xserver-xorg-video-intel: Changes to 'debian-experimental'
ChangeLog | 352 ++++++++++++++++++++++
NEWS | 22 +
configure.ac | 2
debian/changelog | 8
src/i965_render.c | 2
src/i965_video.c | 2
src/intel.h | 2
src/intel_display.c | 1
src/intel_driver.c | 34 --
src/intel_driver.h | 17 -
src/intel_module.c | 33 +-
src/legacy/i810/i810_driver.c | 33 --
src/sna/fb/fbpict.c | 28 +
src/sna/gen4_render.c | 242 +++++++++------
src/sna/gen5_render.c | 5
src/sna/gen7_render.c | 30 +
src/sna/gen7_render.h | 215 ++-----------
src/sna/kgem.c | 8
src/sna/kgem.h | 2
src/sna/sna.h | 3
src/sna/sna_accel.c | 664 ++++++++++++++++++++++++++++++++++--------
src/sna/sna_display.c | 3
src/sna/sna_dri.c | 117 +++++--
src/sna/sna_driver.c | 49 ---
src/sna/sna_reg.h | 1
src/sna/sna_render.c | 4
src/sna/sna_render.h | 4
src/sna/sna_video_hwmc.c | 4
src/sna/sna_video_textured.c | 2
29 files changed, 1314 insertions(+), 575 deletions(-)
New commits:
commit 5aac93a711b13b8a33d5345bfab9d208f0ddf8a2
Author: Julien Danjou <julien@danjou.info>
Date: Sat Jul 28 14:09:15 2012 +0200
New upstream release (2.20.2)
Signed-off-by: Julien Danjou <julien@danjou.info>
diff --git a/ChangeLog b/ChangeLog
index 82dbbf6..588e46d 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,355 @@
+commit a8ee1406244d8b8399bf933d6b61bfd14374b5f9
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Fri Jul 27 09:07:16 2012 +0100
+
+ 2.20.2 release
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit bef73cd9279be3438e467981db39c67bc13104f5
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 21:54:33 2012 +0100
+
+ sna/dri: Select the engine before emitting the wait
+
+ So that if we have a flexible WAIT_FOR_EVENT that can go on either
+ pipeline, we can choose our preferred pipeline for DRI.
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 1ced4f1ddcf30b518e1760c7aa4a5ed4f934b9f5
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 10:50:31 2012 +0100
+
+ Reduce maximum thread count for IVB GT1 to avoid spontaneous combustion
+
+ Somewhere along the way it seems that IVB GT1 was reduced to only allow
+ a maximum of 48 threads, as revealed in the lastest bspecs.
+
+ Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52473
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 8f8f8759111f791ee99adfd87296443fb0e6acad
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 17:39:34 2012 +0100
+
+ sna/gen4: Tweak heuristics for render/blt usage
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit c9dd1401615f0ed9492a0c0f547fb37150e013d1
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 16:31:16 2012 +0100
+
+ sna/gen4: Bump thread counts
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 7f3fdef98c1ab2fa27439c3be9810b7a934017ce
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 15:39:05 2012 +0100
+
+ sna/gen7: IVB requires a complete pipeline stall when changing blend modes
+
+ Similar to how SandyBridge behaves, I had hoped that with IvyBridge they
+ would have made the pipelined operation actually pipelined, but alas.
+
+ Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52473
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 0938b3df8c25178c8ea0012e1ead1061d03a4e7c
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 15:21:59 2012 +0100
+
+ sna/dri: Add an explanatory assertion
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit c621183466aa55a5938027b702069e792df2272d
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 15:18:56 2012 +0100
+
+ sna/dri: Tidy fallback/normal CopyRegion
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 52d2491a1bafb979d79bb970027c55788f199acb
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 15:19:14 2012 +0100
+
+ sna/video: Protect against attempting to use TexturedVideo whilst wedged
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 8c0e3bbb0c301d0fa4652aa38edd84a9fd6b555e
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 14:58:04 2012 +0100
+
+ sna: Force the damage on the DRI pixmap to be flushed to the GPU
+
+ Otherwise nothing will happen if we consider ourselves wedged.
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit f50d7b03b5d0b81b24f0acb4ae555545bbdaa179
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 14:47:44 2012 +0100
+
+ sna: Only recommend not to create bo for a pixmap whilst wedged
+
+ This allows us to continue to map a GPU bo and operate inplace if we are
+ force to create a GPU bo for a compositor.
