The test helps to validate clamping and mount behaviors according to supported file system timestamp ranges.
Note that the test can fail on 32-bit systems for a few file systems. This will be corrected when vfs is transitioned to use 64-bit timestamps.
Signed-off-by: Deepa Dinamani deepa.kernel@gmail.com --- common/attr | 27 ++++++ src/Makefile | 2 +- src/y2038_futimens.c | 61 +++++++++++++ tests/generic/390 | 238 ++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/390.out | 2 + tests/generic/group | 1 + 6 files changed, 330 insertions(+), 1 deletion(-) create mode 100644 src/y2038_futimens.c create mode 100755 tests/generic/390 create mode 100644 tests/generic/390.out
diff --git a/common/attr b/common/attr index ce2d76a..579dc9b 100644 --- a/common/attr +++ b/common/attr @@ -56,6 +56,33 @@ _acl_get_max() esac }
+_filesystem_timestamp_range() +{ + device=${1:-$TEST_DEV} + case $FSTYP in + ext4) #dumpe2fs + if [ $(dumpe2fs -h $device 2>/dev/null | grep "Inode size:" | cut -d: -f2) -gt 128 ]; then + echo "-2147483648 15032385535" + else + echo "-2147483648 2147483647" + fi + ;; + + xfs) + echo "-2147483648 2147483647" + ;; + jfs) + echo "0 4294967295" + ;; + f2fs) + echo "-2147483648 2147483647" + ;; + *) + echo "-1 -1" + ;; + esac +} + _require_acl_get_max() { if [ $(_acl_get_max) -eq 0 ]; then diff --git a/src/Makefile b/src/Makefile index dd51216..0b99ae4 100644 --- a/src/Makefile +++ b/src/Makefile @@ -21,7 +21,7 @@ LINUX_TARGETS = xfsctl bstat t_mtab getdevicesize preallo_rw_pattern_reader \ stale_handle pwrite_mmap_blocked t_dir_offset2 seek_sanity_test \ seek_copy_test t_readdir_1 t_readdir_2 fsync-tester nsexec cloner \ renameat2 t_getcwd e4compact test-nextquota punch-alternating \ - attr-list-by-handle-cursor-test listxattr + attr-list-by-handle-cursor-test listxattr y2038_futimens
SUBDIRS =
diff --git a/src/y2038_futimens.c b/src/y2038_futimens.c new file mode 100644 index 0000000..291e4fa --- /dev/null +++ b/src/y2038_futimens.c @@ -0,0 +1,61 @@ +#include <fcntl.h> +#include <stdio.h> +#include <sys/stat.h> +#include <errno.h> +#include <stdlib.h> + +int +do_utime(int fd, long long time) +{ + struct timespec t[2]; + + /* + * Convert long long to timespec format. + * Seconds precision is assumed here. + */ + t[0].tv_sec = time; + t[0].tv_nsec = 0; + t[1].tv_sec = time; + t[1].tv_nsec = 0; + + /* Call utimens to update time. */ + if (futimens(fd, t)) { + perror("futimens"); + return 1; + } + + return 0; +} + +int +main(int argc, char **argv) +{ + int fd; + long long time; + + if(argc < 3) { + fprintf(stderr, "Usage: %s filename timestamp\n" + "Filename: file to be created or opened in current directory\n" + "Timestamp: is seconds since 1970-01-01 00:00:00 UTC\n", argv[0]); + exit(1); + } + + /* Create the file */ + fd = creat(argv[1], 0666); + if(fd < 0) { + perror("creat"); + exit(1); + } + + /* Get the timestamp */ + time = strtoull(argv[2], NULL, 0); + if (errno) { + perror("strtoull"); + exit(1); + } + + if (do_utime(fd, time)) + return 1; + + return 0; +} diff --git a/tests/generic/390 b/tests/generic/390 new file mode 100755 index 0000000..f069988 --- /dev/null +++ b/tests/generic/390 @@ -0,0 +1,238 @@ +#! /bin/bash +# FS QA Test No. generic/390 +# +# Tests to verify policy for filesystem timestamps for +# supported ranges: +# 1. Verify filesystem rw mount according to sysctl +# timestamp_supported. +# 2. Verify timestamp clamping for timestamps beyond max +# timestamp supported. +# +# Exit status 1: either or both tests above fail. +# Exit status 0: both the above tests pass. + +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" +#echo "output in $seqres.full" +here=`pwd` + +# Get standard environment, filters and checks. +. ./common/rc +. ./common/filter +. ./common/attr + +SRC_GROUPS=`find tests -not -path tests -type d -printf "%f "` + +# Prerequisites for the test run. +_supported_fs generic +_supported_os Linux +_require_scratch + +Y2038_PROG=$here/src/y2038_futimens + +#initialize exit status +status=0 + +# Generic test cleanup function. +_cleanup() +{ + # Remove any leftover files from last run. + rm -f ${SCRATCH_MNT}/test_? +} + +#unmount and mount $SCRATCH_DEV. +_umount_mount_scratch_dev() +{ + #change directory so that you are not using SCRATCH_MNT anymore. + cd $here + + # Unmount the ${SCRATCH_DEV} + _scratch_unmount + if [ $? != 0 ]; then + return $? + fi + + # Mount the ${SCRATCH_DEV} + _scratch_mount + if [ $? != 0 ]; then + return $? + fi + + cd ${SCRATCH_MNT} +} + +# Compare file timestamps obtained from