable, PowerPC or cisco 4500, version 1, statically linked, stripped

/Micke

--

Michael Lundkvist
Epact Technology AB
ml@epact.se

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[David De Ridder]

On 22 Mar 1999, Michael Lundkvist wrote:

> David De Ridder <s970707@uia.ua.ac.be> writes:
> 
> > 
> >  So, the kernel seems to be requesting those files (and receiving them ?).
> >  You could be right about the dynamic libraries. But, before I reinstall
> >  the entire root fs, I want to make sure *that* is the problem. Does
> >  any of you kind people have a statically linked program I could use to
> >  test, e.g. a Hello World program ?
> > 
> 
> Try /bin/rpm.

 Thank you ! The dynamic libraries seem to be the problem. You suggest to
 go for an R5 instead of R4 ? I noticed at several FTP sites that there
 are only so-called 'pre-R5' RPMS available. Are those the ones you mean ?

 Kind regards from rainy and traffic-congested Belgium,

+-----------------------------------------------------------------+
|   David 'Septimus' De Ridder   | "Never underestimate the       |  
|                                |  power of the soul unleashed." |
| <david.de.ridder@bitsmart.com> |   - 'Essence Flare', M:TG      |
+-----------------------------------------------------------------+

--
"Only the bamboo bows when the storm grumbles. Steel breaks !"
 - Yemitsu, last words


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Michael Lundkvist]

David De Ridder <s970707@uia.ua.ac.be> writes:

> 
>  Thank you ! The dynamic libraries seem to be the problem. You suggest to
>  go for an R5 instead of R4 ? I noticed at several FTP sites that there
>  are only so-called 'pre-R5' RPMS available. Are those the ones you mean ?
> 
>  Kind regards from rainy and traffic-congested Belgium,
> 

I used binaries from
http://mac.pvv.ntnu.no/pub/linux/linuxppc/linuxppc-pre-R5/software/RedHat/RPMS/ 


Hope it works.

I'm now trying my limited skills on writing a driver for a PMC DSP
card. It's a lot of fun. Too bad I'm probably the only one who has
that hardware. :)

/Micke

--

Michael Lundkvist
Epact Technology AB
ml@epact.se

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[David De Ridder]

On 22 Mar 1999, Michael Lundkvist wrote:

> >  Thank you ! The dynamic libraries seem to be the problem. You suggest to
> >  go for an R5 instead of R4 ? I noticed at several FTP sites that there
> >  are only so-called 'pre-R5' RPMS available. Are those the ones you mean ?
> > 
> >  Kind regards from rainy and traffic-congested Belgium,
> > 
> 
> I used binaries from
> http://mac.pvv.ntnu.no/pub/linux/linuxppc/linuxppc-pre-R5/software/RedHat/RPMS/ 

 I installed the packages you mentioned early and, as expected, get lots
 of errors because rpm cannot execute the scripts. I understand there
 is some way around this, but how ? Is it possible to extract run
 them manually ?

+-----------------------------------------------------------------+
|   David 'Septimus' De Ridder   | "Never underestimate the       |  
|                                |  power of the soul unleashed." |
| <david.de.ridder@bitsmart.com> |   - 'Essence Flare', M:TG      |
+-----------------------------------------------------------------+

--
"Only the bamboo bows when the storm grumbles. Steel breaks !"
 - Yemitsu, last words


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME 2306 almost up and running

[Michael Lundkvist]

David De Ridder <s970707@uia.ua.ac.be> writes:

> 
>  I installed the packages you mentioned early and, as expected, get lots
>  of errors because rpm cannot execute the scripts. I understand there
>  is some way around this, but how ? Is it possible to extract run
>  them manually ?
> 

Most of the scripts can just be ignored. Check with --scripts to see
what it does. If it's just an ldconfig or similar you can ignore it.

/Micke

--

Michael Lundkvist
Epact Technology AB
ml@epact.se

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

MVME2306 problems dragging on

[David De Ridder]

Hello everyone,

I'm still trying to get our board to run LinuxPPC. So far, I've come to
booting the Linux kernel. This kernel can also mount its NFS fs.

There seems to be a problem with the serial line settings. After the
kernel says : ``Freeing unused memory : [...] open firmware'', the
serial console gets flooded with two kinds of similar messages :

``request_irq(): irq 4 handler c008dae8 name serial dev_id 0000`` and
``request_irq(): irq 4 handler c0000000 name <NULL> dev_id 0000``.

BTW, I'm using a /dev/console with major 5 minor 1. Inbetween those
messages, I see some errors that seem to come from init. I have no update
daemon, and bash cannot be loaded because it cannot find its libtermcap.
When I manually make a link to libtermcap, there are no more errors,
but also no more screen output at all.
  
Does anyone know how to get it correctly running ?

+-----------------------------------------------------------------+
|   David 'Septimus' De Ridder   | "Never underestimate the       |  
|                                |  power of the soul unleashed." |
| <david.de.ridder@bitsmart.com> |   - 'Essence Flare', M:TG      |
+-----------------------------------------------------------------+

--
"I want you to find happiness."
 - Marilyn Monroe in 'Gentlemen prefer Blondes'



[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME2306 problems dragging on

[VALETTE Eric]

>>>>> "David" == David De Ridder <s970707@uia.ua.ac.be> writes:

David> Hello everyone,

David> I'm still trying to get our board to run LinuxPPC. So far, I've come to
David> booting the Linux kernel. This kernel can also mount its NFS fs.

