Introduce a ww transaction utility
Subsystems (2)
v2 cover Introduce a ww transaction utility
A ww transaction utility intended to help removing the obj->mm.lock from the driver and introduce ww transactions in a robust way.
Patch 1/2 breaks the current i915 utilities out to separate files Patch 2/2 introduces the ww transaction utility
A similar utility could easily be introduced in the core ww_mutex code, allowing even for cross-driver ww transactions, and the template argument could then allow for per-ww-class derived ww_acquire_mutex types. To facilitate a core implementation, (since we can never guarantee the contended lock stays alive), we'd need a
void ww_mutex_relax(struct ww_acquire_ctx *)
and its interruptible variant that does the equivalent of locking and unlocking the contended mutex.
With this driver implementation, we can extend the code to take a reference on the object containing the contended lock to make sure it stays alive.
Finally, the drawback of the current implementation is the use of the hash table and corresponding performance cost, but as mentioned in patch 2, a core variant could probably do this in a much more efficient way.
v2: Version 1 of patch 2 was obviously a WIP patch. Fix that.
_______________________________________________
Intel-gfx mailing list [email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
RFC PATCH v2 1/2 drm/i915: Break out dma_resv ww locking utilities to separate files
From: Thomas Hellström <[email protected]>
As we're about to add more ww-related functionality, break out the dma_resv ww locking utilities to their own files
--- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/gem/i915_gem_object.h | 1 + drivers/gpu/drm/i915/gt/intel_renderstate.h | 1 + drivers/gpu/drm/i915/i915_gem.c | 64 ------------------ drivers/gpu/drm/i915/i915_gem.h | 15 ----- drivers/gpu/drm/i915/i915_gem_ww.c | 72 +++++++++++++++++++++ drivers/gpu/drm/i915/i915_gem_ww.h | 23 +++++++ 7 files changed, 98 insertions(+), 79 deletions(-) create mode 100644 drivers/gpu/drm/i915/i915_gem_ww.c create mode 100644 drivers/gpu/drm/i915/i915_gem_ww.h
drivers/gpu/drm/i915/Makefile +1 -0
@@ -45,6 +45,7 @@ i915-y += i915_drv.o \ i915_switcheroo.o \ i915_sysfs.o \ i915_utils.o \ + i915_gem_ww.o \ intel_device_info.o \ intel_dram.o \ intel_memory_region.o \
drivers/gpu/drm/i915/gem/i915_gem_object.h +1 -0
@@ -15,6 +15,7 @@#include "i915_gem_object_types.h" #include "i915_gem_gtt.h" #include "i915_vma_types.h" +#include "i915_gem_ww.h" void i915_gem_init__objects (struct drm_i915_private *i915);
drivers/gpu/drm/i915/gt/intel_renderstate.h +1 -0
@@ -26,6 +26,7 @@#include <linux/types.h> #include "i915_gem.h" +#include "i915_gem_ww.h" struct i915_request ;struct intel_context ;
drivers/gpu/drm/i915/i915_gem.c +0 -64
70 lines not on this page · raw message
drivers/gpu/drm/i915/i915_gem.h +0 -15
@@ -27,8 +27,6 @@#include <linux/bug.h> #include <linux/interrupt.h> -#include <linux/stackdepot.h> -#include <linux/stacktrace.h> #include <drm/drm_drv.h> #include "i915_utils.h"
@@ -117,17 +115,4 @@ static inline bool __tasklet_is_scheduled(struct tasklet_struct *t) return test_bit(TASKLET_STATE_SCHED, &t->state); } -struct i915_gem_ww_ctx { -struct ww_acquire_ctx ctx; -struct list_head obj_list; -bool intr; -struct drm_i915_gem_object *contended; -depot_stack_handle_t contended_bt; -}; - -void i915_gem_ww_ctx_init (struct i915_gem_ww_ctx *ctx,bool intr); -void i915_gem_ww_ctx_fini (struct i915_gem_ww_ctx *ctx); -int __must_check i915_gem_ww_ctx_backoff (struct i915_gem_ww_ctx *ctx); -void i915_gem_ww_unlock_single (struct drm_i915_gem_object *obj); -#endif /* __I915_GEM_H__ */
drivers/gpu/drm/i915/i915_gem_ww.c new file +72 -0
72 lines not on this page · raw message
