-
Notifications
You must be signed in to change notification settings - Fork 689
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add tests for literal overloads of fmod, length, and normalize #6437
Merged
tex3d
merged 4 commits into
microsoft:release-1.8.2403
from
tex3d:rem-assert-on-bad-dxil-overload
Mar 21, 2024
Merged
Add tests for literal overloads of fmod, length, and normalize #6437
tex3d
merged 4 commits into
microsoft:release-1.8.2403
from
tex3d:rem-assert-on-bad-dxil-overload
Mar 21, 2024
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Certain HL ops with no double overload will generate double overloads for literal types. These are lowered to double DXIL overloads, some of which are not legal for these ops. We currently rely on illegal intermediate DXIL op overloads for constant evaluation. Disabling this assert allows test cases to be written to verify the current behavior for these operations.
tex3d
changed the title
Remove assert on illegal DXIL overload
Remove assert on illegal DXIL overload - add tests
Mar 20, 2024
pow2clk
reviewed
Mar 20, 2024
tools/clang/test/HLSLFileCheck/hlsl/intrinsics/basic/fmod_const_eval.hlsl
Outdated
Show resolved
Hide resolved
tools/clang/test/HLSLFileCheck/hlsl/intrinsics/basic/fmod_const_eval.hlsl
Outdated
Show resolved
Hide resolved
- Last minute reordering of NO_FOLD to the end caused unary sqrt op function to be listed first, and the first illegal overload is the only one reported. - Changed NO_FOLD test to directly use full normalize output, instead of just the first component.
What matters is 'error: DXIL intrinsic overload must be valid.' It's unnecessary to check the specific intrinsic reported, partly because it will only report the first one in function declaration order, which could change with changes to other code. The purpose of the test is to make sure the validator gracefully caught the invalid DXIL overload.
Also remove unused, incomplete, and potentally wrong SPECIALS section from fmod_const_eval.hlsl.
tex3d
changed the title
Remove assert on illegal DXIL overload - add tests
Add tests for literal overloads of fmod, length, and normalize
Mar 21, 2024
pow2clk
approved these changes
Mar 21, 2024
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I suppose it's fine, but it has no effect at present.
python3kgae
approved these changes
Mar 21, 2024
Other than testing this scenario across configurations in our build pipeline, and preventing any further changes in the branch from regressing it, right? |
pow2clk
pushed a commit
to pow2clk/DirectXShaderCompiler
that referenced
this pull request
Jul 17, 2024
…soft#6437) Certain HL ops with no double overload will generate double overloads for literal types. These are lowered to double DXIL overloads, some of which are not legal for these ops. We currently rely on illegal intermediate DXIL op overloads for constant evaluation. If constant evaluation fails, we end up with illegal DXIL overloads in final DXIL, which is caught by the validator. The prior revert restored the ability to run these scenarios (with asserts disabled). This change adds tests for microsoft#6419 so automated testing will test across configurations and prevent regressions of this scenario in this branch. (cherry picked from commit 9c2b828)
pow2clk
added a commit
that referenced
this pull request
Jul 17, 2024
#6799) Certain HL ops with no double overload will generate double overloads for literal types. These are lowered to double DXIL overloads, some of which are not legal for these ops. We currently rely on illegal intermediate DXIL op overloads for constant evaluation. If constant evaluation fails, we end up with illegal DXIL overloads in final DXIL, which is caught by the validator. The prior revert restored the ability to run these scenarios (with asserts disabled). This change adds tests for #6419 so automated testing will test across configurations and prevent regressions of this scenario in this branch. (cherry picked from commit 9c2b828) Co-authored-by: Tex Riddell <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Certain HL ops with no double overload will generate double overloads for literal types. These are lowered to double DXIL overloads, some of which are not legal for these ops. We currently rely on illegal intermediate DXIL op overloads for constant evaluation. If constant evaluation fails, we end up with illegal DXIL overloads in final DXIL, which is caught by the validator.
The prior revert restored the ability to run these scenarios (with asserts disabled).
This change adds tests for #6419 so automated testing will test across configurations and prevent regressions of this scenario in this branch.