-
4e55fdb4
by Mike Schuchardt
at 2021-08-30T09:17:43-07:00
build: Update to header 1.2.190
- Update known-good
- Generate source
-
d6a5a7ae
by Charles Giessen
at 2021-08-30T13:10:33-06:00
scripts: Remove loader_icd_init_entries duplicate
Seems to be a copy-paste error that makes the function declaration appear twice.
-
b97855a8
by Charles Giessen
at 2021-08-31T15:04:40-06:00
test: Add TestICD DeviceGroup support
Add support for tests which use device groups by adding a new structure that
holds pointes to PhysicalDevice structs and implementing the necessary vulkan
API calls.
This commit also:
- Formats test_icd.cpp
- Makes vk_icdGetPhysicalDeviceProcAddr return stuff by refactoring the
instance procaddr functions
-
9b66d3ef
by Charles Giessen
at 2021-08-31T15:04:40-06:00
test: Move DeviceGroup tests to regression suite
-
140738f8
by Charles Giessen
at 2021-08-31T15:04:40-06:00
test: Add handle_assert helper functions
Added functions which check handle values using GTEST. Includes:
checking for null, not null, equality of two handles, and variants that
operate on vectors and arrays.
-
4c8bf4cc
by Mark Young
at 2021-09-02T07:51:43-06:00
Update XDG paths
Issue #245 was created because we were not properly using the XDG
paths.
This change updates the path search order to be more correct.
While an argument could be made that the loader really should only
be looking in the DATA folders, we already are looking in the
CONFIG folders. Therefore, just correct the order and also make
sure the proper _HOME variable for CONFIG and DATA is searched
first (but only in non-superuser mode).
-
ee4d7bb1
by Mark Young
at 2021-09-02T14:31:28-06:00
Fix loader not knowing about extensions enabled in layers
This change allows the loader to know extensions enabled in the layers
as well as extensions enabled but handled by layers.
This should fix several issues where the enabled extensions is not
known soon enough for layers to use.
-
20cf221e
by Charles Giessen
at 2021-09-03T11:11:25-06:00
loader: Make all loader functions use snake_case
Before it was a mix of camelCase, PascalCase, and snake_case.
"Vulkan" functions, like terminator_GetPhysicalDeviceProperties, remain in camelCase.
This way it looks visually similar with the actual vulkan functions they correspond to.
-
37100015
by Charles Giessen
at 2021-09-06T18:19:14-06:00
test: Improved Test Wrapper interface
Made InstanceWrapper and DeviceWrapper simpler to use by moving them to test_environment
then making the Create() member functions and doing the gtest assertions in the function
itself.
Additionally:
Removed the `detail` namespace in the test_environment.
Renamed get_new_test_icd/layer to just `reset_icd/layer`. This is clearer about what it
is doing and makes the interface more expressive with code such as
`env->reset_icd().SetMinInterfaceVersion(5);`
-
7d7cdaa1
by Mike Schuchardt
at 2021-09-08T15:22:33-07:00
build: Update to header 1.2.191
- Update known-good
- Generate source
-
d81a521e
by Charles Giessen
at 2021-09-09T11:14:49-06:00
loader: Move layer cleanup to a single function
Previously the logic to free all the components of loader_layer_properties was spread
across a half dozen different places, resulting in leaks. This commit moves the logic
into a single location, centralizing the process and making it easier to add new
fields to the loader_layer_properties struct.
-
9b11eca7
by Charles Giessen
at 2021-09-09T11:14:49-06:00
test: Fix extra } in layer manifest printer
-
3547086a
by Charles Giessen
at 2021-09-09T11:14:49-06:00
test: Add test for Override Meta Layer
-
f30584de
by Charles Giessen
at 2021-09-09T11:28:56-06:00
build: Alphabetize loader source file list
-
e4e0b936
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Update Copyright
-
17624a2d
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Use #pragma once everywhere
The #pragma once include guard is sufficiently widespread enough to allow
usage of it everywhere.
-
7d46bb6d
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Update Copyright for 2021
-
f6760604
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Move loader struct defs into common file
This commit separates the structure definitions from the function declarations.
The reason to do this is to allow splitting of loader.h/.c functionality into
individual headers, which may require access to these structs but don't need to
know about the other functions.
-
370b46d9
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Add gpa_helper.c
Make gpa_helper.h not contain function definitions, moving them into the
appropriate header file.
