From: Hui Zhu <teawater@gmail.com>
To: fengguang.wu@qq.com, linux-kernel@vger.kernel.org
Cc: Hui Zhu <teawater@gmail.com>, Hui Zhu <teawaterz@linux.alibaba.com>
Subject: [PATCH for vm-scalability v2] usemem: Add new option --punch-holes for generating fragmented pages
Date: Thu, 5 Mar 2020 10:36:04 +0800 [thread overview]
Message-ID: <1583375764-20761-1-git-send-email-teawater@gmail.com> (raw)
In-Reply-To: <20200304073016.GB19046@wfg-e595>
Hi Fengguang,
Thanks for your review. This is the new version that was updated
according to you comments.
This commit adds new option --punch-holes. usemem will free every
other page after allocation. Then it will generate size/2/pagesize
fragmented pages with this option.
Its implementation is to use madvise to release a page every other page.
For example:
usemem --punch-holes -s -1 400m
Ideally, this command will generate 200m fragmented pages in the system.
This command can help test anti-fragmentation function and other features
that are affected by fragmentation issues of the Linux kernel.
Signed-off-by: Hui Zhu <teawaterz@linux.alibaba.com>
---
usemem.c | 46 +++++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 45 insertions(+), 1 deletion(-)
diff --git a/usemem.c b/usemem.c
index 9ac36e4..e0f9991 100644
--- a/usemem.c
+++ b/usemem.c
@@ -95,6 +95,7 @@ int opt_sync_free = 0;
int opt_bind_interval = 0;
unsigned long opt_delay = 0;
int opt_read_again = 0;
+int opt_punch_holes = 0;
int nr_task;
int nr_thread;
int nr_cpu;
@@ -153,6 +154,7 @@ void usage(int ok)
" -O|--anonymous mmap with MAP_ANONYMOUS\n"
" -U|--hugetlb allocate hugetlbfs page\n"
" -Z|--read-again read memory again after access the memory\n"
+ " --punch-holes free every other page after allocation\n"
" -h|--help show this message\n"
, ourname);
@@ -191,6 +193,7 @@ static const struct option opts[] = {
{ "delay" , 1, NULL, 'e' },
{ "hugetlb" , 0, NULL, 'U' },
{ "read-again" , 0, NULL, 'Z' },
+ { "punch-holes", 0, NULL, 0 },
{ "help" , 0, NULL, 'h' },
{ NULL , 0, NULL, 0 }
};
@@ -655,6 +658,21 @@ static void timing_free(void *ptrs[], unsigned int nptr,
static void wait_for_sigusr1(int signal) {}
+static void do_punch_holes(void *addr, unsigned long len)
+{
+ unsigned long offset;
+
+ for (offset = 0; offset + 2 * pagesize <= len; offset += 2 * pagesize) {
+ if (madvise(addr + offset, pagesize,
+ MADV_DONTNEED) != 0) {
+ fprintf(stderr,
+ "madvise failed with error %s\n",
+ strerror(errno));
+ exit(1);
+ }
+ }
+}
+
long do_units(void)
{
struct drand48_data rand_data;
@@ -752,6 +770,15 @@ long do_units(void)
}
}
+ if (opt_punch_holes) {
+ if (prealloc)
+ do_punch_holes(prealloc, opt_bytes);
+ else {
+ for (i = 0; i < nptr; i++)
+ do_punch_holes(ptrs[i], lens[i]);
+ }
+ }
+
while (sleep_secs)
sleep_secs = sleep(sleep_secs);
@@ -896,6 +923,7 @@ int do_tasks(void)
int main(int argc, char *argv[])
{
int c;
+ int opt_index = 0;
#ifdef DBG
/* print the command line parameters passed on to main */
@@ -910,9 +938,18 @@ int main(int argc, char *argv[])
pagesize = getpagesize();
while ((c = getopt_long(argc, argv,
- "aAB:f:FPp:gqowRMm:n:t:b:ds:T:Sr:u:j:e:EHDNLWyxOUZh", opts, NULL)) != -1)
+ "aAB:f:FPp:gqowRMm:n:t:b:ds:T:Sr:u:j:e:EHDNLWyxOUZh",
+ opts, &opt_index)) != -1)
{
switch (c) {
+ case 0:
+ if (strcmp(opts[opt_index].name,
+ "punch-holes") == 0) {
+ opt_punch_holes = 1;
+ } else
+ usage(1);
+ break;
+
case 'a':
opt_malloc++;
break;
@@ -1045,6 +1082,13 @@ int main(int argc, char *argv[])
}
}
+ if (opt_punch_holes && opt_malloc) {
+ fprintf(stderr,
+ "%s: malloc options ignored for punch-holes\n",
+ ourname);
+ opt_malloc = 0;
+ }
+
if (opt_malloc) {
if (map_populate|map_anonymous|map_hugetlb)
fprintf(stderr,
--
2.7.4
parent reply other threads:[~2020-03-05 2:36 UTC|newest]
Thread overview: expand[flat|nested] mbox.gz Atom feed
[parent not found: <20200304073016.GB19046@wfg-e595>]
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=1583375764-20761-1-git-send-email-teawater@gmail.com \
--to=teawater@gmail.com \
--cc=fengguang.wu@qq.com \
--cc=linux-kernel@vger.kernel.org \
--cc=teawaterz@linux.alibaba.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).