Starting with SCMIv3.1, the PERFORMANCE_LIMITS_SET command allows a user
to request only one between max and min ranges to be changed, while leaving
the other untouched if set to zero in the request; anyway SCMIv3.1 states
also explicitly that you cannot leave both of those unchanged (zeroed) when
issuing such command: add a proper check for this condition.
Signed-off-by: Cristian Marussi <[email protected]>
---
drivers/firmware/arm_scmi/perf.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/firmware/arm_scmi/perf.c b/drivers/firmware/arm_scmi/perf.c
index 65ffda5495d6..8f4051aca220 100644
--- a/drivers/firmware/arm_scmi/perf.c
+++ b/drivers/firmware/arm_scmi/perf.c
@@ -423,6 +423,9 @@ static int scmi_perf_limits_set(const struct scmi_protocol_handle *ph,
struct scmi_perf_info *pi = ph->get_priv(ph);
struct perf_dom_info *dom = pi->dom_info + domain;
+ if (PROTOCOL_REV_MAJOR(pi->version) >= 0x3 && !max_perf && !min_perf)
+ return -EINVAL;
+
if (dom->fc_info && dom->fc_info->limit_set_addr) {
iowrite32(max_perf, dom->fc_info->limit_set_addr);
iowrite32(min_perf, dom->fc_info->limit_set_addr + 4);
--
2.32.0
On Wed, Mar 30, 2022 at 04:05:51PM +0100, Cristian Marussi wrote:
> Starting with SCMIv3.1, the PERFORMANCE_LIMITS_SET command allows a user
> to request only one between max and min ranges to be changed, while leaving
> the other untouched if set to zero in the request; anyway SCMIv3.1 states
> also explicitly that you cannot leave both of those unchanged (zeroed) when
> issuing such command: add a proper check for this condition.
>
> Signed-off-by: Cristian Marussi <[email protected]>
> ---
> drivers/firmware/arm_scmi/perf.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/firmware/arm_scmi/perf.c b/drivers/firmware/arm_scmi/perf.c
> index 65ffda5495d6..8f4051aca220 100644
> --- a/drivers/firmware/arm_scmi/perf.c
> +++ b/drivers/firmware/arm_scmi/perf.c
> @@ -423,6 +423,9 @@ static int scmi_perf_limits_set(const struct scmi_protocol_handle *ph,
> struct scmi_perf_info *pi = ph->get_priv(ph);
> struct perf_dom_info *dom = pi->dom_info + domain;
>
> + if (PROTOCOL_REV_MAJOR(pi->version) >= 0x3 && !max_perf && !min_perf)
> + return -EINVAL;
> +
Do we really need the version check here ? I agree it was explicitly added
in v3.1, but it makes sense on any version really. No ?
--
Regards,
Sudeep
On Thu, Apr 28, 2022 at 02:49:48PM +0100, Cristian Marussi wrote:
> On Thu, Apr 28, 2022 at 02:13:57PM +0100, Sudeep Holla wrote:
> > On Wed, Mar 30, 2022 at 04:05:51PM +0100, Cristian Marussi wrote:
> > > Starting with SCMIv3.1, the PERFORMANCE_LIMITS_SET command allows a user
> > > to request only one between max and min ranges to be changed, while leaving
> > > the other untouched if set to zero in the request; anyway SCMIv3.1 states
> > > also explicitly that you cannot leave both of those unchanged (zeroed) when
> > > issuing such command: add a proper check for this condition.
> > >
> > > Signed-off-by: Cristian Marussi <[email protected]>
> > > ---
> > > drivers/firmware/arm_scmi/perf.c | 3 +++
> > > 1 file changed, 3 insertions(+)
> > >
> > > diff --git a/drivers/firmware/arm_scmi/perf.c b/drivers/firmware/arm_scmi/perf.c
> > > index 65ffda5495d6..8f4051aca220 100644
> > > --- a/drivers/firmware/arm_scmi/perf.c
> > > +++ b/drivers/firmware/arm_scmi/perf.c
> > > @@ -423,6 +423,9 @@ static int scmi_perf_limits_set(const struct scmi_protocol_handle *ph,
> > > struct scmi_perf_info *pi = ph->get_priv(ph);
> > > struct perf_dom_info *dom = pi->dom_info + domain;
> > >
> > > + if (PROTOCOL_REV_MAJOR(pi->version) >= 0x3 && !max_perf && !min_perf)
> > > + return -EINVAL;
> > > +
> >
> > Do we really need the version check here ? I agree it was explicitly added
> > in v3.1, but it makes sense on any version really. No ?
>
> Indeed seemed a silly patch also to me but given that only in v3.1 it is
> explicitly stated that you cannot issue this command with both min and
> max ZEROED I though this could have broken older fw that allowed
> setting PERF_LIMITS_SET max=0 min=0
>
> ....maybe overthought ...
Hmm, let's keep it unconditional for now. We can add if someone reports
broken firmware. BTW there are no users in the kernel ????.
--
Regards,
Sudeep
On Thu, Apr 28, 2022 at 02:13:57PM +0100, Sudeep Holla wrote:
> On Wed, Mar 30, 2022 at 04:05:51PM +0100, Cristian Marussi wrote:
> > Starting with SCMIv3.1, the PERFORMANCE_LIMITS_SET command allows a user
> > to request only one between max and min ranges to be changed, while leaving
> > the other untouched if set to zero in the request; anyway SCMIv3.1 states
> > also explicitly that you cannot leave both of those unchanged (zeroed) when
> > issuing such command: add a proper check for this condition.
> >
> > Signed-off-by: Cristian Marussi <[email protected]>
> > ---
> > drivers/firmware/arm_scmi/perf.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/drivers/firmware/arm_scmi/perf.c b/drivers/firmware/arm_scmi/perf.c
> > index 65ffda5495d6..8f4051aca220 100644
> > --- a/drivers/firmware/arm_scmi/perf.c
> > +++ b/drivers/firmware/arm_scmi/perf.c
> > @@ -423,6 +423,9 @@ static int scmi_perf_limits_set(const struct scmi_protocol_handle *ph,
> > struct scmi_perf_info *pi = ph->get_priv(ph);
> > struct perf_dom_info *dom = pi->dom_info + domain;
> >
> > + if (PROTOCOL_REV_MAJOR(pi->version) >= 0x3 && !max_perf && !min_perf)
> > + return -EINVAL;
> > +
>
> Do we really need the version check here ? I agree it was explicitly added
> in v3.1, but it makes sense on any version really. No ?
Indeed seemed a silly patch also to me but given that only in v3.1 it is
explicitly stated that you cannot issue this command with both min and
max ZEROED I though this could have broken older fw that allowed
setting PERF_LIMITS_SET max=0 min=0
....maybe overthought ...
Thanks,
Cristian