[codegen 1.5] Fix a pipeline-register insertion bug for I/O & state operands #3634
+0
−0
Conversation
This file contains hidden or 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
copybara-service
bot
force-pushed
the
test_849268379
branch
2 times, most recently
from
084b4a9 to
bc45266
Compare
copybara-service
bot
force-pushed
the
test_849268379
branch
from
bc45266 to
7036f35
Compare
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Outdated suggestions cannot be applied.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.
[codegen 1.5] Fix a pipeline-register insertion bug for I/O & state operands
After I/O lowering, we previously ended up moving the use of the operands into logic outside of the pipeline stages. Since pipeline register insertion happened after that point, it would think the uses did not need pipeline registers, accidentally linking (e.g.) Sends directly to their inputs with no intervening registers.
To fix this, we temporarily route these uses through identity nodes in the stage where the operation was scheduled, and rely on our dataflow & identity cleanup passes to remove them later.
The same principle applies to state-write operations too.
This lets us re-enable the
BasicDatapathResetAndInputFloptest.