[llvm] 96b2c3b - [X86][MC] Report error when the instruction length exceeds 15 bytes (#83708)

[via llvm-commits]

Author: Shengchen Kan
Date: 2024-03-04T16:12:39+08:00
New Revision: 96b2c3bde72ec6b681b03370f9650cfd03e66753

URL: https://github.com/llvm/llvm-project/commit/96b2c3bde72ec6b681b03370f9650cfd03e66753
DIFF: https://github.com/llvm/llvm-project/commit/96b2c3bde72ec6b681b03370f9650cfd03e66753.diff

LOG: [X86][MC] Report error when the instruction length exceeds 15 bytes (#83708)

The instruction-size limit of 15 bytes still applies to APX
instructions.

Note that it is possible for an EVEX-encoded legacy instruction to reach
the 15-byte instruction length limit: 4
bytes of EVEX prefix + 1 byte of opcode + 1 byte of ModRM + 1 byte of
SIB + 4 bytes of displacement + 4 bytes of immediate = 15 bytes in
total, e.g.

```
addq    $184, -96, %rax   # encoding:
[0x62,0xf4,0xfc,0x18,0x81,0x04,0x25,0xa0,0xff,0xff,0xff,0xb8,0x00,0x00,0x00]
```

If we added a segment prefix like fs, the length would be 16.

In such a case, no additional (ASIZE or segment override) prefix can be
used.

To help users find this issue earlier, especially for assembler users,
we change
the internal compiler error to error in this patch.

Diagnostic is aligned with GAS
https://sourceware.org/bugzilla/show_bug.cgi?id=31323

Added: 
    llvm/test/MC/X86/apx/long-instruction-err.s

Modified: 
    llvm/lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp b/llvm/lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp
index fdb11d1a408bb6..1fa676eeb79b0f 100644
--- a/llvm/lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp
+++ b/llvm/lib/Target/X86/MCTargetDesc/X86MCCodeEmitter.cpp
@@ -1924,8 +1924,8 @@ void X86MCCodeEmitter::encodeInstruction(const MCInst &MI,
   if ((TSFlags & X86II::OpMapMask) == X86II::ThreeDNow)
     emitByte(X86II::getBaseOpcodeFor(TSFlags), CB);
 
-  assert(CB.size() - StartByte <= 15 &&
-         "The size of instruction must be no longer than 15.");
+  if (CB.size() - StartByte > 15)
+    Ctx.reportError(MI.getLoc(), "instruction length exceeds the limit of 15");
 #ifndef NDEBUG
   // FIXME: Verify.
   if (/*!Desc.isVariadic() &&*/ CurOp != NumOps) {

diff  --git a/llvm/test/MC/X86/apx/long-instruction-err.s b/llvm/test/MC/X86/apx/long-instruction-err.s
new file mode 100644
index 00000000000000..966c6d893521a2
--- /dev/null
+++ b/llvm/test/MC/X86/apx/long-instruction-err.s
@@ -0,0 +1,25 @@
+# RUN: not llvm-mc -triple x86_64 -show-encoding %s 2>&1 | FileCheck %s
+
+# CHECK: [[#@LINE+2]]:1: error: instruction length exceeds the limit of 15
+# CHECK: addq    $1234, %cs:-96, %rax
+addq    $1234, %cs:-96, %rax
+
+# CHECK: [[#@LINE+2]]:1: error: instruction length exceeds the limit of 15
+# CHECK: subq    $1234, %fs:257(%rbx, %rcx), %rax
+subq    $1234, %fs:257(%rbx, %rcx), %rax
+
+# CHECK: [[#@LINE+2]]:1: error: instruction length exceeds the limit of 15
+# CHECK: orq     $1234, 257(%ebx, %ecx), %rax
+orq     $1234, 257(%ebx, %ecx), %rax
+
+# CHECK: [[#@LINE+2]]:1: error: instruction length exceeds the limit of 15
+# CHECK: xorq    $1234, %gs:257(%ebx), %rax
+xorq    $1234, %gs:257(%ebx), %rax
+
+# CHECK: [[#@LINE+2]]:1: error: instruction length exceeds the limit of 15
+# CHECK: {nf} andq    $1234, %cs:-96
+{nf} andq    $1234, %cs:-96
+
+# CHECK: [[#@LINE+2]]:1: error: instruction length exceeds the limit of 15
+# CHECK: {evex} adcq    $1234, %cs:-96
+{evex} adcq    $1234, %cs:-96