- 17 Jan, 2013 11 commits
-
-
Qiang Huang authored
We check shared lock, exclusive lock and mandatory-read lock for flock file locks. After restore, by parsing /proc/locks file, we check if the process holds the same file locks as it was dumped. Signed-off-by:
Qiang Huang <h.huangqiang@huawei.com> Signed-off-by:
Pavel Emelyanov <xemul@parallels.com>
-
Qiang Huang authored
We check read lock and write lock for posix file locks. After restore, we use fcntl(fd, F_GETLK, &lock), to see if the file is hold the same lock as it was dumped. Signed-off-by:
-
Qiang Huang authored
Signed-off-by:
-
Qiang Huang authored
According to the file lock information from the image, we recall flock or fcntl with proper parameters, so we can rehold the file locks as we were dumped. We only support flock and posix file locks so far. Changelog since the initial version: a. Use prepare_file_locks instead of restore function directly. b. Fix some bugs. Originally-signed-off-by:
Zheng Gu <cengku.gu@huawei.com> Signed-off-by:
-
Qiang Huang authored
Dump file locks' necessary entries to the image, we only support flock and posix file lock right now. Changelog since the initial version: We got file lock info from global list, so the dump_task_file_locks can be much simpler. Originally-signed-off-by:
-
Qiang Huang authored
We collect all file locks to a golbal list, so we can use them easily in dump_one_task. For optimizaton, we only collect file locks hold by tasks in the pstree. Thanks to the ptrace-seize machanism, we can aviod the blocked file lock issue, makes the work simpler. Right now, the check handles only one situation: -- Dumping tasks with file locks hold without the -l option. This covers for the most part. But we still need some more work to make it perfect robust in the future. Signed-off-by:
-
Qiang Huang authored
Sometimes, when we parse some global info files, we can only care about tasks which are taken into dump(such as file locks), which means their pids are in the pstree. So a function like this would be help. Signed-off-by:
-
Qiang Huang authored
Some contents needed by filelock's image. Originally-signed-off-by:
-
Qiang Huang authored
Add a callback for filelock's image show. Originally-signed-off-by:
-
Qiang Huang authored
We need a new protobuf description for file-lock. Originally-signed-off-by:
-
Qiang Huang authored
Add --file-locks/-l option to support handling file locks, for safety, only used for container. Originally-signed-off-by:
-
- 16 Jan, 2013 12 commits
-
-
Alexander Kartashov authored
This workarounds a compilation warning on ARM: packet-sock.pb-c.c: In function 'packet_sock_entry__init': packet-sock.pb-c.c:98:3: error: this decimal constant is unsigned only in ISO C90 [-Werror] packet-sock.pb-c.c: At top level: packet-sock.pb-c.c:318:1: error: this decimal constant is unsigned only in ISO C90 [-Werror] Signed-off-by:
Alexander Kartashov <alekskartashov@parallels.com> Signed-off-by:
-
Alexander Kartashov authored
This patch replaces the format specifier %ld with PRIx64 in the following places: * the format string argument of the functions scanf() and printf(), * in the macros GEN_SYSCTL_*_FUNC. We need explicit specification of the integer size there. Signed-off-by:
-
Alexander Kartashov authored
The layout of the struct rlimit depends on the value of the macro FILE_OFFSET_BITS. If FILE_OFFSET_BITS is 64 the userspace and kernel definitions becomes incoherent on a 32-bit platform. The struct krlimit representing the kernel version of the struct rlimit is introduced to address the issue: the function restore_rlims() is fixed to convert between the userspace and kernel representations of the struct. Signed-off-by:
-
Alexander Kartashov authored
This makes it possible to use TASK_SIZE instead of TASK_SIZE_MAX. Signed-off-by:
-
Alexander Kartashov authored
Signed-off-by:
-
Cyrill Gorcunov authored
I occasionally read FanotifyMarkEntry object as InotifyWdEntry in collect_one_fanotify_mark, this didn't trigger a bug in test since the events are still occured (and before protobuf file refine the formats were close to each other), which means the fanotify00 test-case need to be updated (which is addressed in further patch). And don't forget to init fields. Signed-off-by:
