Re: [LinuxBIOS] s2881 build broken

[Lu, Yinghai]

That should be double skip length problem..., remove one +4 in
copy_and_run.c could make it work.

YH

-----Original Message-----
From: [e-mail]
[mailto:[e-mail]] On Behalf Of Stefan Reinauer
Sent: Wednesday, May 03, 2006 4:12 PM
To: Ward Vandewege
Cc: [e-mail]
Subject: Re: [LinuxBIOS] s2881 build broken

* Ward Vandewege  [060504 00:39]:
> Hi all,
>
> It seems that revision 2288 broke something in the tyan/s2881 build;

2289 and
> 2290 are also bad. 2287 is ok.
>
> Boot log attached; it just keeps rebooting after the 'Jumping to

LinuxBIOS'
> line. This is gcc 3.4.
>
> Suggestions?

Yes, I suggest I owe you a beer next time we meet and you try 2291.

Yinghai Lu has sent a patch that is more complete but also more
intrusive, which will go in later I guess. Until then this should work
fine.

This one is a good reason for automated testing on hardware.. 8)

Stefan

[LinuxBIOS] s2881 build broken

[Ronald G Minnich]

Lu, Yinghai wrote:
> That should be double skip length problem..., remove one +4 in
> copy_and_run.c could make it work.
>
> YH
>
> -----Original Message-----
> From: [e-mail]
> [mailto:[e-mail]] On Behalf Of Stefan Reinauer
> Sent: Wednesday, May 03, 2006 4:12 PM
> To: Ward Vandewege
> Cc: [e-mail]
> Subject: Re: [LinuxBIOS] s2881 build broken
> Ward Vandewege  [060504 00:39]:
> > Hi all,
> >
> > It seems that revision 2288 broke something in the tyan/s2881 build;
>
> 2289 and
> > 2290 are also bad. 2287 is ok.
> >
> > Boot log attached; it just keeps rebooting after the 'Jumping to
>
> LinuxBIOS'
> > line. This is gcc 3.4.
> >
> > Suggestions?
>
> Yes, I suggest I owe you a beer next time we meet and you try 2291.
>
> Yinghai Lu has sent a patch that is more complete but also more
> intrusive, which will go in later I guess. Until then this should work
> fine.
>
> This one is a good reason for automated testing on hardware.. 8)
>
> Stefan

hey guys, sorry but I'm lost. did the compress patch bust Ward's 
linuxbios? Why would that happen? Did the behaviour default to 'new 
feetchur enabled'?

ron

[LinuxBIOS] s2881 build broken

[Stefan Reinauer]

* Ronald G Minnich  [060504 01:59]:
> hey guys, sorry but I'm lost. did the compress patch bust Ward's 
> linuxbios? Why would that happen? Did the behaviour default to 'new 
> feetchur enabled'?

It was indirectly connected to the new feature. Since we're using the
same compression algorithm as we use for CONFIG_COMPRESS I switched
things so that they share code - 
... and tested on a system with CONFIG_COMPRESS disabled.

The bug was that nrv2b puts the length of the decompressed file in the
first 4 bytes of the result file. unrv2b() skips those. And so does
copy_and_run.c in the assembler part which I missed.

I fixed it in 2291 though.

Stefan