Using PGDIR_SHIFT to identify espfix64 addresses on 5-level systems
was wrong, and it resulted in panics due to unhandled double faults.
Use P4D_SHIFT instead, which is correct on 4-level and 5-level
machines.
This fixes a panic when running x86 selftests on 5-level machines.
Fixes: 1d33b219563f ("x86/espfix: Add support for 5-level paging")
Cc: stable@xxxxxxxxxxxxxxx
Cc: "Kirill A. Shutemov" <kirill@xxxxxxxxxxxxx>
Cc: Dave Hansen <dave.hansen@xxxxxxxxx>
Signed-off-by: Andy Lutomirski <luto@xxxxxxxxxx>
---
arch/x86/kernel/traps.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index 74136fd16f49..c4e5b0a7516f 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -360,7 +360,7 @@ dotraplinkage void do_double_fault(struct pt_regs *regs, long error_code)
*
* No need for ist_enter here because we don't use RCU.
*/
- if (((long)regs->sp >> PGDIR_SHIFT) == ESPFIX_PGD_ENTRY &&
+ if (((long)regs->sp >> P4D_SHIFT) == ESPFIX_PGD_ENTRY &&
regs->cs == __KERNEL_CS &&
regs->ip == (unsigned long)native_irq_return_iret)
{
--
2.13.6