This extension allows to use F_UNLCK on query, which currently returns EINVAL. Instead it can be used to query the locks on a particular fd - something that is not currently possible. The basic idea is that on F_OFD_GETLK, F_UNLCK would "conflict" with (or query) any types of the lock on the same fd, and ignore any locks on other fds.
Use-cases:
1. CRIU-alike scenario when you want to read the locking info from an fd for the later reconstruction. This can now be done by setting l_start and l_len to 0 to cover entire file range, and do F_OFD_GETLK. In the loop you need to advance l_start past the returned lock ranges, to eventually collect all locked ranges.
2. Implementing the lock checking/enforcing policy. Say you want to implement an "auditor" module in your program, that checks that the I/O is done only after the proper locking is applied on a file region. In this case you need to know if the particular region is locked on that fd, and if so - with what type of the lock. If you would do that currently (without this extension) then you can only check for the write locks, and for that you need to probe the lock on your fd and then open the same file via another fd and probe there. That way you can identify the write lock on a particular fd, but such trick is non-atomic and complex. As for finding out the read lock on a particular fd - impossible. This extension allows to do such queries without any extra efforts.
3. Implementing the mandatory locking policy. Suppose you want to make a policy where the write lock inhibits any unlocked readers and writers. Currently you need to check if the write lock is present on some other fd, and if it is not there - allow the I/O operation. But because the write lock can appear at any moment, you need to do that under some global lock, which can be released only when the I/O operation is finished. With the proposed extension you can instead just check the write lock on your own fd first, and if it is there - allow the I/O operation on that fd without using any global lock. Only if there is no write lock on this fd, then you need to take global lock and check for a write lock on other fds.
The second patch adds a test-case for OFD locks. It tests both the generic things and the proposed extension.
The third patch is a proposed man page update for fcntl(2) (not for the linux source tree)
Changes in v2: - Dropped the l_pid extension patch and updated test-case accordingly.
Stas Sergeev (2): fs/locks: F_UNLCK extension for F_OFD_GETLK selftests: add OFD lock tests
fs/locks.c | 23 +++- tools/testing/selftests/locking/Makefile | 2 + tools/testing/selftests/locking/ofdlocks.c | 132 +++++++++++++++++++++ 3 files changed, 154 insertions(+), 3 deletions(-) create mode 100644 tools/testing/selftests/locking/ofdlocks.c
CC: Jeff Layton jlayton@kernel.org CC: Chuck Lever chuck.lever@oracle.com CC: Alexander Viro viro@zeniv.linux.org.uk CC: Christian Brauner brauner@kernel.org CC: linux-fsdevel@vger.kernel.org CC: linux-kernel@vger.kernel.org CC: Shuah Khan shuah@kernel.org CC: linux-kselftest@vger.kernel.org CC: linux-api@vger.kernel.org
Currently F_UNLCK with F_OFD_GETLK returns -EINVAL. The proposed extension allows to use it for getting the lock information from the particular fd.
