Re: [PATCH] export e820 table on x86

[Dave Hansen <haveblue@us.ibm.com>]

Linus Torvalds wrote:
> On Wed, 20 Nov 2002, Dave Hansen wrote:
> 
>>I stole a patch that Arjan did a while ago, and ported it up to 2.5:
>>http://www.kernelnewbies.org/kernels/rh80/SOURCES/linux-2.4.0-e820.patch
>>
>>We need this so avoid making BIOS calls when using kexec.
> 
>  - why isn't the info in /proc/iomem good enough - ie wouldn't it be 
>    better to just extend resource handling to 64 bit instead of
>    creating a new file.

It looks good enough.  The only irritating part is turning the "S3 
Inc. Trio 64 3D" or "ACPI Tables" back into the numberic e820 type.  I 
accomplished this in the previous patch by removing the printing of 
the name completely.  I thought it was silly to have the kernel 
printing out a pretty name just to have the userspace program parse it 
back into a number.  It saved a bug hunk of code in both the kernel 
and the kexec utility to skip the name.

What would you think of just adding another field to /proc/iomem which 
contains the e820 field type?  I've never seen any userspace use of 
iomem, but I would imagine that things like kudzu use it.  I wonder if 
they'll get tripped up if
50000000-50000fff : Texas Instruments PCI1450
changes into something like
50000000-50000fff : 2 : Texas Instruments PCI1450

>  - please use the seq_file interfaces for new files if you do end up 
>    creating new files.

I posted one this morning.

-- 
Dave Hansen
haveblue@us.ibm.com