David> There seems to be a problem with the serial line settings. After the
David> kernel says : ``Freeing unused memory : [...] open firmware'', the
David> serial console gets flooded with two kinds of similar messages :

David> ``request_irq(): irq 4 handler c008dae8 name serial dev_id 0000`` and
David> ``request_irq(): irq 4 handler c0000000 name <NULL> dev_id 0000``.

David> BTW, I'm using a /dev/console with major 5 minor 1. Inbetween those
David> messages, I see some errors that seem to come from init. I have no update
David> daemon, and bash cannot be loaded because it cannot find its libtermcap.
David> When I manually make a link to libtermcap, there are no more errors,
David> but also no more screen output at all.
  
David> Does anyone know how to get it correctly running ?

Fisrt I assumes that you are using a serial console to boot? If yes then
change the /etc/inittab file to

1:2345:respawn:/sbin/getty 38400 tty1 => 1:23:respawn:/sbin/getty -L ttyS0 9600 linux

-- 
   __                 
  /  `                   	Eric Valette
 /--   __  o _.          	Canon CRF
(___, / (_(_(__         	Rue de la touche lambert
				35517 Cesson-Sevigne  Cedex
				FRANCE
Tel: +33 (0)2 99 87 68 91	Fax: +33 (0)2 99 84 11 30
E-mail: valette@crf.canon.fr

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME2306 problems dragging on

[David De Ridder]

On Fri, 2 Apr 1999, VALETTE Eric wrote:

> David> serial console gets flooded with two kinds of similar messages :
> 
> David> ``request_irq(): irq 4 handler c008dae8 name serial dev_id 0000`` and
> David> ``request_irq(): irq 4 handler c0000000 name <NULL> dev_id 0000``.
> 
> David> BTW, I'm using a /dev/console with major 5 minor 1. Inbetween those
> David> messages, I see some errors that seem to come from init. I have no update
> David> daemon, and bash cannot be loaded because it cannot find its libtermcap.
> David> When I manually make a link to libtermcap, there are no more errors,
> David> but also no more screen output at all.
>   
> David> Does anyone know how to get it correctly running ?
> 
> Fisrt I assumes that you are using a serial console to boot? If yes then
> change the /etc/inittab file to
> 
> 1:2345:respawn:/sbin/getty 38400 tty1 => 1:23:respawn:/sbin/getty -L ttyS0 9600 linux

 Thanks, that helped a bit. I'm using 

 1:23:respawn:/sbin/getty ttyS0 DT9600 vt220

 since I could not find any agetty (to use the -L option) in RH5.
 So, now the system boots like a real Linux machine, and I *finally* get
 a login prompt (Raw Hide 1.2.7). It says ``localhost login :''.
 Strange enough, when I type anything, the word 'login' appears on the
 next line, and then after a few secs, the login banner resets the screen.
 Also, backspace isn't working. It appears that I can walk through the
 screen. This seems just a simple terminal setting problem. But I don't
 see what's wrong. (BTW, I already added ``ttyS0'' to /etc/securetty).

 Regards,

+-----------------------------------------------------------------+
|   David 'Septimus' De Ridder   | "Never underestimate the       |  
|                                |  power of the soul unleashed." |
| <david.de.ridder@bitsmart.com> |   - 'Essence Flare', M:TG      |
+-----------------------------------------------------------------+

--
"I want you to find happiness."
 - Marilyn Monroe in 'Gentlemen prefer Blondes'


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: MVME2306 problems dragging on

[VALETTE Eric]

>>>>> "David" == David De Ridder <s970707@uia.ua.ac.be> writes:

David> On Fri, 2 Apr 1999, VALETTE Eric wrote:
David> serial console gets flooded with two kinds of similar messages :
>> 
David> ``request_irq(): irq 4 handler c008dae8 name serial dev_id 0000`` and
David> ``request_irq(): irq 4 handler c0000000 name <NULL> dev_id 0000``.
>> 
David> BTW, I'm using a /dev/console with major 5 minor 1. Inbetween those
David> messages, I see some errors that seem to come from init. I have no update
David> daemon, and bash cannot be loaded because it cannot find its libtermcap.
David> When I manually make a link to libtermcap, there are no more errors,
David> but also no more screen output at all.
>> 
David> Does anyone know how to get it correctly running ?
>> 
>> Fisrt I assumes that you are using a serial console to boot? If yes then
>> change the /etc/inittab file to
>> 
>> 1:2345:respawn:/sbin/getty 38400 tty1 => 1:23:respawn:/sbin/getty -L ttyS0 9600 linux

David>  Thanks, that helped a bit. I'm using 

David>  1:23:respawn:/sbin/getty ttyS0 DT9600 vt220

David>  since I could not find any agetty (to use the -L option) in RH5.
David>  So, now the system boots like a real Linux machine, and I *finally* get
David>  a login prompt (Raw Hide 1.2.7). It says ``localhost login :''.
David>  Strange enough, when I type anything, the word 'login' appears on the
David>  next line, and then after a few secs, the login banner resets the screen.
David>  Also, backspace isn't working. It appears that I can walk through the
David>  screen. This seems just a simple terminal setting problem. But I don't
David>  see what's wrong. (BTW, I already added ``ttyS0'' to /etc/securetty).

Other ideas :
      1) Did you also specify console=ttyS0 at boot prompt?
      2) some ksh are buggy and died immediately
      3) uhhhh???

Have a nice week-end...

-- eric

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]