Skip to contentSkip to conversation

Revert "drm/amd/display: dynamically allocate dml2_configuration_options structures"

PATCH In review George Zhang 3 participants 3 messages Jun 3, 2024 – Jun 4, 2024

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Revert "drm/amd/display: dynamically allocate dml2_configuration_options structures"

This reverts commit 416b5c5eec9e708b31c68f00cb79130f2cfaf7ed.

This patch caused a regression on DCN 3.2 on the IGT test assr-links-suspend, with the dmesg warning:

BUG: sleeping function called from invalid context at include/linux/sched/mm.h:306

in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 449, name: kworker/u64:8
preempt_count: 2, expected: 0
RCU nest depth: 0, expected: 0
Preemption disabled at:
ffffffffc0ce1580>] dc_fpu_begin+0x30/0xd0 [amdgpu]

CPU: 5 PID: 449 Comm: kworker/u64:8 Tainted: G        W          6.8.0+ #35
Hardware name: System manufacturer System Product Name/ROG STRIX X570-E GAMING WIFI II, BIOS 4204 02/24/2022
Workqueue: events_unbound async_run_entry_fn

Reverting this patch will re-introduce stack size warnings.

---
 .../display/dc/resource/dcn32/dcn32_resource.c   | 16 +++++-----------
 .../display/dc/resource/dcn321/dcn321_resource.c | 16 +++++-----------
 2 files changed, 10 insertions(+), 22 deletions(-)
drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +5 -11
index 0f11d7c8791c..022d320be1d5 100644
@@ -2007,27 +2007,21 @@ void dcn32_calculate_wm_and_dlg(struct dc *dc, struct dc_state *context,
 
 static void dcn32_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params)
 {
-	struct dml2_configuration_options *dml2_opt;
-
-	dml2_opt = kmemdup(&dc->dml2_options, sizeof(dc->dml2_options), GFP_KERNEL);
-	if (!dml2_opt)
-		return;
+	struct dml2_configuration_options dml2_opt = dc->dml2_options;
 
 	DC_FP_START();
 
 	dcn32_update_bw_bounding_box_fpu(dc, bw_params);
 
-	dml2_opt->use_clock_dc_limits = false;
+	dml2_opt.use_clock_dc_limits = false;
 	if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2)
-		dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2);
+		dml2_reinit(dc, &dml2_opt, &dc->current_state->bw_ctx.dml2);
 
-	dml2_opt->use_clock_dc_limits = true;
+	dml2_opt.use_clock_dc_limits = true;
 	if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2_dc_power_source)
-		dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source);
+		dml2_reinit(dc, &dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source);
 
 	DC_FP_END();
-
-	kfree(dml2_opt);
 }
 
 static struct resource_funcs dcn32_res_pool_funcs = {
drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c +5 -11
index 07ca6f58447d..e4b360d89b3b 100644
@@ -1581,27 +1581,21 @@ static struct dc_cap_funcs cap_funcs = {
 
 static void dcn321_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params)
 {
-	struct dml2_configuration_options *dml2_opt;
-
-	dml2_opt = kmemdup(&dc->dml2_options, sizeof(dc->dml2_options), GFP_KERNEL);
-	if (!dml2_opt)
-		return;
+	struct dml2_configuration_options dml2_opt = dc->dml2_options;
 
 	DC_FP_START();
 
 	dcn321_update_bw_bounding_box_fpu(dc, bw_params);
 
-	dml2_opt->use_clock_dc_limits = false;
+	dml2_opt.use_clock_dc_limits = false;
 	if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2)
-		dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2);
+		dml2_reinit(dc, &dml2_opt, &dc->current_state->bw_ctx.dml2);
 
-	dml2_opt->use_clock_dc_limits = true;
+	dml2_opt.use_clock_dc_limits = true;
 	if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2_dc_power_source)
-		dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source);
+		dml2_reinit(dc, &dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source);
 
 	DC_FP_END();
-
-	kfree(dml2_opt);
 }
 
 static struct resource_funcs dcn321_res_pool_funcs = {
Signature · 1 line
2.34.1
#1 lore raw
Quoted 19 lines · George Zhang

This reverts commit 416b5c5eec9e708b31c68f00cb79130f2cfaf7ed.

This patch caused a regression on DCN 3.2 on the IGT test assr-links-suspend, with
the dmesg warning:

BUG: sleeping function called from invalid context at 

include/linux/sched/mm.h:306
in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 449, name:
kworker/u64:8

static void dcn32_update_bw_bounding_box(struct dc *dc, struct
clk_bw_params *bw_params)
{

drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +0 -3
-	struct dml2_configuration_options *dml2_opt;
-
-	dml2_opt = kmemdup(&dc->dml2_options, sizeof(dc->dml2_options), 

GFP_KERNEL);

drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c void dcn32_calculate_wm_and_dlg(struct dc *dc, struct dc_state *context, · line 2012

Quoted 5 lines
drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +1 -2
-	if (!dml2_opt)
-		return;
+	struct dml2_configuration_options dml2_opt = dc->dml2_options;
 
 	DC_FP_START();

This means we also enable floating-point mode inside of an atomic section. Is that safe?