stat +# with a given timestamp. +_check_stat() #check_stat(file, timestamp) +{ + stat_timestamp=`stat -c"%X;%Y" $1` + + prev_timestamp="$2;$2" + if [ $prev_timestamp != $stat_timestamp ]; then + echo "$prev_timestamp != $stat_timestamp" >> $seqres.full + return 1 + else + return 0 + fi +} + +_run_test_individual() #_run_individual_test(file, timestamp, update_time) +{ + file=$1 + timestamp=$2 + update_time=$3 + + #check if the time needs update + if [ $update_time -eq 1 ]; then + echo "Updating file: $file to timestamp `date -d @$timestamp`" >> $seqres.full + $Y2038_PROG $file $timestamp &>> $seqres.full + if [ $? != 0 ]; then + echo "Failed to run the y2038 program" >> $seqres.full + return 1 + fi + fi + + tsclamp=$(($timestamp>$tsmax?$tsmax:$timestamp)) + _check_stat $file $tsclamp + echo "Checking file: $file Updated timestamp is `date -d @$tsclamp`" >> $seqres.full + + if [ $? != 0 ]; then + echo "Failed to set time to $timestamp" >> $seqres.full + return 1 + fi + + return 0 +} + +_run_test() #_run_test(update_time) +{ + #initialization iterator + n=1 + + for TIME in "${TIMESTAMPS[@]}" + do + #Run the test + _run_test_individual ${SCRATCH_MNT}/test_$n $TIME $1 + + if [ $? != 0 ]; then + echo "file timestamp update failed `date -d @$TIME`" >> $seqres.full + if [ $n -lt 4 ]; then + echo "Test failed" + return 1 + fi + fi + + #update iterator + ((n++)) + done + + return 0 +} + +#Remove log from last run +rm -f $seqres.full + +#install cleaner +trap "_cleanup; exit $status" 0 1 2 3 15 + +_scratch_mkfs &>> $seqres.full 2>&1 || _fail "mkfs failed" +read tsmin tsmax <<<$(_filesystem_timestamp_range $SCRATCH_DEV) + +if [ $tsmin -eq -1 -a $tsmax -eq -1 ]; then + _notrun "filesystem $FSTYP timestamp bounds are unknown" +fi + +echo min supported timestamp $tsmin $(date --date=@$tsmin) >> $seqres.full +echo max supported timestamp $tsmax $(date --date=@$tsmax) >> $seqres.full + +#Test timestamps array + +declare -a TIMESTAMPS=( + $tsmin + 0 + $tsmax + $((tsmax+1)) + 4294967295 + 8589934591 + 34359738367 +) + +# Max timestamp is hardcoded to Mon Jan 18 19:14:07 PST 2038 +sys_tsmax=2147483647 +echo "min timestamp that needs to be supported by fs for rw mount is $sys_tsmax $(date --date=@$sys_tsmax)" >> $seqres.full + +read ts_check <<<$(cat /proc/sys/fs/fs-timestamp-check-on) + +_scratch_mount +result=$? + +if [ $ts_check != 0 ]; then + echo "sysctl filesystem timestamp check is on" >> $seqres.full + if [ $sys_tsmax > $tsmax ]; then + if [ $result != -1 ]; then + echo "mount test failed" >> $seqres.full + fi + else + if [ $result != 0 ]; then + echo "mount test failed" >> $seqres.full + fi + fi +else + echo "sysctl filesystem timestamp check is off" >> $seqres.full + if [ $result != 0 ]; then + echo "mount test failed" >> $seqres.full + fi +fi + +#cd to the SCRATCH_MNT to run the tests +cd $SCRATCH_MNT + +# Begin test case 1 +echo "In memory timestamps update test start" >> $seqres.full + +#update time on the file +update_time=1 + +#Run test +_run_test $update_time + +if [ $? != 0 ]; then + echo "In memory timestamps update failed" >> $seqres.full + status=1 + exit +fi + +echo "In memory timestamps update complete" >> $seqres.full + +echo "Unmounting and mounting scratch $SCRATCH_MNT" >> $seqres.full + +#unmount and remount $SCRATCH_DEV +_umount_mount_scratch_dev + +if [ $? != 0 ];then + status=1 + exit +fi + +# Begin test case 2 + +#re-initialize iterator +n=1 + +#Do not update time on the file, just read from disk +update_time=0 + +echo "On disk timestamps update test start" >> $seqres.full + +#Re-run test +_run_test $update_time + +if [ $? != 0 ];then + status=1 + exit +fi + +echo "On disk timestamps update test complete" >> $seqres.full + +echo "y2038 inode filestamp update successful" diff --git a/tests/generic/390.out b/tests/generic/390.out new file mode 100644 index 0000000..355f28e --- /dev/null +++ b/tests/generic/390.out @@ -0,0 +1,2 @@ +QA output created by 390 +y2038 inode filestamp update successful diff --git a/tests/generic/group b/tests/generic/group index 08007d7..d137d01 100644 --- a/tests/generic/group +++ b/tests/generic/group @@ -392,3 +392,4 @@ 387 auto clone 388 auto log metadata 389 auto quick acl +390 auto quick rw
On Wed, Nov 23, 2016 at 04:52:02PM -0800, Deepa Dinamani wrote:
The test helps to validate clamping and mount behaviors according to supported file system timestamp ranges.
Note that the test can fail on 32-bit systems for a few file systems. This will be corrected when vfs is transitioned to use 64-bit timestamps.
