x86: Fix earlyprintk=dbgp for machines without NX

Since parse_early_param() may (e.g. for earlyprintk=dbgp)
involve calls to page table manipulation functions (here
set_fixmap_nocache()), NX hardware support must be determined
before calling that function (so that __supported_pte_mask gets
properly set up).

But the call after parse_early_param() can also not go away, as
that will honor eventual command line specified disabling of
the NX functionality.

( This will then just result in whatever mappings got
  established during parse_early_param() having the NX bit set
  despite it being disabled on the command line, but I think
  that's tolerable).

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Cc: Yinghai Lu <yhlu.kernel@gmail.com>
LKML-Reference: <4A97F3BD02000078000121B9@vpn.id2.novell.com>
[ merged to x86/pat to resolve a conflict. ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>

diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 02643cc3bf26d94e6ec90914eefe0ffd6e77fdc9..eb1f1e6e52b00f988025f4e4d3a20fb8011e02b1 100644 (file)
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -714,6 +714,16 @@ void __init setup_arch(char **cmdline_p)
        strlcpy(command_line, boot_command_line, COMMAND_LINE_SIZE);
        *cmdline_p = command_line;
 
+#ifdef CONFIG_X86_64
+       /*
+        * Must call this twice: Once just to detect whether hardware doesn't
+        * support NX (so that the early EHCI debug console setup can safely
+        * call set_fixmap(), and then again after parsing early parameters to
+        * honor the respective command line option.
+        */
+       check_efer();
+#endif
+
        parse_early_param();
 
        /* VMI may relocate the fixmap; do this before touching ioremap area */