From: "Stephen C. Tweedie" <sct@redhat.com>
To: Andrea Arcangeli <andrea@suse.de>
Cc: linux-kernel <linux-kernel@vger.kernel.org>,
linux-mm <linux-mm@kvack.org>, Andrew Morton <akpm@digeo.com>,
Stephen Tweedie <sct@redhat.com>
Subject: Re: [PATCH] Fix for vma merging refcounting bug
Date: 11 May 2003 21:04:06 +0100 [thread overview]
Message-ID: <1052683446.4609.29.camel@sisko.scot.redhat.com> (raw)
In-Reply-To: <20030510163336.GB15010@dualathlon.random>
[-- Attachment #1: Type: text/plain, Size: 1001 bytes --]
Hi,
On Sat, 2003-05-10 at 17:33, Andrea Arcangeli wrote:
> On Fri, May 09, 2003 at 01:34:21PM +0100, Stephen C. Tweedie wrote:
> > When a new vma can be merged simultaneously with its two immediate
> > neighbours in both directions, vma_merge() extends the predecessor vma
> > and deletes the successor. However, if the vma maps a file, it fails to
> > fput() when doing the delete, leaving the file's refcount inconsistent.
> great catch! nobody could notice it in practice
Yep --- I only noticed it because I was running a quick-and-dirty vma
merging test and wanted to test on a shmfs file, and noticed that the
temporary shmfs filesystem became unmountable afterwards. Test
attached, in case anybody is interested (it's the third test, mapping a
file page by page in two interleaved passes, which triggers this case.)
> I'm attaching for review what I'm applying to my -aa tree, to fix the
> above and the other issue with the non-ram vma merging fixed in 2.5.
Looks OK.
Cheers,
Stephen
[-- Attachment #2: vma-merge.c --]
[-- Type: text/x-c, Size: 2544 bytes --]
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <unistd.h>
#include <fcntl.h>
#include <sys/mman.h>
#include <sys/ipc.h>
#include <sys/shm.h>
static char *testfile = "/tmp/vma-test.dat";
#define TEST_PAGES 1024
int pagesize;
int filesize;
char *map_addr;
void DIE(char *) __attribute__ ((noreturn));
void DIE(char *why)
{
perror(why);
exit(1);
}
#define plural(n) (((n) == 1) ? "" : "s")
void test_maps(char *which)
{
int fd;
int rc;
int count = 0;
char buffer[256];
char filename[128];
FILE *mapfile;
fd = open("/proc/self/maps", O_RDONLY);
if (fd < 0)
DIE("open(/proc/self/maps");
mapfile = fopen("/proc/self/maps", "r");
while (1) {
if (!fgets(buffer, 256, mapfile))
break;
rc = sscanf(buffer,
"%*x-%*x %*4s %*x %*5s %*d %127s\n",
filename);
if (!rc)
continue;
if (!strcmp(testfile, filename))
count++;
}
printf("Testing %s: found %d map%s\n", which, count, plural(count));
}
#define clear_maps() \
err = munmap(map_addr, filesize); \
if (err) \
DIE("munmap"); \
static void map_page(int fd, int i)
{
char *ptr;
ptr = mmap(map_addr + i * pagesize,
pagesize,
PROT_READ,
MAP_SHARED | MAP_FIXED,
fd,
i * pagesize);
if (ptr == MAP_FAILED)
DIE("mmap");
if (ptr != map_addr + i * pagesize) {
fprintf(stderr, "mmap returned unexpected address\n");
exit(1);
}
}
int main(int argc, char *argv[])
{
int fd;
int err;
int i;
if (argc > 1)
testfile = argv[1];
pagesize = getpagesize();
filesize = TEST_PAGES * pagesize;
fd = open(testfile, O_CREAT|O_TRUNC|O_RDWR, 0666);
if (fd < 0)
DIE("open");
err = ftruncate(fd, filesize);
if (err)
DIE("ftuncate");
/* Find a suitable mmap address for the entire file */
map_addr = mmap(0, filesize, PROT_READ, MAP_SHARED, fd, 0);
if (map_addr == MAP_FAILED)
DIE("mmap");
clear_maps();
/* Now map it in piece by piece */
for (i = 0; i < TEST_PAGES; i++)
map_page(fd, i);
test_maps("backwards merging");
clear_maps();
/* Next, map it in backwards */
for (i = TEST_PAGES; i-- > 0; )
map_page(fd, i);
test_maps("forwards merging");
clear_maps();
/* Finally, map it in in two interleaved passes */
for (i = 0; i < TEST_PAGES; i+=2)
map_page(fd, i);
for (i = 1; i < TEST_PAGES; i+=2)
map_page(fd, i);
test_maps("interleaved merging");
close(fd);
unlink(testfile);
return 0;
}
next prev parent reply other threads:[~2003-05-11 20:04 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2003-05-09 12:34 Stephen C. Tweedie
2003-05-10 16:33 ` Andrea Arcangeli
2003-05-11 20:04 ` Stephen C. Tweedie [this message]
2003-05-13 22:52 ` Andrea Arcangeli
2003-05-13 23:05 ` Andrea Arcangeli
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=1052683446.4609.29.camel@sisko.scot.redhat.com \
--to=sct@redhat.com \
--cc=akpm@digeo.com \
--cc=andrea@suse.de \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
/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