Arnd

#2 lore raw
In reply to George Zhang
Quoted 30 lines · George Zhang

This reverts commit 416b5c5eec9e708b31c68f00cb79130f2cfaf7ed.

This patch caused a regression on DCN 3.2 on the IGT test assr-links-suspend, with the dmesg warning:

BUG: sleeping function called from invalid context at include/linux/sched/mm.h:306

in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 449, name: kworker/u64:8
preempt_count: 2, expected: 0
RCU nest depth: 0, expected: 0
Preemption disabled at:
ffffffffc0ce1580>] dc_fpu_begin+0x30/0xd0 [amdgpu]

CPU: 5 PID: 449 Comm: kworker/u64:8 Tainted: G        W          6.8.0+ #35
Hardware name: System manufacturer System Product Name/ROG STRIX X570-E GAMING WIFI II, BIOS 4204 02/24/2022
Workqueue: events_unbound async_run_entry_fn

Reverting this patch will re-introduce stack size warnings.

---
 .../display/dc/resource/dcn32/dcn32_resource.c   | 16 +++++-----------
 .../display/dc/resource/dcn321/dcn321_resource.c | 16 +++++-----------
 2 files changed, 10 insertions(+), 22 deletions(-)
drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +0 -0
index 0f11d7c8791c..022d320be1d5 100644

drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c void dcn32_calculate_wm_and_dlg(struct dc *dc, struct dc_state *context, · line 2012

Quoted 6 lines
drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +0 -3
@@ -2007,27 +2007,21 @@ void dcn32_calculate_wm_and_dlg(struct dc *dc, struct dc_state *context,
 
 static void dcn32_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params)
 {
-       struct dml2_configuration_options *dml2_opt;
-
-       dml2_opt = kmemdup(&dc->dml2_options, sizeof(dc->dml2_options), GFP_KERNEL);

Instead of reverting, you could change this from GFP_KERNEL to GFP_ATOMIC. This aligns with similar allocations in the other resource files. Same comment on the allocation below.

That said, to Arnd's point, I'm not sure about FP in atomic contexts.

Alex

Quoted 66 lines
drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +5 -8
-       if (!dml2_opt)
-               return;
+       struct dml2_configuration_options dml2_opt = dc->dml2_options;
 
        DC_FP_START();
 
        dcn32_update_bw_bounding_box_fpu(dc, bw_params);
 
-       dml2_opt->use_clock_dc_limits = false;
+       dml2_opt.use_clock_dc_limits = false;
        if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2)
-               dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2);
+               dml2_reinit(dc, &dml2_opt, &dc->current_state->bw_ctx.dml2);
 
-       dml2_opt->use_clock_dc_limits = true;
+       dml2_opt.use_clock_dc_limits = true;
        if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2_dc_power_source)
-               dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source);
+               dml2_reinit(dc, &dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source);
 
        DC_FP_END();
-
-       kfree(dml2_opt);
 }
 
 static struct resource_funcs dcn32_res_pool_funcs = {
drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c +5 -12
index 07ca6f58447d..e4b360d89b3b 100644
@@ -1581,27 +1581,21 @@ static struct dc_cap_funcs cap_funcs = {
 
 static void dcn321_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params)
 {
-       struct dml2_configuration_options *dml2_opt;
-
-       dml2_opt = kmemdup(&dc->dml2_options, sizeof(dc->dml2_options), GFP_KERNEL);
-       if (!dml2_opt)
-               return;
+       struct dml2_configuration_options dml2_opt = dc->dml2_options;
 
        DC_FP_START();
 
        dcn321_update_bw_bounding_box_fpu(dc, bw_params);
 
-       dml2_opt->use_clock_dc_limits = false;
+       dml2_opt.use_clock_dc_limits = false;
        if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2)
-               dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2);
+               dml2_reinit(dc, &dml2_opt, &dc->current_state->bw_ctx.dml2);
 
-       dml2_opt->use_clock_dc_limits = true;
+       dml2_opt.use_clock_dc_limits = true;
        if (dc->debug.using_dml2 && dc->current_state && dc->current_state->bw_ctx.dml2_dc_power_source)
-               dml2_reinit(dc, dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source);
+               dml2_reinit(dc, &dml2_opt, &dc->current_state->bw_ctx.dml2_dc_power_source);
 
        DC_FP_END();
-
-       kfree(dml2_opt);
 }
 
 static struct resource_funcs dcn321_res_pool_funcs = {
--

2.34.1