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 49403ddd1bd9a23e1b32a10e7d0757ae2897a579
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 14:44:14 2012 +0100
+
+ sna/dri: Mark a flush required for any new DRI already on exec/dirty lists
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit efe3c8ff48738f2a274e1d4514d32499fc4aadbc
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 14:43:33 2012 +0100
+
+ sna: Allow DRI to force allocation of a GPU bo and backing pixmap
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 2069384f9f06d3ef9dbb0c3f2c64cac4b24e10fc
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 14:15:45 2012 +0100
+
+ sna/dri: Implement fallback handling for CopyRegion whilst wedged
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit b18143de47e060b67a46d9c68590a2d35df9fca6
+Author: Paul Menzel <paulepanter@users.sourceforge.net>
+Date: Thu Jul 26 12:51:57 2012 +0200
+
+ NEWS: Correct release version: s/2.12.0/2.20.0/
+
+ Signed-off-by: Paul Menzel <paulepanter@users.sourceforge.net>
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit c262d02fb5defe9100df54cf83cc00e11e335745
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 01:12:11 2012 +0100
+
+ Limit PCI matching to VGA devices
+
+ Fixes X -configure
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit ad6355311b8b80777bc0fec95f6bf6cd1b4969d9
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 01:02:53 2012 +0100
+
+ sna: Compile against xorg-1.10
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit b6c7c490d76c683b7b5c89d20f902603b85bd3bc
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Thu Jul 26 00:48:59 2012 +0100
+
+ sna: Check for a NULL scanout after DPMS off with shadow enabled
+
+ We may mark the scanout as detached when all outputs are turned off (for
+ example during rotation) and so in the subsequent block handler we need
+ to be careful in case we are handling a detached shadow.
+
+ Reported-by: chr.ohm@gmx.net
+ Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52514
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit f4acc01bb09f68edbad4c6cc7e04c271c92661c1
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Wed Jul 25 22:43:32 2012 +0100
+
+ sna: Restore inplace upload for DRI compositors
+
+ With a DRI compositor we have to flush the output after every request,
+ which leads to major inefficiencies. This can be mitigated slightly if
+ we know we will have to upload shortly, which we track using the pixmap
+ flush flag.
+
+ In particular PutImage is meant to upload inplace to an active DRI
+ buffer, however this was accidentally dropped in
+
+ commit a253c95ec63b2b075e66ae7380fed6a73469eba5
+ Author: Chris Wilson <chris@chris-wilson.co.uk>
+ Date: Sun Jul 15 13:32:35 2012 +0100
+
+ sna: Prefer uploads to be staged in snoopable bo
+
+ Performace of putimage500 on ivb i7-3720qm:
+ bare: 4610
+ gnome-shell: 3000
+ patched: 3390
+
+ Reported-by: Michael Larabel <Michael@phoronix.com>
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 5c969a05ef815b261e157fe8d1172aebfd7f5841
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Wed Jul 25 20:28:41 2012 +0100
+
+ sna/gen7: Move the PS thread count definition into the constant struct
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 901bb618215d65747eb0a8d481c77c1201f69362
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Wed Jul 25 20:13:15 2012 +0100
+
+ sna/gen7: Remove duplicated constants
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 8c5077e4ed055a97bf9deda59c0e9a45e42317ca
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Wed Jul 25 16:59:11 2012 +0100
+
+ Assume all unknown chipsets are future gen
+
+ I think the likelihood of a new product being launched based on a 8xx
+ design is remote enough not to worry about.
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 954d9c6aca792098cb16a46b4ec8d5fc104b15a9
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Wed Jul 25 15:41:29 2012 +0100
+
+ sna: Initialise single-shot tile offsets before use
+
+ As noted by the compiler amidst all the noise.
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 484b072072c8297a87940c9e32097923f0a77c8f
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Wed Jul 25 12:22:47 2012 +0100
+
+ sna: Add 'gmux_backlight' to list of known devices for backwards compatability
+
+ Reported-by: Austin Lund <austin.lund@gmail.com>
+ References: https://bugs.freedesktop.org/show_bug.cgi?id=52423
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 0ab6d7a50d37cf4454577cb8c333d4b8683aa054
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Wed Jul 25 12:20:36 2012 +0100
+
+ sna: Prefer platform backlight devices over firmware
+
+ This is in contrast to libbacklight but closer to our original code as
+ we prefer a known custom backlight controller over the ACPI interfaces.
+ As only the ACPI interfaces are marked as firmware, and the custom
+ backlight controllers as platform, we therefore need to prefer platform
+ backlight devices.
