Re: [PATCH v4: {linux-4.14.y} ] dm cache: submit writethrough writes in parallel to origin and cache

[John Donnelly]

> On Aug 4, 2020, at 7:07 PM, john.p.donnelly@xxxxxxxxxx wrote:
> 
> From: Mike Snitzer <snitzer@xxxxxxxxxx>
> 
> Discontinue issuing writethrough write IO in series to the origin and
> then cache.
> 
> Use bio_clone_fast() to create a new origin clone bio that will be
> mapped to the origin device and then bio_chain() it to the bio that gets
> remapped to the cache device.  The origin clone bio does _not_ have a
> copy of the per_bio_data -- as such check_if_tick_bio_needed() will not
> be called.
> 
> The cache bio (parent bio) will not complete until the origin bio has
> completed -- this fulfills bio_clone_fast()'s requirements as well as
> the requirement to not complete the original IO until the write IO has
> completed to both the origin and cache device.
> 
> Signed-off-by: Mike Snitzer <snitzer@xxxxxxxxxx>
> 
> (cherry picked from commit 2df3bae9a6543e90042291707b8db0cbfbae9ee9)
> 
> Fixes: 4ec34f2196d125ff781170ddc6c3058c08ec5e73 (dm bio record:
> save/restore bi_end_io and bi_integrity )
> 
> 4ec34f21 introduced a mkfs.ext4 hang on a LVM device that has been
> modified with lvconvert --cachemode=writethrough.
> 
> CC:stable@xxxxxxxxxxxxxxx for 4.14.y
> 
> Signed-off-by: John Donnelly <john.p.donnelly@xxxxxxxxxx>
> Reviewed-by: Somasundaram Krishnasamy <somasundaram.krishnasamy@xxxxxxxxxx>
> ---
> drivers/md/dm-cache-target.c | 92 ++++++++++++++++++--------------------------
> 1 file changed, 37 insertions(+), 55 deletions(-)
> 
> diff --git a/drivers/md/dm-cache-target.c b/drivers/md/dm-cache-target.c
> index 69cdb29ef6be..2732d1df05fa 100644
> --- a/drivers/md/dm-cache-target.c
> +++ b/drivers/md/dm-cache-target.c
> @@ -450,6 +450,7 @@ struct cache {
> 	struct work_struct migration_worker;
> 	struct delayed_work waker;
> 	struct dm_bio_prison_v2 *prison;
> +	struct bio_set *bs;
> 
> 	mempool_t *migration_pool;
> 
> @@ -537,11 +538,6 @@ static void wake_deferred_bio_worker(struct cache *cache)
> 	queue_work(cache->wq, &cache->deferred_bio_worker);
> }
> 
> -static void wake_deferred_writethrough_worker(struct cache *cache)
> -{
> -	queue_work(cache->wq, &cache->deferred_writethrough_worker);
> -}
> -
> static void wake_migration_worker(struct cache *cache)
> {
> 	if (passthrough_mode(&cache->features))
> @@ -868,16 +864,23 @@ static void check_if_tick_bio_needed(struct cache *cache, struct bio *bio)
> 	spin_unlock_irqrestore(&cache->lock, flags);
> }
> 
> -static void remap_to_origin_clear_discard(struct cache *cache, struct bio *bio,
> -					  dm_oblock_t oblock)
> +static void __remap_to_origin_clear_discard(struct cache *cache, struct bio *bio,
> +					    dm_oblock_t oblock, bool bio_has_pbd)
> {
> -	// FIXME: this is called way too much.
> -	check_if_tick_bio_needed(cache, bio);
> +	if (bio_has_pbd)
> +		check_if_tick_bio_needed(cache, bio);
> 	remap_to_origin(cache, bio);
> 	if (bio_data_dir(bio) == WRITE)
> 		clear_discard(cache, oblock_to_dblock(cache, oblock));
> }
> 
> +static void remap_to_origin_clear_discard(struct cache *cache, struct bio *bio,
> +					  dm_oblock_t oblock)
> +{
> +	// FIXME: check_if_tick_bio_needed() is called way too much through this interface
> +	__remap_to_origin_clear_discard(cache, bio, oblock, true);
> +}
> +
> static void remap_to_cache_dirty(struct cache *cache, struct bio *bio,
> 				 dm_oblock_t oblock, dm_cblock_t cblock)
> {
> @@ -937,57 +940,26 @@ static void issue_op(struct bio *bio, void *context)
> 	accounted_request(cache, bio);
> }
> 
> -static void defer_writethrough_bio(struct cache *cache, struct bio *bio)
> -{
> -	unsigned long flags;
> -
> -	spin_lock_irqsave(&cache->lock, flags);
> -	bio_list_add(&cache->deferred_writethrough_bios, bio);
> -	spin_unlock_irqrestore(&cache->lock, flags);
> -
> -	wake_deferred_writethrough_worker(cache);
> -}
> -
> -static void writethrough_endio(struct bio *bio)
> -{
> -	struct per_bio_data *pb = get_per_bio_data(bio, PB_DATA_SIZE_WT);
> -
> -	dm_unhook_bio(&pb->hook_info, bio);
> -
> -	if (bio->bi_status) {
> -		bio_endio(bio);
> -		return;
> -	}
> -
> -	dm_bio_restore(&pb->bio_details, bio);
> -	remap_to_cache(pb->cache, bio, pb->cblock);
> -
> -	/*
> -	 * We can't issue this bio directly, since we're in interrupt
> -	 * context.  So it gets put on a bio list for processing by the
> -	 * worker thread.
> -	 */
> -	defer_writethrough_bio(pb->cache, bio);
> -}
> -
> /*
> - * FIXME: send in parallel, huge latency as is.
>  * When running in writethrough mode we need to send writes to clean blocks
> - * to both the cache and origin devices.  In future we'd like to clone the
> - * bio and send them in parallel, but for now we're doing them in
> - * series as this is easier.
> + * to both the cache and origin devices.  Clone the bio and send them in parallel.
>  */
> -static void remap_to_origin_then_cache(struct cache *cache, struct bio *bio,
> -				       dm_oblock_t oblock, dm_cblock_t cblock)
> +static void remap_to_origin_and_cache(struct cache *cache, struct bio *bio,
> +				      dm_oblock_t oblock, dm_cblock_t cblock)
> {
> -	struct per_bio_data *pb = get_per_bio_data(bio, PB_DATA_SIZE_WT);
> +	struct bio *origin_bio = bio_clone_fast(bio, GFP_NOIO, cache->bs);
> +
> +	BUG_ON(!origin_bio);
> 
> -	pb->cache = cache;
> -	pb->cblock = cblock;
> -	dm_hook_bio(&pb->hook_info, bio, writethrough_endio, NULL);
> -	dm_bio_record(&pb->bio_details, bio);
> +	bio_chain(origin_bio, bio);
> +	/*
> +	 * Passing false to __remap_to_origin_clear_discard() skips
> +	 * all code that might use per_bio_data (since clone doesn't have it)
> +	 */
> +	__remap_to_origin_clear_discard(cache, origin_bio, oblock, false);
> +	submit_bio(origin_bio);
> 
> -	remap_to_origin_clear_discard(pb->cache, bio, oblock);
> +	remap_to_cache(cache, bio, cblock);
> }
> 
> /*----------------------------------------------------------------
> @@ -1873,7 +1845,7 @@ static int map_bio(struct cache *cache, struct bio *bio, dm_oblock_t block,
> 		} else {
> 			if (bio_data_dir(bio) == WRITE && writethrough_mode(&cache->features) &&
> 			    !is_dirty(cache, cblock)) {
> -				remap_to_origin_then_cache(cache, bio, block, cblock);
> +				remap_to_origin_and_cache(cache, bio, block, cblock);
> 				accounted_begin(cache, bio);
> 			} else
> 				remap_to_cache_dirty(cache, bio, block, cblock);
> @@ -2132,6 +2104,9 @@ static void destroy(struct cache *cache)
> 		kfree(cache->ctr_args[i]);
> 	kfree(cache->ctr_args);
> 
> +	if (cache->bs)
> +		bioset_free(cache->bs);
> +
> 	kfree(cache);
> }
> 
> @@ -2589,6 +2564,13 @@ static int cache_create(struct cache_args *ca, struct cache **result)
> 	cache->features = ca->features;
> 	ti->per_io_data_size = get_per_bio_data_size(cache);
> 
> +	if (writethrough_mode(&cache->features)) {
> +		/* Create bioset for writethrough bios issued to origin */
> +		cache->bs = bioset_create(BIO_POOL_SIZE, 0, 0);
> +		if (!cache->bs)
> +			goto bad;
> +	}
> +
> 	cache->callbacks.congested_fn = cache_is_congested;
> 	dm_table_add_target_callbacks(ti->table, &cache->callbacks);
> 
> -- 
> 1.8.3.1
> 

Hi Greg,

I was working out my sendmail setup   

Hopefully this is better

JD