Re: [PATCH v2] zram: support REQ_DISCARD

[Sergey Senozhatsky <sergey.senozhatsky@gmail.com>]

On (02/26/14 14:44), Jerome Marchand wrote:
> On 02/26/2014 02:16 PM, Sergey Senozhatsky wrote:
> > Hello,
> > 
> > On (02/26/14 14:23), Joonsoo Kim wrote:
> >> zram is ram based block device and can be used by backend of filesystem.
> >> When filesystem deletes a file, it normally doesn't do anything on data
> >> block of that file. It just marks on metadata of that file. This behavior
> >> has no problem on disk based block device, but has problems on ram based
> >> block device, since we can't free memory used for data block. To overcome
> >> this disadvantage, there is REQ_DISCARD functionality. If block device
> >> support REQ_DISCARD and filesystem is mounted with discard option,
> >> filesystem sends REQ_DISCARD to block device whenever some data blocks are
> >> discarded. All we have to do is to handle this request.
> >>
> >> This patch implements to flag up QUEUE_FLAG_DISCARD and handle this
> >> REQ_DISCARD request. With it, we can free memory used by zram if it isn't
> >> used.
> >>
> >> v2: handle unaligned case commented by Jerome
> >>
> >> Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
> >>
> >> diff --git a/drivers/block/zram/zram_drv.c b/drivers/block/zram/zram_drv.c
> >> index 5ec61be..5364c1e 100644
> >> --- a/drivers/block/zram/zram_drv.c
> >> +++ b/drivers/block/zram/zram_drv.c
> >> @@ -501,6 +501,36 @@ static int zram_bvec_rw(struct zram *zram, struct bio_vec *bvec, u32 index,
> >>  	return ret;
> >>  }
> >>  
> >> +static void zram_bio_discard(struct zram *zram, struct bio *bio)
> >> +{
> >> +	u32 index = bio->bi_iter.bi_sector >> SECTORS_PER_PAGE_SHIFT;
> >> +	size_t n = bio->bi_iter.bi_size;
> >> +	size_t misalign;
> >> +
> >> +	/*
> >> +	 * On some arch, logical block (4096) aligned request couldn't be
> >> +	 * aligned to PAGE_SIZE, since their PAGE_SIZE aren't 4096.
> >> +	 * Therefore we should handle this misaligned case here.
> >> +	 */
> >> +	misalign = (bio->bi_iter.bi_sector &
> >> +			(SECTORS_PER_PAGE - 1)) << SECTOR_SHIFT;
> >> +	if (misalign) {
> >> +		if (n < misalign)
> >> +			return;
> >> +
> >> +		n -= misalign;
> >> +		index++;
> >> +	}
> >> +
> >> +	while (n >= PAGE_SIZE) {
> >> +		write_lock(&zram->meta->tb_lock);
> >> +		zram_free_page(zram, index);
> >> +		write_unlock(&zram->meta->tb_lock);
> >> +		index++;
> >> +		n -= PAGE_SIZE;
> >> +	}
> >> +}
> >> +
> > 
> > a side note, do we need zram_bio_discard() function? I mean, can we handle
> > discard request in zram_bvec_rw(), where we already know index, etc. (passed
> > from __zram_make_request())?
> > 
> 
> We'd still have to make sure not to discard pages that are still partially
> used, but it might simplify the code: __zram_make_request() already takes
> care of splitting the request.
> 
> > for example:
> > 
> > @@ -510,6 +510,11 @@ static int zram_bvec_rw(struct zram *zram, struct bio_vec *bvec, u32 index,
> >                 ret = zram_bvec_write(zram, bvec, index, offset);
> >         }
> >  
> > +       if (unlikely(bio->bi_rw & REQ_DISCARD)) {
> 
> +		if (!is_partial_io(bvec) {
> 
> > +               write_lock(&zram->meta->tb_lock);
> > +               zram_free_page(zram, index);
> > +               write_unlock(&zram->meta->tb_lock);
> 
> +		}
> 
> Also this code might still call zram_bvec_read() and increase num_reads
> for discard request: I guess bio_data_dir(bio) == READ == 0 in this case.
> 
> Btw, why __zram_make_request() has an that rw argument? All the information
> it needs is passed by the bio argument already. I kind of recollect to have
> seen a cleanup patch that get rid of it or is it just my imagination?
> 

it doesn't. cleanup patch 'do not pass rw argument to __zram_make_request()'
is in linux-next.

	-ss

> Jerome
> 
> > +       }
> >         return ret;
> >  }
> > 
> > 	-ss
> > 
> >>  static void zram_reset_device(struct zram *zram, bool reset_capacity)
> >>  {
> >>  	size_t index;
> >> @@ -618,6 +648,12 @@ static void __zram_make_request(struct zram *zram, struct bio *bio)
> >>  	struct bio_vec bvec;
> >>  	struct bvec_iter iter;
> >>  
> >> +	if (unlikely(bio->bi_rw & REQ_DISCARD)) {
> >> +		zram_bio_discard(zram, bio);
> >> +		bio_endio(bio, 0);
> >> +		return;
> >> +	}
> >> +
> >>  	index = bio->bi_iter.bi_sector >> SECTORS_PER_PAGE_SHIFT;
> >>  	offset = (bio->bi_iter.bi_sector &
> >>  		  (SECTORS_PER_PAGE - 1)) << SECTOR_SHIFT;
> >> @@ -784,6 +820,10 @@ static int create_device(struct zram *zram, int device_id)
> >>  					ZRAM_LOGICAL_BLOCK_SIZE);
> >>  	blk_queue_io_min(zram->disk->queue, PAGE_SIZE);
> >>  	blk_queue_io_opt(zram->disk->queue, PAGE_SIZE);
> >> +	zram->disk->queue->limits.discard_granularity = PAGE_SIZE;
> >> +	zram->disk->queue->limits.max_discard_sectors = UINT_MAX;
> >> +	zram->disk->queue->limits.discard_zeroes_data = 1;
> >> +	queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, zram->disk->queue);
> >>  
> >>  	add_disk(zram->disk);
> >>  
> >> -- 
> >> 1.7.9.5
> >>
>