+
+ Reported-by: Austin Lund <austin.lund@gmail.com>
+ References: https://bugs.freedesktop.org/show_bug.cgi?id=52423
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit b6d82ab07661aba98963f239f9501b50c3910962
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Wed Jul 25 10:40:07 2012 +0100
+
+ sna: Reduce ping-pong for Composite with render disabled
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit de707b7dc853a3b315ead9789d5ca541412c99bc
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Wed Jul 25 09:05:46 2012 +0100
+
+ uxa: Add Apple's gmux to the list of known preferred backlights
+
+ Reported-by: Austin Lund <austin.lund@gmail.com>
+ Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52423
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 3d254e4010d0753f433cfe62c6a7546b02482847
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Mon Jul 23 23:45:33 2012 +0100
+
+ sna: Use SETUP_BLT to reduce overheads for tiled BLT
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 2ed44149eaa27b9632ec83a776f3ee67b0eec7b0
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Mon Jul 23 23:45:33 2012 +0100
+
+ sna: Use SETUP_BLT to reduce overheads for stippled BLT
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit c1785aaca2c9347114d28f114ee59ef8206d829b
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Mon Jul 23 23:00:25 2012 +0100
+
+ sna: Remove a surplus function
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 554fce8a65b2518cae032f1eadf58830559299c8
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Mon Jul 23 22:59:16 2012 +0100
+
+ uxa: Pass the correct parameters to intel_detect_chipset()
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 40d90dfd8674c255a45b46bbdc09d497af5b3f50
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Mon Jul 23 21:55:46 2012 +0100
+
+ intel: Refactor the common chipset detection/override
+
+ Reduce the duplicate messages for which type of chip we by
+ amalgamating the common code.
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit e3f6c48d18c316899c71b6fc34971039c6f9e5f8
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Mon Jul 23 14:43:23 2012 +0100
+
+ sna: Refactor PutImage to avoid calling drawable_gc_flags() too early
+
+ drawable_gc_flags() asserts that the gc has been moved to the CPU prior
+ to its calls so that it can read the reduced raster operation.
+
+ Reported-by: Zdenek Kabelac <zkabelac@redhat.com>
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit 1af26ea4228a9d7768b475b4f9164d2c7620d4fd
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Mon Jul 23 14:39:44 2012 +0100
+
+ sna: Flesh out tiled operations using the BLT
+
+ Before enabling the RENDER pipeline for this operation, let's just see
+ what is required to fully use the BLT pipeline as well.
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
+commit ac182a006732525a921a9c539e5ebfb537ad3b52
+Author: Chris Wilson <chris@chris-wilson.co.uk>
+Date: Mon Jul 23 11:05:05 2012 +0100
+
+ sna: Hold a reference to the full stipple pattern for repeated tiles
+
+ Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
+
commit 83f683b47063eab8cfb5037d02133dd977c3fc25
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Sun Jul 22 23:20:23 2012 +0100
diff --git a/debian/changelog b/debian/changelog
index efe5457..62c5fa0 100644
--- a/debian/changelog
+++ b/debian/changelog
@@ -1,5 +1,6 @@
-xserver-xorg-video-intel (2:2.20.1-1) UNRELEASED; urgency=low
+xserver-xorg-video-intel (2:2.20.2-1) experimental; urgency=low
+ [ Maarten Lankhorst ]
* New upstream release:
- First official release with sna
* Update to 2.20.1 point release:
@@ -21,7 +22,10 @@ xserver-xorg-video-intel (2:2.20.1-1) UNRELEASED; urgency=low
- Shadow support was dropped from UXA as it was neither complete nor
correct, use SNA instead.
- -- Maarten Lankhorst <maarten.lankhorst@canonical.com> Mon, 16 Jul 2012 16:21:28 +0200
+ [ Julien Danjou ]
+ * New upstream release (2.20.2)
+
+ -- Julien Danjou <acid@debian.org> Sun, 29 Jul 2012 00:06:11 +0200
xserver-xorg-video-intel (2:2.18.0-1+exp1) experimental; urgency=low
commit a8ee1406244d8b8399bf933d6b61bfd14374b5f9
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Fri Jul 27 09:07:16 2012 +0100
2.20.2 release
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
diff --git a/NEWS b/NEWS
index 08340f8..7e267a6 100644
--- a/NEWS
+++ b/NEWS
@@ -1,3 +1,23 @@
+Release 2.20.2 (2012-07-27)
+===========================
+For the last 9 months, since 2.16.901, we have been shipping a driver that
+does not work on IvyBridge GT1 systems (HD2500 graphics); we were telling
+the GPU to use an invalid number of threads for the pixel shader and this
+in turned caused the GPU to hang.
+
+Also fixed since the last release just a few days ago:
+
+* Support for the gmux backlight controller on Apple laptops
+ https://bugs.freedesktop.org/show_bug.cgi?id=52423
+
+* Fix X -configure not to list this driver as matching any Intel device,
+ just the VGA class devices will do!
+
+* A crash in SNA when repeatedly switching xrandr rotations
+
+* Corruption in SNA observed in kwin on IvyBridge
+ https://bugs.freedesktop.org/show_bug.cgi?id=52473
+
Release 2.20.1 (2012-07-22)
===========================
A week in, grab the brown paper bags, for it is time to reveal a couple
diff --git a/configure.ac b/configure.ac
index 3cdacdd..9945d5b 100644
--- a/configure.ac
+++ b/configure.ac
@@ -23,7 +23,7 @@
# Initialize Autoconf
AC_PREREQ([2.60])
AC_INIT([xf86-video-intel],
- [2.20.1],
+ [2.20.2],
[https://bugs.freedesktop.org/enter_bug.cgi?product=xorg],
[xf86-video-intel])
AC_CONFIG_SRCDIR([Makefile.am])
commit bef73cd9279be3438e467981db39c67bc13104f5
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Thu Jul 26 21:54:33 2012 +0100
sna/dri: Select the engine before emitting the wait
So that if we have a flexible WAIT_FOR_EVENT that can go on either
pipeline, we can choose our preferred pipeline for DRI.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
diff --git a/src/sna/sna_dri.c b/src/sna/sna_dri.c
index 1daf1c4..8d6c305 100644
--- a/src/sna/sna_dri.c
+++ b/src/sna/sna_dri.c
@@ -546,6 +546,14 @@ sna_dri_copy_to_front(struct sna *sna, DrawablePtr draw, RegionPtr region,
}
}
+ if (!wedged(sna)) {
+ if (sync)
+ sync = sna_pixmap_is_scanout(sna, pixmap);
+
+ sna_dri_select_mode(sna, src_bo, sync);
+ } else
+ sync = false;
+
dx = dy = 0;
if (draw->type != DRAWABLE_PIXMAP) {
WindowPtr win = (WindowPtr)draw;
@@ -569,7 +577,7 @@ sna_dri_copy_to_front(struct sna *sna, DrawablePtr draw, RegionPtr region,
region = &clip;
}
- if (sync && sna_pixmap_is_scanout(sna, pixmap)) {
+ if (sync) {
crtc = sna_covering_crtc(sna->scrn, &clip.extents, NULL);
if (crtc)
flush = sna_wait_for_scanline(sna, pixmap, crtc,
@@ -595,8 +603,6 @@ sna_dri_copy_to_front(struct sna *sna, DrawablePtr draw, RegionPtr region,
dst_bo, dx, dy,
boxes, n);
} else {
- sna_dri_select_mode(sna, src_bo, flush);
-
sna->render.copy_boxes(sna, GXcopy,
(PixmapPtr)draw, src_bo, -draw->x, -draw->y,
pixmap, dst_bo, dx, dy,
commit 1ced4f1ddcf30b518e1760c7aa4a5ed4f934b9f5
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Thu Jul 26 10:50:31 2012 +0100
Reduce maximum thread count for IVB GT1 to avoid spontaneous combustion
Somewhere along the way it seems that IVB GT1 was reduced to only allow
a maximum of 48 threads, as revealed in the lastest bspecs.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=52473
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
diff --git a/src/i965_render.c b/src/i965_render.c
index 9d45944..2182df8 100644
--- a/src/i965_render.c
+++ b/src/i965_render.c
@@ -2703,7 +2703,7 @@ gen7_composite_wm_state(intel_screen_private *intel,
OUT_BATCH((1 << GEN7_PS_SAMPLER_COUNT_SHIFT) |
(num_surfaces << GEN7_PS_BINDING_TABLE_ENTRY_COUNT_SHIFT));
OUT_BATCH(0); /* scratch space base offset */
- OUT_BATCH(((86 - 1) << GEN7_PS_MAX_THREADS_SHIFT) |
+ OUT_BATCH(((48 - 1) << GEN7_PS_MAX_THREADS_SHIFT) |
