Skip to contentSkip to conversation

drm/i915: Unpin vma iomapping when fbdev is destroyed

PATCH In review Tvrtko Ursulin 4 participants 7 messages Jul 4, 2017

Subsystems (2)
lore
Share & tools
lore

Tvrtko Ursulin

author maintainer
#0 lore raw

PATCH 1/1 drm/i915: Unpin vma iomapping when fbdev is destroyed

From: Tvrtko Ursulin <[email protected]>

If we don't release the iomapping we are not able to unpin the vma which then gets leaked.

---
Some time last year we talked about a pending patch to get rid of the fbdev VMA leak. I lost track of what happened with that.

Compile tested only.

---
 drivers/gpu/drm/i915/intel_fbdev.c | 1 +
 1 file changed, 1 insertion(+)
drivers/gpu/drm/i915/intel_fbdev.c +1 -0
index 03347c6ae599..46831021236f 100644
@@ -537,6 +537,7 @@ static void intel_fbdev_destroy(struct intel_fbdev *ifbdev)
 
 	if (ifbdev->fb) {
 		mutex_lock(&ifbdev->helper.dev->struct_mutex);
+		i915_vma_unpin_iomap(ifbdev->vma);
 		intel_unpin_fb_vma(ifbdev->vma);
 		mutex_unlock(&ifbdev->helper.dev->struct_mutex);
 
Signature · 6 lines
2.9.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw

✓ Fi.CI.BAT: success for drm/i915: Unpin vma iomapping when fbdev is destroyed

== Series Details ==

== Series Details ==

Series: drm/i915: Unpin vma iomapping when fbdev is destroyed
URL : https://patchwork.freedesktop.org/series/26802/
State : success

== Summary ==

Series 26802v1 drm/i915: Unpin vma iomapping when fbdev is destroyed
https://patchwork.freedesktop.org/api/1.0/series/26802/revisions/1/mbox/

Test gem_exec_flush:

        Subgroup basic-batch-kernel-default-uc:
                pass       -> FAIL       (fi-snb-2600) fdo#100007
Test kms_pipe_crc_basic:
        Subgroup hang-read-crc-pipe-a:
                dmesg-warn -> PASS       (fi-pnv-d510) fdo#101597 +1

fdo#100007 https://bugs.freedesktop.org/show_bug.cgi?id=100007
fdo#101597 https://bugs.freedesktop.org/show_bug.cgi?id=101597

fi-bdw-5557u     total:279  pass:264  dwarn:0   dfail:0   fail:3   skip:11  time:435s
fi-bdw-gvtdvm    total:279  pass:257  dwarn:8   dfail:0   fail:0   skip:14  time:428s
fi-blb-e6850     total:279  pass:224  dwarn:1   dfail:0   fail:0   skip:54  time:356s
fi-bsw-n3050     total:279  pass:239  dwarn:0   dfail:0   fail:3   skip:36  time:516s
fi-bxt-j4205     total:279  pass:256  dwarn:0   dfail:0   fail:3   skip:19  time:505s
fi-byt-j1900     total:279  pass:250  dwarn:1   dfail:0   fail:3   skip:24  time:483s
fi-byt-n2820     total:279  pass:246  dwarn:1   dfail:0   fail:3   skip:28  time:482s
fi-glk-2a        total:279  pass:256  dwarn:0   dfail:0   fail:3   skip:19  time:594s
fi-hsw-4770      total:279  pass:259  dwarn:0   dfail:0   fail:3   skip:16  time:430s
fi-hsw-4770r     total:279  pass:259  dwarn:0   dfail:0   fail:3   skip:16  time:412s
fi-ilk-650       total:279  pass:225  dwarn:0   dfail:0   fail:3   skip:50  time:417s
fi-ivb-3520m     total:279  pass:257  dwarn:0   dfail:0   fail:3   skip:18  time:495s
fi-ivb-3770      total:279  pass:257  dwarn:0   dfail:0   fail:3   skip:18  time:470s
fi-kbl-7500u     total:279  pass:257  dwarn:0   dfail:0   fail:3   skip:18  time:459s
fi-kbl-7560u     total:279  pass:264  dwarn:1   dfail:0   fail:3   skip:10  time:564s
fi-kbl-r         total:279  pass:256  dwarn:1   dfail:0   fail:3   skip:18  time:563s
fi-pnv-d510      total:279  pass:222  dwarn:2   dfail:0   fail:0   skip:55  time:561s
fi-skl-6260u     total:279  pass:265  dwarn:0   dfail:0   fail:3   skip:10  time:459s
fi-skl-6700hq    total:279  pass:219  dwarn:1   dfail:0   fail:33  skip:24  time:302s
fi-skl-6700k     total:279  pass:257  dwarn:0   dfail:0   fail:3   skip:18  time:463s
fi-skl-6770hq    total:279  pass:265  dwarn:0   dfail:0   fail:3   skip:10  time:477s
fi-skl-gvtdvm    total:279  pass:266  dwarn:0   dfail:0   fail:0   skip:13  time:439s
fi-snb-2520m     total:279  pass:247  dwarn:0   dfail:0   fail:3   skip:28  time:536s
fi-snb-2600      total:279  pass:245  dwarn:0   dfail:0   fail:4   skip:29  time:404s

cadaf524db971245c2744e7a76fa6a89143d8cea drm-tip: 2017y-07m-04d-07h-49m-57s UTC integration manifest 9e7f3ca drm/i915: Unpin vma iomapping when fbdev is destroyed

== Logs ==

For more details see: https://intel-gfx-ci.01.org/CI/Patchwork_5104/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#2 lore raw
In reply to Tvrtko Ursulin

Quoting Tvrtko Ursulin (2017-07-04 10:46:40)

Quoted 4 lines

From: Tvrtko Ursulin <[email protected]>

If we don't release the iomapping we are not able to unpin the vma which then gets leaked.

