[PATCH 2/2] drm/tidss: Avoid race condition while handling interrupt registers
Devarsh Thakkar
devarsht at ti.com
Sat Oct 12 15:07:10 UTC 2024
There is a possibility of a race condition between interrupt subroutine
which accesses the interrupt related registers to clear the statuses before
handling the interrupt and other functions such as display soft reset,
runtime resume/suspend etc which also access the interrupt related
registers. To prevent such scenarioes, use a spinlock to serialize access
to interrupt related registers.
Fixes: 32a1795f57ee ("drm/tidss: New driver for TI Keystone platform Display SubSystem")
Signed-off-by: Devarsh Thakkar <devarsht at ti.com>
---
drivers/gpu/drm/tidss/tidss_dispc.c | 4 ++++
drivers/gpu/drm/tidss/tidss_irq.c | 2 ++
2 files changed, 6 insertions(+)
diff --git a/drivers/gpu/drm/tidss/tidss_dispc.c b/drivers/gpu/drm/tidss/tidss_dispc.c
index b04419b24863..cec59deff015 100644
--- a/drivers/gpu/drm/tidss/tidss_dispc.c
+++ b/drivers/gpu/drm/tidss/tidss_dispc.c
@@ -2767,8 +2767,12 @@ static void dispc_init_errata(struct dispc_device *dispc)
*/
static void dispc_softreset_k2g(struct dispc_device *dispc)
{
+ unsigned long flags;
+
+ spin_lock_irqsave(&dispc->tidss->wait_lock, flags);
dispc_set_irqenable(dispc, 0);
dispc_read_and_clear_irqstatus(dispc);
+ spin_unlock_irqrestore(&dispc->tidss->wait_lock, flags);
for (unsigned int vp_idx = 0; vp_idx < dispc->feat->num_vps; ++vp_idx)
VP_REG_FLD_MOD(dispc, vp_idx, DISPC_VP_CONTROL, 0, 0, 0);
diff --git a/drivers/gpu/drm/tidss/tidss_irq.c b/drivers/gpu/drm/tidss/tidss_irq.c
index 604334ef526a..d053dbb9d28c 100644
--- a/drivers/gpu/drm/tidss/tidss_irq.c
+++ b/drivers/gpu/drm/tidss/tidss_irq.c
@@ -60,7 +60,9 @@ static irqreturn_t tidss_irq_handler(int irq, void *arg)
unsigned int id;
dispc_irq_t irqstatus;
+ spin_lock(&tidss->wait_lock);
irqstatus = dispc_read_and_clear_irqstatus(tidss->dispc);
+ spin_unlock(&tidss->wait_lock);
for (id = 0; id < tidss->num_crtcs; id++) {
struct drm_crtc *crtc = tidss->crtcs[id];
--
2.39.1
More information about the dri-devel
mailing list