Signed-off-by: Deepa Dinamani deepa.kernel@gmail.com
common/attr | 27 ++++++ src/Makefile | 2 +- src/y2038_futimens.c | 61 +++++++++++++ tests/generic/390 | 238 ++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/390.out | 2 + tests/generic/group | 1 + 6 files changed, 330 insertions(+), 1 deletion(-) create mode 100644 src/y2038_futimens.c create mode 100755 tests/generic/390 create mode 100644 tests/generic/390.out
diff --git a/common/attr b/common/attr index ce2d76a..579dc9b 100644 --- a/common/attr +++ b/common/attr @@ -56,6 +56,33 @@ _acl_get_max() esac } +_filesystem_timestamp_range() +{
- device=${1:-$TEST_DEV}
- case $FSTYP in
- ext4) #dumpe2fs
if [ $(dumpe2fs -h $device 2>/dev/null | grep "Inode size:" | cut -d: -f2) -gt 128 ]; thenecho "-2147483648 15032385535"elseecho "-2147483648 2147483647"fi
Do ext3 and ext2 follow the same config as ext4?
;;- xfs)
echo "-2147483648 2147483647";;- jfs)
echo "0 4294967295";;- f2fs)
echo "-2147483648 2147483647";;- *)
echo "-1 -1";;- esac
+}
_require_acl_get_max() { if [ $(_acl_get_max) -eq 0 ]; then diff --git a/src/Makefile b/src/Makefile index dd51216..0b99ae4 100644 --- a/src/Makefile +++ b/src/Makefile @@ -21,7 +21,7 @@ LINUX_TARGETS = xfsctl bstat t_mtab getdevicesize preallo_rw_pattern_reader \ stale_handle pwrite_mmap_blocked t_dir_offset2 seek_sanity_test \ seek_copy_test t_readdir_1 t_readdir_2 fsync-tester nsexec cloner \ renameat2 t_getcwd e4compact test-nextquota punch-alternating \
- attr-list-by-handle-cursor-test listxattr
- attr-list-by-handle-cursor-test listxattr y2038_futimens
Need an entry in .gitignore file too.
SUBDIRS = diff --git a/src/y2038_futimens.c b/src/y2038_futimens.c new file mode 100644 index 0000000..291e4fa --- /dev/null +++ b/src/y2038_futimens.c @@ -0,0 +1,61 @@ +#include <fcntl.h> +#include <stdio.h> +#include <sys/stat.h> +#include <errno.h> +#include <stdlib.h>
+int +do_utime(int fd, long long time) +{
- struct timespec t[2];
- /*
* Convert long long to timespec format.* Seconds precision is assumed here.*/- t[0].tv_sec = time;
- t[0].tv_nsec = 0;
- t[1].tv_sec = time;
- t[1].tv_nsec = 0;
- /* Call utimens to update time. */
- if (futimens(fd, t)) {
perror("futimens");return 1;- }
- return 0;
+}
+int +main(int argc, char **argv) +{
- int fd;
- long long time;
- if(argc < 3) {
fprintf(stderr, "Usage: %s filename timestamp\n""Filename: file to be created or opened in current directory\n""Timestamp: is seconds since 1970-01-01 00:00:00 UTC\n", argv[0]);exit(1);
Seems there's no need for an extra level of indention :)
- }
- /* Create the file */
- fd = creat(argv[1], 0666);
- if(fd < 0) {
perror("creat");exit(1);- }
- /* Get the timestamp */
- time = strtoull(argv[2], NULL, 0);
- if (errno) {
From errno(3), errno is never set to zero by any system call or library
function, so errno isn't reset to zero on strtoull and this check always fails for me, with errno = ENOENT, because:
... access("/usr/share/dracut/modules.d/01fips", F_OK) = -1 ENOENT (No such file or directory) ... write(4, "strtoull: No such file or direct"..., 36strtoull: No such file or directory
the errno was from access(2) call in my case.
perror("strtoull");exit(1);- }
- if (do_utime(fd, time))
return 1;- return 0;
+} diff --git a/tests/generic/390 b/tests/generic/390 new file mode 100755 index 0000000..f069988 --- /dev/null +++ b/tests/generic/390 @@ -0,0 +1,238 @@ +#! /bin/bash +# FS QA Test No. generic/390 +# +# Tests to verify policy for filesystem timestamps for +# supported ranges: +# 1. Verify filesystem rw mount according to sysctl +# timestamp_supported. +# 2. Verify timestamp clamping for timestamps beyond max +# timestamp supported. +# +# Exit status 1: either or both tests above fail. +# Exit status 0: both the above tests pass.
Please use "./new" script to generate template, which contains all necessary initial setups and the copyright info.
+seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" +#echo "output in $seqres.full" +here=`pwd`
+# Get standard environment, filters and checks. +. ./common/rc +. ./common/filter +. ./common/attr
+SRC_GROUPS=`find tests -not -path tests -type d -printf "%f "`
What's this for? Seems it's not used in the test.
+# Prerequisites for the test run. +_supported_fs generic +_supported_os Linux +_require_scratch
Need to check the existence of y2038_futimens program, to make sure it's really built, e.g.
_require_test_program "y2038_futimens"
+Y2038_PROG=$here/src/y2038_futimens
+#initialize exit status +status=0
We use 1 as the default value of status, so you can just "exit" on failure (because trap will catch the signal and exit with $status again) and only set status=0 and exit when test passes. "./new" already sets it up for you :)
+# Generic test cleanup function. +_cleanup() +{
- # Remove any leftover files from last run.
- rm -f ${SCRATCH_MNT}/test_?
No need to cleanup files on SCRATCH_DEV, it's meant to be mkfs'ed every time before using it.
+}
+#unmount and mount $SCRATCH_DEV. +_umount_mount_scratch_dev()
There's a helper to do this: _scratch_cycle_mount, only that it doesn't change PWD. And if you don't want to write $SCRATCH_MNT again, this is what we do usually in fstests:
testfile=$SCRATCH_MNT/<somefile> ... do_test $testfile
+{
- #change directory so that you are not using SCRATCH_MNT anymore.
- cd $here
- # Unmount the ${SCRATCH_DEV}
- _scratch_unmount
- if [ $? != 0 ]; then
- return $?
- fi
- # Mount the ${SCRATCH_DEV}
- _scratch_mount
- if [ $? != 0 ]; then
- return $?
- fi
- cd ${SCRATCH_MNT}
+}
+# Compare file timestamps obtained from stat +# with a given timestamp. +_check_stat() #check_stat(file, timestamp)
Name local functions without the leading underscore, that's usually for common helper functions, like functions in common/rc, common/attr.
+{
- stat_timestamp=`stat -c"%X;%Y" $1`
And please use one tab (8 spaces width) for indention.
- prev_timestamp="$2;$2"
- if [ $prev_timestamp != $stat_timestamp ]; then
- echo "$prev_timestamp != $stat_timestamp" >> $seqres.full
- return 1
No need to return 0 or 1, just echo the "... != ..." to stdout, and tee -a the output to $seqres.full if you want verbose debug too.
Because all the outputs, including stdout and stderr, from the test will be captured and compared with the .out file at the end of each test, and test harness reports failure if the two outputs don't match. That way, you don't have to check return value of commands most of the times.
- else
- return 0
- fi
+}
+_run_test_individual() #_run_individual_test(file, timestamp, update_time) +{
- file=$1
- timestamp=$2
- update_time=$3
- #check if the time needs update
- if [ $update_time -eq 1 ]; then
- echo "Updating file: $file to timestamp `date -d @$timestamp`" >> $seqres.full
- $Y2038_PROG $file $timestamp &>> $seqres.full
- if [ $? != 0 ]; then
echo "Failed to run the y2038 program" >> $seqres.fullreturn 1- fi
- fi
Seems $Y2038_PROG outputs nothing on success, so again, just run the program without checking results & redirecting outputs, test harness will report failure if there's error messages from this program. e.g.
$Y2038_PROG $file $timestamp
and just let test run even if something goes wrong.
- tsclamp=$(($timestamp>$tsmax?$tsmax:$timestamp))
- _check_stat $file $tsclamp
- echo "Checking file: $file Updated timestamp is `date -d @$tsclamp`" >> $seqres.full
- if [ $? != 0 ]; then
- echo "Failed to set time to $timestamp" >> $seqres.full
- return 1
- fi
- return 0
+}
+_run_test() #_run_test(update_time) +{
- #initialization iterator
- n=1
- for TIME in "${TIMESTAMPS[@]}"
- do
- #Run the test
- _run_test_individual ${SCRATCH_MNT}/test_$n $TIME $1
- if [ $? != 0 ]; then
echo "file timestamp update failed `date -d @$TIME`" >> $seqres.fullif [ $n -lt 4 ]; then
This is not obvious to me, need some comments on the magic number 4 :)
echo "Test failed"return 1fi- fi
- #update iterator
- ((n++))
- done
- return 0
+}
+#Remove log from last run +rm -f $seqres.full
+#install cleaner +trap "_cleanup; exit $status" 0 1 2 3 15
+_scratch_mkfs &>> $seqres.full 2>&1 || _fail "mkfs failed" +read tsmin tsmax <<<$(_filesystem_timestamp_range $SCRATCH_DEV)
+if [ $tsmin -eq -1 -a $tsmax -eq -1 ]; then
- _notrun "filesystem $FSTYP timestamp bounds are unknown"
+fi
+echo min supported timestamp $tsmin $(date --date=@$tsmin) >> $seqres.full +echo max supported timestamp $tsmax $(date --date=@$tsmax) >> $seqres.full
+#Test timestamps array
+declare -a TIMESTAMPS=(
- $tsmin
- 0
- $tsmax
- $((tsmax+1))
- 4294967295
- 8589934591
- 34359738367
+)
+# Max timestamp is hardcoded to Mon Jan 18 19:14:07 PST 2038 +sys_tsmax=2147483647 +echo "min timestamp that needs to be supported by fs for rw mount is $sys_tsmax $(date --date=@$sys_tsmax)" >> $seqres.full
+read ts_check <<<$(cat /proc/sys/fs/fs-timestamp-check-on)
This fails on systems without this procfs entry, so we need another _require rule, e.g.
_require_proc_attr sys/fs/fs-timestamp-check-on
So test _notrun if kernel has no such procfs entry.
Can you please point me a test kernel tree so I can actually run this test?
+_scratch_mount +result=$?
+if [ $ts_check != 0 ]; then
- echo "sysctl filesystem timestamp check is on" >> $seqres.full
- if [ $sys_tsmax > $tsmax ]; then
- if [ $result != -1 ]; then
echo "mount test failed" >> $seqres.full- fi
- else
- if [ $result != 0 ]; then
echo "mount test failed" >> $seqres.full- fi
- fi
+else
- echo "sysctl filesystem timestamp check is off" >> $seqres.full
- if [ $result != 0 ]; then
- echo "mount test failed" >> $seqres.full
- fi
echo error messages to stdout on mount failure, test harness could catch the failures.