drivers/gpu/drm/i915/i915_gem_ww.h new file +23 -0
@@ -0,0 +1,23 @@ +/* SPDX-License-Identifier: MIT */ +/* +* Copyright © 2020 Intel Corporation +*/ +#ifndef __I915_GEM_WW_H__ +#define __I915_GEM_WW_H__ + +#include <linux/stackdepot.h> +#include <drm/drm_drv.h> + +struct i915_gem_ww_ctx { +struct ww_acquire_ctx ctx; +struct list_head obj_list; +struct drm_i915_gem_object *contended; +depot_stack_handle_t contended_bt; +bool intr; +}; + +void i915_gem_ww_ctx_init (struct i915_gem_ww_ctx *ctx,bool intr); +void i915_gem_ww_ctx_fini (struct i915_gem_ww_ctx *ctx); +int __must_check i915_gem_ww_ctx_backoff (struct i915_gem_ww_ctx *ctx); +void i915_gem_ww_unlock_single (struct drm_i915_gem_object *obj); +#endif
Signature · 6 lines
2.25.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
RFC PATCH v2 2/2 drm/i915: Introduce a i915_gem_do_ww(){} utility
In reply to Thomas Hellström (Intel)From: Thomas Hellström <[email protected]>
With the huge number of sites where multiple-object locking is needed in the driver, it becomes difficult to avoid recursive ww_acquire_ctx initialization, and the function prototypes become bloated passing the ww_acquire_ctx around. Furthermore it's not always easy to get the -EDEADLK handling correct and to follow it.
Introduce a i915_gem_do_ww utility that tries to remedy all these problems by enclosing parts of a ww transaction in the following way:
my_function() {
struct i915_gem_ww_ctx *ww, template;
int err;
bool interruptible = true;
i915_do_ww(ww, &template, err, interruptible) {
err = ww_transaction_part(ww);
}
return err;
}The utility will automatically look up an active ww_acquire_ctx if one is initialized previously in the call chain, and if one found will forward the -EDEADLK instead of handling it, which takes care of the recursive initalization. Using the utility also discourages nested ww unlocking / relocking that is both very fragile and hard to follow.
To look up and register an active ww_acquire_ctx, use a driver-wide hash table for now. But noting that a task could only have a single active ww_acqurie_ctx per ww_class, the active CTX is really task state and a generic version of this utility in the ww_mutex code could thus probably use a quick lookup from a list in the struct task_struct.
--- drivers/gpu/drm/i915/i915_gem_ww.c | 74 ++++++++++++++++++++++++++++- drivers/gpu/drm/i915/i915_gem_ww.h | 56 +++++++++++++++++++++- drivers/gpu/drm/i915/i915_globals.c | 1 + drivers/gpu/drm/i915/i915_globals.h | 1 + 4 files changed, 130 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_gem_ww.c +73 -1
@@ -1,10 +1,12 @@// SPDX-License-Identifier: MIT /* -* Copyright © 2020 Intel Corporation +* Copyright © 2019 Intel Corporation */ +#include <linux/rhashtable.h> #include <linux/dma-resv.h> #include <linux/stacktrace.h> #include "i915_gem_ww.h" +#include "i915_globals.h" #include "gem/i915_gem_object.h" void i915_gem_ww_ctx_init (struct i915_gem_ww_ctx *ww,bool intr)
@@ -70,3 +72,73 @@ int __must_check i915_gem_ww_ctx_backoff(struct i915_gem_ww_ctx *ww)return ret; } + +static struct rhashtable ww_ht; +static const struct rhashtable_params ww_params = { + .key_len =sizeof (struct task_struct *), + .key_offset = offsetof(struct i915_gem_ww_ctx, ctx.task), + .head_offset = offsetof(struct i915_gem_ww_ctx , head), + .min_size =128 , +}; + +static void i915_ww_item_free (void *ptr,void *arg) +{ +WARN_ON_ONCE (1 ); +} + +static void i915_global_ww_exit (void ) +{ +rhashtable_free_and_destroy (&ww_ht, i915_ww_item_free,NULL ); +} + +static void i915_global_ww_shrink (void ) +{ +} + +static struct i915_global global = { + .shrink = i915_global_ww_shrink, + .exit = i915_global_ww_exit, +}; + +int __init i915_global_ww_init (void ) +{ +int ret =rhashtable_init (&ww_ht, &ww_params); + +if (ret) +return ret; + +i915_global_register (&global); + +return 0 ; +} + +void __i915_gem_ww_mark_unused (struct i915_gem_ww_ctx *ww) +{ +GEM_WARN_ON (rhashtable_remove_fast (&ww_ht, &ww->head, ww_params)); +} + +/** +* __i915_gem_ww_locate_or_use - return the task's i915_gem_ww_ctx context +* to use. +* +* @template: The context to use if there was none initialized previously +* in the call chain. +* +* RETURN: The task's i915_gem_ww_ctx context. +*/ +struct i915_gem_ww_ctx * +__i915_gem_ww_locate_or_use (struct i915_gem_ww_ctx *template) +{ +struct i915_gem_ww_ctx *tmp; + +/* ctx.task is the hash key, so set it first. */ + template->ctx.task = current; + +/* +* Ideally we'd just hook the active context to the +* struct task_struct. But for now use a hash table. +*/ + tmp =rhashtable_lookup_get_insert_fast (&ww_ht, &template->head, + ww_params); +return tmp; +}
drivers/gpu/drm/i915/i915_gem_ww.h +55 -1
73 lines not on this page · raw message
drivers/gpu/drm/i915/i915_globals.c +1 -0
@@ -94,6 +94,7 @@ static __initconst int (* const initfn[])(void) = { i915_global_request_init, i915_global_scheduler_init, i915_global_vma_init, + i915_global_ww_init, };int __init i915_globals_init (void )
drivers/gpu/drm/i915/i915_globals.h +1 -0
@@ -34,5 +34,6 @@ int i915_global_objects_init(void);int i915_global_request_init (void );int i915_global_scheduler_init (void );int i915_global_vma_init (void ); +int i915_global_ww_init (void );#endif /* _I915_GLOBALS_H_ */
Signature · 6 lines
2.25.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [RFC PATCH v2 2/2] drm/i915: Introduce a i915_gem_do_ww(){} utility
Quoted 28 lines · Thomas Hellström (Intel)
From: Thomas Hellström <[email protected]>
With the huge number of sites where multiple-object locking is needed in the driver, it becomes difficult to avoid recursive
ww_acquire_ctxinitialization, and the function prototypes become bloated passing theww_acquire_ctxaround. Furthermore it's not always easy to get the-EDEADLKhandling correct and to follow it.Introduce a
i915_gem_do_wwutility that tries to remedy all these problems by enclosing parts of a ww transaction in the following way:my_function() { struct i915_gem_ww_ctx *ww, template; int err; bool interruptible = true; i915_do_ww(ww, &template, err, interruptible) { err = ww_transaction_part(ww); } return err; }The utility will automatically look up an active
ww_acquire_ctxif one is initialized previously in the call chain, and if one found will forward the-EDEADLKinstead of handling it, which takes care of the recursive initalization. Using the utility also discourages nested ww unlocking / relocking that is both very fragile and hard to follow.
Quoted 6 lines
To look up and register an active
ww_acquire_ctx, use a driver-wide hash table for now. But noting that a task could only have a single activeww_acqurie_ctxperww_class, the active CTX is really task state and a generic version of this utility in theww_mutexcode could thus probably use a quick lookup from a list in the structtask_struct.
Maybe a stupid question, but is it safe to assume process context is the only entry point to a ww transaction? I guess I was thinking about things like background scrub/migrate threads, but yes, they would be threads so would work. Other than those I have no ideas who could need locking multiple objects so from this aspect it looks okay.
But it is kind of neat to avoid changing deep hierarchies of function prototypes.
My concern is that the approach isn't to "magicky"? I mean too hidden and too stateful, and that some unwanted surprises could be coming in use with this model. But it is a very vague feeling at this point so don't know.
