diff mbox series

[1/2] firmware: arm_scmi: Use xa_insert() to store opps

Message ID 20240108185050.1628687-1-cristian.marussi@arm.com (mailing list archive)
State New, archived
Headers show
Series [1/2] firmware: arm_scmi: Use xa_insert() to store opps | expand

Commit Message

Cristian Marussi Jan. 8, 2024, 6:50 p.m. UTC 
When storing opps by level or index use xa_insert() instead of xa_store()
and add error-checking to spot bad duplicates indexes possibly wrongly
provided by the platform firmware.

Fixes: 31c7c1397a33 ("firmware: arm_scmi: Add v3.2 perf level indexing mode support")
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
---
 drivers/firmware/arm_scmi/perf.c | 23 ++++++++++++++++++-----
 1 file changed, 18 insertions(+), 5 deletions(-)

Comments

Sudeep Holla Jan. 22, 2024, 2:41 p.m. UTC | #1 
On Mon, 08 Jan 2024 18:50:49 +0000, Cristian Marussi wrote:
> When storing opps by level or index use xa_insert() instead of xa_store()
> and add error-checking to spot bad duplicates indexes possibly wrongly
> provided by the platform firmware.
>

Applied to sudeep.holla/linux (for-next/scmi/fixes), thanks!

[1/2] firmware: arm_scmi: Use xa_insert() to store opps
      https://git.kernel.org/sudeep.holla/c/e8ef4bbe39b9
[2/2] firmware: arm_scmi: Use xa_insert() when saving raw queues
      https://git.kernel.org/sudeep.holla/c/b5dc0ffd3656
--
Regards,
Sudeep
diff mbox series

Patch

diff --git a/drivers/firmware/arm_scmi/perf.c b/drivers/firmware/arm_scmi/perf.c
index cd6b7ea82436..7fd49ea68168 100644
--- a/drivers/firmware/arm_scmi/perf.c
+++ b/drivers/firmware/arm_scmi/perf.c
@@ -350,8 +350,8 @@  process_response_opp(struct scmi_opp *opp, unsigned int loop_idx,
 }
 
 static inline void
-process_response_opp_v4(struct perf_dom_info *dom, struct scmi_opp *opp,
-			unsigned int loop_idx,
+process_response_opp_v4(struct device *dev, struct perf_dom_info *dom,
+			struct scmi_opp *opp, unsigned int loop_idx,
 			const struct scmi_msg_resp_perf_describe_levels_v4 *r)
 {
 	opp->perf = le32_to_cpu(r->opp[loop_idx].perf_val);
@@ -362,10 +362,23 @@  process_response_opp_v4(struct perf_dom_info *dom, struct scmi_opp *opp,
 	/* Note that PERF v4 reports always five 32-bit words */
 	opp->indicative_freq = le32_to_cpu(r->opp[loop_idx].indicative_freq);
 	if (dom->level_indexing_mode) {
+		int ret;
+
 		opp->level_index = le32_to_cpu(r->opp[loop_idx].level_index);
 
-		xa_store(&dom->opps_by_idx, opp->level_index, opp, GFP_KERNEL);
-		xa_store(&dom->opps_by_lvl, opp->perf, opp, GFP_KERNEL);
+		ret = xa_insert(&dom->opps_by_idx, opp->level_index, opp,
+				GFP_KERNEL);
+		if (ret)
+			dev_warn(dev,
+				 "Failed to add opps_by_idx at %d - ret:%d\n",
+				 opp->level_index, ret);
+
+		ret = xa_insert(&dom->opps_by_lvl, opp->perf, opp, GFP_KERNEL);
+		if (ret)
+			dev_warn(dev,
+				 "Failed to add opps_by_lvl at %d - ret:%d\n",
+				 opp->perf, ret);
+
 		hash_add(dom->opps_by_freq, &opp->hash, opp->indicative_freq);
 	}
 }
@@ -382,7 +395,7 @@  iter_perf_levels_process_response(const struct scmi_protocol_handle *ph,
 	if (PROTOCOL_REV_MAJOR(p->version) <= 0x3)
 		process_response_opp(opp, st->loop_idx, response);
 	else
-		process_response_opp_v4(p->perf_dom, opp, st->loop_idx,
+		process_response_opp_v4(ph->dev, p->perf_dom, opp, st->loop_idx,
 					response);
 	p->perf_dom->opp_count++;