All of lore.kernel.org
 help / color / mirror / Atom feed
From: Daniel Vetter <daniel.vetter@ffwll.ch>
To: DRI Development <dri-devel@lists.freedesktop.org>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>,
	Thierry Reding <treding@nvidia.com>
Subject: [PATCH] drm: Rip out totally bogus vga_switcheroo->can_switch locking
Date: Tue, 22 Apr 2014 22:45:19 +0200	[thread overview]
Message-ID: <1398199519-21821-1-git-send-email-daniel.vetter@ffwll.ch> (raw)
In-Reply-To: <20140417145600.GH550@ulmo>

So I just wanted to add a new field to struct drm_device and
accidentally stumbled over something. According to comments
dev->open_count is protected by dev->count_lock, but that's totally
not the case. It's protected by drm_global_mutex.

Unfortunately the vga switcheroo callbacks took this comment at face
value. The problem is that we can't just take the drm_global_mutex
because:
- It would lead to a locking inversion with the driver load/unload
  paths.
- It wouldn't actually protect anything, for that we'd need to wrap
  the entire vga switcheroo code in the drm_global_mutex. And I'm not
  sure whether that would actually solve anything.

What we probably want is a try_to_grab_switcheroo reference kind of
thing which is used in the driver's ->open callback. Then we could
move all that ->can_switch madness into the vga switcheroo core where
it really belongs.

But since that would amount to real work take the easy way out and
just add a comment. It's definitely not going to make anything worse
since doing switcheroo state changes while restarting X just isn't
recommended. Even though the delayed switching code does exactly that.

v2:
- Simplify the ->can_switch implementations more (Thierry)
- Fix comment about the dev->open_count locking (Thierry)

Cc: Thierry Reding <treding@nvidia.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> (v1)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_dma.c        | 11 ++++++-----
 drivers/gpu/drm/nouveau/nouveau_vga.c  | 11 ++++++-----
 drivers/gpu/drm/radeon/radeon_device.c | 11 ++++++-----
 include/drm/drmP.h                     |  2 +-
 4 files changed, 19 insertions(+), 16 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 96177eec0a0e..283ff06001bc 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -1277,12 +1277,13 @@ static void i915_switcheroo_set_state(struct pci_dev *pdev, enum vga_switcheroo_
 static bool i915_switcheroo_can_switch(struct pci_dev *pdev)
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
-	bool can_switch;
 
-	spin_lock(&dev->count_lock);
-	can_switch = (dev->open_count == 0);
-	spin_unlock(&dev->count_lock);
-	return can_switch;
+	/*
+	 * FIXME: open_count is protected by drm_global_mutex but that would lead to
+	 * locking inversion with the driver load path. And the access here is
+	 * completely racy anyway. So don't bother with locking for now.
+	 */
+	return dev->open_count == 0;
 }
 
 static const struct vga_switcheroo_client_ops i915_switcheroo_ops = {
diff --git a/drivers/gpu/drm/nouveau/nouveau_vga.c b/drivers/gpu/drm/nouveau/nouveau_vga.c
index fb84da3cb50d..4f4c3fec6916 100644
--- a/drivers/gpu/drm/nouveau/nouveau_vga.c
+++ b/drivers/gpu/drm/nouveau/nouveau_vga.c
@@ -64,12 +64,13 @@ static bool
 nouveau_switcheroo_can_switch(struct pci_dev *pdev)
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
-	bool can_switch;
 
-	spin_lock(&dev->count_lock);
-	can_switch = (dev->open_count == 0);
-	spin_unlock(&dev->count_lock);
-	return can_switch;
+	/*
+	 * FIXME: open_count is protected by drm_global_mutex but that would lead to
+	 * locking inversion with the driver load path. And the access here is
+	 * completely racy anyway. So don't bother with locking for now.
+	 */
+	return dev->open_count == 0;
 }
 
 static const struct vga_switcheroo_client_ops
diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index 511fe26198e4..9aa1afd1786e 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -1125,12 +1125,13 @@ static void radeon_switcheroo_set_state(struct pci_dev *pdev, enum vga_switchero
 static bool radeon_switcheroo_can_switch(struct pci_dev *pdev)
 {
 	struct drm_device *dev = pci_get_drvdata(pdev);
-	bool can_switch;
 