Signed-off-by: Stas Sergeev stsp2@yandex.ru
CC: Jeff Layton jlayton@kernel.org CC: Chuck Lever chuck.lever@oracle.com CC: Alexander Viro viro@zeniv.linux.org.uk CC: Christian Brauner brauner@kernel.org CC: linux-fsdevel@vger.kernel.org CC: linux-kernel@vger.kernel.org CC: Shuah Khan shuah@kernel.org CC: linux-kselftest@vger.kernel.org CC: linux-api@vger.kernel.org
--- fs/locks.c | 23 ++++++++++++++++++++--- 1 file changed, 20 insertions(+), 3 deletions(-)
diff --git a/fs/locks.c b/fs/locks.c index df8b26a42524..210766007e63 100644 --- a/fs/locks.c +++ b/fs/locks.c @@ -868,6 +868,21 @@ static bool posix_locks_conflict(struct file_lock *caller_fl, return locks_conflict(caller_fl, sys_fl); }
+/* Determine if lock sys_fl blocks lock caller_fl. Used on xx_GETLK + * path so checks for additional GETLK-specific things like F_UNLCK. + */ +static bool posix_test_locks_conflict(struct file_lock *caller_fl, + struct file_lock *sys_fl) +{ + /* F_UNLCK checks any locks on the same fd. */ + if (caller_fl->fl_type == F_UNLCK) { + if (!posix_same_owner(caller_fl, sys_fl)) + return false; + return locks_overlap(caller_fl, sys_fl); + } + return posix_locks_conflict(caller_fl, sys_fl); +} + /* Determine if lock sys_fl blocks lock caller_fl. FLOCK specific * checking before calling the locks_conflict(). */ @@ -901,7 +916,7 @@ posix_test_lock(struct file *filp, struct file_lock *fl) retry: spin_lock(&ctx->flc_lock); list_for_each_entry(cfl, &ctx->flc_posix, fl_list) { - if (!posix_locks_conflict(fl, cfl)) + if (!posix_test_locks_conflict(fl, cfl)) continue; if (cfl->fl_lmops && cfl->fl_lmops->lm_lock_expirable && (*cfl->fl_lmops->lm_lock_expirable)(cfl)) { @@ -2207,7 +2222,8 @@ int fcntl_getlk(struct file *filp, unsigned int cmd, struct flock *flock) if (fl == NULL) return -ENOMEM; error = -EINVAL; - if (flock->l_type != F_RDLCK && flock->l_type != F_WRLCK) + if (cmd != F_OFD_GETLK && flock->l_type != F_RDLCK + && flock->l_type != F_WRLCK) goto out;
error = flock_to_posix_lock(filp, fl, flock); @@ -2414,7 +2430,8 @@ int fcntl_getlk64(struct file *filp, unsigned int cmd, struct flock64 *flock) return -ENOMEM;
error = -EINVAL; - if (flock->l_type != F_RDLCK && flock->l_type != F_WRLCK) + if (cmd != F_OFD_GETLK && flock->l_type != F_RDLCK + && flock->l_type != F_WRLCK) goto out;
error = flock64_to_posix_lock(filp, fl, flock);
Test the basic locking stuff on 2 fds: multiple read locks, conflicts between read and write locks, use of len==0 for queries. Also tests for F_UNLCK F_OFD_GETLK extension.
Signed-off-by: Stas Sergeev stsp2@yandex.ru
CC: Jeff Layton jlayton@kernel.org CC: Chuck Lever chuck.lever@oracle.com CC: Alexander Viro viro@zeniv.linux.org.uk CC: Christian Brauner brauner@kernel.org CC: linux-fsdevel@vger.kernel.org CC: linux-kernel@vger.kernel.org CC: Shuah Khan shuah@kernel.org CC: linux-kselftest@vger.kernel.org CC: linux-api@vger.kernel.org
--- tools/testing/selftests/locking/Makefile | 2 + tools/testing/selftests/locking/ofdlocks.c | 132 +++++++++++++++++++++ 2 files changed, 134 insertions(+) create mode 100644 tools/testing/selftests/locking/ofdlocks.c
diff --git a/tools/testing/selftests/locking/Makefile b/tools/testing/selftests/locking/Makefile index 6e7761ab3536..a83ced1626de 100644 --- a/tools/testing/selftests/locking/Makefile +++ b/tools/testing/selftests/locking/Makefile @@ -7,4 +7,6 @@ all:
TEST_PROGS := ww_mutex.sh
