RFR: 8333393: PhaseCFG::insert_anti_dependences can fail to raise LCAs and to add necessary anti-dependence edges
Daniel Lundén
dlunden at openjdk.org
Fri Dec 20 18:34:48 UTC 2024
When searching for load anti dependences in GCM, it is not always sufficient to just search starting at the direct initial memory input to the load. Specifically, there are cases when we must also search for anti dependences starting at relevant Phi memory nodes in between the load's early block and the initial memory input's block. Here, "in between" refers to blocks in the dominator tree in between the early and initial memory blocks.
#### Example 1
Consider the ideal graph below. The initial memory for 183 loadI is 107 Phi and there is an important anti dependency for node 64 membar_release. To discover this anti dependency, we must rather search from 119 Phi which contains overlapping memory slices with 107 Phi. Looking at the ideal graph block view, we see that both 107 Phi and 119 Phi are in the initial memory block (B7) and thus dominate the early block (B20). If we only search from 107 Phi, we fail to add the anti dependency to 64 membar_release and do not force the load to schedule before 64 membar_release as we should. In the block view, we see that the load is actually scheduled in B24 _after_ a number of anti-dependent stores, the first of which is in block B20 (corresponding to the anti dependency on 64 membar_release). The result is the failure we see in this issue (we load the wrong value).
#### Example 2
There are also situations when we need to start searching from Phis that are strictly in between the initial memory block and early block. Consider the ideal graph below. The initial memory for 100 loadI is 18 MachProj, but we also need to search from 76 Phi to find that we must raise the LCA to the last block on the path between 76 Phi and 75 Phi: B9 (= the load's early block). If we do not search from 76 Phi, the load is again likely scheduled too late (in B11 in the example) after anti-dependent stores (the first of which corresponds to 58 membar_release in B10). Note that the block B6 for 76 Phi is strictly dominated by the initial memory block B2 and also strictly dominates the early block B9.
### Changeset
- Update `PhaseCFG::insert_anti_dependences` to also search for anti dependences from Phi nodes according to the above issue summary. We only apply the additional search when loads have an explicit control input, as the issue seems to only appear for such loads.
- Clean up an old leftover comment relating to loop Phis.
- Add mulitple new regression tests in `TestGCMLoadPlacement.java`.
For reference, I have also developed an [alternative changeset](https://github.com/openjdk/jdk/compare/master...dlunde:jdk:insert-anti-dependences-8333393+rewrite) that solves the issue in a different way (without using dominator information). However, using dominator information leads to a smaller and less invasive fix.
### Testing
- [GitHub Actions](https://github.com/dlunde/jdk/actions/runs/12433411971)
- `tier1` to `tier4` (and additional Oracle-internal testing) on Windows x64, Linux x64, Linux aarch64, macOS x64, and macOS aarch64.
- Performance testing using DaCapo, SPECjbb, and SPECjvm on Windows x64, Linux x64, Linux aarch64, macOS x64, and macOS aarch64. There are no significant improvements nor regressions.
- C2 compilation time testing using DaCapo. There appears to be no significant regression.
-------------
Commit messages:
- First review version
Changes: https://git.openjdk.org/jdk/pull/22852/files
Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=22852&range=00
Issue: https://bugs.openjdk.org/browse/JDK-8333393
Stats: 263 lines in 2 files changed: 256 ins; 5 del; 2 mod
Patch: https://git.openjdk.org/jdk/pull/22852.diff
Fetch: git fetch https://git.openjdk.org/jdk.git pull/22852/head:pull/22852
PR: https://git.openjdk.org/jdk/pull/22852
More information about the hotspot-compiler-dev
mailing list