Intel-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Mathias Krause <minipli@grsecurity.net>
To: intel-gfx@lists.freedesktop.org
Cc: Mathias Krause <minipli@grsecurity.net>
Subject: [Intel-gfx] [PATCH v2 2/2] drm/i915: Clarify type evolution of uabi_node/uabi_engines
Date: Thu, 28 Sep 2023 20:20:19 +0200	[thread overview]
Message-ID: <20230928182019.10256-3-minipli@grsecurity.net> (raw)
In-Reply-To: <20230928182019.10256-1-minipli@grsecurity.net>

Chaining user engines happens in multiple passes during driver
initialization, mutating its type along the way. It starts off with a
simple lock-less linked list (struct llist_node/head) populated by
intel_engine_add_user() which later gets sorted and converted to an
intermediate regular list (struct list_head) just to be converted once
more to its final rb-tree structure (struct rb_node/root) in
intel_engines_driver_register().

All of these types overlay the uabi_node/uabi_engines members which is
unfortunate but safe if one takes care about using the rb-tree based
structure only after the conversion has completed. However, mistakes
happen and commit 1ec23ed7126e ("drm/i915: Use uabi engines for the
default engine map") violated that assumption, as the multiple type
evolution was all to easy hidden behind casts papering over it.

Make the type evolution of uabi_node/uabi_engines more visible by
putting all members into an anonymous union and use the correctly typed
member in its various users. This allows us to drop quite some ugly
casts and, hopefully, make the evolution of the members better
recognisable to avoid future mistakes.

Signed-off-by: Mathias Krause <minipli@grsecurity.net>
---
 drivers/gpu/drm/i915/gt/intel_engine_types.h | 10 +++++++++-
 drivers/gpu/drm/i915/gt/intel_engine_user.c  | 17 +++++++----------
 drivers/gpu/drm/i915/i915_drv.h              | 17 ++++++++++++++++-
 3 files changed, 32 insertions(+), 12 deletions(-)

diff --git a/drivers/gpu/drm/i915/gt/intel_engine_types.h b/drivers/gpu/drm/i915/gt/intel_engine_types.h
index a7e677598004..7585fffac60b 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine_types.h
+++ b/drivers/gpu/drm/i915/gt/intel_engine_types.h
@@ -402,7 +402,15 @@ struct intel_engine_cs {
 
 	unsigned long context_tag;
 
-	struct rb_node uabi_node;
+	/*
+	 * The type evolves during initialization, see related comment for
+	 * struct drm_i915_private's uabi_engines member.
+	 */
+	union {
+		struct llist_node uabi_llist;
+		struct list_head uabi_list;
+		struct rb_node uabi_node;
+	};
 
 	struct intel_sseu sseu;
 
diff --git a/drivers/gpu/drm/i915/gt/intel_engine_user.c b/drivers/gpu/drm/i915/gt/intel_engine_user.c
index dcedff41a825..118164ddbb2e 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine_user.c
+++ b/drivers/gpu/drm/i915/gt/intel_engine_user.c
@@ -38,8 +38,7 @@ intel_engine_lookup_user(struct drm_i915_private *i915, u8 class, u8 instance)
 
 void intel_engine_add_user(struct intel_engine_cs *engine)
 {
-	llist_add((struct llist_node *)&engine->uabi_node,
-		  (struct llist_head *)&engine->i915->uabi_engines);
+	llist_add(&engine->uabi_llist, &engine->i915->uabi_engines_llist);
 }
 
 static const u8 uabi_classes[] = {
@@ -54,9 +53,9 @@ static int engine_cmp(void *priv, const struct list_head *A,
 		      const struct list_head *B)
 {
 	const struct intel_engine_cs *a =
-		container_of((struct rb_node *)A, typeof(*a), uabi_node);
+		container_of(A, typeof(*a), uabi_list);
 	const struct intel_engine_cs *b =
-		container_of((struct rb_node *)B, typeof(*b), uabi_node);
+		container_of(B, typeof(*b), uabi_list);
 
 	if (uabi_classes[a->class] < uabi_classes[b->class])
 		return -1;
@@ -73,7 +72,7 @@ static int engine_cmp(void *priv, const struct list_head *A,
 
 static struct llist_node *get_engines(struct drm_i915_private *i915)
 {
-	return llist_del_all((struct llist_head *)&i915->uabi_engines);
+	return llist_del_all(&i915->uabi_engines_llist);
 }
 
 static void sort_engines(struct drm_i915_private *i915,
@@ -83,9 +82,8 @@ static void sort_engines(struct drm_i915_private *i915,
 
 	llist_for_each_safe(pos, next, get_engines(i915)) {
 		struct intel_engine_cs *engine =
-			container_of((struct rb_node *)pos, typeof(*engine),
-				     uabi_node);
-		list_add((struct list_head *)&engine->uabi_node, engines);
+			container_of(pos, typeof(*engine), uabi_llist);
+		list_add(&engine->uabi_list, engines);
 	}
 	list_sort(NULL, engines, engine_cmp);
 }
@@ -213,8 +211,7 @@ void intel_engines_driver_register(struct drm_i915_private *i915)
 	p = &i915->uabi_engines.rb_node;
 	list_for_each_safe(it, next, &engines) {
 		struct intel_engine_cs *engine =
-			container_of((struct rb_node *)it, typeof(*engine),
-				     uabi_node);
+			container_of(it, typeof(*engine), uabi_list);
 
 		if (intel_gt_has_unrecoverable_error(engine->gt))
 			continue; /* ignore incomplete engines */
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 7a8ce7239bc9..c8690d1d5e51 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -222,7 +222,22 @@ struct drm_i915_private {
 		bool mchbar_need_disable;
 	} gmch;
 
-	struct rb_root uabi_engines;
+	/*
+	 * Chaining user engines happens in multiple stages, starting with a
+	 * simple lock-less linked list created by intel_engine_add_user(),
+	 * which later gets sorted and converted to an intermediate regular
+	 * list, just to be converted once again to its final rb tree structure
+	 * in intel_engines_driver_register().
+	 *
+	 * Make sure to use the right iterator helper, depending on if the code
+	 * in question runs before or after intel_engines_driver_register() --
+	 * for_each_uabi_engine() can only be used afterwards!
+	 */
+	union {
+		struct llist_head uabi_engines_llist;
+		struct list_head uabi_engines_list;
+		struct rb_root uabi_engines;
+	};
 	unsigned int engine_uabi_class_count[I915_LAST_UABI_ENGINE_CLASS + 1];
 
 	/* protects the irq masks */
-- 
2.39.2


  parent reply	other threads:[~2023-09-28 18:20 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-09-28 18:20 [Intel-gfx] [PATCH v2 0/2] drm/i915: fix rb-tree/llist/list confusion Mathias Krause
2023-09-28 18:20 ` [Intel-gfx] [PATCH v2 1/2] drm/i915: Register engines early to avoid type confusion Mathias Krause
2023-09-29 10:57   ` Tvrtko Ursulin
2023-09-28 18:20 ` Mathias Krause [this message]
2023-09-29 11:00   ` [Intel-gfx] [PATCH v2 2/2] drm/i915: Clarify type evolution of uabi_node/uabi_engines Tvrtko Ursulin
2023-10-03  7:32     ` Tvrtko Ursulin
2023-10-04  9:52       ` Mathias Krause
2023-10-04 10:44         ` Tvrtko Ursulin
2023-09-28 19:29 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915: fix rb-tree/llist/list confusion (rev2) Patchwork
2023-09-28 19:29 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2023-09-28 19:48 ` [Intel-gfx] ✗ Fi.CI.BAT: failure " Patchwork
2023-09-29  7:55 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915: fix rb-tree/llist/list confusion (rev3) Patchwork
2023-09-29  7:55 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2023-09-29  8:07 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2023-09-29 12:39 ` [Intel-gfx] ✗ Fi.CI.IGT: failure " Patchwork
2023-10-02 19:02 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for drm/i915: fix rb-tree/llist/list confusion (rev4) Patchwork
2023-10-02 19:02 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2023-10-02 19:12 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2023-10-02 20:42 ` [Intel-gfx] ✗ Fi.CI.IGT: failure " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230928182019.10256-3-minipli@grsecurity.net \
    --to=minipli@grsecurity.net \
    --cc=intel-gfx@lists.freedesktop.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox