[PATCH v2 2/5] drm/etnaviv: assert GPU lock held in perfmon pipe_*_read functions

Lucas Stach l.stach at pengutronix.de
Fri Jul 5 20:00:10 UTC 2024


The perf counter read functions don't just read registers, but they
also mutate state to direct the reads towards the correct pipe and
engine. Assert that the GPU mutex is held at this point, so that
those state changes don't interfere with others.

Signed-off-by: Lucas Stach <l.stach at pengutronix.de>
---
v2: new patch
---
 drivers/gpu/drm/etnaviv/etnaviv_perfmon.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/gpu/drm/etnaviv/etnaviv_perfmon.c b/drivers/gpu/drm/etnaviv/etnaviv_perfmon.c
index dc9dea664a28..d53a5c293373 100644
--- a/drivers/gpu/drm/etnaviv/etnaviv_perfmon.c
+++ b/drivers/gpu/drm/etnaviv/etnaviv_perfmon.c
@@ -62,6 +62,8 @@ static u32 pipe_perf_reg_read(struct etnaviv_gpu *gpu,
 	u32 value = 0;
 	unsigned i;
 
+	lockdep_assert_held(&gpu->lock);
+
 	for (i = 0; i < gpu->identity.pixel_pipes; i++) {
 		pipe_select(gpu, clock, i);
 		value += perf_reg_read(gpu, domain, signal);
@@ -81,6 +83,8 @@ static u32 pipe_reg_read(struct etnaviv_gpu *gpu,
 	u32 value = 0;
 	unsigned i;
 
+	lockdep_assert_held(&gpu->lock);
+
 	for (i = 0; i < gpu->identity.pixel_pipes; i++) {
 		pipe_select(gpu, clock, i);
 		value += gpu_read(gpu, signal->data);
-- 
2.39.2



More information about the dri-devel mailing list