[PATCH 0/2] zram: cleanup stream's ->private allocation code

linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed

* [PATCH 0/2] zram: cleanup stream's ->private allocation code
@ 2015-11-27  4:23 Sergey Senozhatsky
  2015-11-27  4:23 ` [PATCH 1/2] zram: pass gfp from zcomp frontend to backend Sergey Senozhatsky
  2015-11-27  4:23 ` [PATCH 2/2] zram/zcomp: do not zero out zcomp private pages Sergey Senozhatsky
  0 siblings, 2 replies; 6+ messages in thread
From: Sergey Senozhatsky @ 2015-11-27  4:23 UTC (permalink / raw)
  To: Andrew Morton
  Cc: Minchan Kim, linux-kernel, Sergey Senozhatsky, Sergey Senozhatsky

Hello,
	Two patches that 'cleanup' compression streams allocation.
Patch #1 consolidates GFP flags used for stream's ->private
allocation in underlying compression backends.

Patch #2 drops GFP_ZERO flag from ->private allocation.


This patch set depends on the patch set published here
http://marc.info/?l=linux-kernel&m=144859748015849


Minchan Kim (1):
  zram: pass gfp from zcomp frontend to backend

Sergey Senozhatsky (1):
  zram/zcomp: do not zero out zcomp private pages

 drivers/block/zram/zcomp.c     | 24 ++++++++++++++++--------
 drivers/block/zram/zcomp.h     |  2 +-
 drivers/block/zram/zcomp_lz4.c | 16 +++-------------
 drivers/block/zram/zcomp_lzo.c | 16 +++-------------
 4 files changed, 23 insertions(+), 35 deletions(-)

-- 
2.6.3.368.gf34be46


^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH 1/2] zram: pass gfp from zcomp frontend to backend
  2015-11-27  4:23 [PATCH 0/2] zram: cleanup stream's ->private allocation code Sergey Senozhatsky
@ 2015-11-27  4:23 ` Sergey Senozhatsky
  2015-12-01  0:10   ` Andrew Morton
  2015-11-27  4:23 ` [PATCH 2/2] zram/zcomp: do not zero out zcomp private pages Sergey Senozhatsky
  1 sibling, 1 reply; 6+ messages in thread
From: Sergey Senozhatsky @ 2015-11-27  4:23 UTC (permalink / raw)
  To: Andrew Morton
  Cc: Minchan Kim, linux-kernel, Sergey Senozhatsky, Sergey Senozhatsky

From: Minchan Kim <minchan@kernel.org>

Each zcomp backend uses own gfp flag but it's pointless
because the context they could be called is driven by upper
layer(ie, zcomp frontend). As well, zcomp frondend could
call them in different context. One context(ie, zram init part)
is it should be better to make sure successful allocation
other context(ie, further stream allocation part for accelarating
I/O speed) is just optional so let's pass gfp down from driver
(ie, zcomp frontend) like normal MM convention.

[sergey: add missing __vmalloc zero and highmem gfps]
Signed-off-by: Minchan Kim <minchan@kernel.org>
Acked-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
---
 drivers/block/zram/zcomp.c     | 24 ++++++++++++++++--------
 drivers/block/zram/zcomp.h     |  2 +-
 drivers/block/zram/zcomp_lz4.c | 16 +++-------------
 drivers/block/zram/zcomp_lzo.c | 16 +++-------------
 4 files changed, 23 insertions(+), 35 deletions(-)

diff --git a/drivers/block/zram/zcomp.c b/drivers/block/zram/zcomp.c
index c536177..2bc4aea 100644
--- a/drivers/block/zram/zcomp.c
+++ b/drivers/block/zram/zcomp.c
@@ -74,18 +74,18 @@ static void zcomp_strm_free(struct zcomp *comp, struct zcomp_strm *zstrm)
  * allocate new zcomp_strm structure with ->private initialized by
  * backend, return NULL on error
  */
-static struct zcomp_strm *zcomp_strm_alloc(struct zcomp *comp)
+static struct zcomp_strm *zcomp_strm_alloc(struct zcomp *comp, gfp_t flags)
 {
-	struct zcomp_strm *zstrm = kmalloc(sizeof(*zstrm), GFP_NOIO);
+	struct zcomp_strm *zstrm = kmalloc(sizeof(*zstrm), flags);
 	if (!zstrm)
 		return NULL;
 