Oh, we still do unpin the vma on closing the object and we don't hold any extra object reference for the iomap. It is still a good patch for the symmetry, except it doesn't do what you say :-)

Quoted 5 lines

---
Some time last year we talked about a pending patch to get rid of the fbdev VMA leak. I lost track of what happened with that.

That was fixed by the intel_unpin_fb_vma() in there. -Chris
_______________________________________________
Intel-gfx mailing list [email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#3 lore raw
Quoted 9 lines · Chris Wilson

Quoting Tvrtko Ursulin (2017-07-04 10:46:40)

Quoted 4 lines

From: Tvrtko Ursulin <[email protected]>

If we don't release the iomapping we are not able to unpin the vma which then gets leaked.

Oh, we still do unpin the vma on closing the object and we don't hold any extra object reference for the iomap. It is still a good patch for the symmetry, except it doesn't do what you say :-)

What do you mean that we don't hold any extra reference for the iomap? I see i915_vma_pin_iomap -> __i915_vma_pin -> vma->flags++. I can't spot the place which would override this and still unbind it at some point.

Quoted 7 lines
Quoted 5 lines

---
Some time last year we talked about a pending patch to get rid of the fbdev VMA leak. I lost track of what happened with that.

That was fixed by the intel_unpin_fb_vma() in there.

I suppose I don't see how since I don't see what defeats the elevated pin count from the above.

Regards,

Tvrtko
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#4 lore raw

Quoting Tvrtko Ursulin (2017-07-04 12:26:31)

Quoted 8 lines · Chris Wilson
Quoted 6 lines

Quoting Tvrtko Ursulin (2017-07-04 10:46:40)

Quoted 4 lines

From: Tvrtko Ursulin <[email protected]>

If we don't release the iomapping we are not able to unpin the vma which then gets leaked.

Quoted 7 lines
Quoted 3 lines

Oh, we still do unpin the vma on closing the object and we don't hold any extra object reference for the iomap. It is still a good patch for the symmetry, except it doesn't do what you say :-)

What do you mean that we don't hold any extra reference for the iomap? I see i915_vma_pin_iomap -> __i915_vma_pin -> vma->flags++. I can't spot the place which would override this and still unbind it at some point.

free_objects:

	vma->flags &= ~I915_VMA_PIN_MASK;
	i915_vma_close(vma);

-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#5 lore raw
Quoted 13 lines · Chris Wilson

Quoting Tvrtko Ursulin (2017-07-04 12:26:31)

Quoted 12 lines · Chris Wilson
Quoted 9 lines

Quoting Tvrtko Ursulin (2017-07-04 10:46:40)

Quoted 4 lines

From: Tvrtko Ursulin <[email protected]>

If we don't release the iomapping we are not able to unpin the vma which then gets leaked.

Oh, we still do unpin the vma on closing the object and we don't hold any extra object reference for the iomap. It is still a good patch for the symmetry, except it doesn't do what you say :-)

Quoted 7 lines
Quoted 3 lines

What do you mean that we don't hold any extra reference for the iomap? I see i915_vma_pin_iomap -> __i915_vma_pin -> vma->flags++. I can't spot the place which would override this and still unbind it at some point.

free_objects:

	vma->flags &= ~I915_VMA_PIN_MASK;
	i915_vma_close(vma);

Blast.. :) Do we need the unpin in the API at all then?

Regards,

Tvrtko
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#6 lore raw

Quoting Tvrtko Ursulin (2017-07-04 13:33:28)

Quoted 19 lines · Chris Wilson
Quoted 17 lines

Quoting Tvrtko Ursulin (2017-07-04 12:26:31)

Quoted 15 lines · Chris Wilson
Quoted 9 lines

Quoting Tvrtko Ursulin (2017-07-04 10:46:40)

Quoted 4 lines

From: Tvrtko Ursulin <[email protected]>

If we don't release the iomapping we are not able to unpin the vma which then gets leaked.

Oh, we still do unpin the vma on closing the object and we don't hold any extra object reference for the iomap. It is still a good patch for the symmetry, except it doesn't do what you say :-)

What do you mean that we don't hold any extra reference for the iomap? I see i915_vma_pin_iomap -> __i915_vma_pin -> vma->flags++. I can't spot the place which would override this and still unbind it at some point.

Quoted 5 lines
Quoted 3 lines

free_objects:

      vma->flags &= ~I915_VMA_PIN_MASK;
      i915_vma_close(vma);

Blast.. :) Do we need the unpin in the API at all then?

Yes, free_objects is a little late in general :)

If we fix up fbdev, we should be close enough to make that a if (WARN_ON(i915_vma_is_pinned(vma)))

	vma->flags &= ~I915_VMA_PIN_MASK;

-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx