public inbox for intel-gfx@lists.freedesktop.org
 help / color / mirror / Atom feed
From: Daniel Vetter <daniel@ffwll.ch>
To: Ben Widawsky <benjamin.widawsky@intel.com>
Cc: Intel GFX <intel-gfx@lists.freedesktop.org>,
	Ben Widawsky <ben@bwidawsk.net>
Subject: Re: [PATCH 38/48] drm/i915: Clean up VMAs before freeing
Date: Wed, 18 Dec 2013 15:55:05 +0100	[thread overview]
Message-ID: <20131218145505.GJ26371@phenom.ffwll.local> (raw)
In-Reply-To: <1386367941-7131-38-git-send-email-benjamin.widawsky@intel.com>

On Fri, Dec 06, 2013 at 02:11:23PM -0800, Ben Widawsky wrote:
> From: Ben Widawsky <ben@bwidawsk.net>
> 
> It's quite common for an object to simply be on the inactive list (and
> not unbound) when we want to free the context. This of course happens
> with lazy unbinding. Simply, this is needed when an object isn't fully
> unbound but we want to free one VMA of the object, for whatever reason.
> 
> NOTE: The aliasing PPGTT is not a proper VM, so it needs special casing.
> 
> This addresses the fixup requirement mentioned in:
> drm/915: Better reset handling for contexts
> 
> In the flink, and dmabuf case, we can't assert that the object isn't
> still active. To keep it more generic, just check the vma's link in the
> object vma list. If we wanted to do a better job, we could track last
> seqno (and active) per VMA. It was decided not to do this in the last
> iteration. Unfortunately this means the assertion can miss real bugs
> when using flink/dmabuf.
> 
> v2: Use the newer introduced i915_gem_evict_vm(). Note that handling the
> aliasing PPGTT is special.
> 
> Signed-off-by: Ben Widawsky <ben@bwidawsk.net>

Presuming we have proper refcounting for the stuff involved this just
papers over bugs. I've added a big FIXME here.
-Daniel

> ---
>  drivers/gpu/drm/i915/i915_drv.h | 48 ++++++++++++++++++++++++++++++-----------
>  1 file changed, 36 insertions(+), 12 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> index ff30f3f..fe13023 100644
> --- a/drivers/gpu/drm/i915/i915_drv.h
> +++ b/drivers/gpu/drm/i915/i915_drv.h
> @@ -2257,6 +2257,17 @@ int i915_gem_context_create_ioctl(struct drm_device *dev, void *data,
>  int i915_gem_context_destroy_ioctl(struct drm_device *dev, void *data,
>  				   struct drm_file *file);
>  
> +/* i915_gem_evict.c */
> +int __must_check i915_gem_evict_something(struct drm_device *dev,
> +					  struct i915_address_space *vm,
> +					  int min_size,
> +					  unsigned alignment,
> +					  unsigned cache_level,
> +					  bool mappable,
> +					  bool nonblock);
> +int i915_gem_evict_vm(struct i915_address_space *vm, bool do_idle);
> +int i915_gem_evict_everything(struct drm_device *dev);
> +
>  /* i915_gem_gtt.c */
>  void i915_check_and_clear_faults(struct drm_device *dev);
>  void i915_gem_suspend_gtt_mappings(struct drm_device *dev);
> @@ -2294,22 +2305,35 @@ static inline bool intel_enable_ppgtt(struct drm_device *dev, bool full)
>  static inline void ppgtt_release(struct kref *kref)
>  {
>  	struct i915_hw_ppgtt *ppgtt = container_of(kref, struct i915_hw_ppgtt, ref);
> +	struct drm_device *dev = ppgtt->base.dev;
> +	struct drm_i915_private *dev_priv = dev->dev_private;
> +	struct i915_address_space *vm = &ppgtt->base;
> +
> +	if (ppgtt == dev_priv->mm.aliasing_ppgtt ||
> +	    (list_empty(&vm->active_list) && list_empty(&vm->inactive_list))) {
> +		ppgtt->base.cleanup(&ppgtt->base);
> +		return;
> +	}
> +
> +	/* Make sure vmas are unbound before we take down the drm_mm
> +	 */
> +	if (!list_empty(&vm->active_list)) {
> +		struct i915_vma *vma;
> +
> +		list_for_each_entry(vma, &vm->active_list, mm_list)
> +			if (WARN_ON(list_empty(&vma->vma_link) ||
> +				    list_is_singular(&vma->vma_link)))
> +				break;
> +
> +		i915_gem_evict_vm(&ppgtt->base, true);
> +	} else {
> +		i915_gem_retire_requests(dev);
> +		i915_gem_evict_vm(&ppgtt->base, false);
> +	}
>  
>  	ppgtt->base.cleanup(&ppgtt->base);
>  }
>  
> -
> -/* i915_gem_evict.c */
> -int __must_check i915_gem_evict_something(struct drm_device *dev,
> -					  struct i915_address_space *vm,
> -					  int min_size,
> -					  unsigned alignment,
> -					  unsigned cache_level,
> -					  bool mappable,
> -					  bool nonblock);
> -int i915_gem_evict_vm(struct i915_address_space *vm, bool do_idle);
> -int i915_gem_evict_everything(struct drm_device *dev);
> -
>  /* i915_gem_stolen.c */
>  int i915_gem_init_stolen(struct drm_device *dev);
>  int i915_gem_stolen_setup_compression(struct drm_device *dev, int size);
> -- 
> 1.8.4.2
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch

  reply	other threads:[~2013-12-18 14:54 UTC|newest]

Thread overview: 119+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-12-06 21:55 [PULL] PPGTT Ben Widawsky
2013-12-06 22:10 ` [PATCH 01/48] drm/i915: Fix bad refcounting on execbuf failures Ben Widawsky
2013-12-06 22:10   ` [PATCH 02/48] drm/i915: Provide PDP updates via MMIO Ben Widawsky
2013-12-06 22:10   ` [PATCH 03/48] drm/i915: Don't unconditionally try to deref aliasing ppgtt Ben Widawsky
2013-12-18 13:50     ` Daniel Vetter
2013-12-06 22:10   ` [PATCH 04/48] drm/i915: Allow ggtt lookups to not WARN Ben Widawsky
2013-12-06 22:10   ` [PATCH 05/48] drm/i915: Takedown drm_mm on failed gtt setup Ben Widawsky
2013-12-06 22:10   ` [PATCH 06/48] drm/i915: Handle inactivating objects for all VMAs Ben Widawsky
2013-12-06 22:10   ` [PATCH 07/48] drm/i915: Add vm to error BO capture Ben Widawsky
2013-12-06 22:10   ` [PATCH 08/48] drm/i915: Don't use gtt mapping for !gtt error objects Ben Widawsky
2013-12-06 22:10   ` [PATCH 09/48] drm/i915: Identify active VM for batchbuffer capture Ben Widawsky
2013-12-06 22:10   ` [PATCH 10/48] drm/i915: Make pin count per VMA Ben Widawsky
2013-12-06 22:10   ` [PATCH 11/48] drm/i915: Create bind/unbind abstraction for VMAs Ben Widawsky
2013-12-06 22:10   ` [PATCH 12/48] drm/i915: Remove vm arg from relocate entry Ben Widawsky
2013-12-06 22:10   ` [PATCH 13/48] drm/i915: Add a context open function Ben Widawsky
2013-12-06 22:10   ` [PATCH 14/48] drm/i915: relax context alignment Ben Widawsky
2013-12-06 22:11   ` [PATCH 15/48] drm/i915: Simplify ring handling in execbuf Ben Widawsky
2013-12-06 22:11   ` [PATCH 16/48] drm/i915: Permit contexts on all rings Ben Widawsky
2013-12-06 22:11   ` [PATCH 17/48] drm/i915: Track which ring a context ran on Ben Widawsky
2014-04-18  9:51     ` Chris Wilson
2014-04-22 14:25       ` Daniel Vetter
2014-04-22 14:54         ` Chris Wilson
2013-12-06 22:11   ` [PATCH 18/48] drm/i915: Better reset handling for contexts Ben Widawsky
2013-12-18 14:21     ` Daniel Vetter
2013-12-06 22:11   ` [PATCH 19/48] drm/i915: Split context enabling from init Ben Widawsky
2013-12-06 22:11   ` [PATCH 20/48] drm/i915: Generalize default context setup Ben Widawsky
2013-12-06 22:11   ` [PATCH 21/48] drm/i915: PPGTT vfuncs should take a ppgtt argument Ben Widawsky
2013-12-06 22:11   ` [PATCH 22/48] drm/i915: Use drm_mm for PPGTT PDEs Ben Widawsky
2013-12-06 22:11   ` [PATCH 23/48] drm/i915: One hopeful eviction on PPGTT alloc Ben Widawsky
2013-12-06 22:11   ` [PATCH 24/48] drm/i915: Use platform specific ppgtt enable Ben Widawsky
2013-12-06 22:11   ` [PATCH 25/48] drm/i915: Extract mm switching to function Ben Widawsky
2013-12-06 22:11   ` [PATCH 26/48] drm/i915: Use LRI for switching PP_DIR_BASE Ben Widawsky
2013-12-06 22:11   ` [PATCH 27/48] drm/i915: Flush TLBs after !RCS PP_DIR_BASE Ben Widawsky
2013-12-06 22:11   ` [PATCH 28/48] drm/i915: Generalize PPGTT init Ben Widawsky
2013-12-06 22:11   ` [PATCH 29/48] drm/i915: Reorganize intel_enable_ppgtt Ben Widawsky
2013-12-06 22:11   ` [PATCH 30/48] drm/i915: Add VM to context Ben Widawsky
2013-12-06 22:11   ` [PATCH 31/48] drm/i915: Write PDEs at init instead of enable Ben Widawsky
2013-12-06 22:11   ` [PATCH 32/48] drm/i915: Restore PDEs for all VMs Ben Widawsky
2013-12-06 22:11   ` [PATCH 33/48] drm/i915: Do aliasing PPGTT init with contexts Ben Widawsky
2013-12-06 22:11   ` [PATCH 34/48] drm/i915: Create a per file_priv default context Ben Widawsky
2013-12-06 22:11   ` [PATCH 35/48] drm/i915: Piggy back hangstats off of contexts Ben Widawsky
2013-12-06 22:11   ` [PATCH 36/48] drm/i915: Get context early in execbuf Ben Widawsky
2013-12-06 22:11   ` [PATCH 37/48] drm/i915: Defer request freeing Ben Widawsky
2013-12-12 11:08     ` Chris Wilson
2013-12-18 14:39       ` Daniel Vetter
2013-12-06 22:11   ` [PATCH 38/48] drm/i915: Clean up VMAs before freeing Ben Widawsky
2013-12-18 14:55     ` Daniel Vetter [this message]
2013-12-18 14:56       ` Daniel Vetter
2013-12-06 22:11   ` [PATCH 39/48] drm/i915: Do not allow buffers at offset 0 Ben Widawsky
2013-12-12 10:59     ` Chris Wilson
2013-12-18 14:58       ` Daniel Vetter
2013-12-06 22:11   ` [PATCH 40/48] drm/i915: Add a tracepoint for new VMs Ben Widawsky
2013-12-18 14:59     ` Daniel Vetter
2013-12-06 22:11   ` [PATCH 41/48] drm/i915: Use multiple VMs -- the point of no return Ben Widawsky
2013-12-06 22:11   ` [PATCH 42/48] drm/i915: Remove extraneous mm_switch in ppgtt enable Ben Widawsky
2013-12-06 22:11   ` [PATCH 43/48] drm/i915: Warn on gem_pin usage Ben Widawsky
2013-12-18 15:25     ` Daniel Vetter
2013-12-06 22:11   ` [PATCH 44/48] drm/i915: Add PPGTT dumper Ben Widawsky
2013-12-06 22:11   ` [PATCH 45/48] drm/i915: Dump all ppgtt Ben Widawsky
2013-12-18 15:26     ` Daniel Vetter
2013-12-06 22:11   ` [PATCH 46/48] drm: Optionally create mm blocks from top-to-bottom Ben Widawsky
2013-12-08 14:28     ` David Herrmann
2013-12-06 22:11   ` [PATCH 47/48] drm/i915: Use topdown allocation for PPGTT Ben Widawsky
2013-12-06 22:11   ` [PATCH 48/48] page allocator: Tmp OOM deadlock w/a from Chris Ben Widawsky
2013-12-06 22:11   ` [PATCH 01/48] drm/i915: Fix bad refcounting on execbuf failures Ben Widawsky
2013-12-06 22:11   ` [PATCH 02/48] drm/i915: Provide PDP updates via MMIO Ben Widawsky
2013-12-06 22:11   ` [PATCH 03/48] drm/i915: Don't unconditionally try to deref aliasing ppgtt Ben Widawsky
2013-12-06 22:11   ` [PATCH 04/48] drm/i915: Allow ggtt lookups to not WARN Ben Widawsky
2013-12-06 22:11   ` [PATCH 05/48] drm/i915: Takedown drm_mm on failed gtt setup Ben Widawsky
2013-12-06 22:11   ` [PATCH 06/48] drm/i915: Handle inactivating objects for all VMAs Ben Widawsky
2013-12-06 22:11   ` [PATCH 07/48] drm/i915: Add vm to error BO capture Ben Widawsky
2013-12-06 22:11   ` [PATCH 08/48] drm/i915: Don't use gtt mapping for !gtt error objects Ben Widawsky
2013-12-06 22:11   ` [PATCH 09/48] drm/i915: Identify active VM for batchbuffer capture Ben Widawsky
2013-12-06 22:11   ` [PATCH 10/48] drm/i915: Make pin count per VMA Ben Widawsky
2013-12-06 22:11   ` [PATCH 11/48] drm/i915: Create bind/unbind abstraction for VMAs Ben Widawsky
2013-12-06 22:11   ` [PATCH 12/48] drm/i915: Remove vm arg from relocate entry Ben Widawsky
2013-12-06 22:11   ` [PATCH 13/48] drm/i915: Add a context open function Ben Widawsky
2013-12-06 22:11   ` [PATCH 14/48] drm/i915: relax context alignment Ben Widawsky
2013-12-06 22:11   ` [PATCH 15/48] drm/i915: Simplify ring handling in execbuf Ben Widawsky
2013-12-06 22:11   ` [PATCH 16/48] drm/i915: Permit contexts on all rings Ben Widawsky
2013-12-06 22:11   ` [PATCH 17/48] drm/i915: Track which ring a context ran on Ben Widawsky
2013-12-06 22:11   ` [PATCH 18/48] drm/i915: Better reset handling for contexts Ben Widawsky
2013-12-06 22:11   ` [PATCH 19/48] drm/i915: Split context enabling from init Ben Widawsky
2013-12-06 22:11   ` [PATCH 20/48] drm/i915: Generalize default context setup Ben Widawsky
2013-12-06 22:11   ` [PATCH 21/48] drm/i915: PPGTT vfuncs should take a ppgtt argument Ben Widawsky
