Fix - program loading with effective slot at epoch boundary #35283
Conversation
Lichtso
force-pushed
the
fix/program_load_effective_slot_at_epoch_boundary
branch
4 times, most recently
from
8e4e6e8
to
3d4dcb7
Compare
|
Backports to the beta branch are to be avoided unless absolutely necessary for fixing bugs, security issues, and perf regressions. Changes intended for backport should be structured such that a minimum effective diff can be committed separately from any refactoring, plumbing, cleanup, etc that are not strictly necessary to achieve the goal. Any of the latter should go only into master and ride the normal stabilization schedule. Exceptions include CI/metrics changes, CLI improvements and documentation updates on a case by case basis. |
|
much more readable! 😍 thoughts on taking this to 1.17? we don't want our hands tied should we need to emergency patch something that affects jit output |
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #35283 +/- ##
=======================================
Coverage 81.6% 81.6%
=======================================
Files 834 834
Lines 224798 224835 +37
=======================================
+ Hits 183485 183550 +65
+ Misses 41313 41285 -28 |
|
Since we are planning to backport this to v1.17, is there a smaller patch that fixes the problem? The cleanup can be done in a follow on PR, that we don't backport. Also, the change that's going into to v1.17 should be audited before making a release out of it. |
Saying what you do is great, but please always say why (and when a change is complex, say how). I don't think the test change belongs to this PR? Does the (unmodified) test pass with the effective slot change? Changing existing tests when fixing something is not ideal. Why can't we add a new regression test for the specific issue we found, which clearly fails without the fix? Then we can refactor other tests separately. |
|
(The actual code change looks fine to me, but I'd like to see a regression test :P) |
Lichtso
force-pushed
the
fix/program_load_effective_slot_at_epoch_boundary
branch
from
b15170f
to
78f7d6a
Compare
|
Backports to the stable branch are to be avoided unless absolutely necessary for fixing bugs, security issues, and perf regressions. Changes intended for backport should be structured such that a minimum effective diff can be committed separately from any refactoring, plumbing, cleanup, etc that are not strictly necessary to achieve the goal. Any of the latter should go only into master and ride the normal stabilization schedule. |
Lichtso
deleted the
fix/program_load_effective_slot_at_epoch_boundary
branch
* Always limit effective slot to the begin of the current epoch. * Adds comments. * Optimizes to avoid having two entries if there is no relevant feature activation. * Adds test_feature_activation_loaded_programs_epoch_transition(). (cherry picked from commit 2891ce8) # Conflicts: # svm/src/transaction_processor.rs
* Always limit effective slot to the begin of the current epoch. * Adds comments. * Optimizes to avoid having two entries if there is no relevant feature activation. * Adds test_feature_activation_loaded_programs_epoch_transition(). (cherry picked from commit 2891ce8) # Conflicts: # svm/src/transaction_processor.rs
* Always limit effective slot to the begin of the current epoch. * Adds comments. * Optimizes to avoid having two entries if there is no relevant feature activation. * Adds test_feature_activation_loaded_programs_epoch_transition(). (cherry picked from commit 2891ce8)
* Always limit effective slot to the begin of the current epoch. * Adds comments. * Optimizes to avoid having two entries if there is no relevant feature activation. * Adds test_feature_activation_loaded_programs_epoch_transition(). (cherry picked from commit 2891ce8)
* Always limit effective slot to the begin of the current epoch. * Adds comments. * Optimizes to avoid having two entries if there is no relevant feature activation. * Adds test_feature_activation_loaded_programs_epoch_transition(). (cherry picked from commit 2891ce8)
* Always limit effective slot to the begin of the current epoch. * Adds comments. * Optimizes to avoid having two entries if there is no relevant feature activation. * Adds test_feature_activation_loaded_programs_epoch_transition(). (cherry picked from commit 2891ce8)
…ackport of #35283) (#35302) Fix - program loading with effective slot at epoch boundary (#35283) * Always limit effective slot to the begin of the current epoch. * Adds comments. * Optimizes to avoid having two entries if there is no relevant feature activation. * Adds test_feature_activation_loaded_programs_epoch_transition(). (cherry picked from commit 2891ce8) Co-authored-by: Alexander Meißner <[email protected]>
…abs#35283) * Always limit effective slot to the begin of the current epoch. * Adds comments. * Optimizes to avoid having two entries if there is no relevant feature activation. * Adds test_feature_activation_loaded_programs_epoch_transition().
Problem
The testnet outage 2024/02/21 was caused by a feature activation which affected RBPF and thus changed the program runtime environment. A bug it the program loading caused programs of the new environment after the epoch boundary to be discarded as they had the same effective slot as programs of the old environment before the epoch boundary.
Summary of Changes
test_feature_activation_loaded_programs_epoch_transition().