Cyrill Gorcunov <gorcunov@openvz.org> Signed-off-by:
-
Cyrill Gorcunov authored
The object is allocated with malloc. The lack of initialization is not problem at moment since we assign members in collect_inotify_mark unconditionally but it might cause problems in future so better to init it as early as possible. Signed-off-by:
-
Cyrill Gorcunov authored
Just reading if event entry exist in buffer is not enough we rathee need to be sure that all flags and bits are recovered well. For this sake we parse fdinfo before and after the c/r sessions and compare results. Signed-off-by:
-
Cyrill Gorcunov authored
Just missed it in first place. Signed-off-by:
-
Cyrill Gorcunov authored
To be consistent with naming (we have collect_one_fanotify_mark helper already). Signed-off-by:
-
Cyrill Gorcunov authored
This removes code duplication. There is no strong need for free_fhandle but I decided to add it to be consistent with alloc/free handlers. Signed-off-by:
-
Cyrill Gorcunov authored
Signed-off-by:
-
- 15 Jan, 2013 17 commits
-
-
Pavel Emelyanov authored
These are structs that (now) tie together ns string and the CLONE_ flag. It's nice to have one (some code becomes simpler) and will help us with auto-namespaces detection. Signed-off-by: -
Cyrill Gorcunov authored
As Pavel proposed we can refine fanotify image objects squeezing common part in separate entry. Finally the objects are grouped as enum mark_type { INODE = 1; MOUNT = 2; } message fanotify_inode_mark_entry { required uint64 i_ino = 1; required fh_entry f_handle = 2; } message fanotify_mount_mark_entry { required uint32 mnt_id = 1; } message fanotify_mark_entry { required uint32 id = 1; required mark_type type = 2; required uint32 mflags = 3; required uint32 mask = 4; required uint32 ignored_mask = 5; required uint32 s_dev = 6; optional fanotify_inode_mark_entry ie = 7; optional fanotify_mount_mark_entry me = 8; } This required some tuning in fdinfo parsing and fsnotify code itself, but result looks good to me. Signed-off-by: -
Pavel Emelyanov authored
Signed-off-by: -
Cyrill Gorcunov authored
As Andrew pointed the logically it's more correct to use err() instead of fail() if test_daemon() has not been used. Signed-off-by:
-
Andrey Vagin authored
Signed-off-by:
Andrey Vagin <avagin@openvz.org> Signed-off-by:
-
Andrey Vagin authored
Signed-off-by:
-
Andrey Vagin authored
Restore of netns uses the file descriptor on the root netns. This descriptor is opened before forking. We can add one more service fd, but I think this approach is better. Signed-off-by:
-
Cyrill Gorcunov authored
This one is pretty trivial and need to be extented for more masks and flags, but fine for now. Signed-off-by:
-
Cyrill Gorcunov authored
Signed-off-by:
-
Cyrill Gorcunov authored
Signed-off-by:
-
Cyrill Gorcunov authored
For fanotify dumping we need to find mount points the path lay on, moreover we save a mount point device number in the image, thus collect mount point information on dump stage. Signed-off-by:
-
Cyrill Gorcunov authored
The same way as done for inotify we walk over all marks hooked on file entry and restore them with sys_fanotify_mark call. Signed-off-by:
-
Cyrill Gorcunov authored
No real restore yet, but only collecting of file entries and marks for simplicity. Ideally, once everything will be ready and tested we will unitfy generic fsnotify restore engine. But for a while for debug reason inotify and fanotify kept separated. Signed-off-by:
-
Cyrill Gorcunov authored
fanotify and inotify will be gathered to separate lists. Signed-off-by:
-
Cyrill Gorcunov authored
We will need collect_fanotify_mark helper so unify naming scheme. Signed-off-by:
-
Cyrill Gorcunov authored
fanotify will have collect_one_fanotify helper, so to unify naming scheme rename this one. Signed-off-by:
-
Cyrill Gorcunov authored
fanotify will have own descriptor so to avoid confusion rename it. Signed-off-by:
-
