uprobes/x86: Only rep+nop can be emulated correctly

__skip_sstep() correctly detects the "nontrivial" nop insns,
but since it doesn't update regs->ip we can not really skip
"0x0f 0x1f | 0x0f 0x19 | 0x87 0xc0", the probed application
is killed by SIGILL'ed handle_swbp().

Remove these additional checks. If we want to implement this
correctly we need to know the full insn length to update ->ip.

rep* + nop is fine even without updating ->ip.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>

1 files changed, 2 insertions, 14 deletions

diff --git a/arch/x86/kernel/uprobes.c b/arch/x86/kernel/uprobes.c
index 9538f00827a9..aafa5557b396 100644
--- a/arch/x86/kernel/uprobes.c
+++ b/arch/x86/kernel/uprobes.c
@@ -651,31 +651,19 @@ void arch_uprobe_abort_xol(struct arch_uprobe *auprobe, struct pt_regs *regs)
 
 /*
  * Skip these instructions as per the currently known x86 ISA.
- * 0x66* { 0x90 | 0x0f 0x1f | 0x0f 0x19 | 0x87 0xc0 }
+ * rep=0x66*; nop=0x90
  */
 static bool __skip_sstep(struct arch_uprobe *auprobe, struct pt_regs *regs)
 {
         int i;
 
         for (i = 0; i < MAX_UINSN_BYTES; i++) {
-                if ((auprobe->insn[i] == 0x66))
+                if (auprobe->insn[i] == 0x66)
                         continue;
 
                 if (auprobe->insn[i] == 0x90)
                         return true;
 
-                if (i == (MAX_UINSN_BYTES - 1))
-                        break;
-
-                if ((auprobe->insn[i] == 0x0f) && (auprobe->insn[i+1] == 0x1f))
-                        return true;
-
-                if ((auprobe->insn[i] == 0x0f) && (auprobe->insn[i+1] == 0x19))
-                        return true;
-
-                if ((auprobe->insn[i] == 0x87) && (auprobe->insn[i+1] == 0xc0))
-                        return true;
-
                 break;
         }
         return false;