GEN7_PS_ATTRIBUTE_ENABLE |
GEN7_PS_16_DISPATCH_ENABLE);
OUT_BATCH((6 << GEN7_PS_DISPATCH_START_GRF_SHIFT_0));
diff --git a/src/i965_video.c b/src/i965_video.c
index d9350ce..bcd6063 100644
--- a/src/i965_video.c
+++ b/src/i965_video.c
@@ -1658,7 +1658,7 @@ gen7_upload_wm_state(ScrnInfoPtr scrn, Bool is_packed)
OUT_BATCH(0); /* scratch space base offset */
OUT_BATCH(
- ((86 - 1) << GEN7_PS_MAX_THREADS_SHIFT) |
+ ((48 - 1) << GEN7_PS_MAX_THREADS_SHIFT) |
GEN7_PS_ATTRIBUTE_ENABLE |
GEN7_PS_16_DISPATCH_ENABLE);
OUT_BATCH(
diff --git a/src/sna/gen7_render.c b/src/sna/gen7_render.c
index 167a5e6..d06b791 100644
--- a/src/sna/gen7_render.c
+++ b/src/sna/gen7_render.c
@@ -77,7 +77,7 @@ struct gt_info {
static const struct gt_info gt1_info = {
.max_vs_threads = 36,
.max_gs_threads = 36,
- .max_wm_threads = (86-1) << GEN7_PS_MAX_THREADS_SHIFT,
+ .max_wm_threads = (48-1) << GEN7_PS_MAX_THREADS_SHIFT,
.urb = { 128, 512, 192 },
};
commit 8f8f8759111f791ee99adfd87296443fb0e6acad
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Thu Jul 26 17:39:34 2012 +0100
sna/gen4: Tweak heuristics for render/blt usage
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
diff --git a/src/sna/gen4_render.c b/src/sna/gen4_render.c
index d62d744..25229e1 100644
--- a/src/sna/gen4_render.c
+++ b/src/sna/gen4_render.c
@@ -1125,7 +1125,7 @@ static bool gen4_rectangle_begin(struct sna *sna,
static int gen4_get_rectangles__flush(struct sna *sna,
const struct sna_composite_op *op)
{
- if (!kgem_check_batch(&sna->kgem, 25))
+ if (!kgem_check_batch(&sna->kgem, (FLUSH_EVERY_VERTEX || op->need_magic_ca_pass) ? 25 : 6))
return 0;
if (!kgem_check_reloc_and_exec(&sna->kgem, 1))
return 0;
@@ -1145,9 +1145,9 @@ inline static int gen4_get_rectangles(struct sna *sna,
start:
rem = vertex_space(sna);
- if (rem < 3*op->floats_per_vertex) {
+ if (rem < op->floats_per_rect) {
DBG(("flushing vbo for %s: %d < %d\n",
- __FUNCTION__, rem, 3*op->floats_per_vertex));
+ __FUNCTION__, rem, op->floats_per_rect));
rem = gen4_get_rectangles__flush(sna, op);
if (unlikely(rem == 0))
goto flush;
@@ -1157,8 +1157,8 @@ start:
!gen4_rectangle_begin(sna, op)))
goto flush;
- if (want > 1 && want * op->floats_per_vertex*3 > rem)
- want = rem / (3*op->floats_per_vertex);
+ if (want > 1 && want * op->floats_per_rect > rem)
+ want = rem / op->floats_per_rect;
sna->render.vertex_index += 3*want;
return want;
@@ -1173,8 +1173,9 @@ flush:
goto start;
}
-static uint32_t *gen4_composite_get_binding_table(struct sna *sna,
- uint16_t *offset)
+static uint32_t *
+gen4_composite_get_binding_table(struct sna *sna,
+ uint16_t *offset)
{
sna->kgem.surface -=
sizeof(struct gen4_surface_state_padded) / sizeof(uint32_t);
@@ -1288,7 +1289,7 @@ static void
gen4_align_vertex(struct sna *sna, const struct sna_composite_op *op)
{
if (op->floats_per_vertex != sna->render_state.gen4.floats_per_vertex) {
- if (sna->render.vertex_size - sna->render.vertex_used < 6*op->floats_per_vertex)
+ if (sna->render.vertex_size - sna->render.vertex_used < 2*op->floats_per_rect)
gen4_vertex_finish(sna);
DBG(("aligning vertex: was %d, now %d floats per vertex, %d->%d\n",
@@ -1335,8 +1336,8 @@ gen4_emit_pipelined_pointers(struct sna *sna,
