libnvdimm: fix badblock range handling of ARS range
authorToshi Kani <toshi.kani@hpe.com>
Fri, 7 Jul 2017 23:44:26 +0000 (17:44 -0600)
committerDan Williams <dan.j.williams@intel.com>
Mon, 17 Jul 2017 18:43:58 +0000 (11:43 -0700)
__add_badblock_range() does not account sector alignment when
it sets 'num_sectors'.  Therefore, an ARS error record range
spanning across two sectors is set to a single sector length,
which leaves the 2nd sector unprotected.

Change __add_badblock_range() to set 'num_sectors' properly.

Cc: <stable@vger.kernel.org>
Fixes: 0caeef63e6d2 ("libnvdimm: Add a poison list and export badblocks")
Signed-off-by: Toshi Kani <toshi.kani@hpe.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
drivers/nvdimm/core.c

index 7cd99b1f8596b9856bf6c02bdddafcdd7910c21d..75bc08c6838ccebe1c01e3cec8e9fe0db0fba33d 100644 (file)
@@ -421,14 +421,15 @@ static void set_badblock(struct badblocks *bb, sector_t s, int num)
 static void __add_badblock_range(struct badblocks *bb, u64 ns_offset, u64 len)
 {
        const unsigned int sector_size = 512;
-       sector_t start_sector;
+       sector_t start_sector, end_sector;
        u64 num_sectors;
        u32 rem;
 
        start_sector = div_u64(ns_offset, sector_size);
-       num_sectors = div_u64_rem(len, sector_size, &rem);
+       end_sector = div_u64_rem(ns_offset + len, sector_size, &rem);
        if (rem)
-               num_sectors++;
+               end_sector++;
+       num_sectors = end_sector - start_sector;
 
        if (unlikely(num_sectors > (u64)INT_MAX)) {
                u64 remaining = num_sectors;