diff options
author | Patrick Rudolph <patrick.rudolph@9elements.com> | 2020-12-17 14:42:29 +0100 |
---|---|---|
committer | Tim Wawrzynczak <twawrzynczak@chromium.org> | 2020-12-22 22:22:21 +0000 |
commit | 7a66ffb34aba44453c5a7fea75aff4d58322e94d (patch) | |
tree | 002daa79525d3a3c810c02e4f3cfe98e8c2ea1a8 /payloads/libpayload/.gitignore | |
parent | 8d127846bc944fb416689ca1c93fd24f487f0bee (diff) |
soc/intel/common/block/acpi: Fix get_cores_per_package
Current implementation uses CPUID 0Bh function that returns the number
of logical cores of requested level. The problem with this approach is
that this value doesn't change when HyperThreading is disabled (it's in
the Intel docs), so it breaks generate_cpu_entries() because `numcpus`
ends up being zero due to integer division truncation.
- Use MSR 0x35 instead, which returns the correct number of logical
processors with and without HT.
- Use cpu_read_topology() to gather the required information
Tested on Prodrive Hermes, the ACPI code is now generated even with
HyperThreading disabled.
Change-Id: Id9b985a07cd3f99a823622f766c80ff240ac1188
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/48691
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Diffstat (limited to 'payloads/libpayload/.gitignore')
0 files changed, 0 insertions, 0 deletions