diff options
| author | Angel Pons <th3fanbus@gmail.com> | 2020-10-01 22:50:12 +0200 |
|---|---|---|
| committer | Angel Pons <th3fanbus@gmail.com> | 2020-11-03 17:04:03 +0000 |
| commit | 3dea2b63eeb8f97b31571f6f0eb37f38f9967b6b (patch) | |
| tree | 6d9843c11049d2ce89c55bd8deb451d542caee3c /src/soc/intel | |
| parent | 88991caf00d3b8862bc3c1aa4497baeb67e910e6 (diff) | |
| download | coreboot-3dea2b63eeb8f97b31571f6f0eb37f38f9967b6b.tar.xz | |
soc/intel/common/block/systemagent/memmap.c: Align cached region
When asked to place cbmem_top(), FSP does not seem to care about
alignment. It can return an address that is MTRR poison, which will
exhaust all variable MTRRs when trying to set up caching for CBMEM.
This will make memory-mapped flash and TSEG caching fail as well.
Safeguard against this by aligning the region to cache to half of its
size, and move it upwards to compensate. It is assumed that caching
memory above the provided bootloader TOLUM address is inconsequential.
TEST=Boot Purism Librem Mini WHL, observe no MTRR exhaustion error
messages in console. The boot process also feels more fluid.
Change-Id: Ic64fd6d3d9e8ab4c78d68b910a476f9c4eb2d353
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45930
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Diffstat (limited to 'src/soc/intel')
| -rw-r--r-- | src/soc/intel/common/block/systemagent/memmap.c | 11 |
1 files changed, 6 insertions, 5 deletions
diff --git a/src/soc/intel/common/block/systemagent/memmap.c b/src/soc/intel/common/block/systemagent/memmap.c index 985f2c4814..27870b0cf7 100644 --- a/src/soc/intel/common/block/systemagent/memmap.c +++ b/src/soc/intel/common/block/systemagent/memmap.c @@ -6,6 +6,7 @@ #include <cpu/x86/mtrr.h> #include <cpu/x86/smm.h> #include <intelblocks/systemagent.h> +#include <types.h> /* * Expected Host Memory Map (we don't know 100% and not all regions are present on all SoCs): @@ -55,18 +56,18 @@ void smm_region(uintptr_t *start, size_t *size) void fill_postcar_frame(struct postcar_frame *pcf) { - uintptr_t top_of_ram; + /* FSP does not seem to bother w.r.t. alignment when asked to place cbmem_top() */ + uintptr_t top_of_ram = ALIGN_UP((uintptr_t)cbmem_top(), 8 * MiB); /* * We need to make sure ramstage will be run cached. At this * point exact location of ramstage in cbmem is not known. - * Instruct postcar to cache 16 megs under cbmem top which is + * Instruct postcar to cache 16 megs below cbmem top which is * a safe bet to cover ramstage. */ - top_of_ram = (uintptr_t) cbmem_top(); printk(BIOS_DEBUG, "top_of_ram = 0x%lx\n", top_of_ram); - top_of_ram -= 16*MiB; - postcar_frame_add_mtrr(pcf, top_of_ram, 16*MiB, MTRR_TYPE_WRBACK); + + postcar_frame_add_mtrr(pcf, top_of_ram - 16 * MiB, 16 * MiB, MTRR_TYPE_WRBACK); /* Cache the TSEG region */ postcar_enable_tseg_cache(pcf); |