[PATCH] Rewrite of EFI memory map walk code

From: Khalid Aziz <khalid_aziz_at_hp.com>
Date: 2005-01-08 11:00:52
In an attempt to simplify EFI emmory range management and EFI memory map
walk, I have rewritten how EFI memory map is managed in the kernel. Here
is how I have approached this code. I am walking the EFI memory map and
from there I am constructing a linked list of valid memory ranges for
kernel to use. To find space to store this linked list in, from within
reserve_memory(), I walk the EFI memory map and look for the smallest
granule aligned block that will be sufficient to hold the linked list.
Amount of space I am allocating for linked list is the number of entries
in efi_memmap + another few for safety margin. I then mark this memory
reserved.

I have added two new functions - efi_gather_memory() which walks the EFI
memory map and builds a linked list of kernel usable memory map, and
efi_trim_memory() which applies the "max_addr=" and "mem=" restrictions
by splitting the linked list of kernel uable memory ranges
appropriately. Each node in this linked list has a field called
"ignore". To trim a memory range, I can split the list at approrpiate
point and mark every node in the list from that point onwards as ignored
by setting this field.

I rewrote efi_memmap_walk() to use the new kern_memmap list. I have also
replaced occurences of "md->num_pages << EFI_PAGE_SHIFT" with a macro.

One issue I have not been able to come up with a good solution for is
what to do if I fail to allocate memory for the linked list of memory
ranges. The only time this will happen is if I can not find a single
memory range that is larger than granule size. If that were to happen, I
do not see how the machine will boot even if we did not need space for
EFI memory map. So for now I have shelved this issue and simply do a
machine_restart() if this were to happen ever.

Tony, please apply.

-- 
Khalid

====================================================================
Khalid Aziz                                Linux and Open Source Lab
(970)898-9214                                        Hewlett-Packard
khalid_aziz@hp.com                                  Fort Collins, CO

"The Linux kernel is subject to relentless development" 
				- Alessandro Rubini



-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Received on Fri Jan 7 19:13:24 2005

This archive was generated by hypermail 2.1.8 : 2005-08-02 09:20:34 EST