-	zstrm->private = comp->backend->create();
+	zstrm->private = comp->backend->create(flags);
 	/*
 	 * allocate 2 pages. 1 for compressed data, plus 1 extra for the
 	 * case when compressed size is larger than the original one
 	 */
-	zstrm->buffer = (void *)__get_free_pages(GFP_NOIO | __GFP_ZERO, 1);
+	zstrm->buffer = (void *)__get_free_pages(flags | __GFP_ZERO, 1);
 	if (!zstrm->private || !zstrm->buffer) {
 		zcomp_strm_free(comp, zstrm);
 		zstrm = NULL;
@@ -120,8 +120,16 @@ static struct zcomp_strm *zcomp_strm_multi_find(struct zcomp *comp)
 		/* allocate new zstrm stream */
 		zs->avail_strm++;
 		spin_unlock(&zs->strm_lock);
-
-		zstrm = zcomp_strm_alloc(comp);
+		/*
+		 * This function can be called in swapout/fs write path
+		 * so we can't use GFP_FS|IO. And it assumes we already
+		 * have at least one stream in zram initialization so we
+		 * don't do best effort to allocate more stream in here.
+		 * A default stream will work well without further multiple
+		 * streams. That's why we use NORETRY | NOWARN | NOMEMALLOC.
+		 */
+		zstrm = zcomp_strm_alloc(comp, GFP_NOIO | __GFP_NORETRY |
+					__GFP_NOWARN | __GFP_NOMEMALLOC);
 		if (!zstrm) {
 			spin_lock(&zs->strm_lock);
 			zs->avail_strm--;
@@ -209,7 +217,7 @@ static int zcomp_strm_multi_create(struct zcomp *comp, int max_strm)
 	zs->max_strm = max_strm;
 	zs->avail_strm = 1;
 
-	zstrm = zcomp_strm_alloc(comp);
+	zstrm = zcomp_strm_alloc(comp, GFP_KERNEL);
 	if (!zstrm) {
 		kfree(zs);
 		return -ENOMEM;
@@ -259,7 +267,7 @@ static int zcomp_strm_single_create(struct zcomp *comp)
 
 	comp->stream = zs;
 	mutex_init(&zs->strm_lock);
-	zs->zstrm = zcomp_strm_alloc(comp);
+	zs->zstrm = zcomp_strm_alloc(comp, GFP_KERNEL);
 	if (!zs->zstrm) {
 		kfree(zs);
 		return -ENOMEM;
diff --git a/drivers/block/zram/zcomp.h b/drivers/block/zram/zcomp.h
index 46e2b9f..b7d2a4b 100644
--- a/drivers/block/zram/zcomp.h
+++ b/drivers/block/zram/zcomp.h
@@ -33,7 +33,7 @@ struct zcomp_backend {
 	int (*decompress)(const unsigned char *src, size_t src_len,
 			unsigned char *dst);
 
-	void *(*create)(void);
+	void *(*create)(gfp_t flags);
 	void (*destroy)(void *private);
 
 	const char *name;
diff --git a/drivers/block/zram/zcomp_lz4.c b/drivers/block/zram/zcomp_lz4.c
index f2bfced..dc2338d 100644
--- a/drivers/block/zram/zcomp_lz4.c
+++ b/drivers/block/zram/zcomp_lz4.c
@@ -15,24 +15,14 @@
 
 #include "zcomp_lz4.h"
 
-static void *zcomp_lz4_create(void)
+static void *zcomp_lz4_create(gfp_t flags)
 {
 	void *ret;
 
-	/*
-	 * This function can be called in swapout/fs write path
-	 * so we can't use GFP_FS|IO. And it assumes we already
-	 * have at least one stream in zram initialization so we
-	 * don't do best effort to allocate more stream in here.
-	 * A default stream will work well without further multiple
-	 * streams. That's why we use NORETRY | NOWARN | NOMEMALLOC.
-	 */
-	ret = kzalloc(LZ4_MEM_COMPRESS, GFP_NOIO | __GFP_NORETRY |
-					__GFP_NOWARN | __GFP_NOMEMALLOC);
+	ret = kzalloc(LZ4_MEM_COMPRESS, flags);
 	if (!ret)
 		ret = __vmalloc(LZ4_MEM_COMPRESS,
-				GFP_NOIO | __GFP_NORETRY | __GFP_NOWARN |
-				__GFP_NOMEMALLOC | __GFP_ZERO | __GFP_HIGHMEM,
+				flags | __GFP_ZERO | __GFP_HIGHMEM,
 				PAGE_KERNEL);
 	return ret;
 }
diff --git a/drivers/block/zram/zcomp_lzo.c b/drivers/block/zram/zcomp_lzo.c
index 7fbb4a3..0ab6fce 100644
--- a/drivers/block/zram/zcomp_lzo.c
+++ b/drivers/block/zram/zcomp_lzo.c
@@ -15,24 +15,14 @@
 
 #include "zcomp_lzo.h"
 
-static void *lzo_create(void)
+static void *lzo_create(gfp_t flags)
 {
 	void *ret;
 
-	/*
-	 * This function can be called in swapout/fs write path
-	 * so we can't use GFP_FS|IO. And it assumes we already
-	 * have at least one stream in zram initialization so we
-	 * don't do best effort to allocate more stream in here.
-	 * A default stream will work well without further multiple
-	 * streams. That's why we use NORETRY | NOWARN | NOMEMALLOC.
-	 */
-	ret = kzalloc(LZO1X_MEM_COMPRESS, GFP_NOIO | __GFP_NORETRY |
-					__GFP_NOWARN | __GFP_NOMEMALLOC);
+	ret = kzalloc(LZO1X_MEM_COMPRESS, flags);
 	if (!ret)
 		ret = __vmalloc(LZO1X_MEM_COMPRESS,
-				GFP_NOIO | __GFP_NORETRY | __GFP_NOWARN |
-				__GFP_NOMEMALLOC | __GFP_ZERO | __GFP_HIGHMEM,
+				flags | __GFP_ZERO | __GFP_HIGHMEM,
 				PAGE_KERNEL);
 	return ret;
 }
-- 
2.6.3.368.gf34be46


^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH 1/2] zram: pass gfp from zcomp frontend to backend
  2015-11-27  4:23 ` [PATCH 1/2] zram: pass gfp from zcomp frontend to backend Sergey Senozhatsky
@ 2015-12-01  0:10   ` Andrew Morton
  2015-12-01  0:44     ` Sergey Senozhatsky
  0 siblings, 1 reply; 6+ messages in thread
From: Andrew Morton @ 2015-12-01  0:10 UTC (permalink / raw)
  To: Sergey Senozhatsky; +Cc: Minchan Kim, linux-kernel, Sergey Senozhatsky

On Fri, 27 Nov 2015 13:23:13 +0900 Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com> wrote:

> From: Minchan Kim <minchan@kernel.org>
> 
> Each zcomp backend uses own gfp flag but it's pointless
> because the context they could be called is driven by upper
> layer(ie, zcomp frontend). As well, zcomp frondend could
> call them in different context. One context(ie, zram init part)
> is it should be better to make sure successful allocation
> other context(ie, further stream allocation part for accelarating
> I/O speed) is just optional so let's pass gfp down from driver
> (ie, zcomp frontend) like normal MM convention.
> 
> [sergey: add missing __vmalloc zero and highmem gfps]
> Signed-off-by: Minchan Kim <minchan@kernel.org>
> Acked-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>

I changed this to signed-off-by:you.  Documentation/SubmittingPatches
section 11 explains why ;)

(Hm, rule (c) is recursive.  Nice!)

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH 1/2] zram: pass gfp from zcomp frontend to backend
  2015-12-01  0:10   ` Andrew Morton
@ 2015-12-01  0:44     ` Sergey Senozhatsky
  0 siblings, 0 replies; 6+ messages in thread
From: Sergey Senozhatsky @ 2015-12-01  0:44 UTC (permalink / raw)
  To: Andrew Morton
  Cc: Sergey Senozhatsky, Minchan Kim, linux-kernel, Sergey Senozhatsky

On (11/30/15 16:10), Andrew Morton wrote:
> > From: Minchan Kim <minchan@kernel.org>
> > 
> > Each zcomp backend uses own gfp flag but it's pointless
> > because the context they could be called is driven by upper
> > layer(ie, zcomp frontend). As well, zcomp frondend could
> > call them in different context. One context(ie, zram init part)
> > is it should be better to make sure successful allocation
> > other context(ie, further stream allocation part for accelarating
> > I/O speed) is just optional so let's pass gfp down from driver
> > (ie, zcomp frontend) like normal MM convention.
> > 
> > [sergey: add missing __vmalloc zero and highmem gfps]
> > Signed-off-by: Minchan Kim <minchan@kernel.org>
> > Acked-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
> 
> I changed this to signed-off-by:you.  Documentation/SubmittingPatches
> section 11 explains why ;)

Thank you.

> (Hm, rule (c) is recursive.  Nice!)

	-ss

^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH 2/2] zram/zcomp: do not zero out zcomp private pages
  2015-11-27  4:23 [PATCH 0/2] zram: cleanup stream's ->private allocation code Sergey Senozhatsky
  2015-11-27  4:23 ` [PATCH 1/2] zram: pass gfp from zcomp frontend to backend Sergey Senozhatsky
@ 2015-11-27  4:23 ` Sergey Senozhatsky
  2015-11-30  7:14   ` Minchan Kim
  1 sibling, 1 reply; 6+ messages in thread
From: Sergey Senozhatsky @ 2015-11-27  4:23 UTC (permalink / raw)
  To: Andrew Morton
  Cc: Minchan Kim, linux-kernel, Sergey Senozhatsky, Sergey Senozhatsky

Do not __GFP_ZERO allocated zcomp ->private pages. We keep
allocated streams around and use them for read/write requests,
so we supply a zeroed out ->private to compression algorithm
as a scratch buffer only once -- the first time we use that
stream. For the rest of IO requests served by this stream
->private usually contains some temporarily data from the
previous requests.

Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
---
 drivers/block/zram/zcomp_lz4.c | 4 ++--
 drivers/block/zram/zcomp_lzo.c | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/block/zram/zcomp_lz4.c b/drivers/block/zram/zcomp_lz4.c
index dc2338d..0110086 100644
--- a/drivers/block/zram/zcomp_lz4.c
+++ b/drivers/block/zram/zcomp_lz4.c
@@ -19,10 +19,10 @@ static void *zcomp_lz4_create(gfp_t flags)
 {
 	void *ret;
 
-	ret = kzalloc(LZ4_MEM_COMPRESS, flags);
+	ret = kmalloc(LZ4_MEM_COMPRESS, flags);
 	if (!ret)
 		ret = __vmalloc(LZ4_MEM_COMPRESS,
-				flags | __GFP_ZERO | __GFP_HIGHMEM,
+				flags | __GFP_HIGHMEM,
 				PAGE_KERNEL);
 	return ret;
 }
diff --git a/drivers/block/zram/zcomp_lzo.c b/drivers/block/zram/zcomp_lzo.c
index 0ab6fce..ed7a1f0 100644
--- a/drivers/block/zram/zcomp_lzo.c
+++ b/drivers/block/zram/zcomp_lzo.c
@@ -19,10 +19,10 @@ static void *lzo_create(gfp_t flags)
 {
 	void *ret;
 
-	ret = kzalloc(LZO1X_MEM_COMPRESS, flags);
+	ret = kmalloc(LZO1X_MEM_COMPRESS, flags);
 	if (!ret)
 		ret = __vmalloc(LZO1X_MEM_COMPRESS,
-				flags | __GFP_ZERO | __GFP_HIGHMEM,
+				flags | __GFP_HIGHMEM,
 				PAGE_KERNEL);
 	return ret;
 }
-- 
2.6.3.368.gf34be46


^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH 2/2] zram/zcomp: do not zero out zcomp private pages
  2015-11-27  4:23 ` [PATCH 2/2] zram/zcomp: do not zero out zcomp private pages Sergey Senozhatsky
@ 2015-11-30  7:14   ` Minchan Kim
  0 siblings, 0 replies; 6+ messages in thread
From: Minchan Kim @ 2015-11-30  7:14 UTC (permalink / raw)
  To: Sergey Senozhatsky; +Cc: Andrew Morton, linux-kernel, Sergey Senozhatsky

On Fri, Nov 27, 2015 at 01:23:14PM +0900, Sergey Senozhatsky wrote:
> Do not __GFP_ZERO allocated zcomp ->private pages. We keep
> allocated streams around and use them for read/write requests,
> so we supply a zeroed out ->private to compression algorithm
> as a scratch buffer only once -- the first time we use that
> stream. For the rest of IO requests served by this stream
> ->private usually contains some temporarily data from the
> previous requests.
> 
> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Acked-by: Minchan Kim <minchan@kernel.org>

Thanks.

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2015-12-01  0:43 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2015-11-27  4:23 [PATCH 0/2] zram: cleanup stream's ->private allocation code Sergey Senozhatsky
2015-11-27  4:23 ` [PATCH 1/2] zram: pass gfp from zcomp frontend to backend Sergey Senozhatsky
2015-12-01  0:10   ` Andrew Morton
2015-12-01  0:44     ` Sergey Senozhatsky
2015-11-27  4:23 ` [PATCH 2/2] zram/zcomp: do not zero out zcomp private pages Sergey Senozhatsky
2015-11-30  7:14   ` Minchan Kim

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).