[PATCH 1/2] drm/amd/display: use pre-allocated temp structure for bounding box

[PATCH 1/2] drm/amd/display: use pre-allocated temp structure for bounding box

[Alex Deucher]

This mirrors what the driver does for older DCN generations.

Should fix:

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

Fixes: 88c61827cedc ("drm/amd/display: dynamically allocate dml2_configuration_options structures")
Tested-by: George Zhang <George.zhang@amd.com>
Suggested-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: George Zhang <george.zhang@amd.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: harry.wentland@amd.com
Cc: sunpeng.li@amd.com
Cc: Rodrigo.Siqueira@amd.com
---
 drivers/gpu/drm/amd/display/dc/dc.h                       | 1 +
 .../drm/amd/display/dc/resource/dcn32/dcn32_resource.c    | 8 +++-----
 .../drm/amd/display/dc/resource/dcn321/dcn321_resource.c  | 8 +++-----
 3 files changed, 7 insertions(+), 10 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h
index d0ed01ac460d..d843933ad731 100644
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@@ -1444,6 +1444,7 @@ struct dc {
 	} scratch;
 
 	struct dml2_configuration_options dml2_options;
+	struct dml2_configuration_options dml2_tmp;
 	enum dc_acpi_cm_power_state power_state;
 
 };
diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c
index 0f11d7c8791c..7a8aa9396dea 100644
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c
@@ -2007,11 +2007,9 @@ 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;
+	struct dml2_configuration_options *dml2_opt = &dc->dml2_tmp;
 
-	dml2_opt = kmemdup(&dc->dml2_options, sizeof(dc->dml2_options), GFP_KERNEL);
-	if (!dml2_opt)
-		return;
+	memcpy(dml2_opt, &dc->dml2_options, sizeof(dc->dml2_options));
 
 	DC_FP_START();
 
@@ -2027,7 +2025,7 @@ static void dcn32_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw
 
 	DC_FP_END();
 
-	kfree(dml2_opt);
+	memcpy(&dc->dml2_options, dml2_opt, sizeof(dc->dml2_options));
 }
 
 static struct resource_funcs dcn32_res_pool_funcs = {
diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c
index 07ca6f58447d..ef30e8632607 100644
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c
@@ -1581,11 +1581,9 @@ 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;
+	struct dml2_configuration_options *dml2_opt = &dc->dml2_tmp;
 
-	dml2_opt = kmemdup(&dc->dml2_options, sizeof(dc->dml2_options), GFP_KERNEL);
-	if (!dml2_opt)
-		return;
+	memcpy(dml2_opt, &dc->dml2_options, sizeof(dc->dml2_options));
 
 	DC_FP_START();
 
@@ -1601,7 +1599,7 @@ static void dcn321_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *b
 
 	DC_FP_END();
 
-	kfree(dml2_opt);
+	memcpy(&dc->dml2_options, dml2_opt, sizeof(dc->dml2_options));
 }
 
 static struct resource_funcs dcn321_res_pool_funcs = {
-- 
2.45.1

[PATCH 2/2] drm/amd/display/dcn401: use pre-allocated temp structure for bounding box

[Alex Deucher]

This mirrors what the driver does for older DCN generations.