-
359cd87f
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Dont use TLS in dirent_on_windows
Loader will now pass in the instance handle directly rather than rely on TLS to
pass it. This makes the loader code simpler to reason about and better since
if a second instance is created on the same thread the TLS wont be overwritten.
-
995e78dc
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: cJSON to use regular heap instead of TLS
Refactored the cJSON library to take an loader_instance pointer everywhere it
allocates memory. This way a TLS loader_instance pointer doesn't need to be
set at initialization and then at vkCreateInstance.
-
903a17df
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Remove TLS
The only purpose TLS was serving was as a channel for allocation callbacks to
be useable inside cJSON and dirent_on_windows. TLS didn't solve any specific
threading problem. Instead it served as an information channel. On top of that,
the only time when the TLS was used was inside vkCreateInstance as the
pre-instance calls would have the TLS pointer to loader_instance be set to
NULL.
-
353e1394
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Move allocation functions to a header
Functions pertainting to allocation (malloc, realloc, free) have been moved to
their own translation unit for better organizational purposes.
-
055d1dd4
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Dont include C++ libraries in C++ mode
-
439b453d
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Shift around vk_loader_platform headers
Moved the includes into their own block, instead of living in the block for
each operating system. This way is easier to see what includes are in the file
in its entirety.
Also, _GNU_SOURCE was uncommented since it appears in each source file
individually and is needed for alloca and secure_getenv
-
6622bcda
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Remove unneeded parts of loader_common.h
-
1b870bdc
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Make vk_loader_layer.h a proper header
Makes it possible to include this file without worry of not having the required
headers already included.
-
f07c3aee
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Remove duplicate macro defines
-
e13e2f51
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Move alloca into allocation.h
-
0530f8cb
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Make _GNU_SOURCE a cmake definition
Moves the many instances of #define _GNU_SOURCE into a single cmake call which
sets the macro in all translation units for the vulkan library. This way it
cannot be forgotten and makes behavior across the library consistent.
-
8138a8d8
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Make logging and getenv their own headers
This commit pulls the logging capabilities into a single log.h/.c file and does
the same for the getenv functionality.
-
de30d282
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Remove headers from cmake source list
Headers do not do any functional thing here. Better to remove them so they dont
clutter up or make people thing they need to be there.
-
4799c294
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Header include cleanup
Separated headers into common groups, with system headers being first, then
Vulkan API headers, followed by loader headers.
Then alphabetized headers in each group. Due to moving _GNU_SOURCE to being set
by cmake, there isn't any implicit dependency between the include order used.
Made sure headers should be able to be included independently, such as
including the vulkan headers where appropriate.
-
4da6200c
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Move init functions closer together
-
8cb5f492
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Move struct declarations into header
-
6bc20751
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Move Windows logic to dedicated file
Relocated most windows specific functionality into loader_windows.h/.c
This removes the need for loader.c to include windows specific header files.
-
62957141
by Charles Giessen
at 2021-09-09T11:28:56-06:00
gn: Fix gn build for refactor changes
-
5cd8dae5
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Rename SetupLoaderTermPhysDevGroups
Forgot this name when renaming all the functions.
-
310ba146
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Move simple terminators to terminator.c
Keeps the complex terminators in loader.c for the time being.
-
0f897d9f
by Charles Giessen
at 2021-09-09T11:28:56-06:00
loader: Make trampolineGetProcAddr snake_case
-
782cbe6f
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Handle OOM in loader_phys_dev_ext_gpa
Removing some dead code revealed a case where a potencial OOM situation
was not being handled.
Change-Id: Ibc5ad607f6adc0f590f663ab928c1abca6732d96
-
4ecb4210
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Fix potential memory leak
Also remove a redundant setting of an int to 0
Change-Id: Ia099ad75e9da14aacf9874d75267716c12794305
-
ed07c59c
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Use abort to force a crash where needed
Many places in the loader need to intentionally crash. This commit makes the
loader use `abort();` in those places, including replacing numerous
`assert(false)` statements. Using abort clearly signals intent instead of
having it do a null pointer dereference to cause the crash.
Change-Id: I6874eab5702c94af59b7984fab00f6eaf03f2f45
-
163ef9dc
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Format WSI error messages
Change-Id: I3040be426a8e70ca41e7beede1fae6b52e1381af
-
bff4830f
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: assert if trying to malloc 0 bytes
The behavior of malloc is undefined if size is 0. This commit checks for that
and prevents a possible malloc of zero in device extension property enumeration
by only malloc'ing if size is greater than 0.
Change-Id: Id506ab93d404bd4895503c660ef74bcb9c4550a3
-
7cf713bd
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Remove redundant stores
`layer_node = layer_node->next;` is redundant in error paths that goto out.
Change-Id: I889b03c825f896f458a1ed3171b951d64d8827d7
-
692345c7
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Get DeviceGroups early exits if 0 are found
This commit prevents undefined behavior by exiting
`setupLoaderTermPhysDevGroups` if total_count is 0.
Change-Id: Ieef2747724f79ec26a08a74884904bc0e0485c17
-
9d080205
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Prevent stack_alloc of 0
`loader_create_device_chain` would previously attempt to use stack allocations
when `expanded_activated_layer_list.count` was zero, which is undefined behavior
Change-Id: I56fab835a1728914191f1a7adc210dae58afa155
-
81537271
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Remove dead store of total_count
`loaderScanForLayers` total_count is not used after determining
if it is not zero.
Change-Id: I2ff3a1a5d7dde42ce5ad7d57aab63831ccf58156
-
1b5f8c0a
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Don't pass null pointer into strcmp
strcmp wasn't correctly guarded against a null first parameter when
doing the shallow search in loader_name_in_dev_ext_table and
loader_name_in_phys_dev_ext_table.
Change-Id: Ia751e9860334895930210defa2dcb2becbfd6daa
-
f5557ab7
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Prevent use-after-free in error path.
When `setup_loader_term_phys_devs` and `setup_loader_term_phys_dev_groups` hit an
error, they attempt to delete all the alocated elements in `new_phys_devs`. However,
the code first may copy elements from another array into it first before hitting
the error. This causes the deletion code to free those 'good' elements which were
copied, leading to use-after-free's.
Additionally, an error message used total_gpu_count instead of total_icd_count, this
is fixed with this commit.
Change-Id: I8f714f699ff09cc99d094cf2ed8190cedcad684b
-
841d3740
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Remove deadstore when breaking from a loop
Change-Id: Ic72cf396cad918ce255b331fa9a9c3a652b8764d
-
07a8264a
by Charles Giessen
at 2021-09-09T12:15:49-06:00
loader: Fix freeing of uninitialized list
The memset of icd_tramp_list in terminator_EnumerateInstanceExtensionPropeties
was inside an else-branch. If an error occured in the preceding if branch, this
would cause the cleanup code to run on a non-zerod out icd_tramp_list.
Change-Id: I7af1da31580f01c54c4ffababbb9c7ecfa8966d8
-
f9d94bff
by Charles Giessen
at 2021-09-09T12:15:49-06:00
test: Add test for memory leak in setup_tramp_phys_devs
Tries to trick the loader by increasing the number of physical devices
after the call to vkEnumeratePhysicalDevices to get the count.
-
6b28636a
by Charles Giessen
at 2021-09-09T13:46:18-06:00
build: Use generated source everywhere
The previous commit that added a generated header version didn't apply
it to mac or linux specific versioning. This commit fixes that.
-
4c723516
by Charles Giessen
at 2021-09-09T13:53:46-06:00
loader: Move disable_environment into function body
The function `loader_read_layer_json` does not modify the
disable_environment parameter that is passed in. It can safely be made
a variable local to the function.
Change-Id: I25ef3f3e5f008082dc9514be8f607d2acbf212a1
-
5e77efae
by Charles Giessen
at 2021-09-09T14:31:22-06:00
loader: Reverst asserting on size 0 in malloc
-
f7c1fe1b
by Mark Young
at 2021-09-10T13:57:32-06:00
Update more logging info for the Loader
Output more information to the "info" debug target when the loader loads
so we can determine more info about:
- Where loader was loaded from (Windows)
- What Git Branch was used to build it
- What Git tags were most recent when it was built.
Also, add more detailed logging of implicit layers:
- Log manifest file location
- Layer library location
- Type of layer (implicit/explicit)
- Disable enum if layer is implicit
-
087148ac
by Charles Giessen
at 2021-09-10T13:59:45-06:00
test: Add callback to layers create instance and device
This allows some code to be written inside the test which then gets run
during the call to vkCreateInstance and vkCreateDevice.