I also worry that if the graphics subsystem would start thinking it is so special that it needs dedicated handling in task_struct, that it might make it (subsystem) sound a bit pretentious. I had a quick browse through struct task_struct and couldn't spot any precedent to such things so I don't know what core kernel would say.
Regards,
Tvrtko
Quoted 197 lines
Signed-off-by Thomas Hellström <[email protected]>--- drivers/gpu/drm/i915/i915_gem_ww.c | 74 ++++++++++++++++++++++++++++- drivers/gpu/drm/i915/i915_gem_ww.h | 56 +++++++++++++++++++++- drivers/gpu/drm/i915/i915_globals.c | 1 + drivers/gpu/drm/i915/i915_globals.h | 1 + 4 files changed, 130 insertions(+), 2 deletions(-)drivers/gpu/drm/i915/i915_gem_ww.c +73
-1@@ -1,10 +1,12 @@// SPDX-License-Identifier: MIT /* -* Copyright © 2020 Intel Corporation +* Copyright © 2019 Intel Corporation */ +#include <linux/rhashtable.h> #include <linux/dma-resv.h> #include <linux/stacktrace.h> #include "i915_gem_ww.h" +#include "i915_globals.h" #include "gem/i915_gem_object.h" void i915_gem_ww_ctx_init (struct i915_gem_ww_ctx *ww,bool intr)@@ -70,3 +72,73 @@ int __must_check i915_gem_ww_ctx_backoff(struct i915_gem_ww_ctx *ww)return ret; } + +static struct rhashtable ww_ht; +static const struct rhashtable_params ww_params = { + .key_len =sizeof (struct task_struct *), + .key_offset = offsetof(struct i915_gem_ww_ctx, ctx.task), + .head_offset = offsetof(struct i915_gem_ww_ctx , head), + .min_size =128 , +}; + +static void i915_ww_item_free (void *ptr,void *arg) +{ +WARN_ON_ONCE (1 ); +} + +static void i915_global_ww_exit (void ) +{ +rhashtable_free_and_destroy (&ww_ht, i915_ww_item_free,NULL ); +} + +static void i915_global_ww_shrink (void ) +{ +} + +static struct i915_global global = { + .shrink = i915_global_ww_shrink, + .exit = i915_global_ww_exit, +}; + +int __init i915_global_ww_init (void ) +{ +int ret =rhashtable_init (&ww_ht, &ww_params); + +if (ret) +return ret; + +i915_global_register (&global); + +return 0 ; +} + +void __i915_gem_ww_mark_unused (struct i915_gem_ww_ctx *ww) +{ +GEM_WARN_ON (rhashtable_remove_fast (&ww_ht, &ww->head, ww_params)); +} + +/** +* __i915_gem_ww_locate_or_use - return the task's i915_gem_ww_ctx context +* to use. +* +* @template: The context to use if there was none initialized previously +* in the call chain. +* +* RETURN: The task's i915_gem_ww_ctx context. +*/ +struct i915_gem_ww_ctx * +__i915_gem_ww_locate_or_use (struct i915_gem_ww_ctx *template) +{ +struct i915_gem_ww_ctx *tmp; + +/* ctx.task is the hash key, so set it first. */ + template->ctx.task = current; + +/* +* Ideally we'd just hook the active context to the +* struct task_struct. But for now use a hash table. +*/ + tmp =rhashtable_lookup_get_insert_fast (&ww_ht, &template->head, + ww_params); +return tmp; +}drivers/gpu/drm/i915/i915_gem_ww.h +55
-1@@ -6,18 +6,72 @@#define __I915_GEM_WW_H__#include <linux/stackdepot.h> +#include <linux/rhashtable-types.h> #include <drm/drm_drv.h> struct i915_gem_ww_ctx {struct ww_acquire_ctx ctx; +struct rhash_head head;struct list_head obj_list;struct drm_i915_gem_object *contended;depot_stack_handle_t contended_bt; -bool intr; +u32 call_depth; +unsigned short intr; +unsigned short loop; };void i915_gem_ww_ctx_init (struct i915_gem_ww_ctx *ctx,bool intr);void i915_gem_ww_ctx_fini (struct i915_gem_ww_ctx *ctx);int __must_check i915_gem_ww_ctx_backoff (struct i915_gem_ww_ctx *ctx);void i915_gem_ww_unlock_single (struct drm_i915_gem_object *obj); + +/* Internal functions used by the inlines! Don't use. */ +void __i915_gem_ww_mark_unused (struct i915_gem_ww_ctx *ww); +struct i915_gem_ww_ctx * +__i915_gem_ww_locate_or_use (struct i915_gem_ww_ctx *template); + +static inline int __i915_gem_ww_fini (struct i915_gem_ww_ctx *ww,int err) +{ + ww->loop =0 ; +if (ww->call_depth) { + ww->call_depth--; +return err; + } + +if (err == -EDEADLK ) { + err =i915_gem_ww_ctx_backoff (ww); +if (!err) + ww->loop =1 ; + } + +if (!ww->loop) { +i915_gem_ww_ctx_fini (ww); +__i915_gem_ww_mark_unused (ww); + } + +return err; +} + +static inline struct i915_gem_ww_ctx * +__i915_gem_ww_init (struct i915_gem_ww_ctx *template,bool intr) +{ +struct i915_gem_ww_ctx *ww =__i915_gem_ww_locate_or_use (template); + +if (!ww) { + ww = template; + ww->call_depth =0 ; +i915_gem_ww_ctx_init (ww, intr); + }else { + ww->call_depth++; + } + + ww->loop =1 ; + +return ww; +} + +#define i915_gem_do_ww (_ww, _template, _err, _intr) \ + for ((_ww) = __i915_gem_ww_init(_template, _intr); (_ww)->loop; \ + _err = __i915_gem_ww_fini(_ww, _err)) +#endif drivers/gpu/drm/i915/i915_globals.c +1
-0@@ -94,6 +94,7 @@ static __initconst int (* const initfn[])(void) = { i915_global_request_init, i915_global_scheduler_init, i915_global_vma_init, + i915_global_ww_init, };int __init i915_globals_init (void )drivers/gpu/drm/i915/i915_globals.h +0
-0
drivers/gpu/drm/i915/i915_globals.h int i915_global_objects_init(void); · line 39
Quoted 6 lines
drivers/gpu/drm/i915/i915_globals.h +1
-0@@ -34,5 +34,6 @@ int i915_global_objects_init(void);int i915_global_request_init (void );int i915_global_scheduler_init (void );int i915_global_vma_init (void ); +int i915_global_ww_init (void );#endif /* _I915_GLOBALS_H_ */
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] [RFC PATCH v2 2/2] drm/i915: Introduce a i915_gem_do_ww(){} utility
Quoted 37 lines · Tvrtko Ursulin
Quoted 28 lines · Thomas Hellström (Intel)
From: Thomas Hellström <[email protected]>
With the huge number of sites where multiple-object locking is needed in the driver, it becomes difficult to avoid recursive
ww_acquire_ctxinitialization, and the function prototypes become bloated passing theww_acquire_ctxaround. Furthermore it's not always easy to get the-EDEADLKhandling correct and to follow it.Introduce a
i915_gem_do_wwutility that tries to remedy all these problems by enclosing parts of a ww transaction in the following way:
my_function(){
structi915_gem_ww_ctx*ww, template;
int err;
bool interruptible = true;i915_do_ww(ww, &template, err, interruptible) { err =
ww_transaction_part(ww);
}
return err;
}The utility will automatically look up an active
ww_acquire_ctxif one is initialized previously in the call chain, and if one found will forward the-EDEADLKinstead of handling it, which takes care of the recursive initalization. Using the utility also discourages nested ww unlocking / relocking that is both very fragile and hard to follow.Quoted 6 lines
To look up and register an active
ww_acquire_ctx, use a driver-wide hash table for now. But noting that a task could only have a single activeww_acqurie_ctxperww_class, the active CTX is really task state and a generic version of this utility in theww_mutexcode could thus probably use a quick lookup from a list in the structtask_struct.
Quoted 7 lines
Maybe a stupid question, but is it safe to assume process context is the only entry point to a ww transaction? I guess I was thinking about things like background scrub/migrate threads, but yes, they would be threads so would work. Other than those I have no ideas who could need locking multiple objects so from this aspect it looks okay.
