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>
Subject: [PATCH 35/50] drm: Rip out totally bogus vga_switcheroo->can_switch locking
Date: Wed, 11 Dec 2013 11:34:56 +0100	[thread overview]
Message-ID: <1386758111-3446-36-git-send-email-daniel.vetter@ffwll.ch> (raw)
In-Reply-To: <1386758111-3446-1-git-send-email-daniel.vetter@ffwll.ch>

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.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_dma.c        | 7 +++++--
 drivers/gpu/drm/nouveau/nouveau_vga.c  | 7 +++++--
 drivers/gpu/drm/radeon/radeon_device.c | 7 +++++--
 include/drm/drmP.h                     | 2 +-
 4 files changed, 16 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
index 7f653ab59db4..2babb597d23c 100644
--- a/drivers/gpu/drm/i915/i915_dma.c
+++ b/drivers/gpu/drm/i915/i915_dma.c
@@ -1269,9 +1269,12 @@ 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);
+	/*
+	 * 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.
+	 */
 	can_switch = (dev->open_count == 0);
-	spin_unlock(&dev->count_lock);
 	return can_switch;
 }
 
diff --git a/drivers/gpu/drm/nouveau/nouveau_vga.c b/drivers/gpu/drm/nouveau/nouveau_vga.c
index 81638d7f2eff..db721b4d778b 100644
--- a/drivers/gpu/drm/nouveau/nouveau_vga.c
+++ b/drivers/gpu/drm/nouveau/nouveau_vga.c
@@ -64,9 +64,12 @@ nouveau_switcheroo_can_switch(struct pci_dev *pdev)
 	struct drm_device *dev = pci_get_drvdata(pdev);
 	bool can_switch;
 
-	spin_lock(&dev->count_lock);
+	/*
+	 * 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.
+	 */
 	can_switch = (dev->open_count == 0);
-	spin_unlock(&dev->count_lock);
 	return can_switch;
 }
 
diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index 39b033b441d2..d67140811ca5 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -1119,9 +1119,12 @@ 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);
+	/*
+	 * 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.
+	 */
 	can_switch = (dev->open_count == 0);
-	spin_unlock(&dev->count_lock);
 	return can_switch;
 }
 
diff --git a/include/drm/drmP.h b/include/drm/drmP.h
index 1016ecc8de95..ee65c25e25d3 100644
--- a/include/drm/drmP.h
+++ b/include/drm/drmP.h
@@ -1085,7 +1085,7 @@ struct drm_device {
 
 	/** \name Usage Counters */
 	/*@{ */
-	int open_count;			/**< Outstanding files open */
+	int open_count;			/**< Outstanding files open, protected by drm_global_lock. */
 	atomic_t ioctl_count;		/**< Outstanding IOCTLs pending */
 	atomic_t vma_count;		/**< Outstanding vma areas open */
 	int buf_use;			/**< Buffers in use -- cannot alloc */
-- 
1.8.4.3

  parent reply	other threads:[~2013-12-11 10:35 UTC|newest]