-
813fc548
by Charles Giessen
at 2021-09-10T13:59:45-06:00
test: Add ICD Interface V6 TestICD binary
-
2ca9c3a9
by Charles Giessen
at 2021-09-10T13:59:45-06:00
test: Add test which calls GetPDProps inside CreateInstance
This test makes sure that any layer calling GetPDProps inside of CreateInstance
succeeds and doesn't crash.
-
e1dc2228
by Charles Giessen
at 2021-09-10T14:25:29-06:00
loader: Move struct decls to loader_common.h
Not all structs declarations in loader.h were moved to loader_common.h. This
caused compilation errors due to not having the struct declaration available
in the headers it was needed.
-
244d0300
by Charles Giessen
at 2021-09-11T10:38:56-06:00
test: Fix 32 bit test suite build error
-
4c84b822
by Jamie Madill
at 2021-09-13T09:37:03-06:00
loader: Add loader.rc to Windows GN build
This omission was causing the build dll to be missing version info.
-
6e345ec1
by Mike Schuchardt
at 2021-09-15T09:41:50-07:00
build: Update to header 1.2.192
- Update known-good
- Generate source
-
2d69d5c7
by Natalie Chouinard
at 2021-09-16T15:06:49-06:00
loader: Add cJSON prefix to update function
-
a1256b51
by Mike Schuchardt
at 2021-09-21T10:49:42-07:00
build: Update to header 1.2.193
- Update known-good
- Generate source
-
909ef587
by unknown
at 2021-09-21T13:25:38-06:00
build: Fix using wrong Windows SDK for VS 2015
CMake 3.19 disallowed VS 2015 builds from selecting an SDK higher than
10.0.14393. Unfortunately, the loader requires SDK version 10.0.17763 or
higher. This commit uses the 3.19 capability to disable that limit set in
place by CMake so that we can use an SDK that suits our needs.
-
75431e4b
by Charles Giessen
at 2021-09-22T16:26:19-06:00
test: Isolate Env-Vars from test framework
Previously if any of the env-vars the loader looks for were set on the
system, the loader would find and use them, leading to highly confusing
logs. Removing them preemptively at tests start make this a non-issue.
-
393e75eb
by Nathaniel Cesario
at 2021-09-23T13:19:14-06:00
build: Add UPDATE_DEPS cmake arg
Add an UPDATE_DEPS cmake arg that can be used to automatically run
update_deps.py as needed.
-
3277ecde
by Charles Giessen
at 2021-09-24T11:47:30-06:00
build: Set DLL version automatically
The loader.rc file was used to set the version used in the loader. However,
this was always done manually (either by hand or a script to edit the file).
This commit makes the version number needed generated automatically based on
the current generated header version or uses a provided version, such as in
the case of SDK builds with a 'build revision', eg "1.2.191.2".
This commit also renames the atrosciously named
loader_generated_header_version_generator to a nicer
loader_versioning_generator.
-
7ff42c87
by Charles Giessen
at 2021-09-27T17:09:46-06:00
Use CMake to define SECURE_GETENV variables
Use `target_compile_definitions` to provide the HAVE_SECURE_GETENV and
HAVE___SECURE_GETENV macro defines. This removes the need to configure
a file and makes the definition global to the loader, rather than just
from whoever included the header file.
-
1bf496e1
by Charles Giessen
at 2021-09-27T17:44:01-06:00
build: Add RELEASE_BUILD option for Windows SDK builds
This CMake option controls the information put in the loader.rc file,
specifically the "Dev Build" string in the version info.
-
11b37c16
by Mike Schuchardt
at 2021-09-28T11:44:36-06:00
build: Update to header 1.2.194
- Update known-good
- Generate source
-
5f67bbd6
by Charles Giessen
at 2021-09-28T12:54:43-06:00
build: Update update_deps.py to match VVL
Was out of date with the feature set VVL contained.
Fixed an issue where backslashes were used on windows, which caused issues when
cmake decided to escape a string one more time than necessary.
"C:\\Users\\..." -> "C:\Users" which has a `\U` escape symbol mucking things up.
-
69fc7ecb
by Charles Giessen
at 2021-09-28T12:54:43-06:00
build: Move test dependencies to known_good.json
This makes the requisite changes to make the dependencies needed for testing automatically
downloaded by cmake when the UPDATE_DEPS option is given.
-
bcd56925
by Charles Giessen
at 2021-09-29T13:19:25-06:00
loader: Only include loader_windows.h on windows
This header should only be included on windows platforms, since non-windows platforms
may not know that the header should be present (which is the case for GN builds).
-
0313a16d
by Ralf Habacker
at 2021-09-29T14:59:11-06:00
loader: Fix setting `libdir` in generated vulkan.pc on Windows
Fixes #668
-
edccdb34
by Charles Giessen
at 2021-09-30T15:07:32-06:00
build: Revert generated loader.rc
Revert the commit that uses the vulkan-headers version and instead uses the
1.0.1111.2222 version number while still allowing for the CMake build arg
BUILD_DLL_VERSIONINFO to dictate the actual version.
-
83e58774
by Charles Giessen
at 2021-09-30T22:39:41-06:00
test: Fixup XDG path handling, add XDG tests
-
d2bad77b
by Charles Giessen
at 2021-09-30T22:39:41-06:00
test: Add XDG & VK_LAYER_PATH tests
Tests check to see if the paths are searched by looking at the log.
Makde FrameworkEnvironment contain a debug_log, this is to prevent any
lifetime issues. This prevents the possibility of creating a DebugUtilsLogger
after the InstWrapper.
Had to redo the test framework a bit to make them work.
-
0cc98ec7
by Charles Giessen
at 2021-09-30T22:39:41-06:00
loader: Remove old debug path printing
This commit only prints paths using the new, more structured, output.
It wasn't removed until now due to the paths being used in tests which
have since been modernized, thus no longer needing to keep the old output
style.
-
d47266c7
by Charles Giessen
at 2021-10-04T12:02:49-06:00
ci: Enable tests on MacOS GithubActions
Tests only get built if you enable them now, instead of when
googletest is in the external folder.
-
9bcea4ab
by Charles Giessen
at 2021-10-05T12:09:50-06:00
test: Move more old tests to new framework
More tests can now be written with the new framework. Several of the
shell scripts used tests in vk_loader_validation_tests which were phased
out so no longer necessary.
-
a15e471c
by Charles Giessen
at 2021-10-05T12:09:50-06:00
test: Update ExtraLoaderTests
Updated the enable/disable env-var tests to use the test framework.
Removed the run_extra_loader_tests.sh.
-
64527ac2
by Charles Giessen
at 2021-10-05T12:09:50-06:00
test: Use correct path in json files
Previously a Windows path in a json file did not include the correct
backslahses. This commit rectifies that.
-
b7a494bf
by Charles Giessen
at 2021-10-05T12:09:50-06:00
test: Use correct name in def files
-
9189b9cb
by Charles Giessen
at 2021-10-05T12:09:50-06:00
test: remove stub functions
Test code should not use stub functions because its undefined behavior. This
resulted in SEH exceptions on Win32.
-
4d52412b
by Mike Schuchardt
at 2021-10-05T11:20:01-07:00
build: Update to header 1.2.195
- Update known-good
- Generate source
-
8c0ee92e
by Graeme Leese
at 2021-10-06T10:06:20-06:00
Only parse git information if .git/HEAD exists
If the source comes from somewhere else (eg. the release tarballs) then
.git/HEAD does not exist, so the cmake step was failing. Only try to
colelct the git information if the file exists.
Fixes #714.
-
77162477
by Charles Giessen
at 2021-10-07T10:00:08-06:00
Zero-init values passed into EnumPhysDevs
-
acfcfd89
by Charles Giessen
at 2021-10-07T10:00:08-06:00
test: Update WrapObjects test to new framework
Moved over the WrapObjects test logic into the new framework, removing all
the old tests infrastructure that is no longer needed.
-
aeffbbd6
by Charles Giessen
at 2021-10-07T13:50:17-06:00
Update README.md to remove mention of Vk-Headers
-
6d9e5b07
by Mike Schuchardt
at 2021-10-14T10:12:55-07:00
build: Update to header 1.2.196
- Update known-good
- Generate source
-
b9d0a4ff
by Charles Giessen
at 2021-10-15T12:34:50-06:00
Only return pre-instance functions when instance is NULL
vkGetInstanceProcAddr should only return the pre-instance, aka global
functions, when the VkInstance handle is NULL. This brings the desktop
vulkan loader into alignment on the spec and with the android loader.
https://github.com/KhronosGroup/Vulkan-Docs/issues/1605
The repo previously did follow this rule, but was changes in the
1.0.42 header version timeframe for unknown reasons.
