Delete dead code in ExceptionTracker and leftover old EH code #114121
Conversation
dotnet-policy-service
bot
added
the
community-contribution
|
The diff is not well matched, it is all just deleted lines. |
filipnavara
changed the title
| T_CONTEXT* pContextRecord, ExceptionTracker *pTracker, bool* pfAborting = NULL); | ||
| UINT_PTR CallCatchHandler(T_CONTEXT* pContextRecord, bool* pfAborting = NULL); | ||
|
|
||
| static bool FindNonvolatileRegisterPointers(Thread* pThread, UINT_PTR uOriginalSP, REGDISPLAY* pRegDisplay, TADDR uResumeFrameFP); | ||
| static void UpdateNonvolatileRegisters(T_CONTEXT* pContextRecord, REGDISPLAY *pRegDisplay, bool fAborting); | ||
|
|
||
| PTR_Frame GetLimitFrame() |
There was a problem hiding this comment.
This still seems to be called from ResetThreadAbortState.
There was a problem hiding this comment.
I'll submit a follow up PR once this is merged. Rough preview: 7891236
| @@ -673,10 +563,6 @@ private: ; | |||
|
|
|||
| void ReleaseResources(); | |||
There was a problem hiding this comment.
It's a bit harder to reason about this since it's not purely unreferenced code.
There was a problem hiding this comment.
Basically anything that references non-static methods in the ExceptionTracker is a dead code. No ExceptionTracker instance gets ever created with the new EH. The ExInfo has replaced it.
| void SetEnclosingClauseInfo(bool fEnclosingClauseIsFunclet, | ||
| DWORD dwEnclosingClauseOffset, | ||
| UINT_PTR uEnclosingClauseCallerSP); | ||
|
|
||
| struct EnclosingClauseInfo |
There was a problem hiding this comment.
There's a rather long usage chain for this going from StackFrameIterator::Filter -> GetCallerOfEnclosingClause.
There was a problem hiding this comment.
The StackFrameIterator::Filter contains dead code too. Anything that executes when pTracker is not null.
|
@janvorli FWIW This is not comprehensive. I can address the other things you pointed out but I would prefer to do it separately from just deleting unreferenced code (to keep it reviewable). |
I am fine with that approach. |
|
/ba-g the test failures occur on all PRs recently |
Ref: #114018 (comment)