2.6.26 does not boot on Pegasos

	Hi,

I just tried to boot 2.6.26 on a Pegasos and the kernel does not boot.
The last message I have is:
gunzip (0xffffffff <----- some more hex digits)

The configuration has been created from a working 2.6.25 one with
make oldconfig and answering N to new config options.

Anybody has seen this or do I have to start digging deeper?

	Regards,
	Gabriel

On Tue, 2008-07-15 at 12:00 +0200, Gabriel Paubert wrote:

	Hi,

I just tried to boot 2.6.26 on a Pegasos and the kernel does not boot.
The last message I have is:
gunzip (0xffffffff <----- some more hex digits)

The configuration has been created from a working 2.6.25 one with
make oldconfig and answering N to new config options.

Anybody has seen this or do I have to start digging deeper?

Unfortunately, I don't have a pegasos anymore. David, did you get a
chance to test anything recent on yours ? Matt ?

Cheers,
Ben.

On Tue, Jul 15, 2008 at 08:05:49PM +1000, Benjamin Herrenschmidt wrote:

On Tue, 2008-07-15 at 12:00 +0200, Gabriel Paubert wrote:

quoted

	Hi,

I just tried to boot 2.6.26 on a Pegasos and the kernel does not boot.
The last message I have is:
gunzip (0xffffffff <----- some more hex digits)

The configuration has been created from a working 2.6.25 one with
make oldconfig and answering N to new config options.

Anybody has seen this or do I have to start digging deeper?

Unfortunately, I don't have a pegasos anymore. David, did you get a
chance to test anything recent on yours ? Matt ?

I seem to remember that 2.6.26-rc5 booted fine. Compiling it right now
before I start a bisection.

	Regards,
	Gabriel

On Tue, 2008-07-15 at 20:05 +1000, Benjamin Herrenschmidt wrote:

On Tue, 2008-07-15 at 12:00 +0200, Gabriel Paubert wrote:

quoted

	Hi,

I just tried to boot 2.6.26 on a Pegasos and the kernel does not boot.
The last message I have is:
gunzip (0xffffffff <----- some more hex digits)

The configuration has been created from a working 2.6.25 one with
make oldconfig and answering N to new config options.

Anybody has seen this or do I have to start digging deeper?

Unfortunately, I don't have a pegasos anymore. David, did you get a
chance to test anything recent on yours ? Matt ?

I'm away from home for the next couple of weeks, but can test when I get
home after OLS. I've run relatively recent kernels on Pegasos within the
last 2-3 weeks.

-- 
dwmw2

If you built this kernel yourself, you need to do it from a system with
an up-to-date binutils (2.18) otherwise, it does this.

If you got it from somewhere (like a distribution) then please tell us
where, as there are some other troubles with load-base location for
Fedora and so on...

-- 
Matt Sealey [off-list ref]
Genesi, Manager, Developer Relations

Gabriel Paubert wrote:

	Hi,

I just tried to boot 2.6.26 on a Pegasos and the kernel does not boot.
The last message I have is:
gunzip (0xffffffff <----- some more hex digits)

The configuration has been created from a working 2.6.25 one with
make oldconfig and answering N to new config options.

Anybody has seen this or do I have to start digging deeper?

	Regards,
	Gabriel
_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@ozlabs.org
https://ozlabs.org/mailman/listinfo/linuxppc-dev

On Tue, Jul 15, 2008 at 04:27:49PM +0100, Matt Sealey wrote:

If you built this kernel yourself, you need to do it from a system with
an up-to-date binutils (2.18) otherwise, it does this.

Thanks, this is likely the problem. The distribution is Debian stable 
with all security udates but the binutils are still 2.17. I was also 
wondering why the zImage was exploding from 2 to 5MB and why it spent
so much time in gzip at the end of a build.

Trying to add Lenny source does not help, bummer. Aptitude fails 
with an error message like "Dynamic MMap ran out of room".

	Regards,
	Gabriel

Gabriel Paubert wrote:

On Tue, Jul 15, 2008 at 04:27:49PM +0100, Matt Sealey wrote:

quoted

If you built this kernel yourself, you need to do it from a system with
an up-to-date binutils (2.18) otherwise, it does this.

Note to the linux-ppc guys; is there any changelog entry which reports
this requirement somewhere? I didn't find one...

Thanks, this is likely the problem. The distribution is Debian stable 
with all security udates but the binutils are still 2.17.

That's generally the problem we had with SuSE 10.3. Debian stable
should be updated by Christmas.. :D

Trying to add Lenny source does not help, bummer. Aptitude fails 
with an error message like "Dynamic MMap ran out of room".

Ouch. Well, all I can recommend is grabbing the source files from;

http://packages.debian.org/source/testing/binutils

And compiling and installing it manually. There is a bit of a
disparity between glibc support between testing and stable that
the prebuild packages will complain about (you have to install
half of Lenny to do it, which is a bad idea just to update one
package).

Once you have the files you can just use "dpkg -x binutils-blah.dsc"
to get going.

-- 
Matt Sealey [off-list ref]
Genesi, Manager, Developer Relations

quoted

quoted

If you built this kernel yourself, you need to do it from a system 
with
an up-to-date binutils (2.18) otherwise, it does this.

Note to the linux-ppc guys; is there any changelog entry which reports
this requirement somewhere? I didn't find one...

That's because there is no such requirement.  2.18 is the latest
version, so it's always wise to test with that in case you have
some problems; but it wouldn't be nice if the kernel can be built
with only one binutils major version.

Please try my build patch for a similar problem, it went into
BenH's tree today.  Would be nice if it was the same problem;
if not, send a proper, _detailed_ bug report here and I'll look
into it.


Segher

On 7/15/08, Matt Sealey [off-list ref] wrote:

If you built this kernel yourself, you need to do it from a system with
 an up-to-date binutils (2.18) otherwise, it does this.

 If you got it from somewhere (like a distribution) then please tell us
 where, as there are some other troubles with load-base location for
 Fedora and so on...

Is it hard to put a check into the kernel make system to look for
binutils 2.17 and refuse to build with a nice error message? I poked
around in the makefile but I'm not 100% sure of how it works.

 --
 Matt Sealey [off-list ref]
 Genesi, Manager, Developer Relations


 Gabriel Paubert wrote:

quoted

       Hi,

I just tried to boot 2.6.26 on a Pegasos and the kernel does not boot.
The last message I have is:
gunzip (0xffffffff <----- some more hex digits)

The configuration has been created from a working 2.6.25 one with
make oldconfig and answering N to new config options.

Anybody has seen this or do I have to start digging deeper?

       Regards,
       Gabriel
_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@ozlabs.org
https://ozlabs.org/mailman/listinfo/linuxppc-dev

 _______________________________________________
 Linuxppc-dev mailing list
 Linuxppc-dev@ozlabs.org
 https://ozlabs.org/mailman/listinfo/linuxppc-dev

-- 
Jon Smirl
jonsmirl@gmail.com