Should fix:
[   26.924055] BUG: sleeping function called from invalid context at include/linux/sched/mm.h:306
[   26.924060] in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 1022, name: modprobe
[   26.924063] preempt_count: 2, expected: 0
[   26.924064] RCU nest depth: 0, expected: 0
[   26.924066] Preemption disabled at:
[   26.924067] [<ffffffffc089e5e0>] dc_fpu_begin+0x30/0xd0 [amdgpu]
[   26.924322] CPU: 9 PID: 1022 Comm: modprobe Not tainted 6.8.0+ #20
[   26.924325] Hardware name: System manufacturer System Product Name/CROSSHAIR VI HERO, BIOS 6302 10/23/2018
[   26.924326] Call Trace:
[   26.924327]  <TASK>
[   26.924329]  dump_stack_lvl+0x37/0x50
[   26.924333]  ? dc_fpu_begin+0x30/0xd0 [amdgpu]
[   26.924589]  dump_stack+0x10/0x20
[   26.924592]  __might_resched+0x16a/0x1c0
[   26.924596]  __might_sleep+0x42/0x70
[   26.924598]  __kmalloc_node_track_caller+0x2ad/0x4b0
[   26.924601]  ? dm_helpers_allocate_gpu_mem+0x12/0x20 [amdgpu]
[   26.924855]  ? dcn401_update_bw_bounding_box+0x2a/0xf0 [amdgpu]
[   26.925122]  kmemdup+0x20/0x50
[   26.925124]  ? kernel_fpu_begin_mask+0x6b/0xe0
[   26.925127]  ? kmemdup+0x20/0x50
[   26.925129]  dcn401_update_bw_bounding_box+0x2a/0xf0 [amdgpu]
[   26.925393]  dc_create+0x311/0x670 [amdgpu]
[   26.925649]  amdgpu_dm_init+0x2aa/0x1fa0 [amdgpu]
[   26.925903]  ? irq_work_queue+0x38/0x50
[   26.925907]  ? vprintk_emit+0x1e7/0x270
[   26.925910]  ? dev_printk_emit+0x83/0xb0
[   26.925914]  ? amdgpu_device_rreg+0x17/0x20 [amdgpu]
[   26.926133]  dm_hw_init+0x14/0x30 [amdgpu]

Fixes: 669d6b078ed8 ("drm/amd/display: avoid large on-stack structures")
Suggested-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: George Zhang <george.zhang@amd.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: harry.wentland@amd.com
Cc: sunpeng.li@amd.com
Cc: Rodrigo.Siqueira@amd.com
---
 drivers/gpu/drm/amd/display/dc/core/dc_state.c      | 13 ++++++-------
 .../display/dc/resource/dcn401/dcn401_resource.c    |  8 +++-----
 2 files changed, 9 insertions(+), 12 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_state.c b/drivers/gpu/drm/amd/display/dc/core/dc_state.c
index 8ea9391c60b7..aaf6817aaf43 100644
--- a/drivers/gpu/drm/amd/display/dc/core/dc_state.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_state.c
@@ -192,15 +192,14 @@ static void init_state(struct dc *dc, struct dc_state *state)
 /* Public dc_state functions */
 struct dc_state *dc_state_create(struct dc *dc, struct dc_state_create_params *params)
 {
+	struct dc_state *state;
 #ifdef CONFIG_DRM_AMD_DC_FP
-	struct dml2_configuration_options *dml2_opt;
+	struct dml2_configuration_options *dml2_opt = &dc->dml2_tmp;
 
-	dml2_opt = kmemdup(&dc->dml2_options, sizeof(*dml2_opt), GFP_KERNEL);
-	if (!dml2_opt)
-		return NULL;
+	memcpy(dml2_opt, &dc->dml2_options, sizeof(dc->dml2_options));
 #endif
-	struct dc_state *state = kvzalloc(sizeof(struct dc_state),
-			GFP_KERNEL);
+
+	state = kvzalloc(sizeof(struct dc_state), GFP_KERNEL);
 
 	if (!state)
 		return NULL;
@@ -218,7 +217,7 @@ struct dc_state *dc_state_create(struct dc *dc, struct dc_state_create_params *p
 		dml2_create(dc, dml2_opt, &state->bw_ctx.dml2_dc_power_source);
 	}
 
-	kfree(dml2_opt);
+	memcpy(&dc->dml2_options, dml2_opt, sizeof(dc->dml2_options));
 #endif
 
 	kref_init(&state->refcount);
diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c
index 8dfb0a3d21cb..7a61dc83b423 100644
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c
@@ -1581,11 +1581,9 @@ static struct dc_cap_funcs cap_funcs = {
 
 static void dcn401_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params)
 {
-	struct dml2_configuration_options *dml2_opt;
+	struct dml2_configuration_options *dml2_opt = &dc->dml2_tmp;
 
-	dml2_opt = kmemdup(&dc->dml2_options, sizeof(*dml2_opt), GFP_KERNEL);
-	if (!dml2_opt)
-		return;
+	memcpy(dml2_opt, &dc->dml2_options, sizeof(dc->dml2_options));
 
 	DC_FP_START();
 
@@ -1601,7 +1599,7 @@ static void dcn401_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *b
 
 	DC_FP_END();
 
-	kfree(dml2_opt);
+	memcpy(&dc->dml2_options, dml2_opt, sizeof(dc->dml2_options));
 }
 
 enum dc_status dcn401_patch_unknown_plane_state(struct dc_plane_state *plane_state)
-- 
2.45.1

