In the cases where the power domain connected to logics is allowed to transition from a level(L)-->power collapse(0)-->retention(1) or vice versa retention(1)-->power collapse(0)-->level(L) will cause the logic to lose the configurations. The ARC does not support retention to collapse transition on MxC rails.
The targets from SM8450 onwards the PLL logics of clock controllers are connected to MxC rails and the recommended configurations are carried out during the clock controller probes. The MxC transition as mentioned above should be skipped to ensure the PLL settings are intact across clock controller power on & off.
On older targets that do not split MX into MxA and MxC does not collapse the logic and it is parked always at RETENTION, thus this issue is never observed on those targets.
Cc: stable@vger.kernel.org # v5.17 Reviewed-by: Bjorn Andersson andersson@kernel.org Signed-off-by: Taniya Das quic_tdas@quicinc.com --- [Changes in v2]: Incorporate the comments in the commit text. --- drivers/pmdomain/qcom/rpmhpd.c | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/drivers/pmdomain/qcom/rpmhpd.c b/drivers/pmdomain/qcom/rpmhpd.c index de9121ef4216..d2cb4271a1ca 100644 --- a/drivers/pmdomain/qcom/rpmhpd.c +++ b/drivers/pmdomain/qcom/rpmhpd.c @@ -40,6 +40,7 @@ * @addr: Resource address as looped up using resource name from * cmd-db * @state_synced: Indicator that sync_state has been invoked for the rpmhpd resource + * @skip_retention_level: Indicate that retention level should not be used for the power domain */ struct rpmhpd { struct device *dev; @@ -56,6 +57,7 @@ struct rpmhpd { const char *res_name; u32 addr; bool state_synced; + bool skip_retention_level; };
struct rpmhpd_desc { @@ -173,6 +175,7 @@ static struct rpmhpd mxc = { .pd = { .name = "mxc", }, .peer = &mxc_ao, .res_name = "mxc.lvl", + .skip_retention_level = true, };
static struct rpmhpd mxc_ao = { @@ -180,6 +183,7 @@ static struct rpmhpd mxc_ao = { .active_only = true, .peer = &mxc, .res_name = "mxc.lvl", + .skip_retention_level = true, };
static struct rpmhpd nsp = { @@ -819,6 +823,9 @@ static int rpmhpd_update_level_mapping(struct rpmhpd *rpmhpd) return -EINVAL;
for (i = 0; i < rpmhpd->level_count; i++) { + if (rpmhpd->skip_retention_level && buf[i] == RPMH_REGULATOR_LEVEL_RETENTION) + continue; + rpmhpd->level[i] = buf[i];
/* Remember the first corner with non-zero level */
--- base-commit: 62c97045b8f720c2eac807a5f38e26c9ed512371 change-id: 20240625-avoid_mxc_retention-b095a761d981
Best regards,
On Tue, 25 Jun 2024 at 06:33, Taniya Das quic_tdas@quicinc.com wrote:
In the cases where the power domain connected to logics is allowed to transition from a level(L)-->power collapse(0)-->retention(1) or vice versa retention(1)-->power collapse(0)-->level(L) will cause the logic to lose the configurations. The ARC does not support retention to collapse transition on MxC rails.
The targets from SM8450 onwards the PLL logics of clock controllers are connected to MxC rails and the recommended configurations are carried out during the clock controller probes. The MxC transition as mentioned above should be skipped to ensure the PLL settings are intact across clock controller power on & off.
On older targets that do not split MX into MxA and MxC does not collapse the logic and it is parked always at RETENTION, thus this issue is never observed on those targets.
Cc: stable@vger.kernel.org # v5.17 Reviewed-by: Bjorn Andersson andersson@kernel.org Signed-off-by: Taniya Das quic_tdas@quicinc.com
Applied for fixes, thanks!
Kind regards Uffe
[Changes in v2]: Incorporate the comments in the commit text.
drivers/pmdomain/qcom/rpmhpd.c | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/drivers/pmdomain/qcom/rpmhpd.c b/drivers/pmdomain/qcom/rpmhpd.c index de9121ef4216..d2cb4271a1ca 100644 --- a/drivers/pmdomain/qcom/rpmhpd.c +++ b/drivers/pmdomain/qcom/rpmhpd.c @@ -40,6 +40,7 @@
- @addr: Resource address as looped up using resource name from
cmd-db- @state_synced: Indicator that sync_state has been invoked for the rpmhpd resource
*/
- @skip_retention_level: Indicate that retention level should not be used for the power domain
struct rpmhpd { struct device *dev; @@ -56,6 +57,7 @@ struct rpmhpd { const char *res_name; u32 addr; bool state_synced;
bool skip_retention_level;};
struct rpmhpd_desc { @@ -173,6 +175,7 @@ static struct rpmhpd mxc = { .pd = { .name = "mxc", }, .peer = &mxc_ao, .res_name = "mxc.lvl",
.skip_retention_level = true,};
static struct rpmhpd mxc_ao = { @@ -180,6 +183,7 @@ static struct rpmhpd mxc_ao = { .active_only = true, .peer = &mxc, .res_name = "mxc.lvl",
.skip_retention_level = true,};
static struct rpmhpd nsp = { @@ -819,6 +823,9 @@ static int rpmhpd_update_level_mapping(struct rpmhpd *rpmhpd) return -EINVAL;
for (i = 0; i < rpmhpd->level_count; i++) {
if (rpmhpd->skip_retention_level && buf[i] == RPMH_REGULATOR_LEVEL_RETENTION)continue;rpmhpd->level[i] = buf[i]; /* Remember the first corner with non-zero level */
base-commit: 62c97045b8f720c2eac807a5f38e26c9ed512371 change-id: 20240625-avoid_mxc_retention-b095a761d981
Best regards,
Taniya Das quic_tdas@quicinc.com
linux-stable-mirror@lists.linaro.org
-
Taniya Das -
Ulf Hansson