+fi
+#cd to the SCRATCH_MNT to run the tests +cd $SCRATCH_MNT
+# Begin test case 1 +echo "In memory timestamps update test start" >> $seqres.full
+#update time on the file +update_time=1
+#Run test +_run_test $update_time
+if [ $? != 0 ]; then
- echo "In memory timestamps update failed" >> $seqres.full
- status=1
- exit
+fi
+echo "In memory timestamps update complete" >> $seqres.full
+echo "Unmounting and mounting scratch $SCRATCH_MNT" >> $seqres.full
+#unmount and remount $SCRATCH_DEV +_umount_mount_scratch_dev
+if [ $? != 0 ];then
- status=1
- exit
+fi
+# Begin test case 2
+#re-initialize iterator +n=1
+#Do not update time on the file, just read from disk +update_time=0
+echo "On disk timestamps update test start" >> $seqres.full
+#Re-run test +_run_test $update_time
+if [ $? != 0 ];then
- status=1
- exit
+fi
+echo "On disk timestamps update test complete" >> $seqres.full
+echo "y2038 inode filestamp update successful"
I may have more comments after you sent v2 and I can actually run the test :)
Thanks, Eryu
diff --git a/tests/generic/390.out b/tests/generic/390.out new file mode 100644 index 0000000..355f28e --- /dev/null +++ b/tests/generic/390.out @@ -0,0 +1,2 @@ +QA output created by 390 +y2038 inode filestamp update successful diff --git a/tests/generic/group b/tests/generic/group index 08007d7..d137d01 100644 --- a/tests/generic/group +++ b/tests/generic/group @@ -392,3 +392,4 @@ 387 auto clone 388 auto log metadata 389 auto quick acl
+390 auto quick rw
2.7.4
-- To unsubscribe from this list: send the line "unsubscribe fstests" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Nov 23, 2016 at 04:52:02PM -0800, Deepa Dinamani wrote:
The test helps to validate clamping and mount behaviors according to supported file system timestamp ranges.
Note that the test can fail on 32-bit systems for a few file systems. This will be corrected when vfs is transitioned to use 64-bit timestamps.
Thanks for doing this, Deepa. A few comments below - Eryu has already given lots of feedback, so I won't repeat all that...
Signed-off-by: Deepa Dinamani deepa.kernel@gmail.com
common/attr | 27 ++++++ src/Makefile | 2 +- src/y2038_futimens.c | 61 +++++++++++++ tests/generic/390 | 238 ++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/390.out | 2 + tests/generic/group | 1 + 6 files changed, 330 insertions(+), 1 deletion(-) create mode 100644 src/y2038_futimens.c create mode 100755 tests/generic/390 create mode 100644 tests/generic/390.out
diff --git a/common/attr b/common/attr index ce2d76a..579dc9b 100644 --- a/common/attr +++ b/common/attr @@ -56,6 +56,33 @@ _acl_get_max() esac } +_filesystem_timestamp_range() +{
- device=${1:-$TEST_DEV}
- case $FSTYP in
- ext4) #dumpe2fs
if [ $(dumpe2fs -h $device 2>/dev/null | grep "Inode size:" | cut -d: -f2) -gt 128 ]; thenecho "-2147483648 15032385535"elseecho "-2147483648 2147483647"fi;;- xfs)
echo "-2147483648 2147483647";;- jfs)
echo "0 4294967295";;- f2fs)
echo "-2147483648 2147483647";;- *)
echo "-1 -1";;- esac
+}
This is better off in common/rc right now - common/attr is for extended attribute test code, not generic filesystem stuff.
diff --git a/src/y2038_futimens.c b/src/y2038_futimens.c new file mode 100644 index 0000000..291e4fa --- /dev/null +++ b/src/y2038_futimens.c @@ -0,0 +1,61 @@ +#include <fcntl.h> +#include <stdio.h> +#include <sys/stat.h> +#include <errno.h> +#include <stdlib.h>
+int +do_utime(int fd, long long time) +{
- struct timespec t[2];
- /*
* Convert long long to timespec format.* Seconds precision is assumed here.*/- t[0].tv_sec = time;
- t[0].tv_nsec = 0;
- t[1].tv_sec = time;
- t[1].tv_nsec = 0;
- /* Call utimens to update time. */
- if (futimens(fd, t)) {
perror("futimens");return 1;- }
- return 0;
+}
+int +main(int argc, char **argv) +{
- int fd;
- long long time;
- if(argc < 3) {
fprintf(stderr, "Usage: %s filename timestamp\n""Filename: file to be created or opened in current directory\n""Timestamp: is seconds since 1970-01-01 00:00:00 UTC\n", argv[0]);exit(1);- }
- /* Create the file */
- fd = creat(argv[1], 0666);
- if(fd < 0) {
perror("creat");exit(1);- }
- /* Get the timestamp */
- time = strtoull(argv[2], NULL, 0);
- if (errno) {
perror("strtoull");exit(1);- }
- if (do_utime(fd, time))
return 1;- return 0;
+}
This might be useful to add to xfs_io rather than a one-off helper for xfstest - that avoids the need to create files, and it can be used to change times on existing files....
....
+_run_test_individual() #_run_individual_test(file, timestamp, update_time) +{
- file=$1
- timestamp=$2
- update_time=$3
No need for comments after the function declaration - the prototype is obvious from the local variable assignments....
...