kernel, blend, op->has_component_alpha, (int)op->dst.format));
sp = SAMPLER_OFFSET(op->src.filter, op->src.repeat,
- op->mask.filter, op->mask.repeat,
- kernel);
+ op->mask.filter, op->mask.repeat,
+ kernel);
bp = gen4_get_blend(blend, op->has_component_alpha, op->dst.format);
key = op->mask.bo != NULL;
@@ -1371,7 +1372,7 @@ gen4_emit_drawing_rectangle(struct sna *sna, const struct sna_composite_op *op)
sna->render_state.gen4.drawrect_limit = limit;
OUT_BATCH(GEN4_3DSTATE_DRAWING_RECTANGLE | (4 - 2));
- OUT_BATCH(0x00000000);
+ OUT_BATCH(0);
OUT_BATCH(limit);
OUT_BATCH(offset);
}
@@ -1713,11 +1714,14 @@ gen4_render_video(struct sna *sna,
tmp.src.filter = SAMPLER_FILTER_BILINEAR;
tmp.src.repeat = SAMPLER_EXTEND_PAD;
+ tmp.src.bo = frame->bo;
+ tmp.mask.bo = NULL;
tmp.u.gen4.wm_kernel =
is_planar_fourcc(frame->id) ? WM_KERNEL_VIDEO_PLANAR : WM_KERNEL_VIDEO_PACKED;
+ tmp.u.gen4.ve_id = 1;
tmp.is_affine = true;
tmp.floats_per_vertex = 3;
- tmp.u.gen4.ve_id = 1;
+ tmp.floats_per_rect = 9;
tmp.priv = frame;
if (!kgem_check_bo(&sna->kgem, tmp.dst.bo, frame->bo, NULL)) {
@@ -2078,13 +2082,7 @@ picture_is_cpu(PicturePtr picture)
if (!picture->pDrawable)
return false;
- /* If it is a solid, try to use the render paths */
- if (picture->pDrawable->width == 1 &&
- picture->pDrawable->height == 1 &&
- picture->repeat)
- return false;
-
- return is_cpu(picture->pDrawable);
+ return is_cpu(picture->pDrawable) || is_dirty(picture->pDrawable);
}
static inline bool prefer_blt(struct sna *sna)
@@ -2099,7 +2097,7 @@ static inline bool prefer_blt(struct sna *sna)
static bool
try_blt(struct sna *sna,
- PicturePtr source,
+ PicturePtr dst, PicturePtr src,
int width, int height)
{
if (prefer_blt(sna)) {
@@ -2113,8 +2111,15 @@ try_blt(struct sna *sna,
return true;
}
+ if (too_large(dst->pDrawable->width, dst->pDrawable->height))
+ return true;
+
+ /* The blitter is much faster for solids */
+ if (sna_picture_is_solid(src, NULL))
+ return true;
+
/* is the source picture only in cpu memory e.g. a shm pixmap? */
- return picture_is_cpu(source);
+ return picture_is_cpu(src);
}
static bool
@@ -2144,7 +2149,7 @@ untransformed(PicturePtr p)
static bool
need_upload(PicturePtr p)
{
- return p->pDrawable && unattached(p->pDrawable) && untransformed(p);
+ return p->pDrawable && untransformed(p) && is_cpu(p->pDrawable);
}
static bool
@@ -2234,12 +2239,12 @@ gen4_composite_fallback(struct sna *sna,
return false;
}
- if (!src_fallback) {
+ if (src_pixmap && !src_fallback) {
DBG(("%s: src is already on the GPU, try to use GPU\n",
__FUNCTION__));
return false;
}
- if (mask && !mask_fallback) {
+ if (mask_pixmap && !mask_fallback) {
DBG(("%s: mask is already on the GPU, try to use GPU\n",
__FUNCTION__));
return false;
@@ -2255,7 +2260,7 @@ gen4_composite_fallback(struct sna *sna,
return true;
}
- if (mask && mask_fallback) {
+ if (mask_fallback) {
DBG(("%s: dst is on the CPU and mask will fallback\n",
__FUNCTION__));
return true;
@@ -2358,7 +2363,7 @@ gen4_render_composite(struct sna *sna,
#endif
if (mask == NULL &&
- try_blt(sna, src, width, height) &&
+ try_blt(sna, dst, src, width, height) &&
sna_blt_composite(sna, op,
src, dst,
src_x, src_y,
@@ -2464,11 +2469,9 @@ gen4_render_composite(struct sna *sna,