-
61688e09
by Mark Young
at 2021-10-15T13:42:00-06:00
Re-arrange/update loader docs
Update the loader documentation in the following ways:
1. Move documentation to new "docs" folder"
2. Split LoaderAndLayerInterface.md into multiple docs to focus
on target audience:
- Top-Level LoaderInterfaceArchitecture.md
- Applications LoaderApplicationInterface.md
- Layers LoaderLayerInterface.md
- ICDs LoaderImplementationInterface.md
3. Upload newer images and their corresponding original Inkscape files.
4. Cleanup and update sections on Linux directory search
5. Add new sections to detail items missing
- VkConfig
- Handling undef Fuchsia
6. Language cleanup
Thanks to @charles-lunarg and @smcv for feedback and various section language
corrections.
-
f5243495
by Charles Giessen
at 2021-10-18T15:14:20-06:00
test: Don't add null paths for redirection
-
b3552792
by Eleni Maria Stea
at 2021-10-20T10:29:26-06:00
loader: include FreeBSD in supported platforms
Added FreeBSD in supported platforms in vk_loader_platform.h, and tests,
defined BSD_VISIBLE and included some necessary header files for
Vulkan-Loader to compile on FreeBSD.
v2: did the same in loader/loader.c (Charles Giessen)
v3: did the same in the tests (Charles Giessen)
v4: applied this downstream patch from Jan Beich because there's no /proc
on FreeBSD: https://github.com/freebsd/freebsd-ports/blob/7caf52e37ba2174dc50a7440d63eec70d7d58ebc/graphics/vulkan-loader/files/patch-loader_vk__loader__platform.h (Jan Beich)
v4: rebased to master and modified tests/loader_regression_tests.cpp,
tests/loader_testing_main.cpp
v5: moved the definition of __BSD_VISIBLE 1 (required for alloca to
work) in CMakeLists.txt from header files (Charles Giessen)
v6: replaced ${CMAKE_SYSTEM_NAME} with CMAKE_SYSTEM_NAME in
loader/CMakeLists.txt to match the style of previous checks (Jan Beich)
v7: replaced add_compile_definitions with target_compile_definitions to
prevent the macro from being applied unnecessarily (Charles Giessen)
-
0162d164
by Mark Young
at 2021-10-20T16:27:28-06:00
Change Implementation->Drivers
Feedback from Khronos was that "Implementation" was for the entire
set of Vulkan components underlying the application (including the
loader, layers, and drivers). Instead, what I had been calling
implementation should be called "Driver" including software
implementations of Vulkan at that level.
Also update the LOADER_DEBUG option "implem" to take "driver" as well.
All images also updated.
-
a9bc611d
by Mark Young
at 2021-10-21T15:33:01-06:00
Fix broken doc links in new docs
-
5a4c6917
by Charles Giessen
at 2021-10-22T16:39:01-06:00
Fix vkGetInstanceProcAddr not handling null instance
The issue was that trying to get a non-pre-instance function with a NULL
instance handle would cause a crash. This change also adds a test that
would of caught the oversight.
-
fab0feba
by Mark Young
at 2021-10-27T07:31:43-06:00
Modify loader warning about physical device count
Modify the loader warning about one or more layers modifying the
physical device count. Originally, it was as a general warning,
but modified it to also output during layer debug.
-
830a0724
by Mark Young
at 2021-10-27T14:00:40-06:00
Added layer debug section and pre-instance info
Add a section in the application documentation on how to debug possible
layer issues.
Also add a small note on why a layer may need to intercept pre-instance
functions.
-
ad393f9b
by Tomek Bury
at 2021-11-02T15:57:35-06:00
Fix terminator for GetPhysicalDeviceToolPropertiesEXT
Set the pToolCount as this function returns VK_SUCCESS.
-
7e3866fa
by Mike Schuchardt
at 2021-11-03T09:21:38-07:00
build: Update to header 1.2.197
- Update known-good
- Generate source
-
bf74db88
by Mark Young
at 2021-11-04T09:39:34-06:00
Add bundling section and rename "desktop loader"
Add a section warning about possible issues with bundling the
Khronos loader with an application/engine.
Also, rename "desktop loader" to "Khronos loader" since it is now
used for more than the original desktop targets.
-
77f3ea47
by Eleni Maria Stea
at 2021-11-04T09:47:34-06:00
loader, asm_offset: define BSD_VISIBLE and fix ifdef bug
In vk_loader_platform.h there was a leftover #if defined (__FreeBSD__)
... elif #defined(__FreeBSD__) that needed fixing. After that it was
necessary to define BSD_VISIBLE before including <sys/types.h> required
by <sys/sysctl.h> required by asm_offset and loader CMakeLists.txt had
to be modified accordingly.