+#Remove log from last run +rm -f $seqres.full
+#install cleaner +trap "_cleanup; exit $status" 0 1 2 3 15
+_scratch_mkfs &>> $seqres.full 2>&1 || _fail "mkfs failed" +read tsmin tsmax <<<$(_filesystem_timestamp_range $SCRATCH_DEV)
This is all test setup preamble, so should be at the top.
+if [ $tsmin -eq -1 -a $tsmax -eq -1 ]; then
- _notrun "filesystem $FSTYP timestamp bounds are unknown"
+fi
This should be in a _requires_timestamp_range() function, I think.
Cheers,
Dave.
On Thursday, November 24, 2016 6:40:55 PM CET Eryu Guan wrote:
On Wed, Nov 23, 2016 at 04:52:02PM -0800, Deepa Dinamani wrote:
+_filesystem_timestamp_range() +{
- device=${1:-$TEST_DEV}
- case $FSTYP in
- ext4) #dumpe2fs
if [ $(dumpe2fs -h $device 2>/dev/null | grep "Inode size:" | cut -d: -f2) -gt 128 ]; thenecho "-2147483648 15032385535"elseecho "-2147483648 2147483647"fiDo ext3 and ext2 follow the same config as ext4?
Those two only support the second case with 128 byte inodes, but the same check should work on all three.
I have an overview of the limits on https://kernelnewbies.org/y2038/vfs, though I'd probably check all of them again, as some of them turned out to be wrong. In particular, identifying whether the on-disk timestamps are meant to be signed or unsigned can be a matter of interpretation and there may be a specification that disagrees with the implementation.
Arnd
Need an entry in .gitignore file too.
Will add this.
SUBDIRS =
diff --git a/src/y2038_futimens.c b/src/y2038_futimens.c new file mode 100644 index 0000000..291e4fa --- /dev/null +++ b/src/y2038_futimens.c @@ -0,0 +1,61 @@ +#include <fcntl.h> +#include <stdio.h> +#include <sys/stat.h> +#include <errno.h> +#include <stdlib.h>
+int +do_utime(int fd, long long time) +{
struct timespec t[2];/** Convert long long to timespec format.* Seconds precision is assumed here.*/t[0].tv_sec = time;t[0].tv_nsec = 0;t[1].tv_sec = time;t[1].tv_nsec = 0;/* Call utimens to update time. */if (futimens(fd, t)) {perror("futimens");return 1;}return 0;+}
+int +main(int argc, char **argv) +{
int fd;long long time;if(argc < 3) {fprintf(stderr, "Usage: %s filename timestamp\n""Filename: file to be created or opened in current directory\n""Timestamp: is seconds since 1970-01-01 00:00:00 UTC\n", argv[0]);exit(1);Seems there's no need for an extra level of indention :)
Ok, will fix this.
}/* Create the file */fd = creat(argv[1], 0666);if(fd < 0) {perror("creat");exit(1);}/* Get the timestamp */time = strtoull(argv[2], NULL, 0);if (errno) {From errno(3), errno is never set to zero by any system call or library function, so errno isn't reset to zero on strtoull and this check always fails for me, with errno = ENOENT, because:
... access("/usr/share/dracut/modules.d/01fips", F_OK) = -1 ENOENT (No such file or directory) ... write(4, "strtoull: No such file or direct"..., 36strtoull: No such file or directory
the errno was from access(2) call in my case.
Right. Man page for strtoull() says:
Since strtoul() can legitimately return 0 or ULONG_MAX (ULLONG_MAX for strtoull()) on both success and failure, the calling program should set errno to 0 before the call, and then determine if an error occurred by checking whether errno has a nonzero value after the call.
I will set errno to 0 before strtoull(). Thanks.
diff --git a/tests/generic/390 b/tests/generic/390 new file mode 100755 index 0000000..f069988 --- /dev/null +++ b/tests/generic/390 @@ -0,0 +1,238 @@ +#! /bin/bash +# FS QA Test No. generic/390 +# +# Tests to verify policy for filesystem timestamps for +# supported ranges: +# 1. Verify filesystem rw mount according to sysctl +# timestamp_supported. +# 2. Verify timestamp clamping for timestamps beyond max +# timestamp supported. +# +# Exit status 1: either or both tests above fail. +# Exit status 0: both the above tests pass.
Please use "./new" script to generate template, which contains all necessary initial setups and the copyright info.
Thanks, will do.
+seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" +#echo "output in $seqres.full" +here=`pwd`
+# Get standard environment, filters and checks. +. ./common/rc +. ./common/filter +. ./common/attr
+SRC_GROUPS=`find tests -not -path tests -type d -printf "%f "`
What's this for? Seems it's not used in the test.
Yes, this is not required. Leftovers from some previous version of the test. Thanks.
+# Prerequisites for the test run. +_supported_fs generic +_supported_os Linux +_require_scratch
Need to check the existence of y2038_futimens program, to make sure it's really built, e.g.
_require_test_program "y2038_futimens"
Will add this
+Y2038_PROG=$here/src/y2038_futimens
+#initialize exit status +status=0
We use 1 as the default value of status, so you can just "exit" on failure (because trap will catch the signal and exit with $status again) and only set status=0 and exit when test passes. "./new" already sets it up for you :)
Will take care of this.
+# Generic test cleanup function. +_cleanup() +{
- # Remove any leftover files from last run.
- rm -f ${SCRATCH_MNT}/test_?
No need to cleanup files on SCRATCH_DEV, it's meant to be mkfs'ed every time before using it.
Ok, makes sense. I will remove this clean up.
