Hi Konrad,
Thanks for the review.
On 12/4/21 8:34 AM, Konrad Dybcio wrote:
Hi,
On 02.12.2021 23:38, Thara Gopinath wrote:
Add compatible to support LMh for sm8150 SoC.
sm8150 does not require explicit enabling for various LMh subsystems.
Move this piece of code under condition that it is executed only
for sdm845 SoC.
Signed-off-by: Thara Gopinath <thara.gopinath@xxxxxxxxxx>
---
drivers/thermal/qcom/lmh.c | 61 ++++++++++++++++++++------------------
1 file changed, 32 insertions(+), 29 deletions(-)
[...]
- return ret;
+ if (of_device_is_compatible(np, "qcom,sdm845-lmh")) {
+ if (!qcom_scm_lmh_dcvsh_available())
+ return -EINVAL;
I don't believe this is the correct approach, as different SoCs may
require different sequences of these writes (for example SDM660/MSM8998
seems to only enable the thermal algorithm), and there will (hopefully) be interest
in adding LMH support for more platforms, so perhaps separating this somehow
could keep this a bit cleaner and easier to work with for the next person..
I have not looked at SDM660/MSM8998. Are you telling me that these SoCs
don't enable the current and BCL portion of LMh. Maybe they have an
earlier version of Lmh which does not support all the features. The
right approach in this case will be to add a match table with flags for
init based on SoC. I can send v2, adding a match table with a flag to
specify whether to do the init sequence or not. Since I am not adding
the support for any other SoC at the moment, I cannot put in flags
separating out thermal , current and BCL init.
Konrad
--
Warm Regards
Thara (She/Her/Hers)
