On Fri, Nov 13, 2020 at 11:47:05AM -0500, Masayoshi Mizuma wrote:
On Fri, Nov 13, 2020 at 10:59:06AM +0100, Karel Zak wrote:
On Tue, Nov 10, 2020 at 08:24:00PM -0500, Masayoshi Mizuma wrote:
> sys-utils/lscpu-arm.c | 22 ++++++++--------------
> sys-utils/lscpu.c | 4 ++++
> sys-utils/lscpu.h | 2 ++
> 3 files changed, 14 insertions(+), 14 deletions(-)
Ported to the new lscpu and merged. Thanks (and extra thanks for
your patience :-).
Thank you so much!
I see one issue with the stuff from BIOS. It seems it reads the
information always from the same place. How does it work if I have two
different CPUs? I guess it always return the same model name, but what
if I have Cortex-A53 and Cortex-A72 on the same board?
The first entry, /sys/firmware/dmi/entries/4-0/raw, is only used,
so lscpu will show the same information of the SMBIOS type4 from
the entry even if the system has two diffrent cpu models, like
Cortex-A53 and Cortex-A72.
I'm not familiar with the heterogenous SoC, probably each
CPU model has own SMBIOS type4, right? for example, /sys/firmware/dmi/entries/4-0/raw
is for Cortex-A53 and /sys/firmware/dmi/entries/4-1/raw is for Cortex-A72.
That's correct: there will be a 4-x/raw file for each CPU type. So far
it's been safe to assume that all CPUs are the same in a system, but
with ARM big.LITTLE, that's no longer a safe assumption. It will take a
bit more work to associate each SMBIOS entry with each CPU in
/proc/cpuinfo.
https://en.wikipedia.org/wiki/ARM_big.LITTLE
--
Jeff Bastian
Kernel QE - Hardware Enablement
Red Hat