+TEST_GEN_PROGS := ofdlocks + include ../lib.mk diff --git a/tools/testing/selftests/locking/ofdlocks.c b/tools/testing/selftests/locking/ofdlocks.c new file mode 100644 index 000000000000..1ccb2b9b5ead --- /dev/null +++ b/tools/testing/selftests/locking/ofdlocks.c @@ -0,0 +1,132 @@ +// SPDX-License-Identifier: GPL-2.0 + +#define _GNU_SOURCE +#include <fcntl.h> +#include <assert.h> +#include <stdio.h> +#include <unistd.h> +#include <string.h> +#include "../kselftest.h" + +static int lock_set(int fd, struct flock *fl) +{ + int ret; + + fl->l_pid = 0; // needed for OFD locks + fl->l_whence = SEEK_SET; + ret = fcntl(fd, F_OFD_SETLK, fl); + if (ret) + perror("fcntl()"); + return ret; +} + +static int lock_get(int fd, struct flock *fl) +{ + int ret; + + fl->l_pid = 0; // needed for OFD locks + fl->l_whence = SEEK_SET; + ret = fcntl(fd, F_OFD_GETLK, fl); + if (ret) + perror("fcntl()"); + return ret; +} + +int main(void) +{ + int rc; + struct flock fl, fl2; + int fd = open("/tmp/aa", O_RDWR | O_CREAT | O_EXCL, 0600); + int fd2 = open("/tmp/aa", O_RDONLY); + + unlink("aa"); + assert(fd != -1); + assert(fd2 != -1); + ksft_print_msg("[INFO] opened fds %i %i\n", fd, fd2); + + /* Set some read lock */ + fl.l_type = F_RDLCK; + fl.l_start = 5; + fl.l_len = 3; + rc = lock_set(fd, &fl); + if (rc == 0) { + ksft_print_msg + ("[SUCCESS] set OFD read lock on first fd\n"); + } else { + ksft_print_msg("[FAIL] to set OFD read lock on first fd\n"); + return -1; + } + /* Make sure read locks do not conflict on different fds. */ + fl.l_type = F_RDLCK; + fl.l_start = 5; + fl.l_len = 1; + rc = lock_get(fd2, &fl); + if (rc != 0) + return -1; + if (fl.l_type != F_UNLCK) { + ksft_print_msg("[FAIL] read locks conflicted\n"); + return -1; + } + /* Make sure read/write locks do conflict on different fds. */ + fl.l_type = F_WRLCK; + fl.l_start = 5; + fl.l_len = 1; + rc = lock_get(fd2, &fl); + if (rc != 0) + return -1; + if (fl.l_type != F_UNLCK) { + ksft_print_msg + ("[SUCCESS] read and write locks conflicted\n"); + } else { + ksft_print_msg + ("[SUCCESS] read and write locks not conflicted\n"); + return -1; + } + /* Get info about the lock on first fd. */ + fl.l_type = F_UNLCK; + fl.l_start = 5; + fl.l_len = 1; + rc = lock_get(fd, &fl); + if (rc != 0) { + ksft_print_msg + ("[FAIL] F_OFD_GETLK with F_UNLCK not supported\n"); + return -1; + } + if (fl.l_type != F_UNLCK) { + ksft_print_msg + ("[SUCCESS] F_UNLCK test returns: locked, type %i pid %i len %zi\n", + fl.l_type, fl.l_pid, fl.l_len); + } else { + ksft_print_msg + ("[FAIL] F_OFD_GETLK with F_UNLCK did not return lock info\n"); + return -1; + } + /* Try the same but by locking everything by len==0. */ + fl2.l_type = F_UNLCK; + fl2.l_start = 0; + fl2.l_len = 0; + rc = lock_get(fd, &fl2); + if (rc != 0) { + ksft_print_msg + ("[FAIL] F_OFD_GETLK with F_UNLCK not supported\n"); + return -1; + } + if (memcmp(&fl, &fl2, sizeof(fl))) { + ksft_print_msg + ("[FAIL] F_UNLCK test returns: locked, type %i pid %i len %zi\n", + fl.l_type, fl.l_pid, fl.l_len); + return -1; + } + ksft_print_msg("[SUCCESS] F_UNLCK with len==0 returned the same\n"); + /* Get info about the lock on second fd - no locks on it. */ + fl.l_type = F_UNLCK; + fl.l_start = 0; + fl.l_len = 0; + lock_get(fd2, &fl); + if (fl.l_type != F_UNLCK) { + ksft_print_msg + ("[FAIL] F_OFD_GETLK with F_UNLCK return lock info from another fd\n"); + return -1; + } + return 0; +}
On Wed, 2023-06-21 at 20:22 +0500, Stas Sergeev wrote:
Test the basic locking stuff on 2 fds: multiple read locks, conflicts between read and write locks, use of len==0 for queries. Also tests for F_UNLCK F_OFD_GETLK extension.
Signed-off-by: Stas Sergeev stsp2@yandex.ru
CC: Jeff Layton jlayton@kernel.org CC: Chuck Lever chuck.lever@oracle.com CC: Alexander Viro viro@zeniv.linux.org.uk CC: Christian Brauner brauner@kernel.org CC: linux-fsdevel@vger.kernel.org CC: linux-kernel@vger.kernel.org CC: Shuah Khan shuah@kernel.org CC: linux-kselftest@vger.kernel.org CC: linux-api@vger.kernel.org
tools/testing/selftests/locking/Makefile | 2 + tools/testing/selftests/locking/ofdlocks.c | 132 +++++++++++++++++++++ 2 files changed, 134 insertions(+) create mode 100644 tools/testing/selftests/locking/ofdlocks.c
diff --git a/tools/testing/selftests/locking/Makefile b/tools/testing/selftests/locking/Makefile index 6e7761ab3536..a83ced1626de 100644 --- a/tools/testing/selftests/locking/Makefile +++ b/tools/testing/selftests/locking/Makefile @@ -7,4 +7,6 @@ all: TEST_PROGS := ww_mutex.sh +TEST_GEN_PROGS := ofdlocks
include ../lib.mk
I'm not sure this really belongs in the "locking" directory. Given that there is only the ww_mutex test in there, that's more for internal synchronization mechanisms, I think.
Can you create a new "filelock" directory and drop this into there instead?
diff --git a/tools/testing/selftests/locking/ofdlocks.c b/tools/testing/selftests/locking/ofdlocks.c new file mode 100644 index 000000000000..1ccb2b9b5ead --- /dev/null +++ b/tools/testing/selftests/locking/ofdlocks.c @@ -0,0 +1,132 @@ +// SPDX-License-Identifier: GPL-2.0
+#define _GNU_SOURCE +#include <fcntl.h> +#include <assert.h> +#include <stdio.h> +#include <unistd.h> +#include <string.h> +#include "../kselftest.h"
+static int lock_set(int fd, struct flock *fl) +{
- int ret;
- fl->l_pid = 0; // needed for OFD locks
- fl->l_whence = SEEK_SET;
- ret = fcntl(fd, F_OFD_SETLK, fl);
- if (ret)
perror("fcntl()");- return ret;
+}
+static int lock_get(int fd, struct flock *fl) +{
- int ret;
- fl->l_pid = 0; // needed for OFD locks
- fl->l_whence = SEEK_SET;
- ret = fcntl(fd, F_OFD_GETLK, fl);
- if (ret)
perror("fcntl()");- return ret;
+}
+int main(void) +{
- int rc;
- struct flock fl, fl2;
- int fd = open("/tmp/aa", O_RDWR | O_CREAT | O_EXCL, 0600);
- int fd2 = open("/tmp/aa", O_RDONLY);
- unlink("aa");
- assert(fd != -1);
- assert(fd2 != -1);
- ksft_print_msg("[INFO] opened fds %i %i\n", fd, fd2);
- /* Set some read lock */
- fl.l_type = F_RDLCK;
- fl.l_start = 5;
- fl.l_len = 3;
- rc = lock_set(fd, &fl);
- if (rc == 0) {
ksft_print_msg("[SUCCESS] set OFD read lock on first fd\n");- } else {
ksft_print_msg("[FAIL] to set OFD read lock on first fd\n");return -1;- }
- /* Make sure read locks do not conflict on different fds. */
- fl.l_type = F_RDLCK;
- fl.l_start = 5;
- fl.l_len = 1;
- rc = lock_get(fd2, &fl);
- if (rc != 0)
return -1;- if (fl.l_type != F_UNLCK) {
ksft_print_msg("[FAIL] read locks conflicted\n");return -1;- }
- /* Make sure read/write locks do conflict on different fds. */
- fl.l_type = F_WRLCK;
- fl.l_start = 5;
- fl.l_len = 1;
