I wrote some code which needed Round(); it isn't in ES2 so it wasn't in
the DSL intrinsic set yet.
Change-Id: I304f61b502a68d255d15899181e34fcae2ef16a0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/447017
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
This test fails on my work laptop (Radeon 5300M) so simply disabling it
on the tree isn't sufficient in this case. We don't seem to have a test
opt-out mechanism that automatically applies to local runs of dm.
Change-Id: I83582f0cdc3c9e6a61e6dc8e07851b9705143423
Bug: skia:12434
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/447057
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
This reduces our memory footprint slightly (smaller dehydrated data,
fewer IRNodes in the rehydrated GPU module).
This required a small tweak to our built-in types. $squareMat and
$squareHMat are now aligned with our $vec/$genType generics. (The "1x1
matrix" case is explicitly marked as invalid, allowing `float2x2` to
match up with `float2` and `half2x2` to match up with `half2`.)
Change-Id: I219a7be42102afbf60bb6857b53ac2b4f469113d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/447187
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Metal doesn't natively offer this intrinsic at all, but we now write an
equivalent template function whenever the intrinsic is encountered.
Proper testing will be added in a followup CL (when outerProduct is made
available in public SkSL).
Change-Id: Ie8d6bf8d735d0ab45b7559be68036b08c5802365
Bug: skia:12202
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/447296
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
* Update capture method to use latest interface
* Add simple hooks to enable command buffer capture for dm, nanobench, etc.
Change-Id: Ia69f1d10f1f80084342f4f38d12aca92f4dc710c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/447193
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Bug: chromium:1241134
Change-Id: If99fbd62aaa9e2b5d064fdec6a010d3ab0f9dd66
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446817
Commit-Queue: Greg Daniel <egdaniel@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
This includes compile-time optimization and tests.
The unit test is disabled in a followup CL
(http://review.skia.org/447057) because it exposes a Radeon 5300M bug
in OpenGL.
Change-Id: I8b2f0411358aeb68c4edfeb0bd7a2814c4be1f40
Bug: skia:12202
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/447056
Reviewed-by: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Change-Id: I1f3e5e0bdc216910be6b522f460b2c277134635c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/447190
Auto-Submit: Brian Osman <brianosman@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Bug: skia:10412
Bug: skia:12437
Change-Id: I93077bbd2ed40252966305df1b93ceb813218828
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446181
Reviewed-by: Eric Boren <borenet@google.com>
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
It's dangerous to leave constant variable expressions lying around, as
the temptation to access them via getConstantSubexpression is just too
great, but this currently isn't supported. This bit us in `inverse`
(see http://review.skia.org/440260) and it seems like we had a similar
latent bug with `cross` as well.
Fortunately, this change seemed to simplify the code in general.
Change-Id: Idd3e84ca202a1f05f91c1d6d5e94a3bc6da0d8d7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/447016
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Bug: dawn:22
Change-Id: I7c0a4f77bc2c2df81ff4be091bf86569898a1ade
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/447216
Auto-Submit: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Stephen White <senorblanco@google.com>
Reviewed-by: Stephen White <senorblanco@google.com>
Change-Id: I75e93820bd1df57d076560f1b2991f1756ca5389
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446996
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
The RenderCommandEncoder created for LoadMSAAFromResolve needs to be
able to merge with the encoder that follows it -- otherwise, the load
will do nothing. Hence if the following encoder has stencil, we need to
add stencil to the LoadFromMSAAResolve encoder.
Bug: skia:12086
Change-Id: Idf1249e386e1a24c4f18477e7ff67910bc21aa3e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446716
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
It turned out that Metal had equivalent intrinsics/casts all along; we
just needed to emit them.
Tests will be improved in a followup CL which adds the ES3-compatible
packing intrinsics into sksl_public.
Change-Id: Iec8a20b9f9fe9b1badea2944eb0b1f0a17c74560
Bug: skia:12351
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446744
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
The GLSL ES 3.0 Specification added a new section in the intrinsics
list, "8.4 : Floating-Point Pack and Unpack Functions." This shifts all
the following sections forward by one. I've updated various comments
which referred to the spec by section number. (The modern-day GLSL spec
is consistent with this new numbering, except for 8.9 "Fragment
Processing Functions," which leaps ahead all the way to 8.14 in today's
spec.)
Also, a handful of post-ES2 intrinsics in FunctionCall.cpp were lumped
together in a catch-all "not required by ES2" section. These have been
moved to the proper ES3 section.
Change-Id: If56c7307a06e6131a2a6d062f7560314362ba265
Bug: skia:12202
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446743
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
The test has been improved and now covers a variety of values. Because
this intrinsic has side-effects (an out-param), we do not support
optimizing it or treating it as a constant-expression.