else if (tmp->src.is_affine)
tmp->prim_emit = gen4_emit_composite_primitive_affine_source;
- tmp->mask.filter = SAMPLER_FILTER_NEAREST;
- tmp->mask.repeat = SAMPLER_EXTEND_NONE;
-
tmp->floats_per_vertex = 3 + !tmp->is_affine;
}
+ tmp->floats_per_rect = 3*tmp->floats_per_vertex;
tmp->u.gen4.wm_kernel =
gen4_choose_composite_kernel(tmp->op,
@@ -2705,7 +2708,7 @@ gen4_render_composite_spans_done(struct sna *sna,
static bool
gen4_check_composite_spans(struct sna *sna,
uint8_t op, PicturePtr src, PicturePtr dst,
- int16_t width, int16_t height,
+ int16_t width, int16_t height,
unsigned flags)
{
if ((flags & COMPOSITE_SPANS_RECTILINEAR) == 0)
@@ -2717,7 +2720,7 @@ gen4_check_composite_spans(struct sna *sna,
if (gen4_composite_fallback(sna, src, NULL, dst))
return false;
- if (!is_gpu(dst->pDrawable))
+ if (need_tiling(sna, width, height) && !is_gpu(dst->pDrawable))
return false;
return true;
@@ -2999,9 +3002,8 @@ fallback_blt:
extents.x1 + src_dx,
extents.y1 + src_dy,
extents.x2 - extents.x1,
- extents.y2 - extents.y1)) {
+ extents.y2 - extents.y1))
goto fallback_tiled_dst;
- }
} else {
tmp.src.bo = kgem_bo_reference(src_bo);
tmp.src.width = src->drawable.width;
@@ -3011,10 +3013,9 @@ fallback_blt:
tmp.src.scale[1] = 1.f/src->drawable.height;
}
- tmp.mask.bo = NULL;
-
tmp.is_affine = true;
tmp.floats_per_vertex = 3;
+ tmp.floats_per_rect = 9;
tmp.u.gen4.wm_kernel = WM_KERNEL;
tmp.u.gen4.ve_id = 1;
@@ -3041,6 +3042,8 @@ fallback_blt:
box->x1 + dst_dx, box->y1 + dst_dy);
box++;
} while (--n);
+
+ gen4_vertex_flush(sna);
sna_render_composite_redirect_done(sna, &tmp);
kgem_bo_destroy(&sna->kgem, tmp.src.bo);
return true;
@@ -3115,6 +3118,7 @@ fallback:
dst->drawable.bitsPerPixel,
op);
}
+
if (dst->drawable.depth == src->drawable.depth) {
op->base.dst.format = sna_render_format_for_depth(dst->drawable.depth);
op->base.src.pict_format = op->base.dst.format;
@@ -3142,10 +3146,9 @@ fallback:
op->base.src.filter = SAMPLER_FILTER_NEAREST;
op->base.src.repeat = SAMPLER_EXTEND_NONE;
- op->base.mask.bo = NULL;
-
op->base.is_affine = true;
op->base.floats_per_vertex = 3;
+ op->base.floats_per_rect = 9;
op->base.u.gen4.wm_kernel = WM_KERNEL;
op->base.u.gen4.ve_id = 1;
@@ -3155,6 +3158,15 @@ fallback:
goto fallback;
}
+ if (kgem_bo_is_dirty(src_bo)) {
+ if (sna_blt_compare_depth(&src->drawable, &dst->drawable) &&
+ sna_blt_copy(sna, alu,
+ src_bo, dst_bo,
+ dst->drawable.bitsPerPixel,
+ op))
+ return true;
+ }
+
gen4_copy_bind_surfaces(sna, &op->base);
gen4_align_vertex(sna, &op->base);
@@ -3296,6 +3308,7 @@ gen4_render_fill_boxes(struct sna *sna,
tmp.is_affine = true;
tmp.floats_per_vertex = 3;
+ tmp.floats_per_rect = 9;
tmp.u.gen4.wm_kernel = WM_KERNEL;
tmp.u.gen4.ve_id = 1;
@@ -3315,6 +3328,7 @@ gen4_render_fill_boxes(struct sna *sna,
box++;
} while (--n);
+ gen4_vertex_flush(sna);
kgem_bo_destroy(&sna->kgem, tmp.src.bo);
return true;
}
@@ -3397,6 +3411,9 @@ gen4_render_fill(struct sna *sna, uint8_t alu,
op->base.dst.bo = dst_bo;
op->base.dst.x = op->base.dst.y = 0;
+ op->base.need_magic_ca_pass = 0;
+ op->base.has_component_alpha = 0;
+
op->base.src.bo =
sna_render_get_solid(sna,
sna_rgba_for_color(color,
@@ -3410,8 +3427,7 @@ gen4_render_fill(struct sna *sna, uint8_t alu,
