[BISECTED] 164UX Regression, 2.6.22->2.6.23

[BISECTED] 164UX Regression, 2.6.22->2.6.23

[Matt Turner]

[-- Attachment #1: Type: text/plain, Size: 2343 bytes --]

Hello,
Alpha/Linux user Ian Las reported that since 2.6.22, the Kernel fails to load
on his 164UX using MILO. (MILO is the only way to load Linux, as it only has
ARCSBIOS.)

It fails with

...Image loaded into memory, entry @ 0xfffffc0000310000
Setting up Milo reboot parameter block at 0xfffffc0000000140
Boot line is bootdevice=sdb2 bootfile=vmlinux.gz root=/dev/sdb6
Bootstrap complete, kernel @ 0xfffffc0000310000, sp @ 0xfffffc0000304000
...turning on virtual addressing and jumping to the Linux Kernel
--------------------------------------------------------------------------------
Kernel panic - not syncing: Attempted to kill init!

Ian successfully bisected the failing commit, inline below. We tried reverting
the commit, but it causes linker failures due to overlapping sections.

How can we further debug and fix this?

Thanks!
Matt


commit caf45dd92677b2001123cc06b5835052ccfee76f
Author: Roland McGrath <roland@redhat.com>
Date:   Thu Jul 19 01:48:37 2007 -0700

    alpha: Put allocated ELF notes in read-only data segment
    
    This changes the alpha linker script to use the asm-generic NOTES macro so
    that ELF note sections with SHF_ALLOC set are linked into the kernel image
    along with other read-only data.  The PT_NOTE also points to their location.
    
    This paves the way for putting useful build-time information into ELF notes
    that can be found easily later in a kernel memory dump.
    
    Signed-off-by: Roland McGrath <roland@redhat.com>
    Cc: Richard Henderson <rth@twiddle.net>
    Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
    Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
    Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/arch/alpha/kernel/vmlinux.lds.S b/arch/alpha/kernel/vmlinux.lds.S
index 6f4f037..fe13daa 100644
--- a/arch/alpha/kernel/vmlinux.lds.S
+++ b/arch/alpha/kernel/vmlinux.lds.S
@@ -3,7 +3,7 @@
 OUTPUT_FORMAT("elf64-alpha")
 OUTPUT_ARCH(alpha)
 ENTRY(__start)
-PHDRS { kernel PT_LOAD ; }
+PHDRS { kernel PT_LOAD; note PT_NOTE; }
 jiffies = jiffies_64;
 SECTIONS
 {
@@ -28,6 +28,9 @@ SECTIONS
   __ex_table : { *(__ex_table) }
   __stop___ex_table = .;
 
+  NOTES :kernel :note
+  .dummy : { *(.dummy) } :kernel
+
   RODATA
 
   /* Will be freed after init */

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [BISECTED] 164UX Regression, 2.6.22->2.6.23

[Roland McGrath]

Best would be someone who thoroughly understands that boot loader's
expectations examining the vmlinux image to see what's wrong.  I'm
not that someone, but if you send me the images made before and after
that patch, I can examine them and speculate what the problem might be.


Thanks,
Roland

Re: [BISECTED] 164UX Regression, 2.6.22->2.6.23

[Roland McGrath]

> Hi Roland,
> attached are a working vml2622.gz and a failing
> vml2623.gz images.
> As far as possible identical config has been used for compilation.

I was looking for an apples-to-apples comparison, i.e. the same exact
everything except reverting this one patch.  Two entirely different kernel
source versions is not what I meant at all.  But I'll see if I can tell
what is going on from looking at these.


Thanks,
Roland

Re: [BISECTED] 164UX Regression, 2.6.22->2.6.23

[Ian Las]

Hi Roland,
sorry for my mistake but with this patch reverted it is not possible to compile the kernel any more.
Thanks,
Ian

--- Roland McGrath <roland@redhat.com> schrieb am Fr, 1.10.2010:

> Von: Roland McGrath <roland@redhat.com>
> Betreff: Re: [BISECTED] 164UX Regression, 2.6.22->2.6.23
> An: "Ian Las" <ialas@yahoo.com>
> CC: "Matt Turner" <mattst88@gmail.com>, linux-alpha@vger.kernel.org, "Richard Henderson" <rth@twiddle.net>, "Ivan Kokshaysky" <ink@jurassic.park.msu.ru>
> Datum: Freitag, 1. Oktober, 2010 22:37 Uhr
> > Hi Roland,
> > attached are a working vml2622.gz and a failing
> > vml2623.gz images.
> > As far as possible identical config has been used for
> compilation.
> 
> I was looking for an apples-to-apples comparison, i.e. the
> same exact
> everything except reverting this one patch.  Two
> entirely different kernel
> source versions is not what I meant at all.  But I'll
> see if I can tell
> what is going on from looking at these.
> 
> 
> Thanks,
> Roland
> 


--
To unsubscribe from this list: send the line "unsubscribe linux-alpha" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [BISECTED] 164UX Regression, 2.6.22->2.6.23

[Roland McGrath]

There is nothing obviously anomalous in those section headers, just the
expected addition of the .notes section.  The only thing that comes to mind
is if this boot loader is reading the ELF phdrs and barfs on anything that
is not a PT_LOAD, since the PT_NOTE is now there too (which a loader should
be ignoring).  Somebody who understands the boot loader in question will
probably need to look at this.


Thanks,
Roland