The modf documentation doesn't mention anything about constant-
expression support or lack thereof. Experimentally, modf is also not
treated as a constant-expression by Apple GLSL or glslang:
http://screen/4RWwYKr6vCjxCPQhttp://screen/45ttDTVAFGDRyxP
Change-Id: I15bb1de80e90fa97ddf8e9d3803352603b9608d0
Bug: skia:12202
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446396
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
The code in writeFunctionCall which supported swizzled out-params has
been factored out into a pair of helper functions. The code for emitting
intrinsics now relies on these helpers when emitting out-params.
Change-Id: I4436185ae107d70b529e7e1ea0dd89f844c6a673
Bug: skia:11052
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446719
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This reverts commit 0f4304e6e7.
Reason for revert: breaks Adreno 6xx
Original change's description:
> Add RelaxedPrecision decoration to function-call temp vars.
>
> This is really same basic issue as http://review.skia.org/446640. We
> were creating a temp variable but ignoring its type's precision.
>
> Change-Id: I9a5fedd7ada864d36757fc196f42ff95bac7d706
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446718
> Commit-Queue: John Stiles <johnstiles@google.com>
> Auto-Submit: John Stiles <johnstiles@google.com>
> Reviewed-by: Brian Osman <brianosman@google.com>
Change-Id: I6ae4e264b60f7f38a1abb5f1d0324461a33c896d
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446742
Auto-Submit: John Stiles <johnstiles@google.com>
Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
An unused child that used its coordinates would triggers asserts in the
GP - we expect that all FPs are sampled/invoked. This generates slightly
sub-optimal code (we compute and interpolate the local coordinates for
that child FP), but it's not likely to happen in real code.
Bug: skia:12429
Change-Id: Ic2ddc65d16a7e1f47af8c4192e5ff9ea329bf335
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446836
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
This is really same basic issue as http://review.skia.org/446640. We
were creating a temp variable but ignoring its type's precision.
Change-Id: I9a5fedd7ada864d36757fc196f42ff95bac7d706
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446718
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Bodymovin uses tt: [1-4] to encode a layer track matte type, and skips
the property when a track matte is not present.
Flow OTOH uses explicit tt: 0 to signal the absence of a track matte.
Update the parser to not error out on tt: 0.
Change-Id: I2d456ac3479e356ba1fc5320589848f8872775e6
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/445957
Commit-Queue: Florin Malita <fmalita@google.com>
Reviewed-by: Jorge Betancourt <jmbetancourt@google.com>
Bug: skia:11837
Change-Id: I522636c6dbf977929b0d368e18beb9c3a01fdeb1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446184
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
skgpu::v2::Device will also need both of these so move them somewhere accessible
Bug: skia:11837
Change-Id: Iacaf0332da933515451056eeda62949b049378b7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446180
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Robert Phillips <robertphillips@google.com>
We don't need to initialize input values for `out` params. (We were
treating them the same as `inout`.)
Change-Id: Ib447d15de237a6a03740ad012180691dc60a50bd
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446717
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Bug: skia:12086
Change-Id: I84f05c874f7f097b9ff7e4ea1070f168964c7afe
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446158
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
Previously we were not honoring the variable precision at all (passing
null to nextId).
Change-Id: I0e217e6c0d3d701dc0540d4d4069a3597abdad11
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446640
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
48da7e7e7c..5a7b8c6125
2021-09-08 b.schade@samsung.com Vulkan: Removed size check when handling mismatched vertex attributes
If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/angle-skia-autoroll
Please CC erikrose@google.com on the revert to ensure that a human
is aware of the problem.
To file a bug in ANGLE: https://bugs.chromium.org/p/angleproject/issues/entry
To file a bug in Skia: https://bugs.chromium.org/p/skia/issues/entry
To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md
Cq-Include-Trybots: skia/skia.primary:Build-Debian10-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-Golo-GPU-QuadroP400-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC8i5BEK-GPU-IntelIris655-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE
Tbr: erikrose@google.com
Test: Test:
Change-Id: Ie023217dc60ebe769c212e3acd9e1fce84431093
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446536
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
fefdf81e09..48da7e7e7c
2021-09-08 geofflang@google.com Add a gn arg to enable collecting Vulkan system info on Linux
2021-09-08 syoussefi@chromium.org Translator: Fix precision of gl_PointSize on ES3+
2021-09-08 b.schade@samsung.com Vulkan: Make descriptor set bindings consistent across shader stages
2021-09-08 cclao@google.com Vulkan: Skip but keep incompatible updates.
2021-09-07 cclao@google.com Vulkan: Simplify TextureVk::initImage() call.
If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/angle-skia-autoroll
Please CC erikrose@google.com,robertphillips@google.com on the revert to ensure that a human
is aware of the problem.
