[llvm] [BOLT][AArch64] Handle OpNegateRAState to enable optimizing binaries with pac-ret hardening (PR #120064)

Tue Sep 9 09:05:55 PDT 2025

================
@@ -0,0 +1,154 @@
+# Optimizing binaries with pac-ret hardening
+
+This is a design document about processing the `DW_CFA_AARCH64_negate_ra_state` DWARF instruction in BOLT. As is describes internal design decisions, the intended audiance is BOLT developers. The document is an updated version of the [RFC posted on the LLVM Discourse](https://discourse.llvm.org/t/rfc-bolt-aarch64-handle-opnegaterastate-to-enable-optimizing-binaries-with-pac-ret-hardening/86594).
+
+
+`DW_CFA_AARCH64_negate_ra_state` is also referred to as  `.cfi_negate_ra_state` in assembly, or `OpNegateRAState` is BOLT sources. In this document, I will use **negate-ra-state** as a shorthand.
+
+## Introduction
+
+### Pointer Authentication
+
+Refer to the [pac-ret section of the BOLT-binary-analysis document](BinaryAnalysis.md#pac-ret-analysis).
+
+### DW_CFA_AARCH64_negate_ra_state
+
+The negate-ra-state CFI is a vendor-specific Call Frame Instruction defined in the [Arm ABI](https://github.com/ARM-software/abi-aa/blob/main/aadwarf64/aadwarf64.rst#id1).
+
+```
+The DW_CFA_AARCH64_negate_ra_state operation negates bit[0] of the RA_SIGN_STATE pseudo-register.
+```
+
+This bit indicates to the unwinder whether the current return address is signed or not (hence the name). The unwinder uses this information to authenticate the pointer, and remove the Pointer Authentication Code (PAC) bits. Incorrect negate-ra-state placement can lead to the unwinder trying to authenticate an unsigned pointer (which segfaults), or skipping authenticating a signed pointer, and trying to access an incorrect location (also leading to a segfault).
+
+(Note: not *all* unwinders do this. Some use the `xpac` instruction to strip the PAC bits without authenticating the pointer. This is incorrect, as it allows control-flow modification in the case of unwinding.)
+
+There are no DWARF instructions to directly set or clear the RA State. However, two other CFIs can also affect the RA state:
+- `DW_CFA_remember_state`: this CFI stores register rules onto an implicit stack.
+- `DW_CFA_restore_state`:  this CFI pops rules from this stack.
+
+Example:
+
+| CFI                            | Effect on RA state             |
+| ------------------------------ | ------------------------------ |
+| (default)                      | 0                              |
+| DW_CFA_AARCH64_negate_ra_state | 0 -> 1                         |
+| DW_CFA_remember_state          | 1 pushed to the stack          |
+| DW_CFA_AARCH64_negate_ra_state | 1 -> 0                         |
+| DW_CFA_restore_state           | 0 -> 1 (popped from the stack) |
+
+The Arm ABI also defines the DW_CFA_AARCH64_negate_ra_state_with_pc CFI, but it is not widely used, and is [likely to become deprecated](https://github.com/ARM-software/abi-aa/issues/327).
+
+### Where are these CFIs needed?
+
+In all locations, where two consecutive instructions have different RA state, this need to be indicated to the unwinder. This happens at pointer signing and authenticating. The other case where two consecutive instructions have different RA state, but neither of them is signing or authenticating means that they are not next to each other in control flow. One is part of an execution path with signed RA, the other is part of a path with an unsigned RA.
+
+In the example below, the first BasicBlock ends in a conditional branch, and jumps to two different BasicBlocks, each with their own authentication, and return. The return at the end of the second BasicBlock is next to the first instruction of the third BasicBlock, but they have different RA statas: the return address has already been authenticated for the return, hence it has Unsigned state. The third BasicBlock is past the pointer signing, but not the authentication. In this case, a negate-ra-state is needed after the return.
+
+```
+        +----------------+
+        |     paciasp    |
+        |                |
+        |      b.cc      |
+        +--------+-------+
+                 |
++----------------+
+|                |
+|       +--------v-------+
+|       |                |
+|       |    autiasp     |
+|       |      ret       |
+|       +----------------+
++----------------+
+                 |
+        +--------v-------+
+        |                |
+        |     autiasp    |
+        |      ret       |
+        +----------------+
+```
+
+> [!important]
+> The unwinder does not follow the control flow graph. It reads unwind information in the layout order.
+
+Because these locations are dependent on how the function layout looks, negate-ra-state CFIs will become invalid during BasicBlock reordering.
+
+## Solution design
+
+The patch introduces two new passes:
+1. `MarkRAStatesPass`: assigns the RA state to each instruction based on the CFIs in the input binary
+2. `InsertNegateRAStatePass`: reads those assigned instruction RA states after optimizations, and emits `DW_CFA_AARCH64_negate_ra_state` CFIs at the correct places: wherever there is a state change between two consecutive instructions in the layout order.
+
+To track metadata on individual instructions, the `MCAnnotation` class was extended. These also have helper function in `MCPlusBuilder`.
+
+### Saving annotations at CFI reading
+
+CFIs are read and added to BinaryFunctions in `CFIReaderWriter::FillCFIInfoFor`. At this point, we add MCAnnotations about negate-ra-state, remember-state and restore-state CFIs to the instructions they refer to. This is to not interfere with the CFI processing that already happens in BOLT (e.g. remember-state and restore-state CFIs are removed in `normalizeCFIState` for reasons unrelated to PAC).
+
+As we add the MCAnnotations *to instructions*, we have to account for the case where the function starts with a CFI altering the RA state. If a function starts with a negate-ra-state CFI for example, we cannot save the annotation on the first instruction, because that itself should already be signed. This is why all BinaryFunctions have an `initialRAState` bool. If the `Offset` the CFI refers to is zero, we don't store an annotation, but set the `initialRAState` in `FillCFIInfoFor`. This info is then used in `MarkRAStates`.
+
+### Binaries without DWARF info
+
+In some cases, the DWARF tables are stripped from the binary. These programs usually have some other unwind-mechanism. To account for code that uses Pointer Authentication, but does not have DWARF CFIs, the passes only run on functions that had at least one negate-ra-state CFI. This is marked during CFI reading.
+
+This also makes sure that the passes don't run on functions that do not store the return address to the stack, and don't need Pointer Authentication, saving on runtime overhead.
+
+In summary:
+- pointer auth is not used: no change, the new passes do not run.
+- pointer auth is used, but DWARF info is stripped: no change, the new passes do not run.
+- pointer auth is used, and we have DWARF CFIs: passes run, and rewrite the negate-ra-state CFI.
+
+### MarkRAStates Pass
+
+This pass runs before optimizations reorder anything.
+
+It processes MCAnnotations generated during the CFI reading stage to check if instructions have either of the three CFIs that can modify RA state:
+- negate-ra-state
+- remember-state
+- restore-state
+
+Then it adds new MCAnnotations to each instruction, indicating their RA state. Those annotations are:
+- Signed
+- Unsigned
+
+Below is a simple example, that shows the two different type of annotations: what we have before the pass, and after it.
+
+| Instruction                 | Before          |  After   |
----------------
paschalis-mpeis wrote:

nitpick: consider adding a few spaces so it aligns on the raw text view, too.

https://github.com/llvm/llvm-project/pull/120064
