[LinuxBIOS] copy_secondary_start_to_1m_below() and non-smp

[Richard Smith]

While merging in my 440bx patchs to the latest revision I got
unresolved references to the
copy_secondary_start_to_1m_below() function.

if CONFIG_SMP != 1 then this function never gets defined but
initialize_cpus() calls it regardless.

Will copy_secondary_start_to_1m_below() ever be used in the non-smp
case?  If so then it needs a CONFIG_SMP wrapper or needs to get
defined as empty in the #else case of CONFIG_SMP.

[Stefan Reinauer]

* Richard Smith  [060109 07:45]:
> While merging in my 440bx patchs to the latest revision I got
> unresolved references to the
> copy_secondary_start_to_1m_below() function.

This went bad in r2142. See 
http://snapshots.linuxbios.org/stats/commit-LinuxBIOSv2-2142.log
and http://snapshots.linuxbios.org/stats/abuild-LinuxBIOSv2-2142.log
for more details.

Author: yhlu
Date: 2005-12-14 03:39:33 +0100 (Wed, 14 Dec 2005)
New Revision: 2142

Modified:
   trunk/LinuxBIOSv2/src/config/linuxbios_ram.ld
   trunk/LinuxBIOSv2/src/cpu/amd/car/clear_1m_ram.c
   trunk/LinuxBIOSv2/src/cpu/amd/car/post_cache_as_ram.c
   trunk/LinuxBIOSv2/src/cpu/amd/model_fxx/init_cpus.c
   trunk/LinuxBIOSv2/src/cpu/x86/lapic/lapic_cpu_init.c
   trunk/LinuxBIOSv2/src/cpu/x86/lapic/secondary.S
   trunk/LinuxBIOSv2/src/cpu/x86/pae/pgtbl.c
Log:
issue 51 and 52: set mtrr for ap before stop it, and _RAMBASE above 1M
support and pgtbl after 1M support

Stefan

[Ronald G Minnich]

Stefan Reinauer wrote:
> Richard Smith  [060109 07:45]:
> > While merging in my 440bx patchs to the latest revision I got
> > unresolved references to the
> > copy_secondary_start_to_1m_below() function.
>
> This went bad in r2142. See 
> http://snapshots.linuxbios.org/stats/commit-LinuxBIOSv2-2142.log
> and http://snapshots.linuxbios.org/stats/abuild-LinuxBIOSv2-2142.log
> for more details.
>
> Author: yhlu
> Date: 2005-12-14 03:39:33 +0100 (Wed, 14 Dec 2005)
> New Revision: 2142
>
> Modified:
>    trunk/LinuxBIOSv2/src/config/linuxbios_ram.ld
>    trunk/LinuxBIOSv2/src/cpu/amd/car/clear_1m_ram.c
>    trunk/LinuxBIOSv2/src/cpu/amd/car/post_cache_as_ram.c
>    trunk/LinuxBIOSv2/src/cpu/amd/model_fxx/init_cpus.c
>    trunk/LinuxBIOSv2/src/cpu/x86/lapic/lapic_cpu_init.c
>    trunk/LinuxBIOSv2/src/cpu/x86/lapic/secondary.S
>    trunk/LinuxBIOSv2/src/cpu/x86/pae/pgtbl.c
> Log:
> issue 51 and 52: set mtrr for ap before stop it, and _RAMBASE above 1M
> support and pgtbl after 1M support
>
> Stefan

Yh Lu, this is your patch that broke things, I assume you will fix?

thanks

ron

[Richard Smith]

> Yh Lu, this is your patch that broke things, I assume you will fix?

I don't mind fixing it. (I already have for my tree) I just don't know
which fix is prefered.  Either add a wrapper around the call or pull
the definition out and make it compile to a empty function like it
does when _RAMBASE is set to a large value.

[Stefan Reinauer]

* Richard Smith  [060109 17:23]:
> > Yh Lu, this is your patch that broke things, I assume you will fix?
>
> I don't mind fixing it. (I already have for my tree) I just don't know
> which fix is prefered.  Either add a wrapper around the call or pull
> the definition out and make it compile to a empty function like it
> does when _RAMBASE is set to a large value.

IMHO the second one would be fine, we did that a couple of times in the
tree.  Both would be fine though I guess. Basically the fix that fixes
the other broken motherboards as well is the preferred one ;-)

Stefan