hwmon: (dimmtemp) Fix bitmap handling

Building arm:allmodconfig may fail with the following error.

In function 'fortify_memcpy_chk',
    inlined from 'bitmap_copy' at include/linux/bitmap.h:261:2,
    inlined from 'bitmap_copy_clear_tail' at include/linux/bitmap.h:270:2,
    inlined from 'bitmap_from_u64' at include/linux/bitmap.h:622:2,
    inlined from 'check_populated_dimms' at
	drivers/hwmon/peci/dimmtemp.c:284:2:
include/linux/fortify-string.h:344:25: error:
	call to '__write_overflow_field' declared with attribute warning:
	detected write beyond size of field (1st parameter)

The problematic code is
	bitmap_from_u64(priv->dimm_mask, dimm_mask);

dimm_mask is declared as u64, but the bitmap in priv->dimm_mask is only
24 bit wide. On 32-bit systems, this results in writes over the end of
the bitmap.

Fix the problem by using u32 instead of u64 for dimm_mask. This is
currently sufficient, and a compile time check to ensure that the number
of dimms does not exceed the bit map size is already in place.

Fixes: 73bc1b885d ("hwmon: peci: Add dimmtemp driver")
Cc: Iwona Winiarska <iwona.winiarska@intel.com>
Reviewed-by: Iwona Winiarska <iwona.winiarska@intel.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
This commit is contained in:
Guenter Roeck 2022-05-17 13:08:25 -07:00
parent 45988d9078
commit 9baabde04d

View file

@ -220,7 +220,7 @@ static int check_populated_dimms(struct peci_dimmtemp *priv)
int chan_rank_max = priv->gen_info->chan_rank_max;
int dimm_idx_max = priv->gen_info->dimm_idx_max;
u32 chan_rank_empty = 0;
u64 dimm_mask = 0;
u32 dimm_mask = 0;
int chan_rank, dimm_idx, ret;
u32 pcs;
@ -279,9 +279,9 @@ static int check_populated_dimms(struct peci_dimmtemp *priv)
return -EAGAIN;
}
dev_dbg(priv->dev, "Scanned populated DIMMs: %#llx\n", dimm_mask);
dev_dbg(priv->dev, "Scanned populated DIMMs: %#x\n", dimm_mask);
bitmap_from_u64(priv->dimm_mask, dimm_mask);
bitmap_from_arr32(priv->dimm_mask, &dimm_mask, DIMM_NUMS_MAX);
return 0;
}