parsing of some early cmdline options

Kumar Gala galak at kernel.crashing.org
Fri Feb 24 18:14:41 EST 2006


On Feb 23, 2006, at 5:40 PM, Michael Ellerman wrote:

> On Fri, 24 Feb 2006 10:19, Benjamin Herrenschmidt wrote:
>> On Thu, 2006-02-23 at 17:02 -0600, Kumar Gala wrote:
>>> Guys,
>>>
>>> I noticed that we have code in arch/powerpc/kernel/prom_init.c to  
>>> handle
>>> parsing iommu=, mem=, and crashkernel=.
>>>
>>> The problem I have is that we only use/build prom_init.c on  
>>> MULTIPLATFORM
>>> (coming from true OF).
>
>>> Any ideas on how to handle this for the !MULTIPLATFORM cases.   
>>> I'm most
>>> concerned about mem=.
>
> mem= is a bit tricky because you need to know about it really  
> early, otherwise
> your bootloader might violate the limit. But we could certainly  
> check for it
> again in prom.c. And you can always just set /chosen/linux,memory- 
> limit.

Agreed, I'm looking at adding "early" cmd_line parsing in  
early_init_dt_scan_chosen().  I assume we agree that mem= on command  
line overrides "linux,memory-limit".  I guessing that "linux,memory- 
limit" is only ever created in prom_init.c which would be from a mem=  
parsing.

- k



More information about the Linuxppc-dev mailing list