- User Since
- Nov 22 2019, 10:17 AM (174 w, 6 d)
Zero out the original jump table entries
The plan is to explicitly zero out the original jump table entries to trigger a crash if we end up using them.
I'm testing it internally, will get back to it tomorrow
Mon, Mar 27
LGTM with small nits
Wed, Mar 22
Wed, Mar 15
Fix multimap erase
Tue, Mar 14
Revert deleteTempFiles change
Thank you for working on it! Would you want to add a line to bolt/CODE_OWNERS.TXT?
I’m fully supportive of this change. BOLT uses this interface which helps maintaibilty as it clearly states the intent and is robust wrt container type change. It’s a clear win to also have the same runtime characteristics as find.
Mon, Mar 13
Rename isSupportedRelocation to shouldRecordCodeRelocation, add comment.
Yes, one accept is enough. Please provide a commit author and email and I'll push it for you.
Sun, Mar 12
Sat, Mar 11
Fri, Mar 10
BOLT changes LGTM
Please also take care of BOLT call sites.
Tue, Mar 7
Mon, Mar 6
Fri, Mar 3
Thanks for a review!
What's the reason behind Counts/JumpTableCounts being a MapVector<StringRef, uint64_t in ReorderData::assignMemData but NamesToBranchesMapTy being an std::map<StringRef, FuncBranchData>?
Thu, Mar 2
Feb 27 2023
Is that primarily needed for "X out of Y samples in the binary (Z%) belong to functions with invalid (possibly stale) profile" log message? Can you add a test using yaml profile, e.g. based on bolt/test/X86/pre-aggregated-perf.test?
Rebase, reduce changes
Feb 24 2023
Feb 23 2023
Nevermind, bolt/test/X86/lit.local.cfg actually has the required check.
Did you build without LLVM_ENABLE_TARGETS=X86? The test passes on aarch64 buildbot running here: https://lab.llvm.org/buildbot/#/builders/221/builds/10483
Let's keep these tests in common folder and add a proper requires statement.
Feb 22 2023
Simplify debug message
Nice, thank you for reducing it!
Feb 17 2023
Feb 16 2023
Reject stripped binaries as a policy (D142686).
Feb 13 2023
@ptr1337 what name and email should I use in a commit?
Feb 12 2023
Feb 11 2023
Thanks, looks good to me. Will need to backport to release/16.x branch