x86/asm/entry/64: Remove a bogus 'ret_from_fork' optimization
authorAndy Lutomirski <luto@amacapital.net>
Thu, 5 Mar 2015 00:09:44 +0000 (01:09 +0100)
committerIngo Molnar <mingo@kernel.org>
Thu, 5 Mar 2015 00:12:23 +0000 (01:12 +0100)
'ret_from_fork' checks TIF_IA32 to determine whether 'pt_regs' and
the related state make sense for 'ret_from_sys_call'.  This is
entirely the wrong check.  TS_COMPAT would make a little more
sense, but there's really no point in keeping this optimization
at all.

This fixes a return to the wrong user CS if we came from int
0x80 in a 64-bit task.

Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: <stable@vger.kernel.org>
Link: http://lkml.kernel.org/r/4710be56d76ef994ddf59087aad98c000fbab9a4.1424989793.git.luto@amacapital.net
[ Backported from tip:x86/asm. ]
Signed-off-by: Ingo Molnar <mingo@kernel.org>
arch/x86/kernel/entry_64.S

index 10074ad9ebf85ed82e552f055baebfa6bb3169f2..1d74d161687c9f2a71f334b5530067356310af18 100644 (file)
@@ -269,11 +269,14 @@ ENTRY(ret_from_fork)
        testl $3, CS-ARGOFFSET(%rsp)            # from kernel_thread?
        jz   1f
 
-       testl $_TIF_IA32, TI_flags(%rcx)        # 32-bit compat task needs IRET
-       jnz  int_ret_from_sys_call
-
-       RESTORE_TOP_OF_STACK %rdi, -ARGOFFSET
-       jmp ret_from_sys_call                   # go to the SYSRET fastpath
+       /*
+        * By the time we get here, we have no idea whether our pt_regs,
+        * ti flags, and ti status came from the 64-bit SYSCALL fast path,
+        * the slow path, or one of the ia32entry paths.
+        * Use int_ret_from_sys_call to return, since it can safely handle
+        * all of the above.
+        */
+       jmp  int_ret_from_sys_call
 
 1:
        subq $REST_SKIP, %rsp   # leave space for volatiles