But it is kind of neat to avoid changing deep hierarchies of function prototypes.
My concern is that the approach isn't to "magicky"? I mean too hidden and too stateful, and that some unwanted surprises could be coming in use with this model. But it is a very vague feeling at this point so don't know.
I also worry that if the graphics subsystem would start thinking it is so special that it needs dedicated handling in
task_struct, that it might make it (subsystem) sound a bit pretentious. I had a quick browse through structtask_structand couldn't spot any precedent to such things so I don't know what core kernel would say.
Thanks for taking a look Tvrtko.
Yes, I've discussed with Maarten and we'll favour changing prototype for now, although for code clarity and simplicity we could still use the i915_do_ww() notation.
For any upsteam move we would have to move the utility to locking / ww_mutex, (with driver-specific adaption still possible, so any additional task struct fields will not belong to graphics but rather to locking) so I'll post a RFC for a generic utility on dri-devel, time permitting, to see if there is any interest.
/Thomas
Quoted 197 lines
Regards,
Tvrtko
Quoted 192 lines
Signed-off-by Thomas Hellström <[email protected]>---
drivers/gpu/drm/i915/i915_gem_ww.c | 74 ++++++++++++++++++++++++++++- drivers/gpu/drm/i915/i915_gem_ww.h | 56 +++++++++++++++++++++-drivers/gpu/drm/i915/i915_globals.c| 1 +drivers/gpu/drm/i915/i915_globals.h| 1 +4 files changed, 130 insertions(+), 2 deletions(-)drivers/gpu/drm/i915/i915_gem_ww.c +50
-1@@ -1,10 +1,12 @@ // SPDX-License-Identifier: MIT /* - * Copyright © 2020 Intel Corporation +* Copyright © 2019 Intel Corporation */ +#include <linux/rhashtable.h> #include <linux/dma-resv.h> #include <linux/stacktrace.h> #include "i915_gem_ww.h" +#include "i915_globals.h" #include "gem/i915_gem_object.h" void i915_gem_ww_ctx_init(struct i915_gem_ww_ctx *ww, bool intr)@@ -70,3 +72,73 @@ int __must_check i915_gem_ww_ctx_backoff(struct i915_gem_ww_ctx *ww) return ret; } + +static struct rhashtable ww_ht; +static const struct rhashtable_params ww_params = { + .key_len = sizeof(struct task_struct *), + .key_offset = offsetof(struct i915_gem_ww_ctx, ctx.task), + .head_offset = offsetof(struct i915_gem_ww_ctx, head), + .min_size = 128, +}; + +static void i915_ww_item_free (void *ptr,void *arg) +{ + WARN_ON_ONCE(1); +} + +static void i915_global_ww_exit (void ) +{ + rhashtable_free_and_destroy(&ww_ht, i915_ww_item_free, NULL); +} + +static void i915_global_ww_shrink (void ) +{ +} + +static struct i915_global global = { + .shrink = i915_global_ww_shrink, + .exit = i915_global_ww_exit, +}; + +int __init i915_global_ww_init (void ) +{ + int ret = rhashtable_init(&ww_ht, &ww_params); + + if (ret) + return ret; + + i915_global_register(&global); + + return 0; +} + +void __i915_gem_ww_mark_unused (struct i915_gem_ww_ctx *ww) +{ + GEM_WARN_ON(rhashtable_remove_fast(&ww_ht, &ww->head, ww_params)); +} + +/** +* __i915_gem_ww_locate_or_use - return the task's i915_gem_ww_ctx context
+ * to use.
+ *
+ * @template: The context to use if there was none initialized previously
+ * in the call chain.
+ *
+ * RETURN: The task'si915_gem_ww_ctxcontext.
