[CGData][MachineOutliner] Global Outlining #90074
Draft
+3,816
−223
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
kyulee-com
force-pushed
the
globaloutline
branch
from
30a4de9
to
286b142
Compare
kyulee-com
force-pushed
the
globaloutline
branch
2 times, most recently
from
35b9da2
to
80df5a2
Compare
This defines the OutlinedHashTree class. It contains sequences of stable hash values of instructions that have been outlined. This OutlinedHashTree can be used to track the outlined instruction sequences across modules. A trie structure is used in its implementation, allowing for a compact sharing of common prefixes.
The llvm-cgdata tool has been introduced to handle reading and writing of codegen data. This data includes an optimistic codegen summary that can be utilized to enhance subsequent codegen. Currently, the tool supports saving and restoring the outlined hash tree, facilitating machine function outlining across modules. Additional codegen summaries can be incorporated into separate sections as required. This patch primarily establishes basic support for the reader and writer, similar to llvm-profdata. The high-level operations of llvm-cgdata are as follows: 1. It reads local raw codegen data from a custom section (for example, __llvm_outline) embedded in native binary files 2. It merges local raw codegen data into an indexed codegen data, complete with a suitable header. 3. It handles reading and writing of the indexed codegen data into a standalone file.
This commit introduces support for outlining functions across modules using codegen data generated from previous codegen. The codegen data currently manages the outlined hash tree, which records outlining instances that occurred locally in the past.
The machine outliner now operates in one of three modes:
1. CGDataMode::None: This is the default outliner mode that uses the suffix tree to identify (local) outlining candidates within a module. This mode is also used by (full)LTO to maintain optimal behavior with the combined module.
2. CGDataMode::Write (`codegen-data-generate`): This mode is identical to the default mode, but it also publishes the stable hash sequences of instructions in the outlined functions into a local outlined hash tree. It then encodes this into the `__llvm_outline` section, which will be dead-stripped at link time.
3. CGDataMode::Read (`codegen-data-use-path={.cgdata}`): This mode reads a codegen data file (.cgdata) and initializes a global outlined hash tree. This tree is used to generate global outlining candidates. Note that the codegen data file has been post-processed with the raw `__llvm_outline` sections from all native objects using the `llvm-cgdata` tool (or a linker, `LLD`, or a new ThinLTO pipeline later).
kyulee-com
changed the title
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.
This commit introduces support for outlining functions across modules using codegen data generated from previous codegen. The codegen data currently manages the outlined hash tree, which records outlining instances that occurred locally in the past.
The machine outliner now operates in one of three modes:
-codegen-data-generate): This mode is identical to the default mode, but it also publishes the stable hash sequences of instructions in the outlined functions into a local outlined hash tree. It then encodes this into the__llvm_outlinesection, which will be dead-stripped at link time.-codegen-data-use-path={.cgdata}): This mode reads a codegen data file (.cgdata) and initializes a global outlined hash tree. This tree is used to generate global outlining candidates. Note that the codegen data file has been post-processed with the raw__llvm_outlinesections from all native objects using thellvm-cgdatatool (or a linker,LLD, or a new ThinLTO pipeline later).This is a patch for https://discourse.llvm.org/t/rfc-enhanced-machine-outliner-part-2-thinlto-nolto/78753.