[PATCH 6/7] proc/kcore: optimize multiple page reads

From: Omar Sandoval <osandov@osandov.com>
To: linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
	Alexey Dobriyan <adobriyan@gmail.com>
Cc: Eric Biederman <ebiederm@xmission.com>, kernel-team@fb.com
Subject: [PATCH 6/7] proc/kcore: optimize multiple page reads
Date: Fri,  6 Jul 2018 12:32:37 -0700	[thread overview]
Message-ID: <ed665d06dcb0bf654cff3637161d7b040cf9e4c5.1530904769.git.osandov@fb.com> (raw)
In-Reply-To: <cover.1530904769.git.osandov@fb.com>

From: Omar Sandoval <osandov@fb.com>

The current code does a full search of the segment list every time for
every page. This is wasteful, since it's almost certain that the next
page will be in the same segment. Instead, check if the previous segment
covers the current page before doing the list search.

Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 fs/proc/kcore.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/fs/proc/kcore.c b/fs/proc/kcore.c
index de225b61f34f..f1e21579cd22 100644
--- a/fs/proc/kcore.c
+++ b/fs/proc/kcore.c
@@ -426,10 +426,18 @@ read_kcore(struct file *file, char __user *buffer, size_t buflen, loff_t *fpos)
 	if ((tsz = (PAGE_SIZE - (start & ~PAGE_MASK))) > buflen)
 		tsz = buflen;
 
+	m = NULL;
 	while (buflen) {
-		list_for_each_entry(m, &kclist_head, list) {
-			if (start >= m->addr && start < (m->addr+m->size))
-				break;
+		/*
+		 * If this is the first iteration or the address is not within
+		 * the previous entry, search for a matching entry.
+		 */
+		if (!m || start < m->addr || start >= m->addr + m->size) {
+			list_for_each_entry(m, &kclist_head, list) {
+				if (start >= m->addr &&
+				    start < m->addr + m->size)
+					break;
+			}
 		}
 
 		if (&m->list == &kclist_head) {
-- 
2.18.0