2013-12-06 22:11   ` [PATCH 22/48] drm/i915: Use drm_mm for PPGTT PDEs Ben Widawsky
2014-03-20 11:10     ` Chris Wilson
2014-03-24 19:36       ` Ben Widawsky
2014-03-24 19:45         ` Chris Wilson
2014-03-24 20:02           ` Ben Widawsky
2014-03-25 13:41             ` Chris Wilson
2014-03-25 15:33               ` Daniel Vetter
2013-12-06 22:11   ` [PATCH 23/48] drm/i915: One hopeful eviction on PPGTT alloc Ben Widawsky
2014-03-20 11:12     ` Chris Wilson
2013-12-06 22:11   ` [PATCH 24/48] drm/i915: Use platform specific ppgtt enable Ben Widawsky
2013-12-06 22:11   ` [PATCH 25/48] drm/i915: Extract mm switching to function Ben Widawsky
2013-12-06 22:11   ` [PATCH 26/48] drm/i915: Use LRI for switching PP_DIR_BASE Ben Widawsky
2013-12-06 22:12   ` [PATCH 27/48] drm/i915: Flush TLBs after !RCS PP_DIR_BASE Ben Widawsky
2013-12-06 22:12   ` [PATCH 28/48] drm/i915: Generalize PPGTT init Ben Widawsky
2013-12-06 22:12   ` [PATCH 29/48] drm/i915: Reorganize intel_enable_ppgtt Ben Widawsky
2013-12-06 22:12   ` [PATCH 30/48] drm/i915: Add VM to context Ben Widawsky
2013-12-06 22:12   ` [PATCH 31/48] drm/i915: Write PDEs at init instead of enable Ben Widawsky
2013-12-06 22:12   ` [PATCH 32/48] drm/i915: Restore PDEs for all VMs Ben Widawsky
2013-12-06 22:12   ` [PATCH 33/48] drm/i915: Do aliasing PPGTT init with contexts Ben Widawsky
2013-12-06 22:12   ` [PATCH 34/48] drm/i915: Create a per file_priv default context Ben Widawsky
2013-12-06 22:12   ` [PATCH 35/48] drm/i915: Piggy back hangstats off of contexts Ben Widawsky
2013-12-06 22:12   ` [PATCH 36/48] drm/i915: Get context early in execbuf Ben Widawsky
2013-12-06 22:12   ` [PATCH 37/48] drm/i915: Defer request freeing Ben Widawsky
2013-12-06 22:12   ` [PATCH 38/48] drm/i915: Clean up VMAs before freeing Ben Widawsky
2013-12-06 22:12   ` [PATCH 39/48] drm/i915: Do not allow buffers at offset 0 Ben Widawsky
2013-12-06 22:12   ` [PATCH 40/48] drm/i915: Add a tracepoint for new VMs Ben Widawsky
2013-12-06 22:12   ` [PATCH 41/48] drm/i915: Use multiple VMs -- the point of no return Ben Widawsky
2013-12-06 22:12   ` [PATCH 42/48] drm/i915: Remove extraneous mm_switch in ppgtt enable Ben Widawsky
2013-12-06 22:12   ` [PATCH 43/48] drm/i915: Warn on gem_pin usage Ben Widawsky
2013-12-06 22:12   ` [PATCH 44/48] drm/i915: Add PPGTT dumper Ben Widawsky
2013-12-06 22:12   ` [PATCH 45/48] drm/i915: Dump all ppgtt Ben Widawsky
2013-12-06 22:12   ` [PATCH 46/48] drm: Optionally create mm blocks from top-to-bottom Ben Widawsky
2013-12-06 22:12   ` [PATCH 47/48] drm/i915: Use topdown allocation for PPGTT Ben Widawsky
2013-12-06 22:12   ` [PATCH 48/48] page allocator: Tmp OOM deadlock w/a from Chris Ben Widawsky

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=20131218145505.GJ26371@phenom.ffwll.local \
    --to=daniel@ffwll.ch \
    --cc=ben@bwidawsk.net \
    --cc=benjamin.widawsky@intel.com \
    --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