- rc = lock_get(fd2, &fl);
- if (rc != 0)
return -1;- if (fl.l_type != F_UNLCK) {
ksft_print_msg("[SUCCESS] read and write locks conflicted\n");- } else {
ksft_print_msg("[SUCCESS] read and write locks not conflicted\n");return -1;- }
- /* Get info about the lock on first fd. */
- fl.l_type = F_UNLCK;
- fl.l_start = 5;
- fl.l_len = 1;
- rc = lock_get(fd, &fl);
- if (rc != 0) {
ksft_print_msg("[FAIL] F_OFD_GETLK with F_UNLCK not supported\n");return -1;- }
- if (fl.l_type != F_UNLCK) {
ksft_print_msg("[SUCCESS] F_UNLCK test returns: locked, type %i pid %i len %zi\n",fl.l_type, fl.l_pid, fl.l_len);- } else {
ksft_print_msg("[FAIL] F_OFD_GETLK with F_UNLCK did not return lock info\n");return -1;- }
- /* Try the same but by locking everything by len==0. */
- fl2.l_type = F_UNLCK;
- fl2.l_start = 0;
- fl2.l_len = 0;
- rc = lock_get(fd, &fl2);
- if (rc != 0) {
ksft_print_msg("[FAIL] F_OFD_GETLK with F_UNLCK not supported\n");return -1;- }
- if (memcmp(&fl, &fl2, sizeof(fl))) {
ksft_print_msg("[FAIL] F_UNLCK test returns: locked, type %i pid %i len %zi\n",fl.l_type, fl.l_pid, fl.l_len);return -1;- }
- ksft_print_msg("[SUCCESS] F_UNLCK with len==0 returned the same\n");
- /* Get info about the lock on second fd - no locks on it. */
- fl.l_type = F_UNLCK;
- fl.l_start = 0;
- fl.l_len = 0;
- lock_get(fd2, &fl);
- if (fl.l_type != F_UNLCK) {
ksft_print_msg("[FAIL] F_OFD_GETLK with F_UNLCK return lock info from another fd\n");return -1;- }
- return 0;
+}
I'm not opposed to adding a selftest here, but most filesystem testing is done via xfstests:
https://git.kernel.org/pub/scm/fs/xfs/xfstests-dev.git/
It would be better to add this test to the existing generic/478 test that tests OFD locks. Can you patch that to add a test for the new functionality?
Thanks,
22.06.2023 16:48, Jeff Layton пишет:
I'm not opposed to adding a selftest here, but most filesystem testing is done via xfstests:
https://git.kernel.org/pub/scm/fs/xfs/xfstests-dev.git/It would be better to add this test to the existing generic/478 test that tests OFD locks. Can you patch that to add a test for the new functionality?
I don't actually think its possible. It seems like their script creates 2 t_ofd_locks processes, one for creating the lock, one for testing it. This is not our case. To make it work our way, we'd probably need to hack that test directly into t_ofd_locks.c, so that it can set and test from the same fd. And I don't know how to even run these tests. :) So I am really inclined to limit myself with a kernel selftest.
On Jun 22, 2023, at 12:40 PM, stsp stsp2@yandex.ru wrote:
22.06.2023 16:48, Jeff Layton пишет:
I'm not opposed to adding a selftest here, but most filesystem testing is done via xfstests:
https://git.kernel.org/pub/scm/fs/xfs/xfstests-dev.git/It would be better to add this test to the existing generic/478 test that tests OFD locks. Can you patch that to add a test for the new functionality?
I agree with Jeff, this test needs to go in xfstests.
I don't actually think its possible. It seems like their script creates 2 t_ofd_locks processes, one for creating the lock, one for testing it. This is not our case. To make it work our way, we'd probably need to hack that test directly into t_ofd_locks.c, so that it can set and test from the same fd.
Or you could add a new test program.
And I don't know how to even run these tests. :) So I am really inclined to limit myself with a kernel selftest.
IMO that's not a reason not to do this properly.
You should work with Jeff and the maintainer of xfstests to make it happen.
-- Chuck Lever
22.06.2023 21:58, Chuck Lever III пишет:
IMO that's not a reason not to do this properly.
You should work with Jeff and the maintainer of xfstests to make it happen.
But its not going to be in this patch-set anyway, as its a different source tree... So I should prepare it when this is merged, or?
On Jun 22, 2023, at 1:05 PM, stsp stsp2@yandex.ru wrote:
22.06.2023 21:58, Chuck Lever III пишет:
IMO that's not a reason not to do this properly.
You should work with Jeff and the maintainer of xfstests to make it happen.
But its not going to be in this patch-set anyway, as its a different source tree...
If others agree with me, then please drop the selftests patch from this series. There is a considerably higher probability that the new test will be run frequently by CI if it's in xfstests.
So I should prepare it when this is merged, or?
I don't have a strong preference. A good choice is to push the test before the kernel changes are merged.
-- Chuck Lever
On Thu, Jun 22, 2023 at 10:31:06PM +0500, stsp wrote:
22.06.2023 22:12, Chuck Lever III пишет:
I don't have a strong preference. A good choice is to push the test before the kernel changes are merged.
It will fail though w/o kernel changes. So what exactly is the policy?
filesystem unit test functionality needs to be pushed into fstests and/or ltp. The preference is the former, because just about every filesystem developer and distro QA team is running this as part of their every-day testing workflow.
fstests is written to probe whether the kernel supports a given feature or not before testing it. It will _not_run() a test that doesn't have the required kernel/fs/device support, and this is not considered a test failure.
Yes, it means you have to also write the userspace feature probing code, but that should be trivial to do because userspace already has to be able to safely discover that this extension exists, right?
-Dave.
On Jun 22, 2023, at 1:05 PM, stsp stsp2@yandex.ru wrote:
22.06.2023 21:58, Chuck Lever III пишет:
IMO that's not a reason not to do this properly.
You should work with Jeff and the maintainer of xfstests to make it happen.
But its not going to be in this patch-set anyway, as its a different source tree...
If others agree with me, then please drop the selftests patch from this series. There is a considerably higher probability that the new test will be run frequently by CI if it's in xfstests.
So I should prepare it when this is merged, or?
I don't have a strong preference. A good choice is to push the test before the kernel changes are merged.
As an aside, an additional testing option for OFD locks is the multilock test tool that is in the nfs-ganesha project. In preparation to use OFD locks in Ganesha, I added them to multilock to check them out, and that incidentally also allows testing how the NFS client and server work when OFD locks are taken on a file from an NFS mount.
Frank
Hi,
22.06.2023 21:58, Chuck Lever III пишет:
IMO that's not a reason not to do this properly.
You should work with Jeff and the maintainer of xfstests to make it happen.
If only I knew who is that maintainer and why he ignores all patches...
Can someone else please take a look at this: https://www.spinics.net/lists/fstests/msg23115.html and make a review/ack?
22.06.2023 16:48, Jeff Layton пишет:
I'm not sure this really belongs in the "locking" directory. Given that there is only the ww_mutex test in there, that's more for internal synchronization mechanisms, I think.
Can you create a new "filelock" directory and drop this into there instead?
Done and sent the v3.
F_UNLCK has the special meaning when used as a lock type on input. It returns the information about any lock found in the specified region on that particular file descriptor. Locks on other file descriptors are ignored by F_UNLCK.