+#unmount and mount $SCRATCH_DEV. +_umount_mount_scratch_dev()
There's a helper to do this: _scratch_cycle_mount, only that it doesn't change PWD. And if you don't want to write $SCRATCH_MNT again, this is what we do usually in fstests:
testfile=$SCRATCH_MNT/<somefile> ... do_test $testfile
Thanks, will change it similar to what you suggest.
+{
- #change directory so that you are not using SCRATCH_MNT anymore.
- cd $here
- # Unmount the ${SCRATCH_DEV}
- _scratch_unmount
- if [ $? != 0 ]; then
return $?- fi
- # Mount the ${SCRATCH_DEV}
- _scratch_mount
- if [ $? != 0 ]; then
return $?- fi
- cd ${SCRATCH_MNT}
+}
+# Compare file timestamps obtained from stat +# with a given timestamp. +_check_stat() #check_stat(file, timestamp)
Name local functions without the leading underscore, that's usually for common helper functions, like functions in common/rc, common/attr.
Will do.
+{
- stat_timestamp=`stat -c"%X;%Y" $1`
And please use one tab (8 spaces width) for indention.
Will do.
- prev_timestamp="$2;$2"
- if [ $prev_timestamp != $stat_timestamp ]; then
echo "$prev_timestamp != $stat_timestamp" >> $seqres.fullreturn 1No need to return 0 or 1, just echo the "... != ..." to stdout, and tee -a the output to $seqres.full if you want verbose debug too.
Because all the outputs, including stdout and stderr, from the test will be captured and compared with the .out file at the end of each test, and test harness reports failure if the two outputs don't match. That way, you don't have to check return value of commands most of the times.
Ok, will fix this.
- else
return 0- fi
+}
+_run_test_individual() #_run_individual_test(file, timestamp, update_time) +{
- file=$1
- timestamp=$2
- update_time=$3
- #check if the time needs update
- if [ $update_time -eq 1 ]; then
echo "Updating file: $file to timestamp `date -d @$timestamp`" >> $seqres.full$Y2038_PROG $file $timestamp &>> $seqres.fullif [ $? != 0 ]; thenecho "Failed to run the y2038 program" >> $seqres.fullreturn 1fi- fi
Seems $Y2038_PROG outputs nothing on success, so again, just run the program without checking results & redirecting outputs, test harness will report failure if there's error messages from this program. e.g.
$Y2038_PROG $file $timestampand just let test run even if something goes wrong.
Will do.
- tsclamp=$(($timestamp>$tsmax?$tsmax:$timestamp))
- _check_stat $file $tsclamp
- echo "Checking file: $file Updated timestamp is `date -d @$tsclamp`" >> $seqres.full
- if [ $? != 0 ]; then
echo "Failed to set time to $timestamp" >> $seqres.fullreturn 1- fi
- return 0
+}
+_run_test() #_run_test(update_time) +{
- #initialization iterator
- n=1
- for TIME in "${TIMESTAMPS[@]}"
- do
#Run the test_run_test_individual ${SCRATCH_MNT}/test_$n $TIME $1if [ $? != 0 ]; thenecho "file timestamp update failed `date -d @$TIME`" >> $seqres.fullif [ $n -lt 4 ]; thenThis is not obvious to me, need some comments on the magic number 4 :)
This is leftover from a previous version. Will clean this up. Thanks.
echo "Test failed"return 1fifi#update iterator((n++))- done
- return 0
+}
+#Remove log from last run +rm -f $seqres.full
+#install cleaner +trap "_cleanup; exit $status" 0 1 2 3 15
+_scratch_mkfs &>> $seqres.full 2>&1 || _fail "mkfs failed" +read tsmin tsmax <<<$(_filesystem_timestamp_range $SCRATCH_DEV)
+if [ $tsmin -eq -1 -a $tsmax -eq -1 ]; then
- _notrun "filesystem $FSTYP timestamp bounds are unknown"
+fi
+echo min supported timestamp $tsmin $(date --date=@$tsmin) >> $seqres.full +echo max supported timestamp $tsmax $(date --date=@$tsmax) >> $seqres.full
+#Test timestamps array
+declare -a TIMESTAMPS=(
- $tsmin
- 0
- $tsmax
- $((tsmax+1))
- 4294967295
- 8589934591
- 34359738367
+)
+# Max timestamp is hardcoded to Mon Jan 18 19:14:07 PST 2038 +sys_tsmax=2147483647 +echo "min timestamp that needs to be supported by fs for rw mount is $sys_tsmax $(date --date=@$sys_tsmax)" >> $seqres.full
+read ts_check <<<$(cat /proc/sys/fs/fs-timestamp-check-on)
This fails on systems without this procfs entry, so we need another _require rule, e.g.
_require_proc_attr sys/fs/fs-timestamp-check-on
So test _notrun if kernel has no such procfs entry.
I realized this after I sent out the test. Will update this. Thanks.
Can you please point me a test kernel tree so I can actually run this test?
Will post a pointer along with v2.
+_scratch_mount +result=$?
+if [ $ts_check != 0 ]; then
- echo "sysctl filesystem timestamp check is on" >> $seqres.full
- if [ $sys_tsmax > $tsmax ]; then
if [ $result != -1 ]; thenecho "mount test failed" >> $seqres.fullfi- else
if [ $result != 0 ]; thenecho "mount test failed" >> $seqres.fullfi- fi
+else
- echo "sysctl filesystem timestamp check is off" >> $seqres.full
- if [ $result != 0 ]; then
echo "mount test failed" >> $seqres.full- fi
echo error messages to stdout on mount failure, test harness could catch the failures.
will do.