Thread overview: 79+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-12-11 10:34 [PATCH 00/50] more drm de-midlayering Daniel Vetter
2013-12-11 10:34 ` [PATCH 01/50] drm/rcar: call drm_put_dev directly in the ->remove hook Daniel Vetter
2013-12-11 10:34 ` [PATCH 02/50] drm/exynos: call drm_put_dev directly from ->remove Daniel Vetter
2013-12-11 10:34 ` [PATCH 03/50] drm/imx: directly call drm_put_dev in ->remove Daniel Vetter
2013-12-11 10:34 ` [PATCH 04/50] drm/tilcdc: call drm_put_dev directly from ->remove Daniel Vetter
2013-12-11 19:39   ` Rob Clark
2013-12-11 10:34 ` [PATCH 05/50] drm/omap: call drm_put_dev directly in ->remove Daniel Vetter
2013-12-11 12:02   ` Rob Clark
2013-12-11 13:20     ` [PATCH] " Daniel Vetter
2013-12-11 19:35       ` Rob Clark
2013-12-11 10:34 ` [PATCH 06/50] drm/shmob: call drm_put_dev directly from ->remove hook Daniel Vetter
2013-12-11 12:21   ` Laurent Pinchart
2013-12-11 10:34 ` [PATCH 07/50] drm/armada: directly call drm_put_dev in ->remove Daniel Vetter
2013-12-11 10:34 ` [PATCH 08/50] drm/msm: call drm_put_dev directly " Daniel Vetter
2013-12-11 19:38   ` Rob Clark
2013-12-11 10:34 ` [PATCH 09/50] drm: rip out drm_platform_exit Daniel Vetter
2013-12-11 10:34 ` [PATCH 10/50] drm: restrict the device list for shadow attached drivers Daniel Vetter
2013-12-11 10:34 ` [PATCH 11/50] drm/bufs: remove handling of _DRM_GEM mappings Daniel Vetter
2013-12-11 10:34 ` [PATCH 12/50] drm: kill DRIVER_REQUIRE_AGP Daniel Vetter
2013-12-11 10:34 ` [PATCH 13/50] drm: ->agp_init can't fail Daniel Vetter
2013-12-11 10:34 ` [PATCH 14/50] drm: rip out drm_core_has_AGP Daniel Vetter
2013-12-11 10:34 ` [PATCH 15/50] drm: remove agp_init() bus callback Daniel Vetter
2013-12-11 10:34 ` [PATCH 16/50] drm: inline drm_agp_destroy Daniel Vetter
2013-12-11 10:34 ` [PATCH 17/50] drm: kill the ->agp_destroy callback Daniel Vetter
2013-12-11 10:34 ` [PATCH 18/50] drm: remove global_mutex locking around agp_init Daniel Vetter
2013-12-11 10:34 ` [PATCH 19/50] drm: rip out DRM_AGP_MEM and DRM_AGP_KERN Daniel Vetter
2013-12-11 10:34 ` [PATCH 20/50] drm: Kill DRM_HZ Daniel Vetter
2013-12-11 10:34 ` [PATCH 21/50] drm: Kill DRM_IRQ_ARGS Daniel Vetter
2013-12-11 10:34 ` [PATCH 22/50] drm: Kill DRM_WAKUP and DRM_INIT_WAITQUEUE Daniel Vetter
2013-12-11 10:34 ` [PATCH 23/50] drm: Kill DRM_COPY_(TO|FROM)_USER Daniel Vetter
2013-12-11 10:34 ` [PATCH 24/50] drm: Kill DRM_*MEMORYBARRIER Daniel Vetter
2013-12-11 10:34 ` [PATCH 25/50] drm: Kill DRM_SUSER Daniel Vetter
2013-12-11 10:34 ` [PATCH 26/50] drm/gma500: Remove dead code Daniel Vetter
2013-12-11 10:46   ` Patrik Jakobsson
2013-12-11 10:34 ` [PATCH 27/50] drm/irq: Replace DRM_WAIT_ON with wait_event Daniel Vetter
2013-12-11 10:34 ` [PATCH 28/50] drm: Remove DRM_WAIT_ON from all drivers Daniel Vetter
2013-12-18  1:39   ` Dave Airlie
2013-12-18  8:25     ` Daniel Vetter
2013-12-18  9:37       ` Thomas Hellstrom
2013-12-11 10:34 ` [PATCH 29/50] drm/irq: simplify irq checks in drm_wait_vblank Daniel Vetter
2013-12-12 11:29   ` Thierry Reding
2013-12-12 12:51     ` Daniel Vetter
