54  * With the RAID cache, each stripe works in two phases:
59  *   if STRIPE_R5C_CACHING == 0, the stripe is in writing-out phase
60  *   if STRIPE_R5C_CACHING == 1, the stripe is in caching phase
63  * the stripe is always in writing-out phase.
65  * For write-back journal, the stripe is sent to caching phase on write
70  * writes are committed from the log device. Therefore, a stripe in
160  * Each chunk may contain more than one stripe (for example, a 256kB
173  * It is necessary to remember whether a stripe is counted in
176  * two flags are set, the stripe is counted in big_stripe_tree. This
322 /* Check whether we should flush some stripes to free up stripe cache */
336 	 *   - stripe cache pressure high:  in r5c_check_stripe_cache_usage()
339 	 *   - stripe cache pressure moderate:  in r5c_check_stripe_cache_usage()
360 	 * or a full stripe (chunk size / 4k stripes).  in r5c_check_cached_full_stripe()
383  * In cache flush, the stripe goes through 1 and then 2. For a stripe that
387  * most (conf->group_cnt + 1) stripe that passed 1. So total journal space
445  * Put the stripe into writing-out phase by clearing STRIPE_R5C_CACHING.
884 	/* current implementation is one stripe per flush payload */  in r5l_append_flush_payload()
977 /* add stripe to no_space_stripes, and then wake up reclaim */
1002 	/* Don't support stripe batch */  in r5l_write_stripe()
1005 		/* the stripe is written to log, we start writing it to raid */  in r5l_write_stripe()
1033 	 * The stripe must enter state machine again to finish the write, so  in r5l_write_stripe()
1102 		 * we flush log disk cache first, then write stripe data to  in r5l_handle_flush_request()
1278  * stripe data/parity is in log, we start write stripe to raid disks. stripe
1357  * r5c_flush_stripe moves stripe from cached list to handle_list. When called,
1358  * the stripe must be on r5c_cached_full_stripes or r5c_cached_partial_stripes.
1369 	 * The stripe is not ON_RELEASE_LIST, so it is safe to call  in r5c_flush_stripe()
1442 		 * if stripe cache pressure high, flush all full stripes and  in r5c_do_reclaim()
1450 		 * if stripe cache pressure moderate, or if there is many full  in r5c_do_reclaim()
1472 			 * case, we only want to flush stripe on  in r5c_do_reclaim()
1474 			 * condition makes sure the stripe is on one of the  in r5c_do_reclaim()
1929 		return NULL;  /* no more stripe available */  in r5c_recovery_alloc_stripe()
1992  * before loading data to stripe cache, we need verify checksum for all data,
2139 			 * cannot get stripe from raid5_get_active_stripe  in r5c_recovery_analyze_meta_block()
2150 				pr_debug("md/raid:%s: Increasing stripe cache size to %d to recovery data on journal.\n",  in r5c_recovery_analyze_meta_block()
2200  * Load the stripe into cache. The stripe will be written out later by
2201  * the stripe cache state machine.
2221  * For stripes with data and parity, namely Data-Parity stripe
2224  * For stripes with only data, namely Data-Only stripe
2225  * (STRIPE_R5C_CACHING == 1), we load them to stripe cache state machine.
2227  * For a stripe, if we see data after parity, we should discard all previous
2228  * data and parity for this stripe, as these data are already flushed to
2232  * first data-only stripe on the journal device, or next invalid meta block.
2265 	/* load data-only stripes to stripe cache */  in r5c_recovery_flush_log()
2651 		 *  1. The stripe has some data cached, and it is sent to  in r5c_try_caching_write()
2653 		 *  2. The stripe is clean, and this is the first write  in r5c_try_caching_write()
2674 	 * When a stripe is syncing, the write is also handled in write  in r5c_try_caching_write()
2692 	/* if the stripe is not counted in big_stripe_tree, add it now */  in r5c_try_caching_write()
2723 		 * set STRIPE_R5C_PARTIAL_STRIPE, this shows the stripe is  in r5c_try_caching_write()
2797  * clean up the stripe (clear R5_InJournal for dev[pd_idx] etc.) after the
2798  * stripe is committed to RAID disks.
2847 	/* stop counting this stripe in big_stripe_tree */  in r5c_finish_stripe_write_out()
2880 	/* stripe is flused to raid disks, we can do resync now */  in r5c_finish_stripe_write_out()
2909 	 * The stripe must enter state machine again to call endio, so  in r5c_cache_data()
2940 /* check whether this big stripe is in write back cache. */