An attempt to FH4 - #7
Draft
yasster wants to merge 2 commits into
Draft
Conversation
The FH4 UnwindMap entry offsets were incorrectly calculated using a hardcoded approximation of \5 * (I + 1)\ bytes per entry. However, compressed integers in FH4 can be 1-5 bytes depending on value magnitude, causing incorrect offsets when values require different encoding sizes. This caused crashes (STATUS_STACK_BUFFER_OVERRUN 0xC0000409) with deeply nested try-catch blocks where the accumulated offset errors led to the runtime reading invalid unwind data. Changes: - Add getCompressedIntSize() helper to calculate actual byte sizes - Pre-calculate entry sizes and accumulate actual byte offsets before emitting UnwindMap entries - Fix emitCompressedInt() to properly mask values for multi-byte encodings - Defer IP2State table emission to MC layer (MCWin64EH) where symbol offsets are known after layout - Add FH4IP2StateEntry struct and related fields to WinEH::FrameInfo - Fix SelectionDAGBuilder to use isMSVCCXXPersonality() for both MSVC_CXX (FH3) and MSVC_CXX4 (FH4) personalities Fixes exception handling crashes with 6+ levels of nested try-catch blocks when using __CxxFrameHandler4.
dpaoliello
suggested changes
Jan 12, 2026
| // For FH4, the parent function and catch funclets need a reference to | ||
| // the FuncInfo4 structure ($cppxdata$) following the unwind info. | ||
| StringRef FuncLinkageName = GlobalValue::dropLLVMManglingEscape(F.getName()); | ||
| MCSymbol *FuncInfoXData = Asm->OutContext.getOrCreateSymbol( |
There was a problem hiding this comment.
This looks identical to the FH3 info above, what am I missing?
| } | ||
|
|
||
| static void emitCompressedInt(MCStreamer &OS, uint32_t Value) { | ||
| if (Value < 128) { |
There was a problem hiding this comment.
I think I'd prefer to switch on getCompressedIntSize instead of duplicating the magic value comparisons here?
| if (I == 0) { | ||
| // First entry: offset to "before" the unwind map (state -1) | ||
| // This is the size of the NumEntries field | ||
| NextOffset = getCompressedIntSize(FuncInfo.CxxUnwindMap.size()); |
There was a problem hiding this comment.
Seems odd to be re-calculating this every time, can we store it in AccumulatedOffset instead (and, maybe, collapse this if statement)?
| .concat(Twine(I)) | ||
| .concat("$") | ||
| .concat(FuncLinkageName)); | ||
| HandlerMaps.push_back(HandlerMapXData); |
There was a problem hiding this comment.
Should this be in the if statement? Or is it ok to push null pointers?
| emitCompressedInt(OS, TBME.TryLow); | ||
| emitCompressedInt(OS, TBME.TryHigh); | ||
| emitCompressedInt(OS, TBME.CatchHigh); | ||
| OS.emitValue(create32bitRef(HandlerMapXData), 4); |
There was a problem hiding this comment.
I'm assuming that create32bitRef knows how to handle nulls?
| /// - Values 128-16383: 2 bytes, (value << 2) | 1 (low 2 bits 01) | ||
| /// - Values 16384+: 4 bytes, (value << 3) | 3 (low 3 bits 011) | ||
| /// This matches Microsoft's encoding in CxxFrameHandler4. | ||
| static void emitFH4NetEncodedInt(MCStreamer &Streamer, uint32_t Value) { |
There was a problem hiding this comment.
Is there a common place we can put this?
| if (MaybeOffset && *MaybeOffset >= 0) { | ||
| Delta = static_cast<uint32_t>(*MaybeOffset); | ||
| } else { | ||
| Delta = 1; // Fallback for cross-fragment labels |
There was a problem hiding this comment.
Do we emit a fixup to correct this after relaxation?
|
|
||
| ; CHECK-LABEL: $handlerMap$0$multiple_catch: | ||
| ; Should have 2 handlers | ||
| ; CHECK: .byte |
There was a problem hiding this comment.
Is there a value you should be checking here?
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
No description provided.