Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1727357 > unrolled thread

[PATCH 00/18] lightnvm: pblk patches for 4.14

Back to article view | Back to linux.kernel

Contents

  [PATCH 00/18] lightnvm: pblk patches for 4.14 "Javier González" <jg@lightnvm.io> - 2017-09-06 13:00 +0200
    [PATCH 07/18] lightnvm: pblk: use constant for GC parameter "Javier González" <jg@lightnvm.io> - 2017-09-06 13:00 +0200
    [PATCH 01/18] lightnvm: pblk: improve naming for internal req. "Javier González" <jg@lightnvm.io> - 2017-09-06 13:00 +0200
      Re: [PATCH 01/18] lightnvm: pblk: improve naming for internal req. Christoph Hellwig <hch@infradead.org> - 2017-09-06 15:50 +0200
        Re: [PATCH 01/18] lightnvm: pblk: improve naming for internal req. Javier González <jg@lightnvm.io> - 2017-09-06 16:10 +0200
    Re: [PATCH 00/18] lightnvm: pblk patches for 4.14 Jens Axboe <axboe@kernel.dk> - 2017-09-06 16:10 +0200
      Re: [PATCH 00/18] lightnvm: pblk patches for 4.14 Javier González <jg@lightnvm.io> - 2017-09-06 16:20 +0200

#1727357 — [PATCH 00/18] lightnvm: pblk patches for 4.14

Hi Jens,

Here is the pblk patchset for this window.

The most notable patches are on the read path:
  - ("lightnvm: pblk: check lba sanity on read path") uses the lba
    stored on the out-of-bound area to verify that the read ppa
    corresponds to the lba pointed to by the read bio.
  - ("lightnvm: pblk: guarantee line integrity on reads") guarantees
    that if a line is being GC'ed and a read comes in the middle, that
    line is not being moved into the free list until the read completes.
    Otherwise, the line could be reclaimed to be erased and re-written,
    thus causing data corruption.

The rest of the patches are are basically bug fixes and refactoring to
improve code readability.

The patches apply on your for-4.14/block, and you can be found at:
  https://github.com/OpenChannelSSD/linux/tree/pblk.for-4.14

Thanks,
Javier

Javier González (18):
  lightnvm: pblk: improve naming for internal req.
  lightnvm: pblk: refactor read lba sanity check
  lightnvm: pblk: normalize ppa namings
  lightnvm: pblk: check for failed mempool alloc.
  lightnvm: pblk: initialize debug stat counter
  lightnvm: pblk: use right flag for GC allocation
  lightnvm: pblk: use constant for GC parameter
  lightnvm: pblk: check lba sanity on read path
  lightnvm: pblk: simplify data validity check on GC
  lightnvm: pblk: use bio_copy_kern when possible
  lightnvm: pblk: refactor read path on GC
  lightnvm: pblk: free padded entries in write buffer
  lightnvm: pblk: fix write I/O sync stat
  lightnvm: pblk: simplify path on REQ_PREFLUSH
  lightnvm: pblk: avoid deadlock on low LUN config
  lightnvm: pblk: enable 1 LUN configuration
  lightnvm: pblk: guarantee line integrity on reads
  lightnvm: pblk: remove unnecessary check

 drivers/lightnvm/pblk-cache.c    |  24 ++--
 drivers/lightnvm/pblk-core.c     | 187 ++++++++++++++++--------------
 drivers/lightnvm/pblk-gc.c       | 133 ++++++++++------------
 drivers/lightnvm/pblk-init.c     |  25 ++++-
 drivers/lightnvm/pblk-map.c      |  21 ++--
 drivers/lightnvm/pblk-rb.c       |  14 +--
 drivers/lightnvm/pblk-read.c     | 237 ++++++++++++++++++++++++++-------------
 drivers/lightnvm/pblk-recovery.c |   9 +-
 drivers/lightnvm/pblk-rl.c       |   6 +
 drivers/lightnvm/pblk-write.c    |  37 +++---
 drivers/lightnvm/pblk.h          |  40 +++----
 11 files changed, 423 insertions(+), 310 deletions(-)

-- 
2.7.4

[toc] | [next] | [standalone]