To file a bug in ANGLE: https://bugs.chromium.org/p/angleproject/issues/entry
To file a bug in Skia: https://bugs.chromium.org/p/skia/issues/entry
To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+doc/main/autoroll/README.md
Cq-Include-Trybots: skia/skia.primary:Build-Debian10-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-Golo-GPU-QuadroP400-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC8i5BEK-GPU-IntelIris655-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE
Tbr: erikrose@google.com,robertphillips@google.com
Test: Test: VulkanFramebufferTest.R4G4B4A4TextureSampleOnlyActuallyUses444Format
Change-Id: Iead1e5aa621aa613eee6407dc835b911a2472af7
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446470
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Bot-Commit: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Change-Id: I530373e9b53e76a7cd1c0b3a5a1b05354975790c
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446186
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Unblocks angle roller into skia that was triggering an assert in
EGLImageTest. In that test, there were two egl contexts created that
shared the same egl display. The child context was destroyed first,
and ANGLEGLContext's destroyGLContext() implementation always called
eglTerminate on the display. According to the egl spec you're not
supposed to do anything with the terminated display other than call
makeCurrent() to release any active context.
In this case, the child context's termination would cause the display
to be deferred until the parent context was destroyed. This triggered
a bug in ANGLE's D3D backend, leading to the assert. However, with
that fixed, we then trigger critical error messages about calling
eglDestroySurface, etc. in the parent context's destroyGLContext()
function since the display was terminated out from beneath it.
This change adds a rudimentary ownership tracking mechanism to the
ANGLEGLContext so that when a child context is cleaned up it won't
call eglTerminate on its display. This happens to avoid the bug in
ANGLE's D3D backend as well as silencing the critical warnings if it
were fixed.
Bug: skia:12413
Change-Id: I31f64189315a1921adbee204d11138272d4b40af
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446182
Commit-Queue: Robert Phillips <robertphillips@google.com>
Auto-Submit: Michael Ludwig <michaelludwig@google.com>
Reviewed-by: Robert Phillips <robertphillips@google.com>
Change-Id: I31dd90b2f09876d37c2c202d620cef4028af6266
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446183
Reviewed-by: Herb Derby <herb@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Bug: skia:9282
Change-Id: I04afd352ad8c0c167993f29f4293d857f5af7170
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/445619
Reviewed-by: Robert Phillips <robertphillips@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
DSL was improperly allowing interface blocks in runtime shaders, which
caused PipelineStageGenerator to get upset.
Bug: oss-fuzz:38131
Change-Id: I593e68f2cab3db9151d606e65e2826ffa9c494e2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446324
Commit-Queue: Ravi Mistry <rmistry@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Two minor changes:
- Converting a Block with bad statements will now generate a partial
block instead of nullptr.
The change mirrors how DSL behaves; functions containing invalid
statements will now be created and added to the program. Previously, we
would discard a function definition with any invalid statements inside;
this prevented duplicate-function-definition errors from appearing.
- Converting a return with a bad expression will now generate a
poisoned return instead of nullptr.
This change improves diagnostics for functions with invalid return
statements. If we eliminate the return statements (by returning null),
we report bad return statements as "function can exit without returning
a value" (which is confusing).
Change-Id: I6d998d5c50585f8d96bb7e3cb7f59b63125d6a62
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446325
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Change-Id: I5f0c36b44cd5bfd3f978d892a342a1d008baceb2
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446176
Reviewed-by: Herb Derby <herb@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Herb Derby <herb@google.com>
To do this with a clean conscience, I needed to convert the unroll-
counting logic from a linear time algorithm to constant-time. Getting
all the edge cases correct requires a lot of care, and there are now
plenty of unit tests.
Change-Id: I620909d069ac425b7310e345bf80ec844fe035f8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/445643
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
The motivation for this is that with the new backend info structs the
sample count will be stored on non backend specific classes
(e.g. GrAttachment) and will not be part of the GrVkSurfaceSpec struct.
Thus the info will not be available from just a GrVkImage (or at least
until GrVkImage and GrVkAttachment get merged down the road).
This also removes some uses of GrVkImage::imageInfo() getter since that
struct will also be going away.
Bug: skia:12402 skia:10727
Change-Id: Ie90870e1d834e6ea6ba029d0b480641fb8d2cfa0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/446177
Reviewed-by: Jim Van Verth <jvanverth@google.com>
Commit-Queue: Greg Daniel <egdaniel@google.com>
This allows us to get rid of a lot of .c_str()'s.
Change-Id: I09102f90d69620614dc5a7a2ebc64bd3e9b1c437
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/445816
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Bug: 12145
Test: imagedecoder_png_fuzzer
Test: Codec_PngRoundTrip
If a PNG's sBIT specifies that the significant bits are 565, SkPngCodec
treats the image as defaulting to kRGB_565_SkColorType, rather than the
typical kN32_SkColorType. This feature is likely rarely used in the
wild, but it is used by Skia to encode kRGB_565_SkColorType to a PNG and
then recreate the original SkPixmap. These Skia-created PNGs always use
PNG_COLOR_TYPE_RGB, so only respect this sBIT with PNG_COLOR_TYPE_RGB.
Further, if the PNG is PNG_COLOR_TYPE_PALETTE, treating it as 565 means
that we read it incorrectly.
Change-Id: I81dcce771c8492cdcf5677d3c13167a1231acfe4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/445963
Reviewed-by: Heather Miller <hcm@google.com>
Commit-Queue: Derek Sollenberger <djsollen@google.com>
