Skip to contentSkip to conversation

drm/exec, drm/gpuvm: Prefer u32 over uint32_t

PATCH In review Thomas Hellström 6 participants 7 messages Jan 19, 2024 – Jan 22, 2024

Subsystems (3)
lore
Share & tools
lore

Thomas Hellström

author reviewer
#0 lore raw

PATCH 1/1 drm/exec, drm/gpuvm: Prefer u32 over uint32_t

The relatively recently introduced drm/exec utility was using uint32_t in its interface, which was then also carried over to drm/gpuvm.

Prefer u32 in new code and update drm/exec and drm/gpuvm accordingly.

Cc Christian König <[email protected]>
Cc Danilo Krummrich <[email protected]>
---
 drivers/gpu/drm/drm_exec.c | 2 +-
 include/drm/drm_exec.h     | 4 ++--
 include/drm/drm_gpuvm.h    | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/drm_exec.c +1 -1
index 5d2809de4517..20e59d88218d 100644
@@ -72,7 +72,7 @@ static void drm_exec_unlock_all(struct drm_exec *exec)
  *
  * Initialize the object and make sure that we can track locked objects.
  */
-void drm_exec_init(struct drm_exec *exec, uint32_t flags)
+void drm_exec_init(struct drm_exec *exec, u32 flags)
 {
 	exec->flags = flags;
 	exec->objects = kmalloc(PAGE_SIZE, GFP_KERNEL);
include/drm/drm_exec.h +2 -2
index b5bf0b6da791..187c3ec44606 100644
@@ -18,7 +18,7 @@ struct drm_exec {
 	/**
 	 * @flags: Flags to control locking behavior
 	 */
-	uint32_t		flags;
+	u32                     flags;
 
 	/**
 	 * @ticket: WW ticket used for acquiring locks
@@ -135,7 +135,7 @@ static inline bool drm_exec_is_contended(struct drm_exec *exec)
 	return !!exec->contended;
 }
 
-void drm_exec_init(struct drm_exec *exec, uint32_t flags);
+void drm_exec_init(struct drm_exec *exec, u32 flags);
 void drm_exec_fini(struct drm_exec *exec);
 bool drm_exec_cleanup(struct drm_exec *exec);
 int drm_exec_lock_obj(struct drm_exec *exec, struct drm_gem_object *obj);
include/drm/drm_gpuvm.h +1 -1
index 48311e6d664c..554046321d24 100644
@@ -514,7 +514,7 @@ struct drm_gpuvm_exec {
 	/**
 	 * @flags: the flags for the struct drm_exec
 	 */
-	uint32_t flags;
+	u32 flags;
 
 	/**
 	 * @vm: the &drm_gpuvm to lock its DMA reservations
Signature · 1 line
2.43.0
Reviewed-by on v1 1/1
#1 lore raw

Am 19.01.24 um 10:05 schrieb Thomas Hellström:

Quoted 8 lines

The relatively recently introduced drm/exec utility was using uint32_t in its interface, which was then also carried over to drm/gpuvm.

Prefer u32 in new code and update drm/exec and drm/gpuvm accordingly.

Cc Christian König <[email protected]>
Cc Danilo Krummrich <[email protected]>
Quoted 54 lines
---
  drivers/gpu/drm/drm_exec.c | 2 +-
  include/drm/drm_exec.h     | 4 ++--
  include/drm/drm_gpuvm.h    | 2 +-
  3 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/drm_exec.c +1 -1
index 5d2809de4517..20e59d88218d 100644
@@ -72,7 +72,7 @@ static void drm_exec_unlock_all(struct drm_exec *exec)
   *
   * Initialize the object and make sure that we can track locked objects.
   */
-void drm_exec_init(struct drm_exec *exec, uint32_t flags)
+void drm_exec_init(struct drm_exec *exec, u32 flags)
  {
  	exec->flags = flags;
  	exec->objects = kmalloc(PAGE_SIZE, GFP_KERNEL);
include/drm/drm_exec.h +2 -2
index b5bf0b6da791..187c3ec44606 100644
@@ -18,7 +18,7 @@ struct drm_exec {
  	/**
  	 * @flags: Flags to control locking behavior
  	 */
-	uint32_t		flags;
+	u32                     flags;
 
