mbox series

[0/2] cacheinfo: allow using OF style matchine for cache devices

Message ID 20231231-cpu-cacheinfo-of-v1-0-cd3584d2c7b4@linaro.org
Headers show
Series cacheinfo: allow using OF style matchine for cache devices | expand

Message

Dmitry Baryshkov Dec. 31, 2023, 5:30 p.m. UTC
On Qualcomm Krait (APQ8064, MSM8960, MSM8974, APQ8084) platforms L2
cache frequency should follow CPU cache frequencies. It is logical to
put the correponding OPP table into the L2 device node and then add
simple devfreq driver that aggregates CPU votes and sets the clock
accordingly.

In [1] Rob suggested reusing existing cache devices for
platform-specific cache drivers. This series implements that suggestion.

This is posted as an RFC. If this implementation is found to be
acceptable, corresponding L2 cache driver will be included in the next
revision.

[1] https://lore.kernel.org/linux-arm-msm/20231011154935.GA785564-robh@kernel.org/

Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
---
Dmitry Baryshkov (2):
      cpu: allow matching cpu_subsys devices against OF nodes
      cacheinfo: pass of_node to cpu_device_create()

 drivers/base/cacheinfo.c  | 16 ++++++++++++++--
 drivers/base/cpu.c        | 15 ++++++++++++---
 include/linux/cacheinfo.h |  2 ++
 include/linux/cpu.h       |  3 ++-
 4 files changed, 30 insertions(+), 6 deletions(-)
---
base-commit: 39676dfe52331dba909c617f213fdb21015c8d10
change-id: 20231231-cpu-cacheinfo-of-c4f14013bf31

Best regards,

Comments

Dmitry Baryshkov Jan. 4, 2024, 1:26 a.m. UTC | #1
On Sun, 31 Dec 2023 at 19:30, Dmitry Baryshkov
<dmitry.baryshkov@linaro.org> wrote:
>
> On Qualcomm Krait (APQ8064, MSM8960, MSM8974, APQ8084) platforms L2
> cache frequency should follow CPU cache frequencies. It is logical to
> put the correponding OPP table into the L2 device node and then add
> simple devfreq driver that aggregates CPU votes and sets the clock
> accordingly.
>
> In [1] Rob suggested reusing existing cache devices for
> platform-specific cache drivers. This series implements that suggestion.
>
> This is posted as an RFC. If this implementation is found to be
> acceptable, corresponding L2 cache driver will be included in the next
> revision.

I'd like to recall this series for now. It doesn't work as expected.

>
> [1] https://lore.kernel.org/linux-arm-msm/20231011154935.GA785564-robh@kernel.org/
>
> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
> ---
> Dmitry Baryshkov (2):
>       cpu: allow matching cpu_subsys devices against OF nodes
>       cacheinfo: pass of_node to cpu_device_create()