mirror of
https://github.com/torvalds/linux.git
synced 2024-11-01 17:51:43 +00:00
c4e773764c
The use of a memcpy() during a spinlock operation will cause very long thread context switch delays if the flash chip bandwidth is low and the data to be copied large, because a spinlock will disable preemption. For example: A flash with 6,5 MB/s bandwidth will cause under ubifs, which request sometimes 128 KiB (the flash erase size), a preemption delay of 20 milliseconds. High priority threads will not be served during this time, regardless whether this threads access the flash or not. This behavior breaks real time. The patch changes all the use of spin_lock operations for xxxx->mutex into mutex operations, which is exact what the name says and means. I have checked the code of the drivers and there is no use of atomic pathes like interrupt or timers. The mtdoops facility will also not be used by this drivers. So it is dave to replace the spin_lock against mutex. There is no performance regression since the mutex is normally not acquired. Changelog: 06.03.2010 First release 26.03.2010 Fix mutex[1] issue and tested it for compile failure Signed-off-by: Stefani Seibold <stefani@seibold.net> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com> Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
109 lines
2.8 KiB
C
109 lines
2.8 KiB
C
#ifndef FWH_LOCK_H
|
|
#define FWH_LOCK_H
|
|
|
|
|
|
enum fwh_lock_state {
|
|
FWH_UNLOCKED = 0,
|
|
FWH_DENY_WRITE = 1,
|
|
FWH_IMMUTABLE = 2,
|
|
FWH_DENY_READ = 4,
|
|
};
|
|
|
|
struct fwh_xxlock_thunk {
|
|
enum fwh_lock_state val;
|
|
flstate_t state;
|
|
};
|
|
|
|
|
|
#define FWH_XXLOCK_ONEBLOCK_LOCK ((struct fwh_xxlock_thunk){ FWH_DENY_WRITE, FL_LOCKING})
|
|
#define FWH_XXLOCK_ONEBLOCK_UNLOCK ((struct fwh_xxlock_thunk){ FWH_UNLOCKED, FL_UNLOCKING})
|
|
|
|
/*
|
|
* This locking/unlock is specific to firmware hub parts. Only one
|
|
* is known that supports the Intel command set. Firmware
|
|
* hub parts cannot be interleaved as they are on the LPC bus
|
|
* so this code has not been tested with interleaved chips,
|
|
* and will likely fail in that context.
|
|
*/
|
|
static int fwh_xxlock_oneblock(struct map_info *map, struct flchip *chip,
|
|
unsigned long adr, int len, void *thunk)
|
|
{
|
|
struct cfi_private *cfi = map->fldrv_priv;
|
|
struct fwh_xxlock_thunk *xxlt = (struct fwh_xxlock_thunk *)thunk;
|
|
int ret;
|
|
|
|
/* Refuse the operation if the we cannot look behind the chip */
|
|
if (chip->start < 0x400000) {
|
|
DEBUG( MTD_DEBUG_LEVEL3,
|
|
"MTD %s(): chip->start: %lx wanted >= 0x400000\n",
|
|
__func__, chip->start );
|
|
return -EIO;
|
|
}
|
|
/*
|
|
* lock block registers:
|
|
* - on 64k boundariesand
|
|
* - bit 1 set high
|
|
* - block lock registers are 4MiB lower - overflow subtract (danger)
|
|
*
|
|
* The address manipulation is first done on the logical address
|
|
* which is 0 at the start of the chip, and then the offset of
|
|
* the individual chip is addted to it. Any other order a weird
|
|
* map offset could cause problems.
|
|
*/
|
|
adr = (adr & ~0xffffUL) | 0x2;
|
|
adr += chip->start - 0x400000;
|
|
|
|
/*
|
|
* This is easy because these are writes to registers and not writes
|
|
* to flash memory - that means that we don't have to check status
|
|
* and timeout.
|
|
*/
|
|
mutex_lock(&chip->mutex);
|
|
ret = get_chip(map, chip, adr, FL_LOCKING);
|
|
if (ret) {
|
|
mutex_unlock(&chip->mutex);
|
|
return ret;
|
|
}
|
|
|
|
chip->oldstate = chip->state;
|
|
chip->state = xxlt->state;
|
|
map_write(map, CMD(xxlt->val), adr);
|
|
|
|
/* Done and happy. */
|
|
chip->state = chip->oldstate;
|
|
put_chip(map, chip, adr);
|
|
mutex_unlock(&chip->mutex);
|
|
return 0;
|
|
}
|
|
|
|
|
|
static int fwh_lock_varsize(struct mtd_info *mtd, loff_t ofs, uint64_t len)
|
|
{
|
|
int ret;
|
|
|
|
ret = cfi_varsize_frob(mtd, fwh_xxlock_oneblock, ofs, len,
|
|
(void *)&FWH_XXLOCK_ONEBLOCK_LOCK);
|
|
|
|
return ret;
|
|
}
|
|
|
|
|
|
static int fwh_unlock_varsize(struct mtd_info *mtd, loff_t ofs, uint64_t len)
|
|
{
|
|
int ret;
|
|
|
|
ret = cfi_varsize_frob(mtd, fwh_xxlock_oneblock, ofs, len,
|
|
(void *)&FWH_XXLOCK_ONEBLOCK_UNLOCK);
|
|
|
|
return ret;
|
|
}
|
|
|
|
static void fixup_use_fwh_lock(struct mtd_info *mtd, void *param)
|
|
{
|
|
printk(KERN_NOTICE "using fwh lock/unlock method\n");
|
|
/* Setup for the chips with the fwh lock method */
|
|
mtd->lock = fwh_lock_varsize;
|
|
mtd->unlock = fwh_unlock_varsize;
|
|
}
|
|
#endif /* FWH_LOCK_H */
|