  	/**
  	 * @ticket: WW ticket used for acquiring locks
@@ -135,7 +135,7 @@ static inline bool drm_exec_is_contended(struct drm_exec *exec)
  	return !!exec->contended;
  }
 
-void drm_exec_init(struct drm_exec *exec, uint32_t flags);
+void drm_exec_init(struct drm_exec *exec, u32 flags);
  void drm_exec_fini(struct drm_exec *exec);
  bool drm_exec_cleanup(struct drm_exec *exec);
  int drm_exec_lock_obj(struct drm_exec *exec, struct drm_gem_object *obj);
include/drm/drm_gpuvm.h +1 -1
index 48311e6d664c..554046321d24 100644
@@ -514,7 +514,7 @@ struct drm_gpuvm_exec {
  	/**
  	 * @flags: the flags for the struct drm_exec
  	 */
-	uint32_t flags;
+	u32 flags;
 
  	/**
  	 * @vm: the &drm_gpuvm to lock its DMA reservations
#2 lore raw

✗ CI.Patch_applied: failure for drm/exec, drm/gpuvm: Prefer u32 over uint32_t

In reply to Thomas Hellström
== Series Details ==

== Series Details ==

Series: drm/exec, drm/gpuvm: Prefer u32 over uint32_t
URL : https://patchwork.freedesktop.org/series/128973/
State : failure

== Summary ==

=== Applying kernel patches on branch 'drm-tip' with base: ===
Base commit: ad8fc2d4a drm-tip: 2024y-01m-19d-08h-36m-49s UTC integration manifest === git am output follows ===
error: patch failed: drivers/gpu/drm/drm_exec.c:72
error: drivers/gpu/drm/drm_exec.c: patch does not apply
error: patch failed: include/drm/drm_exec.h:135
error: include/drm/drm_exec.h: patch does not apply
hint: Use 'git am --show-current-patch' to see the failed patch
Applying: drm/exec, drm/gpuvm: Prefer u32 over uint32_t
Patch failed at 0001 drm/exec, drm/gpuvm: Prefer u32 over uint32_t When you have resolved this problem, run "git am --continue".
If you prefer to skip this patch, run "git am --skip" instead.
To restore the original branch and stop patching, run "git am --abort".

Reviewed-by on v1 1/1
#3 lore raw
In reply to Thomas Hellström
Quoted 8 lines · Thomas Hellström

The relatively recently introduced drm/exec utility was using uint32_t in its interface, which was then also carried over to drm/gpuvm.

Prefer u32 in new code and update drm/exec and drm/gpuvm accordingly.

Cc Christian König <[email protected]>
Cc Danilo Krummrich <[email protected]>
Quoted 54 lines
---
  drivers/gpu/drm/drm_exec.c | 2 +-
  include/drm/drm_exec.h     | 4 ++--
  include/drm/drm_gpuvm.h    | 2 +-
  3 files changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/drm_exec.c +1 -1
index 5d2809de4517..20e59d88218d 100644
@@ -72,7 +72,7 @@ static void drm_exec_unlock_all(struct drm_exec *exec)
   *
   * Initialize the object and make sure that we can track locked objects.
   */
-void drm_exec_init(struct drm_exec *exec, uint32_t flags)
+void drm_exec_init(struct drm_exec *exec, u32 flags)
  {
  	exec->flags = flags;
  	exec->objects = kmalloc(PAGE_SIZE, GFP_KERNEL);
include/drm/drm_exec.h +2 -2
index b5bf0b6da791..187c3ec44606 100644
@@ -18,7 +18,7 @@ struct drm_exec {
  	/**
  	 * @flags: Flags to control locking behavior
  	 */
-	uint32_t		flags;
+	u32                     flags;
 
  	/**
  	 * @ticket: WW ticket used for acquiring locks
@@ -135,7 +135,7 @@ static inline bool drm_exec_is_contended(struct drm_exec *exec)
  	return !!exec->contended;
  }
 
