[okl4-developer] problem with init_bootmem

Carl van Schaik carl at ok-labs.com
Wed Mar 26 02:27:23 EST 2008


Hi Charles,

Sorry for the delay, its pretty busy here right now.

The physical memory described in the tools/machines.py is taken by build 
system and provided to elfweaver via an xml file. Elfweaver allocates 
memory for each application and also for the kernel heap. It then writes 
out the memory descriptors in the KIP with this information. One or more 
memory descriptors can describe the kernel heap. The kernel init code 
should read and find this info.

the memdescriptor the kernel looks for has type - physical, arch 
specific, sub-type 0
The mips init_bootmem function in 1.5.2 does not handle the case where a 
descriptor was not found very elegantly, which may be part of what you 
are seeing.

regards,
Carl

Charles.VANWITTENBERGHE at fr.thalesgroup.com wrote:
> Hi !
>
> I have adapted and compiled Pistachio for my platform, and I'm currently
> debugging.
>
> I encountered a problem during initialisation with the function
> init_bootmem(). It seems that the start_bootmem and end_bootmem values are
> pretty strange. I know that they are linked to the KIP and the
> MemoryDescriptors, and that the linker is making some of the work for the
> memory descriptors initialisation , but I can't figure the whole process.
>
> Could you tell me what is that "bootmem" supposed to be and if it's possible
> to place it anywhere in a kernel segment ?
>
> Thanks in advance, and sorry for my low-level english.
>
> Charles.
>
> ----
>
> Charles Van Wittenberghe
> Stagiaire service SES
> THALES COMMUNICATION 
> 110 Avenue du Maréchal Leclerc
> BP 70942- 49309 CHOLET
> FRANCE
> tel : 06.84.12.45.87
> charles.vanwittenberghe at fr.thalesgroup.com
>
>
> _______________________________________________
> Developer mailing list
> Developer at okl4.org
> https://lists.okl4.org/mailman/listinfo/developer
>
>   




More information about the Developer mailing list