ef173fb96f
Append C++14 standard flag instead of prepending.
...
* This stops it from being overridden by the global --std=c++11 in the
root CMakeLists.txt which ends up after it in the command line.
2018-01-18 17:55:52 +00:00
bbac580c3a
Fix some missing whitespace around rst references in docs
2018-01-05 11:45:54 +00:00
7cbb1dfd2a
Set TextureDescription.resType in image_viewer
...
This fixes image_viewer on gl_replay. Previously, images would show up
as <Uninitialised Texture>.
2018-01-05 11:41:20 +00:00
d56fc083b7
Fix typo in in_application_api restructuredText, missing colon on param
2018-01-05 11:41:20 +00:00
0f7fc5fbea
Removing a resource replacement also needs to clear PostVS cache
2018-01-05 11:41:20 +00:00
ef19bd974b
Add intermediate resolve image for multisampled outputs
...
* Technically the resolve doesn't allow format conversion at all, so
instead of resolving directly to the swapchain image we need a middle-
man image of the same format as we use internally for the MSAA target
(RGBA8_SRGB) to resolve to, then blit from that to the actual
backbuffer.
2018-01-05 11:41:20 +00:00
a453307d5e
Account correctly for UAV mip slice when fetching 2D texture data
2018-01-05 11:41:19 +00:00
de7a74c7e5
Disable -Wunused-lambda-capture which complains about nonsense sometimes
2017-12-14 10:55:59 +00:00
78c402bb34
Create uniform translation table from live prog instead of initial prog.
...
The table is used on the live program, and we have seen the driver choose
different locations in the initial program due to the TF varyings.
2017-12-13 22:49:13 +00:00
ea7475f65f
Mesh Viewer: end TF if captured app already used TF on its draw call.
2017-12-11 20:12:34 +00:00
49d39f32a5
Attempt to fix weird SWIG compile error with lacking relocatable flag
...
* Although we're only making binaries so this doesn't seem like it
should be necessary.
2017-12-07 19:00:34 +00:00
b471f95c6a
Prevent recursive calls in CreateProcess hooks
...
* In one case the kernel32.dll version called into an API set version,
and we don't want to hook twice.
2017-12-07 18:37:24 +00:00
ecdbfc9978
Fix sampler slot extraction on gather4_po_c when debugging dxbc shader
...
gather4_po_c has 6 operands and offset of sampler should be 4 instead of 3.
2017-12-07 18:37:24 +00:00
c1c94d14e1
Allow NULL in glClear*Buffer and glClear*Tex functions
...
* The spec says that if data is NULL, it acts like a buffer filled with
zeroes, so we can just memset.
2017-12-07 18:37:23 +00:00
c6ab587875
Add hooking for CreateProcessAsUser and CreateProcessWithLogonW
2017-12-07 18:37:22 +00:00
7bdb3174b4
Check for ARB_texture_storage in CopyTex2DMSToArray and used it
...
* ARB_texture_view only works on textures allocated with
ARB_texture_storage so we have to check for both extensions and then
use texture storage to allocate space for the destination array
texture.
2017-12-07 18:13:21 +00:00
1431fd66cf
Snoop and disable requests for intel DX11 extensions. Refs #816
2017-12-07 18:13:04 +00:00
1d0d49f694
Clear PostVS cache when a resource is replaced. Refs #729
...
* For now, resource replacement implicitly means shader edits, but other
replacements could also affect the results of PostVS (e.g. buffer or
state changes). Trying to capture all this state in the cache key is
not very useful.
* The PostVS cache will be re-filled immediately after, as any
replacement does a force-refresh of the current event.
2017-11-10 18:36:31 +00:00
0e0921271b
Fix emulation of glMapNamedBuffer - pass correct map access bits
2017-11-09 18:33:38 +00:00
d4225b2262
Add some hefty padding to reported image required size on AMD. Refs #795
...
* AMD reports image required sizes with some variation even for the same
creation parameters. This can mean the same image is reported with
less required size during capture compared to replay, meaning memory
binds won't work.
* To get around this, we pad up the requirements to alignment * 4 and
fudge to try and encompass any possible variation.
2017-11-09 18:33:24 +00:00
fecd7dd318
Fix formatting from bad git cherry-pick
...
* Apparently git cherry-pick doesn't apply commit hooks :(.
