x86/mm/mtrr: Fix MTRR lookup to handle an inclusive entry
authorToshi Kani <toshi.kani@hp.com>
Tue, 26 May 2015 08:28:05 +0000 (10:28 +0200)
committerIngo Molnar <mingo@kernel.org>
Wed, 27 May 2015 12:40:56 +0000 (14:40 +0200)
When an MTRR entry is inclusive to a requested range, i.e. the
start and end of the request are not within the MTRR entry range
but the range contains the MTRR entry entirely:

  range_start ... [mtrr_start ... mtrr_end] ... range_end

__mtrr_type_lookup() ignores such a case because both
start_state and end_state are set to zero.

This bug can cause the following issues:

1) reserve_memtype() tracks an effective memory type in case
   a request type is WB (ex. /dev/mem blindly uses WB). Missing
   to track with its effective type causes a subsequent request
   to map the same range with the effective type to fail.

2) pud_set_huge() and pmd_set_huge() check if a requested range
   has any overlap with MTRRs. Missing to detect an overlap may
   cause a performance penalty or undefined behavior.

This patch fixes the bug by adding a new flag, 'inclusive',
to detect the inclusive case.  This case is then handled in
the same way as end_state:1 since the first region is the same.
With this fix, __mtrr_type_lookup() handles the inclusive case
properly.

Signed-off-by: Toshi Kani <toshi.kani@hp.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: Elliott@hp.com
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Luis R. Rodriguez <mcgrof@suse.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dave.hansen@intel.com
Cc: linux-mm <linux-mm@kvack.org>
Cc: pebolle@tiscali.nl
Link: http://lkml.kernel.org/r/1431714237-880-3-git-send-email-toshi.kani@hp.com
Link: http://lkml.kernel.org/r/1432628901-18044-3-git-send-email-bp@alien8.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
arch/x86/kernel/cpu/mtrr/generic.c

index 5b239679cfc9ea04a0aa1595e0e412133a2b89e5..e202d26f64a23476fa1f8e97e348df7364f191ed 100644 (file)
@@ -154,7 +154,7 @@ static u8 __mtrr_type_lookup(u64 start, u64 end, u64 *partial_end, int *repeat)
 
        prev_match = 0xFF;
        for (i = 0; i < num_var_ranges; ++i) {
-               unsigned short start_state, end_state;
+               unsigned short start_state, end_state, inclusive;
 
                if (!(mtrr_state.var_ranges[i].mask_lo & (1 << 11)))
                        continue;
@@ -166,19 +166,27 @@ static u8 __mtrr_type_lookup(u64 start, u64 end, u64 *partial_end, int *repeat)
 
                start_state = ((start & mask) == (base & mask));
                end_state = ((end & mask) == (base & mask));
+               inclusive = ((start < base) && (end > base));
 
-               if (start_state != end_state) {
+               if ((start_state != end_state) || inclusive) {
                        /*
                         * We have start:end spanning across an MTRR.
-                        * We split the region into
-                        * either
-                        * (start:mtrr_end) (mtrr_end:end)
-                        * or
-                        * (start:mtrr_start) (mtrr_start:end)
+                        * We split the region into either
+                        *
+                        * - start_state:1
+                        * (start:mtrr_end)(mtrr_end:end)
+                        * - end_state:1
+                        * (start:mtrr_start)(mtrr_start:end)
+                        * - inclusive:1
+                        * (start:mtrr_start)(mtrr_start:mtrr_end)(mtrr_end:end)
+                        *
                         * depending on kind of overlap.
-                        * Return the type for first region and a pointer to
-                        * the start of second region so that caller will
-                        * lookup again on the second region.
+                        *
+                        * Return the type of the first region and a pointer
+                        * to the start of next region so that caller will be
+                        * advised to lookup again after having adjusted start
+                        * and end.
+                        *
                         * Note: This way we handle multiple overlaps as well.
                         */
                        if (start_state)