+fi
+#cd to the SCRATCH_MNT to run the tests +cd $SCRATCH_MNT
+# Begin test case 1 +echo "In memory timestamps update test start" >> $seqres.full
+#update time on the file +update_time=1
+#Run test +_run_test $update_time
+if [ $? != 0 ]; then
- echo "In memory timestamps update failed" >> $seqres.full
- status=1
- exit
+fi
+echo "In memory timestamps update complete" >> $seqres.full
+echo "Unmounting and mounting scratch $SCRATCH_MNT" >> $seqres.full
+#unmount and remount $SCRATCH_DEV +_umount_mount_scratch_dev
+if [ $? != 0 ];then
- status=1
- exit
+fi
+# Begin test case 2
+#re-initialize iterator +n=1
+#Do not update time on the file, just read from disk +update_time=0
+echo "On disk timestamps update test start" >> $seqres.full
+#Re-run test +_run_test $update_time
+if [ $? != 0 ];then
- status=1
- exit
+fi
+echo "On disk timestamps update test complete" >> $seqres.full
+echo "y2038 inode filestamp update successful"
I may have more comments after you sent v2 and I can actually run the test :)
Sounds good :) Thanks for the comments. Will post v2 addressing the issues.
-Deepa
+_filesystem_timestamp_range() +{
- device=${1:-$TEST_DEV}
- case $FSTYP in
- ext4) #dumpe2fs
if [ $(dumpe2fs -h $device 2>/dev/null | grep "Inode size:" | cut -d: -f2) -gt 128 ]; thenecho "-2147483648 15032385535"elseecho "-2147483648 2147483647"fiDo ext3 and ext2 follow the same config as ext4?
Those two only support the second case with 128 byte inodes, but the same check should work on all three.
I have an overview of the limits on https://kernelnewbies.org/y2038/vfs, though I'd probably check all of them again, as some of them turned out to be wrong. In particular, identifying whether the on-disk timestamps are meant to be signed or unsigned can be a matter of interpretation and there may be a specification that disagrees with the implementation.
Right now I've only added tests for a few filesystems. The plan is to add more filesystems later after we agree on the test.
-Deepa
+_filesystem_timestamp_range() +{
device=${1:-$TEST_DEV}case $FSTYP inext4) #dumpe2fsif [ $(dumpe2fs -h $device 2>/dev/null | grep "Inode size:" | cut -d: -f2) -gt 128 ]; thenecho "-2147483648 15032385535"elseecho "-2147483648 2147483647"fi;;xfs)echo "-2147483648 2147483647";;jfs)echo "0 4294967295";;f2fs)echo "-2147483648 2147483647";;*)echo "-1 -1";;esac+}
This is better off in common/rc right now - common/attr is for extended attribute test code, not generic filesystem stuff.
Ok, will move this to common/rc.
diff --git a/src/y2038_futimens.c b/src/y2038_futimens.c new file mode 100644 index 0000000..291e4fa --- /dev/null +++ b/src/y2038_futimens.c @@ -0,0 +1,61 @@ +#include <fcntl.h> +#include <stdio.h> +#include <sys/stat.h> +#include <errno.h> +#include <stdlib.h>
+int +do_utime(int fd, long long time) +{
struct timespec t[2];/** Convert long long to timespec format.* Seconds precision is assumed here.*/t[0].tv_sec = time;t[0].tv_nsec = 0;t[1].tv_sec = time;t[1].tv_nsec = 0;/* Call utimens to update time. */if (futimens(fd, t)) {perror("futimens");return 1;}return 0;+}
+int +main(int argc, char **argv) +{
int fd;long long time;if(argc < 3) {fprintf(stderr, "Usage: %s filename timestamp\n""Filename: file to be created or opened in current directory\n""Timestamp: is seconds since 1970-01-01 00:00:00 UTC\n", argv[0]);exit(1);}/* Create the file */fd = creat(argv[1], 0666);if(fd < 0) {perror("creat");exit(1);}/* Get the timestamp */time = strtoull(argv[2], NULL, 0);if (errno) {perror("strtoull");exit(1);}if (do_utime(fd, time))return 1;return 0;+}
This might be useful to add to xfs_io rather than a one-off helper for xfstest - that avoids the need to create files, and it can be used to change times on existing files....
There are other tests doing the same thing. I will see if using xfs_io here makes more sense.
+_run_test_individual() #_run_individual_test(file, timestamp, update_time) +{
- file=$1
- timestamp=$2
- update_time=$3
No need for comments after the function declaration - the prototype is obvious from the local variable assignments....
Will remove these comments.
+#Remove log from last run +rm -f $seqres.full
+#install cleaner +trap "_cleanup; exit $status" 0 1 2 3 15
+_scratch_mkfs &>> $seqres.full 2>&1 || _fail "mkfs failed" +read tsmin tsmax <<<$(_filesystem_timestamp_range $SCRATCH_DEV)
This is all test setup preamble, so should be at the top.
Will move this to top.
+if [ $tsmin -eq -1 -a $tsmax -eq -1 ]; then
- _notrun "filesystem $FSTYP timestamp bounds are unknown"
+fi
This should be in a _requires_timestamp_range() function, I think.
Yes, that makes sense. Will move this.
Thanks, Deepa
-
Arnd Bergmann -
Dave Chinner -
Deepa Dinamani -
Eryu Guan