linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Dennis Zhou <dennis@kernel.org>
To: David Sterba <dsterba@suse.com>,
	Josef Bacik <josef@toxicpanda.com>, Chris Mason <clm@fb.com>,
	Omar Sandoval <osandov@osandov.com>,
	Nick Terrell <terrelln@fb.com>,
	Nikolay Borisov <nborisov@suse.com>
Cc: kernel-team@fb.com, linux-btrfs@vger.kernel.org,
	linux-kernel@vger.kernel.org, Dennis Zhou <dennis@kernel.org>
Subject: [PATCH 11/12] btrfs: make zstd memory requirements monotonic
Date: Mon,  4 Feb 2019 15:20:07 -0500	[thread overview]
Message-ID: <20190204202008.51652-12-dennis@kernel.org> (raw)
In-Reply-To: <20190204202008.51652-1-dennis@kernel.org>

It is possible based on the level configurations that a higher level
workspace uses less memory than a lower level workspace. In order to
reuse workspaces, this must be made a monotonic relationship. This
precomputes the required memory for each level and enforces the
monotonicity between level and memory required. This is also done
in upstream zstd in [1].

[1] https://github.com/facebook/zstd/commit/a68b76afefec6876f8e8a538155109a5aeac0143

Signed-off-by: Dennis Zhou <dennis@kernel.org>
Cc: Nick Terrell <terrelln@fb.com>
---
 fs/btrfs/zstd.c | 38 +++++++++++++++++++++++++++++++++-----
 1 file changed, 33 insertions(+), 5 deletions(-)

diff --git a/fs/btrfs/zstd.c b/fs/btrfs/zstd.c
index a951d4fe77f7..2231123fedbe 100644
--- a/fs/btrfs/zstd.c
+++ b/fs/btrfs/zstd.c
@@ -20,6 +20,7 @@
 #define ZSTD_BTRFS_MAX_WINDOWLOG 17
 #define ZSTD_BTRFS_MAX_INPUT (1 << ZSTD_BTRFS_MAX_WINDOWLOG)
 #define ZSTD_BTRFS_DEFAULT_LEVEL 3
+#define ZSTD_BTRFS_MAX_LEVEL 15
 
 static ZSTD_parameters zstd_get_btrfs_parameters(unsigned int level,
 						 size_t src_len)
@@ -44,8 +45,39 @@ struct workspace {
 
 static struct workspace_manager wsm;
 
+static size_t zstd_ws_mem_sizes[ZSTD_BTRFS_MAX_LEVEL];
+
+/*
+ * zstd_calc_ws_mem_sizes - calculate monotonic memory bounds
+ *
+ * It is possible based on the level configurations that a higher level
+ * workspace uses less memory than a lower level workspace.  In order to
+ * reuse workspaces, this must be made a monotonic relationship.  This
+ * precomputes the required memory for each level and enforces the
+ * monotonicity between level and memory required.
+ */
+static void zstd_calc_ws_mem_sizes(void)
+{
+	size_t max_size = 0;
+	unsigned int level;
+
+	for (level = 1; level <= ZSTD_BTRFS_MAX_LEVEL; level++) {
+		ZSTD_parameters params =
+			zstd_get_btrfs_parameters(level, ZSTD_BTRFS_MAX_INPUT);
+		size_t level_size =
+			max_t(size_t,
+			      ZSTD_CStreamWorkspaceBound(params.cParams),
+			      ZSTD_DStreamWorkspaceBound(ZSTD_BTRFS_MAX_INPUT));
+
+		max_size = max_t(size_t, max_size, level_size);
+		zstd_ws_mem_sizes[level - 1] = max_size;
+	}
+}
+
 static void zstd_init_workspace_manager(void)
 {
+	zstd_calc_ws_mem_sizes();
+
 	btrfs_init_workspace_manager(&wsm, &btrfs_zstd_compress);
 }
 
@@ -80,17 +112,13 @@ static void zstd_free_workspace(struct list_head *ws)
 
 static struct list_head *zstd_alloc_workspace(unsigned int level)
 {
-	ZSTD_parameters params =
-			zstd_get_btrfs_parameters(level, ZSTD_BTRFS_MAX_INPUT);
 	struct workspace *workspace;
 
 	workspace = kzalloc(sizeof(*workspace), GFP_KERNEL);
 	if (!workspace)
 		return ERR_PTR(-ENOMEM);
 
-	workspace->size = max_t(size_t,
-			ZSTD_CStreamWorkspaceBound(params.cParams),
-			ZSTD_DStreamWorkspaceBound(ZSTD_BTRFS_MAX_INPUT));
+	workspace->size = zstd_ws_mem_sizes[level - 1];
 	workspace->mem = kvmalloc(workspace->size, GFP_KERNEL);
 	workspace->buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
 	if (!workspace->mem || !workspace->buf)
-- 
2.17.1


  parent reply	other threads:[~2019-02-04 20:21 UTC|newest]

Thread overview: 34+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-02-04 20:19 [PATCH v2 00/12] btrfs: add zstd compression level support Dennis Zhou
2019-02-04 20:19 ` [PATCH 01/12] btrfs: add helpers for compression type and level Dennis Zhou
2019-02-04 20:19 ` [PATCH 02/12] btrfs: rename workspaces_list to workspace_manager Dennis Zhou
2019-02-04 20:19 ` [PATCH 03/12] btrfs: manage heuristic workspace as index 0 Dennis Zhou
2019-02-04 20:20 ` [PATCH 04/12] btrfs: unify compression ops with workspace_manager Dennis Zhou
2019-02-04 20:20 ` [PATCH 05/12] btrfs: add helper methods for workspace manager init and cleanup Dennis Zhou
2019-02-04 20:20 ` [PATCH 06/12] btrfs: add compression interface in (get/put)_workspace() Dennis Zhou
2019-02-04 20:20 ` [PATCH 07/12] btrfs: move to fn pointers for get/put workspaces Dennis Zhou
2019-02-04 20:20 ` [PATCH 08/12] btrfs: plumb level through the compression interface Dennis Zhou
2019-02-04 20:20 ` [PATCH 09/12] btrfs: change set_level() to bound the level passed in Dennis Zhou
2019-02-05 19:06   ` David Sterba
2019-02-05 19:32     ` Dennis Zhou
2019-02-05 19:54       ` David Sterba
2019-02-05 19:59         ` Dennis Zhou
2019-02-06 16:48   ` [PATCH v2 " Dennis Zhou
2019-02-04 20:20 ` [PATCH 10/12] btrfs: zstd use the passed through level instead of default Dennis Zhou
2019-02-04 20:20 ` Dennis Zhou [this message]
2019-02-04 20:20 ` [PATCH 12/12] btrfs: add zstd compression level support Dennis Zhou
2019-02-05 19:26   ` [PATCH v2 " Dennis Zhou
2019-02-06 16:47   ` [PATCH v3 " Dennis Zhou
2019-02-05 14:57 ` [PATCH v2 00/12] " David Sterba
2019-02-05 16:03   ` Dennis Zhou
2019-02-05 16:27     ` David Sterba
2019-02-05 16:30       ` Dennis Zhou
2019-02-05 16:51         ` David Sterba
2019-02-05 17:07           ` David Sterba
2019-02-05 18:27             ` David Sterba
2019-02-05 18:30               ` Dennis Zhou
2019-02-05 20:48                 ` Dennis Zhou
2019-02-06 15:15                   ` David Sterba
2019-02-06 16:51                     ` Dennis Zhou
2019-02-07 16:59 ` David Sterba
2019-02-07 17:36   ` Dennis Zhou
2019-02-08 13:22     ` David Sterba

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190204202008.51652-12-dennis@kernel.org \
    --to=dennis@kernel.org \
    --cc=clm@fb.com \
    --cc=dsterba@suse.com \
    --cc=josef@toxicpanda.com \
    --cc=kernel-team@fb.com \
    --cc=linux-btrfs@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=nborisov@suse.com \
    --cc=osandov@osandov.com \
    --cc=terrelln@fb.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).