x86/xen/64: Fix the reported SS and CS in SYSCALL

When I cleaned up the Xen SYSCALL entries, I inadvertently changed
the reported segment registers. Before my patch, regs->ss was
__USER(32)_DS and regs->cs was __USER(32)_CS. After the patch, they
are FLAT_USER_CS/DS(32).

This had a couple unfortunate effects. It confused the
opportunistic fast return logic. It also significantly increased
the risk of triggering a nasty glibc bug:

https://sourceware.org/bugzilla/show_bug.cgi?id=21269

Update the Xen entry code to change it back.

Reported-by: Brian Gerst <[e-mail]>
Signed-off-by: Andy Lutomirski <[e-mail]>
Cc: Andrew Cooper <[e-mail]>
Cc: Boris Ostrovsky <[e-mail]>
Cc: Borislav Petkov <[e-mail]>
Cc: Juergen Gross <[e-mail]>
Cc: Linus Torvalds <[e-mail]>
Cc: Peter Zijlstra <[e-mail]>
Cc: Thomas Gleixner <[e-mail]>
Cc: [e-mail]
Fixes: 8a9949bc71a7 ("x86/xen/64: Rearrange the SYSCALL entries")
Link: http://lkml.kernel.org/r/[e-mail]
Signed-off-by: Ingo Molnar <[e-mail]>

arch/x86/xen/xen-asm_64.S

···
 ENTRY(xen_syscall_target)
 	popq %rcx
 	popq %r11
+
+	/*
+	 * Neither Xen nor the kernel really knows what the old SS and
+	 * CS were.  The kernel expects __USER_DS and __USER_CS, so
+	 * report those values even though Xen will guess its own values.
+	 */
+	movq $__USER_DS, 4*8(%rsp)
+	movq $__USER_CS, 1*8(%rsp)
+
 	jmp entry_SYSCALL_64_after_hwframe
 ENDPROC(xen_syscall_target)
 
···
 ENTRY(xen_syscall32_target)
 	popq %rcx
 	popq %r11
+
+	/*
+	 * Neither Xen nor the kernel really knows what the old SS and
+	 * CS were.  The kernel expects __USER32_DS and __USER32_CS, so
+	 * report those values even though Xen will guess its own values.
+	 */
+	movq $__USER32_DS, 4*8(%rsp)
+	movq $__USER32_CS, 1*8(%rsp)
+
 	jmp entry_SYSCALL_compat_after_hwframe
 ENDPROC(xen_syscall32_target)