From: Amir Goldstein <amir73il@gmail.com>
To: Cyril Hrubis <chrubis@suse.cz>
Cc: ltp@lists.linux.it, Jan Stancek <jstancek@redhat.com>,
Miklos Szeredi <miklos@szeredi.hu>,
linux-unionfs@vger.kernel.org
Subject: [PATCH v3 5/6] syscalls/readahead02: test readahead using posix_fadvise()
Date: Thu, 11 Oct 2018 02:34:40 +0300 [thread overview]
Message-ID: <20181010233441.5337-6-amir73il@gmail.com> (raw)
In-Reply-To: <20181010233441.5337-1-amir73il@gmail.com>
The call to posix_fadvise(POSIX_FADV_WILLNEED) should have the same
effect as the call to readahead() syscall.
Repeat the test cases for local file and overlayfs file with
posix_fadvise().
The new test case is a regression test for kernel commit b833a3660394
("ovl: add ovl_fadvise()") which fixes a regression of fadvise() on
an overlay file that was introduced by kernel commit 5b910bd615ba
("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
.../kernel/syscalls/readahead/readahead02.c | 41 +++++++++++++++----
1 file changed, 34 insertions(+), 7 deletions(-)
diff --git a/testcases/kernel/syscalls/readahead/readahead02.c b/testcases/kernel/syscalls/readahead/readahead02.c
index 56a8496ff..d5cc7677b 100644
--- a/testcases/kernel/syscalls/readahead/readahead02.c
+++ b/testcases/kernel/syscalls/readahead/readahead02.c
@@ -55,24 +55,49 @@ static struct tst_option options[] = {
static struct tcase {
const char *tname;
int use_overlay;
+ int use_fadvise;
} tcases[] = {
- { "readahead on file", 0 },
- { "readahead on overlayfs file", 1 },
+ { "readahead on file", 0, 0 },
+ { "readahead on overlayfs file", 1, 0 },
+ { "POSIX_FADV_WILLNEED on file", 0, 1 },
+ { "POSIX_FADV_WILLNEED on overlayfs file", 1, 1 },
};
+static int fadvise_willneed(int fd, off_t offset, size_t len)
+{
+ /* Should have the same effect as readahead() syscall */
+ return posix_fadvise(fd, offset, len, POSIX_FADV_WILLNEED);
+}
+
+static int libc_readahead(int fd, off_t offset, size_t len)
+{
+ return readahead(fd, offset, len);
+}
+
+typedef int (*readahead_func_t)(int, off_t, size_t);
+static readahead_func_t readahead_func = libc_readahead;
+
static int readahead_supported = 1;
+static int fadvise_supported = 1;
static int check_ret(struct tcase *tc)
{
if (TST_RET == 0)
return 0;
- if (TST_ERR == EINVAL &&
- (!tc->use_overlay || !readahead_supported)) {
+ /* posix_fadvise returns error number (not in errno) */
+ if (tc->use_fadvise && (TST_ERR = TST_RET) == EINVAL &&
+ (!tc->use_overlay || !fadvise_supported)) {
+ fadvise_supported = 0;
+ tst_res(TCONF, "fadvise not supported on %s",
+ tst_device->fs_type);
+ } else if (TST_ERR == EINVAL &&
+ (!tc->use_overlay || !readahead_supported)) {
readahead_supported = 0;
tst_res(TCONF, "readahead not supported on %s",
tst_device->fs_type);
} else {
- tst_res(TFAIL | TTERRNO, "readahead failed on %s",
+ tst_res(TFAIL | TTERRNO, "%s failed on %s",
+ tc->use_fadvise ? "fadvise" : "readahead",
tc->use_overlay ? "overlayfs" : tst_device->fs_type);
}
return 1;
@@ -152,7 +177,6 @@ static void create_testfile(int use_overlay)
free(tmp);
}
-
/* read_testfile - mmap testfile and read every page.
* This functions measures how many I/O and time it takes to fully
* read contents of test file.
@@ -182,7 +206,7 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
if (do_readahead) {
cached_start = get_cached_size();
do {
- TEST(readahead(fd, offset, fsize - offset));
+ TEST(readahead_func(fd, offset, fsize - offset));
if (TST_RET != 0) {
SAFE_CLOSE(fd);
return;
@@ -264,6 +288,9 @@ static void test_readahead(unsigned int n)
create_testfile(tc->use_overlay);
+ /* Use either readahead() syscall or POSIX_FADV_WILLNEED */
+ readahead_func = tc->use_fadvise ? fadvise_willneed : libc_readahead;
+
/* find out how much can cache hold if we read whole file */
read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
cached_high = get_cached_size();
--
2.17.1
next prev parent reply other threads:[~2018-10-10 23:34 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-10-10 23:34 [PATCH v3 0/6] Tests for readahead() and fadvise() on overlayfs Amir Goldstein
2018-10-10 23:34 ` [PATCH v3 1/6] syscalls/readahead01: Convert to newlib Amir Goldstein
2018-11-02 15:34 ` Cyril Hrubis
2018-10-10 23:34 ` [PATCH v3 2/6] syscalls/readahead02: Convert to newlib and cleanup Amir Goldstein
2018-11-20 16:07 ` Cyril Hrubis
2018-11-20 16:42 ` [LTP] " Cyril Hrubis
2018-11-28 16:50 ` Amir Goldstein
2018-10-10 23:34 ` [PATCH v3 3/6] syscalls/readahead02: abort test if readahead syscall fails Amir Goldstein
2018-11-20 16:30 ` Cyril Hrubis
2018-11-28 9:52 ` Amir Goldstein
2018-10-10 23:34 ` [PATCH v3 4/6] syscalls/readahead02: test readahead() on an overlayfs file Amir Goldstein
2018-10-10 23:34 ` Amir Goldstein [this message]
2018-10-10 23:34 ` [PATCH v3 6/6] syscalls/readahead02: fail test if readahead did not use any cache Amir Goldstein
2018-11-02 10:32 ` [PATCH v3 0/6] Tests for readahead() and fadvise() on overlayfs Amir Goldstein
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=20181010233441.5337-6-amir73il@gmail.com \
--to=amir73il@gmail.com \
--cc=chrubis@suse.cz \
--cc=jstancek@redhat.com \
--cc=linux-unionfs@vger.kernel.org \
--cc=ltp@lists.linux.it \
--cc=miklos@szeredi.hu \
/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).