op->base.is_affine = true;
op->base.floats_per_vertex = 3;
- op->base.need_magic_ca_pass = 0;
- op->base.has_component_alpha = 0;
+ op->base.floats_per_rect = 9;
op->base.u.gen4.wm_kernel = WM_KERNEL;
op->base.u.gen4.ve_id = 1;
@@ -3498,6 +3514,7 @@ gen4_render_fill_one(struct sna *sna, PixmapPtr dst, struct kgem_bo *bo,
tmp.is_affine = true;
tmp.floats_per_vertex = 3;
+ tmp.floats_per_rect = 9;
tmp.has_component_alpha = 0;
tmp.need_magic_ca_pass = false;
@@ -3514,8 +3531,7 @@ gen4_render_fill_one(struct sna *sna, PixmapPtr dst, struct kgem_bo *bo,
gen4_render_fill_rectangle(sna, &tmp, x1, y1, x2 - x1, y2 - y1);
- if (sna->render_state.gen4.vertex_offset)
- gen4_vertex_flush(sna);
+ gen4_vertex_flush(sna);
kgem_bo_destroy(&sna->kgem, tmp.src.bo);
return true;
@@ -3538,6 +3554,31 @@ discard_vbo(struct sna *sna)
sna->render.vertex_index = 0;
}
+static void
+gen4_render_retire(struct kgem *kgem)
+{
+ struct sna *sna;
+
+ sna = container_of(kgem, struct sna, kgem);
+ if (kgem->nbatch == 0 && sna->render.vbo && !kgem_bo_is_busy(sna->render.vbo)) {
+ DBG(("%s: resetting idle vbo\n", __FUNCTION__));
+ sna->render.vertex_used = 0;
+ sna->render.vertex_index = 0;
+ }
+}
+
+static void
+gen4_render_expire(struct kgem *kgem)
+{
+ struct sna *sna;
+
+ sna = container_of(kgem, struct sna, kgem);
+ if (sna->render.vbo && !sna->render.vertex_used) {
+ DBG(("%s: discarding vbo\n", __FUNCTION__));
+ discard_vbo(sna);
+ }
+}
+
static void gen4_render_reset(struct sna *sna)
{
sna->render_state.gen4.needs_invariant = true;
@@ -3807,6 +3848,9 @@ bool gen4_render_init(struct sna *sna)
if (!gen4_render_setup(sna))
return false;
+ sna->kgem.retire = gen4_render_retire;
+ sna->kgem.expire = gen4_render_expire;
+
sna->render.composite = gen4_render_composite;
#if !NO_COMPOSITE_SPANS
sna->render.check_composite_spans = gen4_check_composite_spans;
diff --git a/src/sna/gen5_render.c b/src/sna/gen5_render.c
index db7eb7b..3af9097 100644
--- a/src/sna/gen5_render.c
+++ b/src/sna/gen5_render.c
@@ -2096,8 +2096,6 @@ picture_is_cpu(PicturePtr picture)
if (!picture->pDrawable)
return false;
- if (too_large(picture->pDrawable->width, picture->pDrawable->height))
- return true;
return is_cpu(picture->pDrawable) || is_dirty(picture->pDrawable);
}
@@ -2731,7 +2729,8 @@ gen5_render_composite_spans_done(struct sna *sna,
static bool
gen5_check_composite_spans(struct sna *sna,
uint8_t op, PicturePtr src, PicturePtr dst,
- int16_t width, int16_t height, unsigned flags)
+ int16_t width, int16_t height,
+ unsigned flags)
{
if ((flags & COMPOSITE_SPANS_RECTILINEAR) == 0)
return false;
commit c9dd1401615f0ed9492a0c0f547fb37150e013d1
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Thu Jul 26 16:31:16 2012 +0100
sna/gen4: Bump thread counts
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
diff --git a/src/sna/gen4_render.c b/src/sna/gen4_render.c
index bc37615..d62d744 100644
--- a/src/sna/gen4_render.c
+++ b/src/sna/gen4_render.c
@@ -80,7 +80,7 @@
#define URB_CS_ENTRIES 0
#define URB_VS_ENTRY_SIZE 1 // each 512-bit row
-#define URB_VS_ENTRIES 8 // we needs at least 8 entries
+#define URB_VS_ENTRIES 32 // we needs at least 8 entries
#define URB_GS_ENTRY_SIZE 0
#define URB_GS_ENTRIES 0
Reply to: