| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <d221e784d1f84d1880cf88e556e9ff3a@huawei.com>
Date: Wed, 16 Nov 2022 09:25:02 +0000
From: zhaogongyi <zhaogongyi@...wei.com>
To: "linux-kselftest@...r.kernel.org" <linux-kselftest@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"linux-fsdevel@...r.kernel.org" <linux-fsdevel@...r.kernel.org>
CC: "keescook@...omium.org" <keescook@...omium.org>,
"luto@...capital.net" <luto@...capital.net>,
"wad@...omium.org" <wad@...omium.org>,
"shuah@...nel.org" <shuah@...nel.org>,
"cristian.marussi@....com" <cristian.marussi@....com>,
Matthew Wilcox <willy@...radead.org>
Subject: Re: [PATCH -next 2/2] selftests/mincore: Optimize
TEST(check_file_mmap) accoring to filemap read around
Hi!
>
> TEST(check_file_mmap) will fail when we set the ra_pages through
> blockdev like:
> /sbin/blockdev --setra 8196 /dev/sda
>
> And, for file mmap, kernel will read 'ra_pages/2' pages of the address on
> the left hand or right hand. So, add a checking of ra_pages according to
> the ra_pages setting of the block dev.
>
> I have tested it on my system like:
> i=0
> while [ $i -lt 9000 ]
> do
> /sbin/blockdev --setra $i /dev/openeuler/*
> ./mincore_selftest || { echo "$i test failed"; exit 1; }
> let i=$i+1
> done
>
> Signed-off-by: Zhao Gongyi <zhaogongyi@...wei.com>
> ---
> tools/testing/selftests/kselftest_harness.h | 2 +
> .../selftests/mincore/mincore_selftest.c | 74 +++++++++++++++----
> 2 files changed, 62 insertions(+), 14 deletions(-)
>
> diff --git a/tools/testing/selftests/kselftest_harness.h
> b/tools/testing/selftests/kselftest_harness.h
> index 25f4d54067c0..fa40c85a1099 100644
> --- a/tools/testing/selftests/kselftest_harness.h
> +++ b/tools/testing/selftests/kselftest_harness.h
> @@ -781,6 +781,8 @@
> } \
> }
>
> +#define MIN(a, b) ((a) < (b) ? (a) : (b))
> +
> struct __test_results {
> char reason[1024]; /* Reason for test result */
> };
> diff --git a/tools/testing/selftests/mincore/mincore_selftest.c
> b/tools/testing/selftests/mincore/mincore_selftest.c
> index 287351a599a2..5eace1750fef 100644
> --- a/tools/testing/selftests/mincore/mincore_selftest.c
> +++ b/tools/testing/selftests/mincore/mincore_selftest.c
> @@ -17,6 +17,7 @@
> #include <mntent.h>
> #include <sys/stat.h>
> #include <linux/fs.h>
> +#include <sys/ioctl.h>
>
> #include "../kselftest.h"
> #include "../kselftest_harness.h"
> @@ -218,11 +219,7 @@ static struct mntent* find_mount_point(const
> char *name)
> * Test mincore() behavior on a file-backed page.
> * No pages should be loaded into memory right after the mapping. Then,
> * accessing any address in the mapping range should load the page
> - * containing the address and a number of subsequent pages
> (readahead).
> - *
> - * The actual readahead settings depend on the test environment, so we
> - * can't make a lot of assumptions about that. This test covers the most
> - * general cases.
> + * containing the address and a number of around pages (read around).
> */
> TEST(check_file_mmap)
> {
> @@ -236,6 +233,10 @@ TEST(check_file_mmap)
> int ra_pages = 0;
> struct stat s;
> struct mntent *mount_entry;
> + int ra_size;
> + int exp_ra_pages;
> + int odd = 0;
> + int dev_ra_pages;
>
> mount_entry = find_mount_point(".");
> ASSERT_NE(NULL, mount_entry) {
> @@ -248,11 +249,29 @@ TEST(check_file_mmap)
> "test is not supported");
> }
>
> + errno = 0;
> + fd = open(mount_entry->mnt_fsname, O_RDONLY);
> + ASSERT_LT(0, fd) {
> + TH_LOG("Open %s failed: %s",
> + mount_entry->mnt_fsname, strerror(errno));
> + }
> +
> + errno = 0;
> + retval = ioctl(fd, BLKRAGET, &ra_size);
> + ASSERT_EQ(0, retval) {
> + TH_LOG("Get block readahead size failed: %s", strerror(errno));
> + }
> +
> page_size = sysconf(_SC_PAGESIZE);
> vec_size = FILE_SIZE / page_size;
> if (FILE_SIZE % page_size)
> vec_size++;
>
> + dev_ra_pages = (ra_size * 512) / page_size;
> + exp_ra_pages = MIN(vec_size / 2, dev_ra_pages / 2);
> + if (dev_ra_pages <= vec_size)
> + odd = dev_ra_pages % 2;
> +
> vec = calloc(vec_size, sizeof(unsigned char));
> ASSERT_NE(NULL, vec) {
> TH_LOG("Can't allocate array");
> @@ -296,7 +315,7 @@ TEST(check_file_mmap)
>
> /*
> * Touch a page in the middle of the mapping. We expect the next
> - * few pages (the readahead window) to be populated too.
> + * few pages (the read around window) to be populated too.
> */
> addr[FILE_SIZE / 2] = 1;
> retval = mincore(addr, FILE_SIZE, vec); @@ -310,22 +329,49 @@
> TEST(check_file_mmap)
> ra_pages++;
> i++;
> }
> - EXPECT_GT(ra_pages, 0) {
> - TH_LOG("No read-ahead pages found in memory");
> - }
>
> - EXPECT_LT(i, vec_size) {
> - TH_LOG("Read-ahead pages reached the end of the file");
> + if (!exp_ra_pages) {
> + EXPECT_EQ(ra_pages, exp_ra_pages) {
> + TH_LOG("Check read-around pages failed");
> + }
> + } else {
> + EXPECT_EQ(ra_pages + (odd ? 0 : 1), exp_ra_pages) {
> + TH_LOG("Check read-around pages failed");
> + }
> }
> +
> /*
> - * End of the readahead window. The rest of the pages shouldn't
> - * be in memory.
> + * End of the read around window. The rest of the pages shouldn't
> + * be in memory for the right hand.
> */
> if (i < vec_size) {
> while (i < vec_size && !vec[i])
> i++;
> EXPECT_EQ(vec_size, i) {
> - TH_LOG("Unexpected page in memory beyond readahead
> window");
> + TH_LOG("Unexpected page in memory beyond read around
> window");
> + }
> + }
> +
> + i = FILE_SIZE / 2 / page_size - 1;
> + ra_pages = 0;
> + while (i >= 0 && vec[i]) {
> + ra_pages++;
> + i--;
> + }
> +
> + EXPECT_EQ(ra_pages, exp_ra_pages) {
> + TH_LOG("Check read-around pages failed");
> + }
> +
> + /*
> + * End of the read around window. The rest of the pages shouldn't
> + * be in memory for the left hand.
> + */
> + if (i >= 0) {
> + while (i >= 0 && !vec[i])
> + i--;
> + EXPECT_EQ(-1, i) {
> + TH_LOG("Unexpected page in memory beyond read around
> window");
> }
> }
>
> --
> 2.17.1
Gentele ping.
Regards,
Gongyi
Powered by blists - more mailing lists