2017-11-03 20:02:48 +00:00
522dba1981
Handle vulkan base pipelines properly, creating record dependencies
2017-11-03 15:54:13 +00:00
423938801b
Disable buggy and user-hostile NV optimus layer
2017-11-03 15:54:13 +00:00
93efdf46e1
Fix remapping multiple capture vulkan phys. devs to 1 replay phys. dev
2017-11-03 10:25:12 +00:00
c016855dce
Fix copy-paste error checking wrong count for validity. Refs #789
2017-10-28 14:12:10 +01:00
155571eecf
Ensure descriptors for multi-plane images are valid on replay
2017-10-25 00:24:19 +01:00
4befa9dd22
If a resource is NULL, make sure we don't use an invalid descriptor
...
* This can happen if the resource was not included in the capture, but
there was a descriptor passed (i.e. not NULL desc which is handled
separately) which specifies some parameters, but leaves the format
as UNKNOWN.
2017-10-25 00:24:19 +01:00
6589e1fa9a
Make sure to set refData to NULL whenever it's released
2017-10-24 23:04:10 +01:00
e23ddb98b3
Clamp currentIndex from Qt which can come back as -1 in some cases
...
* If this gets propagated far enough and used as an index it can cause
crashes, so we clamp to 0 when we know the value should be
non-negative.
2017-10-24 23:04:02 +01:00
d887e25b8d
Ignore format truncation warning on utf8printf.cpp
...
* We deliberately truncate in some tests, to check our printf behaves
correctly.
2017-10-24 23:03:57 +01:00
400b9487e1
Check for renderdoc.dll precisely when injecting, not just close match
...
* This means we don't call into a random dll if it happens to contain
the substring 'renderdoc.dll' somewhere.
2017-10-24 23:03:57 +01:00
6fd1a82454
D3D11 context state is cleared before executing a command list.
...
* Not doing this could taint the state and even cause invalid read-write
conflicts to change bindings vs. what should happen, if a RT is left
bound when it shouldn't, and then an SRV is bound as NULL because it
would conflict.
2017-10-23 15:53:14 +01:00
078efb8915
Resample thumbnails to fix JPEG shear distortion if(width % 4 != 0)
2017-10-23 13:39:19 +01:00
9b56227e50
Don't multiply mesh output size by num instances twice in GL
2017-10-23 13:39:07 +01:00
02fc28a36a
Try updating travis osx_image to fix intermittent qmake errors :(.
2017-10-11 12:37:57 +01:00
e275bae307
Add brew update to try and fix OSX intermittent errors on travis
2017-10-11 12:37:57 +01:00
b6b7946d3a
Don't pass internal GL shaders through reflection queries pointlessly
...
* Intel's drivers in particular seem buggy when trying to reflect
shaders, so there's no point in reflecting shaders that aren't from
the capture as all we do is add the potential for crashes.
2017-10-11 11:20:23 +01:00
f62f281415
Fix handling of D3D11_KEEP_{RTs/UAVs} in some edge cases
2017-10-11 11:11:52 +01:00
019b3072a1
When no event is found on >= search, fallback to *last* event not first.
2017-09-29 12:05:09 +01:00
50f03de69b
Add reference to SPIR-V library in AMD library
2017-09-26 11:56:24 +01:00
5ccc62f1fe
Auto-guess format of thumbnail from .jpg file extension
2017-09-26 11:56:23 +01:00
e037ac02ff
Add note to error message about possible startup fault - missing DLLs
2017-09-26 11:02:11 +01:00
9c1de3a876
Get NULL-check the right way around in ?: operator. Refs #754
2017-09-26 11:02:10 +01:00
8bf32ac560
Disable CMAKE_INSTALL_PREFIX message for android builds
2017-09-20 10:17:13 +01:00
04df0d60cd
Fix offscreen render setting GLWindowingData::wnd=0, an error condition.
2017-09-19 17:28:29 +01:00
57a897c5b0
Use ES color-renderable SRGB8_ALPHA8 for output window backbuffer.
2017-09-19 17:28:29 +01:00
fad2ce6964
Don't strip capture's device extensions away with backwards strcmp
2017-09-19 17:14:20 +01:00
4d5ab22e97
Don't display a register/slot index for built-in shader variables
2017-09-19 17:14:20 +01:00
3f2d82acfb
Fixes for compiling with disabled xcb, or disabled xlib & xcb
2017-09-19 17:14:20 +01:00
9e102f801a
Add a warning if only CMAKE_INSTALL_PREFIX is customised
...
* VULKAN_LAYER_FOLDER points outside the prefix to /etc in absolute, so
usually if you customise one you'll want to customise the other.
* This is especially true if you're pointing the install prefix at some
non-root location, and you want make install to succeed instead of
failing to write to /etc.
2017-09-19 17:14:19 +01:00
baldurk