gnc_tx_bits declared as __u8. Change to unsigned long.
The goal is to align gnc_tx_bits[] to a 32 bit boundary.
The initial declaration I believe was not a good choice.
The use of the gnc_tx_bits makes more sense as a long than as a 8 bit
value.
It is used by:
static inline int test_and_clear_bit(int nr, unsigned long *addr);
unsigned long find_next_zero_bit(unsigned long *addr, unsigned long
size, unsigned long offset);
static inline int test_and_set_bit(int nr, unsigned long *addr);
all of which takes a unsigned long pointer.
Test-parameters: trivial
Signed-off-by: Chris Horn <hornc@cray.com>
Change-Id: I559e2a356182f253716d30f69bf675c485fe1b72
Reviewed-on: https://review.whamcloud.com/29897
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Doug Oucharek <dougso@me.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
kgn_fma_memblock_t *gnc_fma_blk; /* pointer to fma block for our mailbox */
gni_smsg_attr_t gnpr_smsg_attr; /* my short msg. attributes */
spinlock_t gnc_tx_lock; /* protect tx alloc/free */
- __u8 gnc_tx_bits[GNILND_MAX_MSG_ID/8]; /* bit table for tx id */
+ unsigned long gnc_tx_bits[(GNILND_MAX_MSG_ID/8)/sizeof(unsigned long)]; /* bit table for tx id */
int gnc_next_tx; /* next tx to use in tx_ref_table */
kgn_tx_t **gnc_tx_ref_table; /* table of TX descriptors for this conn */
int gnc_mbox_id; /* id of mbox in fma_blk */