-void drm_exec_init(struct drm_exec *exec, uint32_t flags);
+void drm_exec_init(struct drm_exec *exec, u32 flags);
  void drm_exec_fini(struct drm_exec *exec);
  bool drm_exec_cleanup(struct drm_exec *exec);
  int drm_exec_lock_obj(struct drm_exec *exec, struct drm_gem_object *obj);
include/drm/drm_gpuvm.h +1 -1
index 48311e6d664c..554046321d24 100644
@@ -514,7 +514,7 @@ struct drm_gpuvm_exec {
  	/**
  	 * @flags: the flags for the struct drm_exec
  	 */
-	uint32_t flags;
+	u32 flags;
 
  	/**
  	 * @vm: the &drm_gpuvm to lock its DMA reservations
Reviewed-by on v1 1/1
#4 lore raw
In reply to Thomas Hellström
Quoted 5 lines · Thomas Hellström

The relatively recently introduced drm/exec utility was using uint32_t in its interface, which was then also carried over to drm/gpuvm.

Prefer u32 in new code and update drm/exec and drm/gpuvm accordingly.

Quoted 8 lines
Cc Christian König <[email protected]>
Cc Danilo Krummrich <[email protected]>

---

drivers/gpu/drm/drm_exec.c | 2 +-
include/drm/drm_exec.h     | 4 ++--
include/drm/drm_gpuvm.h    | 2 +-
3 files changed, 4 insertions(+), 4 deletions(-)

I was surprised we have quite a few places using the c99 types rather than kernel types.

$ git grep -ce uint[0-9][0-9]_t drivers/gpu/drm/*.c
drivers/gpu/drm/drm_atomic.c:1
drivers/gpu/drm/drm_atomic_helper.c:7
drivers/gpu/drm/drm_atomic_state_helper.c:1
drivers/gpu/drm/drm_atomic_uapi.c:17
drivers/gpu/drm/drm_color_mgmt.c:4
drivers/gpu/drm/drm_connector.c:6
drivers/gpu/drm/drm_crtc.c:3
drivers/gpu/drm/drm_damage_helper.c:2
drivers/gpu/drm/drm_debugfs_crc.c:1
drivers/gpu/drm/drm_exec.c:1
drivers/gpu/drm/drm_fb_helper.c:10
drivers/gpu/drm/drm_format_helper.c:6
drivers/gpu/drm/drm_fourcc.c:6
drivers/gpu/drm/drm_framebuffer.c:5
drivers/gpu/drm/drm_gem.c:1
drivers/gpu/drm/drm_gem_dma_helper.c:1
drivers/gpu/drm/drm_gem_shmem_helper.c:1
drivers/gpu/drm/drm_gem_ttm_helper.c:1
drivers/gpu/drm/drm_gem_vram_helper.c:5
drivers/gpu/drm/drm_lease.c:6
drivers/gpu/drm/drm_mipi_dbi.c:3
drivers/gpu/drm/drm_mode_config.c:4
drivers/gpu/drm/drm_mode_object.c:20
drivers/gpu/drm/drm_modeset_helper.c:1
drivers/gpu/drm/drm_modeset_lock.c:1
drivers/gpu/drm/drm_of.c:3
drivers/gpu/drm/drm_plane.c:35
drivers/gpu/drm/drm_plane_helper.c:2
drivers/gpu/drm/drm_prime.c:9
drivers/gpu/drm/drm_probe_helper.c:3
drivers/gpu/drm/drm_property.c:11
drivers/gpu/drm/drm_simple_kms_helper.c:4
drivers/gpu/drm/drm_syncobj.c:26

but maybe not worth the churn for what is already there for a long time?

Lucas De Marchi

Quoted 52 lines
drivers/gpu/drm/drm_exec.c +1 -1
index 5d2809de4517..20e59d88218d 100644
@@ -72,7 +72,7 @@ static void drm_exec_unlock_all(struct drm_exec *exec)
 *
 * Initialize the object and make sure that we can track locked objects.
 */
-void drm_exec_init(struct drm_exec *exec, uint32_t flags)
+void drm_exec_init(struct drm_exec *exec, u32 flags)
 {
 	exec->flags = flags;
 	exec->objects = kmalloc(PAGE_SIZE, GFP_KERNEL);
include/drm/drm_exec.h +2 -2
index b5bf0b6da791..187c3ec44606 100644
@@ -18,7 +18,7 @@ struct drm_exec {
 	/**
 	 * @flags: Flags to control locking behavior
 	 */
-	uint32_t		flags;
+	u32                     flags;
 
 	/**
 	 * @ticket: WW ticket used for acquiring locks
@@ -135,7 +135,7 @@ static inline bool drm_exec_is_contended(struct drm_exec *exec)
 	return !!exec->contended;
 }
 
-void drm_exec_init(struct drm_exec *exec, uint32_t flags);
+void drm_exec_init(struct drm_exec *exec, u32 flags);
 void drm_exec_fini(struct drm_exec *exec);
 bool drm_exec_cleanup(struct drm_exec *exec);
 int drm_exec_lock_obj(struct drm_exec *exec, struct drm_gem_object *obj);
include/drm/drm_gpuvm.h +1 -1
index 48311e6d664c..554046321d24 100644
@@ -514,7 +514,7 @@ struct drm_gpuvm_exec {
 	/**
 	 * @flags: the flags for the struct drm_exec
 	 */
-	uint32_t flags;
+	u32 flags;
 
 	/**
 	 * @vm: the &drm_gpuvm to lock its DMA reservations
Signature · 1 line
2.43.0
#5 lore raw
Quoted 56 lines · Lucas De Marchi
Quoted 5 lines · Thomas Hellström

The relatively recently introduced drm/exec utility was using uint32_t in its interface, which was then also carried over to drm/gpuvm.

Prefer u32 in new code and update drm/exec and drm/gpuvm accordingly.

Quoted 8 lines
Cc Christian König <[email protected]>
Cc Danilo Krummrich <[email protected]>

---

drivers/gpu/drm/drm_exec.c | 2 +-
include/drm/drm_exec.h     | 4 ++--
include/drm/drm_gpuvm.h    | 2 +-
3 files changed, 4 insertions(+), 4 deletions(-)

I was surprised we have quite a few places using the c99 types rather than kernel types.

$ git grep -ce uint[0-9][0-9]_t drivers/gpu/drm/*.c
drivers/gpu/drm/drm_atomic.c:1
drivers/gpu/drm/drm_atomic_helper.c:7
drivers/gpu/drm/drm_atomic_state_helper.c:1
drivers/gpu/drm/drm_atomic_uapi.c:17
drivers/gpu/drm/drm_color_mgmt.c:4
drivers/gpu/drm/drm_connector.c:6
drivers/gpu/drm/drm_crtc.c:3
drivers/gpu/drm/drm_damage_helper.c:2
drivers/gpu/drm/drm_debugfs_crc.c:1
drivers/gpu/drm/drm_exec.c:1
drivers/gpu/drm/drm_fb_helper.c:10
drivers/gpu/drm/drm_format_helper.c:6
drivers/gpu/drm/drm_fourcc.c:6
drivers/gpu/drm/drm_framebuffer.c:5
drivers/gpu/drm/drm_gem.c:1
drivers/gpu/drm/drm_gem_dma_helper.c:1
drivers/gpu/drm/drm_gem_shmem_helper.c:1
drivers/gpu/drm/drm_gem_ttm_helper.c:1
drivers/gpu/drm/drm_gem_vram_helper.c:5
drivers/gpu/drm/drm_lease.c:6
drivers/gpu/drm/drm_mipi_dbi.c:3
drivers/gpu/drm/drm_mode_config.c:4
drivers/gpu/drm/drm_mode_object.c:20
drivers/gpu/drm/drm_modeset_helper.c:1
drivers/gpu/drm/drm_modeset_lock.c:1
drivers/gpu/drm/drm_of.c:3
drivers/gpu/drm/drm_plane.c:35
drivers/gpu/drm/drm_plane_helper.c:2
drivers/gpu/drm/drm_prime.c:9
drivers/gpu/drm/drm_probe_helper.c:3
drivers/gpu/drm/drm_property.c:11
drivers/gpu/drm/drm_simple_kms_helper.c:4
drivers/gpu/drm/drm_syncobj.c:26
Quoted 1 line

but maybe not worth the churn for what is already there for a long time?

Personally, I think the one time churn is worth it to unify and keep the codebase in kernel types only. This is basically what we did in i915 years ago, and new c99 types don't really even creep in because there are zero examples around. It's natural to follow the style around you instead of mixing.

BR,
Jani.

Signature · 1 line
Jani Nikula, Intel
author reviewer
#6 lore raw

Hi,

Quoted 56 lines · Jani Nikula
Quoted 55 lines · Lucas De Marchi
Quoted 5 lines · Thomas Hellström

The relatively recently introduced drm/exec utility was using uint32_t in its interface, which was then also carried over to drm/gpuvm.

Prefer u32 in new code and update drm/exec and drm/gpuvm accordingly.

Quoted 8 lines
Cc Christian König <[email protected]>
Cc Danilo Krummrich <[email protected]>

---

drivers/gpu/drm/drm_exec.c | 2 +-
include/drm/drm_exec.h     | 4 ++--
include/drm/drm_gpuvm.h    | 2 +-
3 files changed, 4 insertions(+), 4 deletions(-)

I was surprised we have quite a few places using the c99 types rather than kernel types.

$ git grep -ce uint[0-9][0-9]_t drivers/gpu/drm/*.c
drivers/gpu/drm/drm_atomic.c:1
drivers/gpu/drm/drm_atomic_helper.c:7
drivers/gpu/drm/drm_atomic_state_helper.c:1
drivers/gpu/drm/drm_atomic_uapi.c:17
drivers/gpu/drm/drm_color_mgmt.c:4
drivers/gpu/drm/drm_connector.c:6
drivers/gpu/drm/drm_crtc.c:3
drivers/gpu/drm/drm_damage_helper.c:2
drivers/gpu/drm/drm_debugfs_crc.c:1
drivers/gpu/drm/drm_exec.c:1
drivers/gpu/drm/drm_fb_helper.c:10
drivers/gpu/drm/drm_format_helper.c:6
drivers/gpu/drm/drm_fourcc.c:6
drivers/gpu/drm/drm_framebuffer.c:5
drivers/gpu/drm/drm_gem.c:1
drivers/gpu/drm/drm_gem_dma_helper.c:1
drivers/gpu/drm/drm_gem_shmem_helper.c:1
drivers/gpu/drm/drm_gem_ttm_helper.c:1
drivers/gpu/drm/drm_gem_vram_helper.c:5
drivers/gpu/drm/drm_lease.c:6
drivers/gpu/drm/drm_mipi_dbi.c:3
drivers/gpu/drm/drm_mode_config.c:4
drivers/gpu/drm/drm_mode_object.c:20
drivers/gpu/drm/drm_modeset_helper.c:1
drivers/gpu/drm/drm_modeset_lock.c:1
drivers/gpu/drm/drm_of.c:3
drivers/gpu/drm/drm_plane.c:35
drivers/gpu/drm/drm_plane_helper.c:2
drivers/gpu/drm/drm_prime.c:9
drivers/gpu/drm/drm_probe_helper.c:3
drivers/gpu/drm/drm_property.c:11
drivers/gpu/drm/drm_simple_kms_helper.c:4
drivers/gpu/drm/drm_syncobj.c:26
Quoted 1 line
Quoted 1 line

but maybe not worth the churn for what is already there for a long time?

This originally dates back to around or slightly after when the drm code was a set of template headers with the objective of sharing code with some bsds, and then I guess it also leaked. The reason I sent this patch was I made a review comment of this for drm_gpuvm and then also promised to send a patch against drm_exec.

Quoted 5 lines

Personally, I think the one time churn is worth it to unify and keep the codebase in kernel types only. This is basically what we did in i915 years ago, and new c99 types don't really even creep in because there are zero examples around. It's natural to follow the style around you instead of mixing.

+1.

/Thomas

Quoted 4 lines

BR,
Jani.