2.6.10, alpha, "Relocation overflows" - .data.percpu ?

[David Monro]

Hi Alan,

well thats interesting... so my revision of the compiler is 2 revisions 
newer (1:3.3.5-8). My ld also identifies as 2.15, dpkg says binutils is 
at 2.15-5 and thats the only version I can see in the debian pool. This 
is all rather puzzling!

I tried a 2.95 build (with the patch for traps.c) (haven't had time to 
get the old RH 2.96 compiler working yet) and it failed building some of 
the modules..

Cheers,

	David

Alan Young wrote:
> The newer version of GCC should be fine.  I've compiled and run a SMP 2.6.10
> on a DP264 - which is pretty much the same board as your DS20.  I'm currently
> using 3.3.5 from unstable (I think).  gcc identifies itself as Debian 1:3.3.5-6.
> 
> I think the relocation record issue might be due to an old version of binutils.
> I would check to see if you an upgrade that to a more recent version.
> I'm not sure which binutils package I'm currently using.  ld identifies itself as 2.15.
> 
> See this message for a possible fix for the spinlock in 2.6.11:
> https://www.redhat.com/archives/axp-list/2005-February/msg00028.htm
> 
> Alan
> 
> On Wed, 23 Mar 2005 01:19:55 +1030 David Monro <davidm at monro.org.uk> wrote:
> 
> 
>>Ugh. Hate it when I figure out more stuff just after sending the 
>>message. Looks like any module which has a .data.percpu section doesn't 
>>load! Modules with a relocation record of type GPRELHIGH pointing to 
>>.data.percpu get the "Relocation overflow vs section X" message,
>>ones which have the section but no relocation records referring to it 
>>get a "Relocation overflow vs symbolname" error instead.
>>
>>Any clues how to fix that?
>>
>>Cheers,
>>
>>	David
>>
>>David Monro wrote:
>>
>>>Hi,
>>>
>>>ok I know this isn't specifically redhat related, but hey there aren't 
>>>many active alpha lists out there!
>>>
>>>I'm trying to get kernel.org 2.6.10 running on my SMP ds20 alpha system. 
>>>Kernel compiles fine, but _some_ modules fail to load once running, in 
>>>particular the ipv6 and scsi ones (there may be more, but those are the 
>>>ones I'm noticing).
>>>
>>>The error is "Relocation overflow vs section 25" for ipv6 (scsi_mod says 
>>>section 27 instead). Adding a few printks, it seems that its always the 
>>>R_ALPHA_GPRELHIGH case (line 265 of arch/alpha/kernel/module.c), and as 
>>>far as I can tell, the problem is that "value" is more than 0x8000 less 
>>>than "gp" which results in a negative value... The weird thing is, it 
>>>doesn't appear to happen for UP-compiled kernels, only SMP ones!
>>>
>>>This is using gcc version 3.3.5 (Debian 1:3.3.5-8). I tried using gcc 
>>>2.95 (gcc version 2.95.4 20011002 (Debian prerelease)), but that gets a 
>>>parse error on line 55 of arch/alpha/kernel/traps.c. (What compiler 
>>>_should_ I be using? The documentation seems to indicate gcc 2.95.3, but 
>>>I'm finding that hard to believe).
>>>
>>>I can't go forward to 2.6.11 because thats missing _raw_read_trylock for 
>>>alpha, and there doesn't appear to be a patch for that yet. 2.6.8 UP 
>>>works, but SMP just hangs when probing the DAC960 raid controller. Is 
>>>SMP alpha just not very well tested at the moment?
>>>
>>>Cheers,
>>>
>>>    David
>>>
>>>_______________________________________________
>>>axp-list mailing list
>>>axp-list at redhat.com
>>>https://www.redhat.com/mailman/listinfo/axp-list
>>
>>_______________________________________________
>>axp-list mailing list
>>axp-list at redhat.com
>>https://www.redhat.com/mailman/listinfo/axp-list
> 
> 
> _______________________________________________
> axp-list mailing list
> axp-list at redhat.com
> https://www.redhat.com/mailman/listinfo/axp-list