[pull] master from tensorflow:master - #8806
Merged
Merged
Conversation
The tf2xla kernels for Qr, Svd, and XlaSvd passed their input straight to the XLA builder libraries without checking its rank. The graph-level shape functions reject inputs of known rank below 2, but an input of unknown static rank, for example one produced by StackPopV2, reaches the kernels unchecked at compile time. QrExplicit then reads dimension rank - 2 = -1 and dies on a fatal ShapeUtil check, and SvdOp dies on the dim_size range check in TensorShape, aborting the process instead of failing compilation. Validate rank >= 2 in all three kernels before any trailing dimensions are read, mirroring the validation and message style of the neighboring MatrixSolve kernel and the errors already produced by the Cholesky and SelfAdjointEig builder libraries. The regression tests feed a rank-1 value through an unknown-rank placeholder, which reproduces the abort on unfixed builds, and assert that a regular InvalidArgumentError is raised instead. Fixes #110798
…motion PiperOrigin-RevId: 975528222
PiperOrigin-RevId: 975532836
PiperOrigin-RevId: 975555261
I could add it to GemmFusion, but I don't think it is a very useful test. Broadcasts would normally get rejected as a user due to IsOutputWorthFusing where output bytes is increased. However, this one doesn't do that because the broadcast is the ROOT and there's nothing else to fuse it to so it says that it's still worth it. V1 actually only rejects it because it doesn't propagate tiling through the broadcast user, but V2 has no issue with this. PiperOrigin-RevId: 975559504
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
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
5 participants
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.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )