optimizer: do not delete statements that may not :terminate
#52999
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
aviatesk
force-pushed
the
avi/optimization-flags-refactor
branch
from
2345411
to
65c7d1d
Compare
aviatesk
force-pushed
the
avi/52991
branch
2 times, most recently
from
10dc95b
to
0b1bda6
Compare
oscardssmith
added
bugfix
|
Not a bugfix. This is an implementation of the #40009 decision. |
|
Added the |
|
Should add the test cases from #40009 |
|
From triage: Nobody supports keeping this UB. Go for it! |
aviatesk
added a commit
that referenced
this pull request
Jun 5, 2024
#54323 ensures that all frames within a cycle have the same, cycle valid effects. However, `src.ssaflags` is calculated using partial effects, so when the effects of a `frame` within the cycle are updated, there would be an inconsistency between `frame.ipo_effects` and `frame.src.ssaflags`. Due to this inconsistency, #54323 breaks the test cases from #51092, when backported to v1.11. On the surface this is because #52999 hasn't been backported to v1.11, but the fundamental issue lies in this inconsistency between cycle effects and `ssaflags`. This commit uses a somewhat hacky approach to resolve this. It identifies statements involved in the cycle comparing `stmt_edges` to `callers_in_cycle`, and updates `ssaflags` according to new cycle valid effects if necessary. This resolves the issue, but ideally, it should be implemented more safely with the new `edges` format that will be implemented in the future. For now, this approach should be okay.
aviatesk
added a commit
that referenced
this pull request
Jun 6, 2024
#54323 ensures that all frames within a cycle have the same, cycle valid effects. However, `src.ssaflags` is calculated using partial effects, so when the effects of a `frame` within the cycle are updated, there would be an inconsistency between `frame.ipo_effects` and `frame.src.ssaflags`. Due to this inconsistency, #54323 breaks the test cases from #51092, when backported to v1.11. On the surface this is because #52999 hasn't been backported to v1.11, but the fundamental issue lies in this inconsistency between cycle effects and `ssaflags`. This commit uses a somewhat hacky approach to resolve this. It identifies statements involved in the cycle comparing `stmt_edges` to `callers_in_cycle`, and updates `ssaflags` according to new cycle valid effects if necessary. This resolves the issue, but ideally, it should be implemented more safely with the new `edges` format that will be implemented in the future. For now, this approach should be okay.
aviatesk
added a commit
that referenced
this pull request
Jun 6, 2024
#54323 ensures that all frames within a cycle have the same, cycle valid effects. However, `src.ssaflags` is calculated using partial effects, so when the effects of a `frame` within the cycle are updated, there would be an inconsistency between `frame.ipo_effects` and `frame.src.ssaflags`. Due to this inconsistency, #54323 breaks the test cases from #51092, when backported to v1.11. On the surface this is because #52999 hasn't been backported to v1.11, but the fundamental issue lies in this inconsistency between cycle effects and `ssaflags`. This commit uses a somewhat hacky approach to resolve this. It identifies statements involved in the cycle comparing `stmt_edges` to `callers_in_cycle`, and updates `ssaflags` according to new cycle valid effects if necessary. This resolves the issue, but ideally, it should be implemented more safely with the new `edges` format that will be implemented in the future. For now, this approach should be okay.
aviatesk
added a commit
that referenced
this pull request
Jun 6, 2024
#54323 ensures that all frames within a cycle have the same, cycle valid effects. However, `src.ssaflags` is calculated using partial effects, so when the effects of a `frame` within the cycle are updated, there would be an inconsistency between `frame.ipo_effects` and `frame.src.ssaflags`. Due to this inconsistency, #54323 breaks the test cases from #51092, when backported to v1.11. On the surface this is because #52999 hasn't been backported to v1.11, but the fundamental issue lies in this inconsistency between cycle effects and `ssaflags`. This commit uses a somewhat hacky approach to resolve this. It identifies statements involved in the cycle comparing `stmt_edges` to `callers_in_cycle`, and updates `ssaflags` according to new cycle valid effects if necessary. This resolves the issue, but ideally, it should be implemented more safely with the new `edges` format that will be implemented in the future. For now, this approach should be okay.
aviatesk
added a commit
that referenced
this pull request
Jun 6, 2024
#54323 ensures that all frames within a cycle have the same, cycle valid effects. However, `src.ssaflags` is calculated using partial effects, so when the effects of a `frame` within the cycle are updated, there would be an inconsistency between `frame.ipo_effects` and `frame.src.ssaflags`. Due to this inconsistency, #54323 breaks the test cases from #51092, when backported to v1.11. On the surface this is because #52999 hasn't been backported to v1.11, but the fundamental issue lies in this inconsistency between cycle effects and `ssaflags`. To resolve this issue, this commit traverses `cycle_backedges` to visit statements involved in the cycle, and updates each `ssaflags` according to new cycle valid effects if necessary.
aviatesk
added a commit
that referenced
this pull request
Jun 7, 2024
…#54689) #54323 ensures that all frames within a cycle have the same, cycle valid effects. However, `src.ssaflags` is calculated using partial effects, so when the effects of a `frame` within the cycle are updated, there would be an inconsistency between `frame.ipo_effects` and `frame.src.ssaflags`. Due to this inconsistency, #54323 breaks the test cases from #51092, when backported to v1.11. On the surface this is because #52999 hasn't been backported to v1.11, but the fundamental issue lies in this inconsistency between cycle effects and `ssaflags`. To resolve this issue, this commit traverses `cycle_backedges` to visit statements involved in the cycle, and updates each `ssaflags` according to new cycle valid effects if necessary.
aviatesk
added a commit
that referenced
this pull request
Jun 7, 2024
…#54689) #54323 ensures that all frames within a cycle have the same, cycle valid effects. However, `src.ssaflags` is calculated using partial effects, so when the effects of a `frame` within the cycle are updated, there would be an inconsistency between `frame.ipo_effects` and `frame.src.ssaflags`. Due to this inconsistency, #54323 breaks the test cases from #51092, when backported to v1.11. On the surface this is because #52999 hasn't been backported to v1.11, but the fundamental issue lies in this inconsistency between cycle effects and `ssaflags`. To resolve this issue, this commit traverses `cycle_backedges` to visit statements involved in the cycle, and updates each `ssaflags` according to new cycle valid effects if necessary.
Closed
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.
Fixes #52991.
Currently this commit marks the test case added in #52954 as
brokensince it has relied on the behavior of #52991.I'm planning to add followup changes in a separate commit.