-	spin_lock(&dev->count_lock);
-	can_switch = (dev->open_count == 0);
-	spin_unlock(&dev->count_lock);
-	return can_switch;
+	/*
+	 * FIXME: open_count is protected by drm_global_mutex but that would lead to
+	 * locking inversion with the driver load path. And the access here is
+	 * completely racy anyway. So don't bother with locking for now.
+	 */
+	return dev->open_count == 0;
 }
 
 static const struct vga_switcheroo_client_ops radeon_switcheroo_ops = {
diff --git a/include/drm/drmP.h b/include/drm/drmP.h
index 9f1fb8d36b67..a20d882ca265 100644
--- a/include/drm/drmP.h
+++ b/include/drm/drmP.h
@@ -1076,7 +1076,7 @@ struct drm_device {
 
 	/** \name Usage Counters */
 	/*@{ */
-	int open_count;			/**< Outstanding files open */
+	int open_count;			/**< Outstanding files open, protected by drm_global_mutex. */
 	int buf_use;			/**< Buffers in use -- cannot alloc */
 	atomic_t buf_alloc;		/**< Buffer allocation in progress */
 	/*@} */
-- 
1.9.2

  reply	other threads:[~2014-04-22 20:45 UTC|newest]

Thread overview: 52+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-04-11 21:35 [PATCH 00/18] drm_bus cleanups and other cruft removal Daniel Vetter
2014-04-11 21:35 ` [PATCH 01/18] drm/omap: fix up pdev_remove Daniel Vetter
2014-04-12 15:26   ` Rob Clark
2014-04-14  6:26   ` Tomi Valkeinen
2014-04-16  8:15     ` Daniel Vetter
2014-04-16  8:27       ` Tomi Valkeinen
2014-04-16 17:05         ` Daniel Vetter
2014-04-11 21:35 ` [PATCH 02/18] drm/irq: simplify irq checks in drm_wait_vblank Daniel Vetter
2014-04-14 20:55   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 03/18] drm/pci: fold in irq_by_busid support Daniel Vetter
2014-04-17 14:33   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 04/18] drm/irq: drm_control is a legacy ioctl, so pci devices only Daniel Vetter
2014-04-17 14:38   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 05/18] drm/irq: remove cargo-culted locking from irq_install/unistall Daniel Vetter
2014-04-17 14:43   ` Thierry Reding
2014-04-17 14:44     ` Thierry Reding
2014-04-11 21:36 ` [PATCH 06/18] drm: remove drm_dev_to_irq from drivers Daniel Vetter
2014-04-17 14:47   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 07/18] drm: kill drm_bus->bus_type Daniel Vetter
2014-04-17 14:50   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 08/18] drm: Rip out totally bogus vga_switcheroo->can_switch locking Daniel Vetter
2014-04-17 14:56   ` Thierry Reding
2014-04-22 20:45     ` Daniel Vetter [this message]
2014-04-23  7:28       ` [PATCH] " Thierry Reding
2014-04-11 21:36 ` [PATCH 09/18] drm: rename dev->count_lock to dev->buf_lock Daniel Vetter
2014-04-17 14:57   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 10/18] drm/irq: track the irq installed in drm_irq_install in dev->irq Daniel Vetter
2014-04-16 21:57   ` Laurent Pinchart
2014-04-16 22:02   ` Laurent Pinchart
2014-04-22  9:46     ` Daniel Vetter
2014-04-22 10:49       ` Thierry Reding
2014-04-17 15:03   ` Thierry Reding
2014-04-22 20:44   ` [PATCH] " Daniel Vetter
2014-04-23  7:27     ` Thierry Reding
2014-04-23  8:31       ` Daniel Vetter
2014-04-11 21:36 ` [PATCH 11/18] drm/irq: Look up the pci irq directly in the drm_control ioctl Daniel Vetter
2014-04-11 21:36 ` [PATCH 12/18] drm: pass the irq explicitly to drm_irq_install Daniel Vetter
2014-04-17 15:06   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 13/18] drm: remove bus->get_irq implementations Daniel Vetter
2014-04-17 15:07   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 14/18] drm: inline drm_pci_set_unique Daniel Vetter
2014-04-17 15:10   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 15/18] drm: rip out dev->devname Daniel Vetter
2014-04-17 15:11   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 16/18] drm: remove drm_bus->get_name Daniel Vetter
2014-04-17 15:12   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 17/18] drm: Remove dev->kdriver Daniel Vetter
2014-04-17 15:13   ` Thierry Reding
2014-04-11 21:36 ` [PATCH 18/18] drm/<drivers>: don't set driver->dev_priv_size to 0 Daniel Vetter
2014-04-17 15:14   ` Thierry Reding
2014-04-16 22:10 ` [PATCH 00/18] drm_bus cleanups and other cruft removal Laurent Pinchart
2014-04-22  7:30 ` David Herrmann

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=1398199519-21821-1-git-send-email-daniel.vetter@ffwll.ch \
    --to=daniel.vetter@ffwll.ch \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=treding@nvidia.com \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.