linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Dennis Zhou <dennis@kernel.org>
To: Dennis Zhou <dennis@kernel.org>, Tejun Heo <tj@kernel.org>,
	Christoph Lameter <cl@linux.com>
Cc: Vlad Buslov <vladbu@mellanox.com>,
	kernel-team@fb.com, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org
Subject: [PATCH 12/12] percpu: use chunk scan_hint to skip some scanning
Date: Wed, 27 Feb 2019 21:18:39 -0500	[thread overview]
Message-ID: <20190228021839.55779-13-dennis@kernel.org> (raw)
In-Reply-To: <20190228021839.55779-1-dennis@kernel.org>

Just like blocks, chunks now maintain a scan_hint. This can be used to
skip some scanning by promoting the scan_hint to be the contig_hint.
The chunk's scan_hint is primarily updated on the backside and relies on
full scanning when a block becomes free or the free region spans across
blocks.

Signed-off-by: Dennis Zhou <dennis@kernel.org>
---
 mm/percpu.c | 36 +++++++++++++++++++++++++++---------
 1 file changed, 27 insertions(+), 9 deletions(-)

diff --git a/mm/percpu.c b/mm/percpu.c
index 197479f2c489..40d49d7fb286 100644
--- a/mm/percpu.c
+++ b/mm/percpu.c
@@ -711,20 +711,31 @@ static void pcpu_block_update_scan(struct pcpu_chunk *chunk, int bit_off,
 /**
  * pcpu_chunk_refresh_hint - updates metadata about a chunk
  * @chunk: chunk of interest
+ * @full_scan: if we should scan from the beginning
  *
  * Iterates over the metadata blocks to find the largest contig area.
- * It also counts the populated pages and uses the delta to update the
- * global count.
+ * A full scan can be avoided on the allocation path as this is triggered
+ * if we broke the contig_hint.  In doing so, the scan_hint will be before
+ * the contig_hint or after if the scan_hint == contig_hint.  This cannot
+ * be prevented on freeing as we want to find the largest area possibly
+ * spanning blocks.
  */
-static void pcpu_chunk_refresh_hint(struct pcpu_chunk *chunk)
+static void pcpu_chunk_refresh_hint(struct pcpu_chunk *chunk, bool full_scan)
 {
 	struct pcpu_block_md *chunk_md = &chunk->chunk_md;
 	int bit_off, bits;
 
-	/* clear metadata */
-	chunk_md->contig_hint = 0;
+	/* promote scan_hint to contig_hint */
+	if (!full_scan && chunk_md->scan_hint) {
+		bit_off = chunk_md->scan_hint_start + chunk_md->scan_hint;
+		chunk_md->contig_hint_start = chunk_md->scan_hint_start;
+		chunk_md->contig_hint = chunk_md->scan_hint;
+		chunk_md->scan_hint = 0;
+	} else {
+		bit_off = chunk_md->first_free;
+		chunk_md->contig_hint = 0;
+	}
 
-	bit_off = chunk_md->first_free;
 	bits = 0;
 	pcpu_for_each_md_free_region(chunk, bit_off, bits) {
 		pcpu_block_update(chunk_md, bit_off, bit_off + bits);
@@ -884,6 +895,13 @@ static void pcpu_block_update_hint_alloc(struct pcpu_chunk *chunk, int bit_off,
 	if (nr_empty_pages)
 		pcpu_update_empty_pages(chunk, -1 * nr_empty_pages);
 
+	if (pcpu_region_overlap(chunk_md->scan_hint_start,
+				chunk_md->scan_hint_start +
+				chunk_md->scan_hint,
+				bit_off,
+				bit_off + bits))
+		chunk_md->scan_hint = 0;
+
 	/*
 	 * The only time a full chunk scan is required is if the chunk
 	 * contig hint is broken.  Otherwise, it means a smaller space
@@ -894,7 +912,7 @@ static void pcpu_block_update_hint_alloc(struct pcpu_chunk *chunk, int bit_off,
 				chunk_md->contig_hint,
 				bit_off,
 				bit_off + bits))
-		pcpu_chunk_refresh_hint(chunk);
+		pcpu_chunk_refresh_hint(chunk, false);
 }
 
 /**
@@ -1005,7 +1023,7 @@ static void pcpu_block_update_hint_free(struct pcpu_chunk *chunk, int bit_off,
 	 * the else condition below.
 	 */
 	if (((end - start) >= PCPU_BITMAP_BLOCK_BITS) || s_index != e_index)
-		pcpu_chunk_refresh_hint(chunk);
+		pcpu_chunk_refresh_hint(chunk, true);
 	else
 		pcpu_block_update(&chunk->chunk_md,
 				  pcpu_block_off_to_off(s_index, start),
@@ -1078,7 +1096,7 @@ static int pcpu_find_block_fit(struct pcpu_chunk *chunk, int alloc_bits,
 	if (bit_off + alloc_bits > chunk_md->contig_hint)
 		return -1;
 
-	bit_off = chunk_md->first_free;
+	bit_off = pcpu_next_hint(chunk_md, alloc_bits);
 	bits = 0;
 	pcpu_for_each_fit_region(chunk, alloc_bits, align, bit_off, bits) {
 		if (!pop_only || pcpu_is_populated(chunk, bit_off, bits,
-- 
2.17.1


  parent reply	other threads:[~2019-02-28  2:19 UTC|newest]

Thread overview: 35+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-02-28  2:18 [PATCH 00/12] introduce percpu block scan_hint Dennis Zhou
2019-02-28  2:18 ` [PATCH 01/12] percpu: update free path with correct new free region Dennis Zhou
2019-03-02 12:56   ` Peng Fan
2019-02-28  2:18 ` [PATCH 02/12] percpu: do not search past bitmap when allocating an area Dennis Zhou
2019-03-02 13:32   ` Peng Fan
2019-03-02 22:23     ` Dennis Zhou
2019-03-03  8:41       ` Peng Fan
2019-02-28  2:18 ` [PATCH 03/12] percpu: introduce helper to determine if two regions overlap Dennis Zhou
2019-03-02 13:37   ` Peng Fan
2019-03-02 22:24     ` Dennis Zhou
2019-02-28  2:18 ` [PATCH 04/12] percpu: manage chunks based on contig_bits instead of free_bytes Dennis Zhou
2019-03-02 13:48   ` Peng Fan
2019-03-02 22:32     ` Dennis Zhou
2019-03-03  8:42       ` Peng Fan
2019-02-28  2:18 ` [PATCH 05/12] percpu: relegate chunks unusable when failing small allocations Dennis Zhou
2019-03-02 13:55   ` Peng Fan
2019-03-02 22:34     ` Dennis Zhou
2019-02-28  2:18 ` [PATCH 06/12] percpu: set PCPU_BITMAP_BLOCK_SIZE to PAGE_SIZE Dennis Zhou
2019-03-03  4:56   ` Peng Fan
2019-02-28  2:18 ` [PATCH 07/12] percpu: add block level scan_hint Dennis Zhou
2019-03-03  6:01   ` Peng Fan
2019-03-03 20:23     ` Dennis Zhou
2019-03-04  9:36       ` Peng Fan
2019-02-28  2:18 ` [PATCH 08/12] percpu: remember largest area skipped during allocation Dennis Zhou
2019-02-28  2:18 ` [PATCH 09/12] percpu: use block scan_hint to only scan forward Dennis Zhou
2019-02-28  2:18 ` [PATCH 10/12] percpu: make pcpu_block_md generic Dennis Zhou
2019-03-03  6:35   ` Peng Fan
2019-02-28  2:18 ` [PATCH 11/12] percpu: convert chunk hints to be based on pcpu_block_md Dennis Zhou
2019-03-03  8:18   ` Peng Fan
2019-03-03 20:22     ` Dennis Zhou
2019-03-04  6:36       ` Peng Fan
2019-02-28  2:18 ` Dennis Zhou [this message]
2019-03-03  8:38   ` [PATCH 12/12] percpu: use chunk scan_hint to skip some scanning Peng Fan
2019-02-28 14:47 ` [PATCH 00/12] introduce percpu block scan_hint Vlad Buslov
2019-03-13 20:19 ` Dennis Zhou

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=20190228021839.55779-13-dennis@kernel.org \
    --to=dennis@kernel.org \
    --cc=cl@linux.com \
    --cc=kernel-team@fb.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=tj@kernel.org \
    --cc=vladbu@mellanox.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).