+ */drivers/gpu/drm/i915/i915_gem_ww.c +16
-0+struct i915_gem_ww_ctx * +__i915_gem_ww_locate_or_use (struct i915_gem_ww_ctx *template) +{ + struct i915_gem_ww_ctx *tmp; + + /* ctx.task is the hash key, so set it first. */ + template->ctx.task = current; + + /* + * Ideally we'd just hook the active context to the + * struct task_struct. But for now use a hash table. + */ + tmp = rhashtable_lookup_get_insert_fast(&ww_ht, &template->head, + ww_params); + return tmp; +}drivers/gpu/drm/i915/i915_gem_ww.h +55
-1@@ -6,18 +6,72 @@ #define __I915_GEM_WW_H__ #include <linux/stackdepot.h> +#include <linux/rhashtable-types.h> #include <drm/drm_drv.h> struct i915_gem_ww_ctx { struct ww_acquire_ctx ctx; + struct rhash_head head; struct list_head obj_list; struct drm_i915_gem_object *contended; depot_stack_handle_t contended_bt; - bool intr; + u32 call_depth; + unsigned short intr; + unsigned short loop; }; void i915_gem_ww_ctx_init(struct i915_gem_ww_ctx *ctx, bool intr); void i915_gem_ww_ctx_fini(struct i915_gem_ww_ctx *ctx); int __must_check i915_gem_ww_ctx_backoff(struct i915_gem_ww_ctx *ctx); void i915_gem_ww_unlock_single(struct drm_i915_gem_object *obj); + +/* Internal functions used by the inlines! Don't use. */ +void __i915_gem_ww_mark_unused (struct i915_gem_ww_ctx *ww); +struct i915_gem_ww_ctx * +__i915_gem_ww_locate_or_use (struct i915_gem_ww_ctx *template); + +static inline int __i915_gem_ww_fini (struct i915_gem_ww_ctx *ww,int err) +{ + ww->loop = 0; + if (ww->call_depth) { + ww->call_depth--; + return err; + } + + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(ww); + if (!err) + ww->loop = 1; + } + + if (!ww->loop) { + i915_gem_ww_ctx_fini(ww); + __i915_gem_ww_mark_unused(ww); + } + + return err; +} + +static inline struct i915_gem_ww_ctx * +__i915_gem_ww_init (struct i915_gem_ww_ctx *template,bool intr) +{ + struct i915_gem_ww_ctx *ww = __i915_gem_ww_locate_or_use(template); + + if (!ww) { + ww = template; + ww->call_depth = 0; + i915_gem_ww_ctx_init(ww, intr); + } else { + ww->call_depth++; + } + + ww->loop = 1; + + return ww; +} + +#define i915_gem_do_ww (_ww, _template, _err, _intr) \ + for ((_ww) = __i915_gem_ww_init(_template, _intr); (_ww)->loop; \ + _err = __i915_gem_ww_fini(_ww, _err)) + #endifdrivers/gpu/drm/i915/i915_globals.c +1
-0@@ -94,6 +94,7 @@ static __initconst int (* const initfn[])(void) = { i915_global_request_init, i915_global_scheduler_init, i915_global_vma_init, + i915_global_ww_init, }; int __init i915_globals_init(void)drivers/gpu/drm/i915/i915_globals.h +0
-1--- a/drivers/gpu/drm/i915/i915_globals.h
Quoted 8 lines
drivers/gpu/drm/i915/i915_globals.h int i915_global_objects_init(void); · line 39
Quoted 8 lines
+++
b/drivers/gpu/drm/i915/i915_globals.hdrivers/gpu/drm/i915/i915_globals.h +1
-0@@ -34,5 +34,6 @@ int i915_global_objects_init(void); int i915_global_request_init(void); int i915_global_scheduler_init(void); int i915_global_vma_init(void); +int i915_global_ww_init (void ); #endif /* _I915_GLOBALS_H_ */
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI[Intel-gfx] ✗ Fi.CI.BUILD: failure for Introduce a ww transaction utility (rev2)
In reply to Thomas Hellström (Intel)== Series Details ==
== Series Details ==
Series: Introduce a ww transaction utility (rev2)
URL : https://patchwork.freedesktop.org/series/81787/
State : failure
== Summary ==
Applying: drm/i915: Break out dma_resv ww locking utilities to separate files
error: sha1 information is lacking or useless (drivers/gpu/drm/i915/gem/i915_gem_object.h).
error: could not build fake ancestor
hint: Use 'git am --show-current-patch=diff' to see the failed patch
Patch failed at 0001 drm/i915: Break out dma_resv ww locking utilities to separate files 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".
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx