Re: [PATCH v3 05/10] x86/ibt: Optimize FineIBT sequence

[Peter Zijlstra]

On Wed, Feb 19, 2025 at 10:01:15AM -0800, Kees Cook wrote:
> On Wed, Feb 19, 2025 at 05:21:12PM +0100, Peter Zijlstra wrote:
> > Scott notes that non-taken branches are faster. Abuse overlapping code
> > that traps instead of explicit UD2 instructions.
> 
> Some kind of commenting is needed in here to explicitly call out the
> embedded EA in the "subl" instruction. There is a tiny hint of it in the
> disassembly dump of fineibt_preamble_start, but it's very subtle for
> someone trying to understand this fresh.

Ah, but you found my clue :-)

How's this?

---
--- a/arch/x86/kernel/alternative.c
+++ b/arch/x86/kernel/alternative.c
@@ -1080,6 +1080,9 @@ early_param("cfi", cfi_parse_cmdline);
  *  4:   41 81 <ea> 78 56 34 12  sub    $0x12345678, %r10d
  *  b:   75 f9                   jne    6 <fineibt_preamble_start+0x6>
  *  d:   0f 1f 00                nopl   (%rax)
+ *
+ * Note that the JNE target is the 0xEA byte inside the SUB, this decodes as
+ * (bad) on x86_64 and raises #UD.
  */
 asm(	".pushsection .rodata				\n"
 	"fineibt_preamble_start:			\n"