Re: [PATCH 1/2] drm/amd/display: use pre-allocated temp structure for bounding box

[Arnd Bergmann]

On Tue, Jun 4, 2024, at 20:06, Alex Deucher wrote:
> This mirrors what the driver does for older DCN generations.
>
> Should fix:
>
> 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
>
> Fixes: 88c61827cedc ("drm/amd/display: dynamically allocate 
> dml2_configuration_options structures")
> Tested-by: George Zhang <George.zhang@amd.com>
> Suggested-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> Cc: George Zhang <george.zhang@amd.com>
> Cc: Arnd Bergmann <arnd@arndb.de>
> Cc: harry.wentland@amd.com
> Cc: sunpeng.li@amd.com
> Cc: Rodrigo.Siqueira@amd.com

That looks nicer than all the other suggestions, thanks!

Acked-by: Arnd Bergmann <arnd@arndb.de>

One part sticks out though:

> @@ -2027,7 +2025,7 @@ static void dcn32_update_bw_bounding_box(struct 
> dc *dc, struct clk_bw_params *bw
> 
>  	DC_FP_END();
> 
> -	kfree(dml2_opt);
> +	memcpy(&dc->dml2_options, dml2_opt, sizeof(dc->dml2_options));
>  }

The driver did not copy the data back before, so this is a
change in behavior. Is that intentional or a mistake?

If the intention is to have the data copied back into
dc->dml2_options in the end, wouldn't it be easier to
just pass a pointer as in the old version before commit
e779f4587f61 ("drm/amd/display: Add handling for DC power
mode")?

     Arnd

RE: [PATCH 1/2] drm/amd/display: use pre-allocated temp structure for bounding box

[Deucher, Alexander]

[Public]

> -----Original Message-----
> From: Arnd Bergmann <arnd@arndb.de>
> Sent: Tuesday, June 4, 2024 3:43 PM
> To: Deucher, Alexander <Alexander.Deucher@amd.com>; amd-
> gfx@lists.freedesktop.org
> Cc: Zhang, George <George.Zhang@amd.com>; Mahfooz, Hamza
> <Hamza.Mahfooz@amd.com>; Wentland, Harry
> <Harry.Wentland@amd.com>; Li, Sun peng (Leo) <Sunpeng.Li@amd.com>;
> Siqueira, Rodrigo <Rodrigo.Siqueira@amd.com>; Aberback, Joshua
> <Joshua.Aberback@amd.com>
> Subject: Re: [PATCH 1/2] drm/amd/display: use pre-allocated temp structure
> for bounding box
>
> On Tue, Jun 4, 2024, at 20:06, Alex Deucher wrote:
> > This mirrors what the driver does for older DCN generations.
> >
> > Should fix:
> >
> > 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
> >
> > Fixes: 88c61827cedc ("drm/amd/display: dynamically allocate
> > dml2_configuration_options structures")
> > Tested-by: George Zhang <George.zhang@amd.com>
> > Suggested-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
> > Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> > Cc: George Zhang <george.zhang@amd.com>
> > Cc: Arnd Bergmann <arnd@arndb.de>
> > Cc: harry.wentland@amd.com
> > Cc: sunpeng.li@amd.com
> > Cc: Rodrigo.Siqueira@amd.com
>
> That looks nicer than all the other suggestions, thanks!
>
> Acked-by: Arnd Bergmann <arnd@arndb.de>
>
> One part sticks out though:
>
> > @@ -2027,7 +2025,7 @@ static void
> dcn32_update_bw_bounding_box(struct
> > dc *dc, struct clk_bw_params *bw
> >
> >     DC_FP_END();
> >
> > -   kfree(dml2_opt);
> > +   memcpy(&dc->dml2_options, dml2_opt, sizeof(dc->dml2_options));
> >  }
>
> The driver did not copy the data back before, so this is a change in behavior. Is
> that intentional or a mistake?

Yes, sorry, was on autopilot based on the patch Hamza had pointed me to.  Will drop in v2.

Thanks,

Alex

>
> If the intention is to have the data copied back into
> dc->dml2_options in the end, wouldn't it be easier to
> just pass a pointer as in the old version before commit
> e779f4587f61 ("drm/amd/display: Add handling for DC power mode")?
>
>      Arnd