2013-12-16 10:30   ` [PATCH] " Daniel Vetter
2013-12-16 11:18     ` Thierry Reding
2013-12-11 10:34 ` [PATCH 30/50] drm/pci: fold in irq_by_busid support Daniel Vetter
2013-12-11 10:34 ` [PATCH 31/50] drm/irq: drm_control is a legacy ioctl, so pci devices only Daniel Vetter
2013-12-11 10:34 ` [PATCH 32/50] drm/irq: remove cargo-culted locking from irq_install/unistall Daniel Vetter
2013-12-11 10:34 ` [PATCH 33/50] drm: remove drm_dev_to_irq from drivers Daniel Vetter
2013-12-11 10:34 ` [PATCH 34/50] drm: kill drm_bus->bus_type Daniel Vetter
2013-12-11 10:34 ` Daniel Vetter [this message]
2013-12-11 10:34 ` [PATCH 36/50] drm: rename dev->count_lock to dev->buf_lock Daniel Vetter
2013-12-12 11:33   ` Thierry Reding
2013-12-12 12:52     ` Daniel Vetter
2013-12-16 10:29   ` [PATCH] " Daniel Vetter
2013-12-16 11:17     ` Thierry Reding
2013-12-11 10:34 ` [PATCH 37/50] drm/irq: track the irq installed in drm_irq_install in dev->irq Daniel Vetter
2013-12-16 10:29   ` [PATCH] " Daniel Vetter
2013-12-16 11:17     ` Thierry Reding
2013-12-11 10:34 ` [PATCH 38/50] drm/irq: Look up the pci irq directly in the drm_control ioctl Daniel Vetter
2013-12-11 10:35 ` [PATCH 39/50] drm: pass the irq explicitly to drm_irq_install Daniel Vetter
2013-12-11 10:35 ` [PATCH 40/50] drm: remove bus->get_irq implementations Daniel Vetter
2013-12-11 10:35 ` [PATCH 41/50] drm: inline drm_pci_set_unique Daniel Vetter
2013-12-11 10:35 ` [PATCH 42/50] drm: rip out dev->devname Daniel Vetter
2013-12-11 10:35 ` [PATCH 43/50] drm: remove drm_bus->get_name Daniel Vetter
2013-12-11 10:35 ` [PATCH 44/50] drm: Remove dev->kdriver Daniel Vetter
2013-12-11 10:35 ` [PATCH 45/50] drm/<drivers>: don't set driver->dev_priv_size to 0 Daniel Vetter
2014-01-10 15:25   ` Damien Lespiau
2013-12-11 10:35 ` [PATCH 46/50] drm: store the gem vma offset manager in a typed pointer Daniel Vetter
2013-12-11 10:53   ` David Herrmann
2013-12-11 13:24     ` [PATCH 1/2] " Daniel Vetter
2013-12-11 13:24       ` [PATCH 2/2] drm/gma500: Remove unused function declaration Daniel Vetter
2013-12-11 16:04         ` Patrik Jakobsson
2013-12-18  1:04       ` [PATCH 1/2] drm: store the gem vma offset manager in a typed pointer Dave Airlie
2013-12-18  1:22         ` Rob Clark
2013-12-11 10:35 ` [PATCH 47/50] drm: rip out dev->ioctl_count tracking Daniel Vetter
2013-12-11 10:35 ` [PATCH 48/50] drm: Kill file_priv->ioctl_count tracking Daniel Vetter
2013-12-11 10:35 ` [PATCH 49/50] drm: remove dev->vma_count Daniel Vetter
2013-12-11 10:35 ` [PATCH 50/50] drm: use memdup_user() as a cleanup Daniel Vetter
2013-12-12 18:08 ` [PATCH 00/50] more drm de-midlayering Jakob Bornecrantz

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=1386758111-3446-36-git-send-email-daniel.vetter@ffwll.ch \
    --to=daniel.vetter@ffwll.ch \
    --cc=dri-devel@lists.freedesktop.org \
    --subject='Re: [PATCH 35/50] drm: Rip out totally bogus vga_switcheroo->can_switch locking' \
    /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

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.