x86, smap: A page fault due to SMAP is an oops
authorH. Peter Anvin <hpa@linux.intel.com>
Fri, 21 Sep 2012 19:43:14 +0000 (12:43 -0700)
committerH. Peter Anvin <hpa@linux.intel.com>
Fri, 21 Sep 2012 19:45:27 +0000 (12:45 -0700)
If we get a page fault due to SMAP, trigger an oops rather than
spinning forever.

Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Link: http://lkml.kernel.org/r/1348256595-29119-11-git-send-email-hpa@linux.intel.com
arch/x86/mm/fault.c

index 76dcd9d..f2fb75d 100644 (file)
@@ -995,6 +995,17 @@ static int fault_in_kernel_space(unsigned long address)
        return address >= TASK_SIZE_MAX;
 }
 
+static inline bool smap_violation(int error_code, struct pt_regs *regs)
+{
+       if (error_code & PF_USER)
+               return false;
+
+       if (!user_mode_vm(regs) && (regs->flags & X86_EFLAGS_AC))
+               return false;
+
+       return true;
+}
+
 /*
  * This routine handles page faults.  It determines the address,
  * and the problem, and then passes it off to one of the appropriate
@@ -1088,6 +1099,13 @@ do_page_fault(struct pt_regs *regs, unsigned long error_code)
        if (unlikely(error_code & PF_RSVD))
                pgtable_bad(regs, error_code, address);
 
+       if (static_cpu_has(X86_FEATURE_SMAP)) {
+               if (unlikely(smap_violation(error_code, regs))) {
+                       bad_area_nosemaphore(regs, error_code, address);
+                       return;
+               }
+       }
+
        perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, regs, address);
 
        /*