#1727358 — [PATCH 07/18] lightnvm: pblk: use constant for GC parameter

Use a constant instead of using meaningless tuning parameters on the
garbage collector algorithms.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
---
 drivers/lightnvm/pblk-gc.c | 4 ++--
 drivers/lightnvm/pblk.h    | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/lightnvm/pblk-gc.c b/drivers/lightnvm/pblk-gc.c
index 6090d28f7995..92694e35afde 100644
--- a/drivers/lightnvm/pblk-gc.c
+++ b/drivers/lightnvm/pblk-gc.c
@@ -93,7 +93,7 @@ static int pblk_gc_move_valid_secs(struct pblk *pblk, struct pblk_gc_rq *gc_rq)
 
 retry:
 	spin_lock(&gc->w_lock);
-	if (gc->w_entries >= PBLK_GC_W_QD) {
+	if (gc->w_entries >= PBLK_GC_RQ_QD) {
 		spin_unlock(&gc->w_lock);
 		pblk_gc_writer_kick(&pblk->gc);
 		usleep_range(128, 256);
@@ -602,7 +602,7 @@ int pblk_gc_init(struct pblk *pblk)
 	spin_lock_init(&gc->w_lock);
 	spin_lock_init(&gc->r_lock);
 
-	sema_init(&gc->gc_sem, 128);
+	sema_init(&gc->gc_sem, PBLK_GC_RQ_QD);
 
 	INIT_LIST_HEAD(&gc->w_list);
 	INIT_LIST_HEAD(&gc->r_list);
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 08a463fe855f..b7f5fa8b49d0 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -818,7 +818,7 @@ int pblk_recov_setup_rq(struct pblk *pblk, struct pblk_c_ctx *c_ctx,
  * pblk gc
  */
 #define PBLK_GC_MAX_READERS 8	/* Max number of outstanding GC reader jobs */
-#define PBLK_GC_W_QD 128	/* Queue depth for inflight GC write I/Os */
+#define PBLK_GC_RQ_QD 128	/* Queue depth for inflight GC requests */
 #define PBLK_GC_L_QD 4		/* Queue depth for inflight GC lines */
 #define PBLK_GC_RSV_LINE 1	/* Reserved lines for GC */
 
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1727359 — [PATCH 01/18] lightnvm: pblk: improve naming for internal req.

Each request type sent to the LightNVM subsystem requires different
metadata. Until now, we have tailored this metadata based on write, read
and erase commands. However, pblk uses different metadata for internal
writes that do not hit the write buffer. Instead of abusing the metadata
for reads, create a new request type - internal write. This improves
code readability.

Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
---
 drivers/lightnvm/pblk-recovery.c | 6 +++---
 drivers/lightnvm/pblk-write.c    | 6 +++---
 drivers/lightnvm/pblk.h          | 5 ++++-
 3 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c
index cb556e06673e..279638309d9a 100644
--- a/drivers/lightnvm/pblk-recovery.c
+++ b/drivers/lightnvm/pblk-recovery.c
@@ -344,7 +344,7 @@ static void pblk_end_io_recov(struct nvm_rq *rqd)
 
 	bio_put(rqd->bio);
 	nvm_dev_dma_free(dev->parent, rqd->meta_list, rqd->dma_meta_list);
-	pblk_free_rqd(pblk, rqd, WRITE);
+	pblk_free_rqd(pblk, rqd, WRITE_INT);
 
 	atomic_dec(&pblk->inflight_io);
 	kref_put(&pad_rq->ref, pblk_recov_complete);
@@ -404,7 +404,7 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
 	ppa_list = (void *)(meta_list) + pblk_dma_meta_size;
 	dma_ppa_list = dma_meta_list + pblk_dma_meta_size;
 
-	rqd = pblk_alloc_rqd(pblk, WRITE);
+	rqd = pblk_alloc_rqd(pblk, WRITE_INT);
 	if (IS_ERR(rqd)) {
 		ret = PTR_ERR(rqd);
 		goto fail_free_meta;
@@ -491,7 +491,7 @@ static int pblk_recov_pad_oob(struct pblk *pblk, struct pblk_line *line,
 fail_free_bio:
 	bio_put(bio);
 fail_free_rqd:
-	pblk_free_rqd(pblk, rqd, WRITE);
+	pblk_free_rqd(pblk, rqd, WRITE_INT);
 fail_free_meta:
 	nvm_dev_dma_free(dev->parent, meta_list, dma_meta_list);
 fail_free_pad:
diff --git a/drivers/lightnvm/pblk-write.c b/drivers/lightnvm/pblk-write.c
index 3ad9e56d2473..920b93fb15d5 100644
--- a/drivers/lightnvm/pblk-write.c
+++ b/drivers/lightnvm/pblk-write.c
@@ -199,7 +199,7 @@ static void pblk_end_io_write_meta(struct nvm_rq *rqd)
 
 	bio_put(rqd->bio);
 	nvm_dev_dma_free(dev->parent, rqd->meta_list, rqd->dma_meta_list);
-	pblk_free_rqd(pblk, rqd, READ);
+	pblk_free_rqd(pblk, rqd, WRITE_INT);
 
 	atomic_dec(&pblk->inflight_io);
 }
@@ -370,7 +370,7 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 	int i, j;
 	int ret;
 
-	rqd = pblk_alloc_rqd(pblk, READ);
+	rqd = pblk_alloc_rqd(pblk, WRITE_INT);
 	if (IS_ERR(rqd)) {
 		pr_err("pblk: cannot allocate write req.\n");
 		return PTR_ERR(rqd);
@@ -434,7 +434,7 @@ int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line)
 	if (likely(l_mg->emeta_alloc_type == PBLK_VMALLOC_META))
 		bio_put(bio);
 fail_free_rqd:
-	pblk_free_rqd(pblk, rqd, READ);
+	pblk_free_rqd(pblk, rqd, WRITE_INT);
 	return ret;
 }
 
diff --git a/drivers/lightnvm/pblk.h b/drivers/lightnvm/pblk.h
index 67e623bd5c2d..08a463fe855f 100644
--- a/drivers/lightnvm/pblk.h
+++ b/drivers/lightnvm/pblk.h
@@ -59,7 +59,10 @@
 		for ((i) = 0, rlun = &(pblk)->luns[0]; \
 			(i) < (pblk)->nr_luns; (i)++, rlun = &(pblk)->luns[(i)])
 
-#define ERASE 2 /* READ = 0, WRITE = 1 */
+/* READ = 0, WRITE (user) = 1 */
+#define ERASE 2
+#define WRITE_INT 3 /* Internal write. Not through write buffer */
+
 
 enum {
 	/* IO Types */
-- 
2.7.4

[toc] | [prev] | [next] | [standalone]

#1727490 — Re: [PATCH 01/18] lightnvm: pblk: improve naming for internal req.

> -#define ERASE 2 /* READ = 0, WRITE = 1 */
> +/* READ = 0, WRITE (user) = 1 */
> +#define ERASE 2
> +#define WRITE_INT 3 /* Internal write. Not through write buffer */

enum {
	PBLK_READ,
	PBLK_WRITE,
	PBLK_ERASE,
	PBLK_WRITE,
};

please.  Don't abuse and overload the messy READ/WRITE macros in any
new code.

[toc] | [prev] | [next] | [standalone]

#1727518 — Re: [PATCH 01/18] lightnvm: pblk: improve naming for internal req.

[Multipart message — attachments visible in raw view] — view raw

> On 6 Sep 2017, at 15.46, Christoph Hellwig <hch@infradead.org> wrote:
> 
>> -#define ERASE 2 /* READ = 0, WRITE = 1 */
>> +/* READ = 0, WRITE (user) = 1 */
>> +#define ERASE 2
>> +#define WRITE_INT 3 /* Internal write. Not through write buffer */
> 
> enum {
> 	PBLK_READ,
> 	PBLK_WRITE,
> 	PBLK_ERASE,
> 	PBLK_WRITE,
> };
> 
> please.  Don't abuse and overload the messy READ/WRITE macros in any
> new code.

Ok. I'll resend.

Javier

[toc] | [prev] | [next] | [standalone]

#1727525

On 09/06/2017 04:50 AM, Javier González wrote:
> Hi Jens,
> 
> Here is the pblk patchset for this window.

You must mean for the next window, surely, since we're in the middle
of the current merge window? Have these patches been posted for
review before? This should have been done (at least) two weeks ago!

Get rid of anything that isn't a bug fix, I don't want to take
anything that looks like a feature or refactoring of code.

-- 
Jens Axboe

[toc] | [prev] | [next] | [standalone]

#1727539

[Multipart message — attachments visible in raw view] — view raw

> On 6 Sep 2017, at 16.04, Jens Axboe <axboe@kernel.dk> wrote:
> 
> On 09/06/2017 04:50 AM, Javier González wrote:
>> Hi Jens,
>> 
>> Here is the pblk patchset for this window.
> 
> You must mean for the next window, surely, since we're in the middle
> of the current merge window? Have these patches been posted for
> review before? This should have been done (at least) two weeks ago!
> 
> Get rid of anything that isn't a bug fix, I don't want to take
> anything that looks like a feature or refactoring of code.
> 

Ok. I'll re-post only bug fixes.

> --
> Jens Axboe

Javier.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web