Signed-off-by: Stas Sergeev stsp2@yandex.ru
CC: Jeff Layton jlayton@kernel.org CC: Chuck Lever chuck.lever@oracle.com CC: Alexander Viro viro@zeniv.linux.org.uk CC: Christian Brauner brauner@kernel.org CC: linux-fsdevel@vger.kernel.org CC: linux-kernel@vger.kernel.org CC: Shuah Khan shuah@kernel.org CC: linux-kselftest@vger.kernel.org CC: linux-api@vger.kernel.org
--- man2/fcntl.2 | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/man2/fcntl.2 b/man2/fcntl.2 index 7b5604e3a..e3e3e7b8c 100644 --- a/man2/fcntl.2 +++ b/man2/fcntl.2 @@ -604,6 +604,13 @@ then details about one of these locks are returned via .IR lock , as described above for .BR F_GETLK . +.B F_UNLCK +has the special meaning when put into +.I l_type +as an input. It returns the information about any lock in the specified +range on that particular file descriptor. The locks on other file +descriptors are ignored by +.BR F_UNLCK . .PP In the current implementation, ." commit 57b65325fe34ec4c917bc4e555144b4a94d9e1f7
On Wed, 2023-06-21 at 20:22 +0500, Stas Sergeev wrote:
F_UNLCK has the special meaning when used as a lock type on input. It returns the information about any lock found in the specified region on that particular file descriptor. Locks on other file descriptors are ignored by F_UNLCK.
Signed-off-by: Stas Sergeev stsp2@yandex.ru
CC: Jeff Layton jlayton@kernel.org CC: Chuck Lever chuck.lever@oracle.com CC: Alexander Viro viro@zeniv.linux.org.uk CC: Christian Brauner brauner@kernel.org CC: linux-fsdevel@vger.kernel.org CC: linux-kernel@vger.kernel.org CC: Shuah Khan shuah@kernel.org CC: linux-kselftest@vger.kernel.org CC: linux-api@vger.kernel.org
man2/fcntl.2 | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/man2/fcntl.2 b/man2/fcntl.2 index 7b5604e3a..e3e3e7b8c 100644 --- a/man2/fcntl.2 +++ b/man2/fcntl.2 @@ -604,6 +604,13 @@ then details about one of these locks are returned via .IR lock , as described above for .BR F_GETLK . +.B F_UNLCK +has the special meaning when put into +.I l_type +as an input. It returns the information about any lock in the specified +range on that particular file descriptor. The locks on other file +descriptors are ignored by +.BR F_UNLCK . .PP In the current implementation, ." commit 57b65325fe34ec4c917bc4e555144b4a94d9e1f7
We need to be pedantic for manpages. A "file description" is the representation of the open file in the kernel (basically, the "struct file" in the kernel). A file _descriptor_ is the numeric identifier returned by open() and similar functions.
The locks are owned by the file description, so that would be the better term to use here. I think you want something like:
"When the l_type is set to F_UNLCK, returned locks are limited to ones set on the given file description. Locks set on other file descriptions are ignored on F_GETLK requests with the l_type set to F_UNLCK."
22.06.2023 17:03, Jeff Layton пишет:
We need to be pedantic for manpages. A "file description" is the representation of the open file in the kernel (basically, the "struct file" in the kernel). A file _descriptor_ is the numeric identifier returned by open() and similar functions.
OK.
The locks are owned by the file description, so that would be the better term to use here. I think you want something like:
"When the l_type is set to F_UNLCK, returned locks are limited to ones set on the given file description.
This is also inaccurate, because "limited" implies other operations act widely. But actually other operations do not consider the "same fd" at all. So the reported sets by F_UNLCK and other ops do not overlap. Which is why I decided to describe it as a "special meaning".
linux-kselftest-mirror@lists.linaro.org
-
Chuck Lever III -
Dave Chinner -
Frank Filz -
Jeff Layton -
Stas Sergeev -
stsp