memrange: Honor limit
in the last step of top-down stealing
We only checked that the resource fits below the given `limit` in memranges_find_entry(), but then accidentally placed it at the top of the found memrange. As most resources have only a coarse limit, e.g. the 4G barrier of 32-bit space, this became only visible when artificially setting an unusual, lower limit on a resource. So, for the final placement, use `MIN(limit, range end)` instead of the range's end alone. Change-Id: I3cc62ac3d427683c00ba0ac9f991fca62e99ce44 Signed-off-by: Nico Huber <nico.h@gmx.de> Reviewed-on: https://review.coreboot.org/c/coreboot/+/76480 Tested-by: build bot (Jenkins) <no-reply@coreboot.org> Reviewed-by: Martin L Roth <gaumless@gmail.com>
This commit is contained in:
parent
8a58483072
commit
b2893e22e6
@ -427,11 +427,12 @@ bool memranges_steal(struct memranges *ranges, resource_t limit, resource_t size
|
||||
return false;
|
||||
|
||||
if (from_top) {
|
||||
limit = MIN(limit, r->end);
|
||||
/* Ensure we're within the range, even aligned down.
|
||||
Proof is simple: If ALIGN_UP(r->begin) would be
|
||||
higher, the stolen range wouldn't fit.*/
|
||||
assert(r->begin <= ALIGN_DOWN(range_entry_end(r) - size, POWER_OF_2(align)));
|
||||
*stolen_base = ALIGN_DOWN(range_entry_end(r) - size, POWER_OF_2(align));
|
||||
assert(r->begin <= ALIGN_DOWN(limit - size + 1, POWER_OF_2(align)));
|
||||
*stolen_base = ALIGN_DOWN(limit - size + 1, POWER_OF_2(align));
|
||||
} else {
|
||||
*stolen_base = ALIGN_UP(r->begin, POWER_OF_2(align));
|
||||
}
|
||||
|
Loading…
x
Reference in New Issue
Block a user