-
3fd34fe1
by vladislemon
at 2021-11-04T09:53:47-06:00
Fix architecture detection on recent versions of GCC
-
059f7c1a
by Mark Young
at 2021-11-04T09:56:58-06:00
Add missing review comment fields on bundling changes.
Somehow this got dropped.
-
47b8d7c0
by Charles Giessen
at 2021-11-04T12:19:14-06:00
test: Fix compiler warnings in MSVC
Warnings/fixes include:
* strcpy/strncpy/wstrcpy
* Dropping static on inline functions
* Use localtime_s
* Name shadowing
* Use wcstombs_s
* 32<->64 bit int conversion
* Silence spurious C4996 generated in vs2015
-
73fa8a57
by Charles Giessen
at 2021-11-04T14:53:29-06:00
Refactor shim function definitions to its own file
The code was written with an assumption that the functions needed to be inlined
so they would appear in the correct places (both the test binaries and shim
library). This can be achieved using static libraries, which is what this
commit does.
-
d5bb7a92
by Mark Young
at 2021-11-04T15:01:10-06:00
Fix GCC 11.2 build
Gcc 11.2 build was failing to build with the GoogleTest version
that was being built. So bump to next version of GoogleTest with
fixes (GoogleTest release-1.11.0), but that also required changing
the build directory to build GoogleTest in so update our Makefiles
to build in the new folder as well.
-
5e849b89
by Charles Giessen
at 2021-11-08T15:15:31-07:00
Add test for unknown physical device functions
Requires adding support in the ICD for unknown instance and physical device
functions separately.
-
0d34e696
by Mike Schuchardt
at 2021-11-09T11:40:47-08:00
build: Update to header 1.2.198
- Update known-good
- Generate source (version bump only)
-
86407894
by Charles Giessen
at 2021-11-09T16:20:03-07:00
Make GetPhysDevToolProps call into drivers
Originally, terminator_GetPhysicalDeviceToolPropertiesEXT simply returned up the
call chain, skipping the driver entirely. Now, it will call into the driver if
the tooling extension is supported. To do that requires querying the list of
extensions the driver supports, because there isn't a list of extensions each
physical device supports stored anywhere.
-
bed7a5c6
by Charles Giessen
at 2021-11-09T16:20:03-07:00
Add test for vkGetPhysicalDeviceToolProperties
Basic tests which just makes sure the trampoline + terminator pass the data
through correctly. Data only originates from TestICD, TestLayer has no
support for the extension.
-
538b2676
by Rémi Bernon
at 2021-11-11T14:15:53-07:00
Link loader-unknown-chain with Vulkan::Headers
-
9e6011fd
by Eric Sullivan
at 2021-11-12T11:01:34-07:00
loader: Update assembly file names to specify their target architecture
-
4f78f14c
by Eric Sullivan
at 2021-11-12T11:01:34-07:00
loader: Add aarch64 unknown ext chain implementation
-
fe8479aa
by Charles Giessen
at 2021-11-17T15:28:51-07:00
Dont call ToolProps on drivers without support
The loader recently added support for calling into drivers in
vkGetPhysicalDeviceToolPropertieesEXT. However, it only used the value of
the function pointer to determine if it was safe. It was found that Mesa
drivers will return a non-null function pointer, even though they do not
support the extension, and so the loader called this function pointer which
would then segfault.
The loader prevents this by first checking if the extension is supported by
the physical device. This necessitates calling
vkEnumerateDeviceExtensionProperties and allocating some memory on each call
but since the number of times this function is called should be low, it is
not an undue performance burden. In the future, the loader should cache the
list of extensions when calling vkEnumeratePhysicalDevices so that checking if
a extension function is supported is fast and easy.
-
0ca96994
by Charles Giessen
at 2021-12-03T13:16:53-07:00
Revert to old behavior in GIPA for old apps
Vulkan header updated 1.2.193 changed the behavior of vkGetInstanceProcAddr for
global entrypoints. They used to always be returned regardless of the value of
the instance paramtere. The spec was amended in this version to only allow
querying global level entrypoints with a NULL instance. However, as to not
break old applications, the new behavior is only applied if the instance passed
in is both valid and minor version is greater than 1.2, which was when this
change in behavior occurred. Only instances with a newer version will get the
new behavior.