slavaim/redirfs

A Linux file system filter based on redirfs

redirfs's Issues

comiple error

/home/user/Desktop/redirfs/redirfs/src/avflt/avflt_check.c:298:47: error: implicit declaration of function ‘current_cred’; did you mean ‘current_chrooted’? [-Werror=implicit-function-declaration]
file = dentry_open(&event->f_path, flags, current_cred());
^~~~~~~~~~~~
current_chrooted
/home/user/Desktop/redirfs/redirfs/src/avflt/avflt_check.c:298:47: warning: passing argument 3 of ‘dentry_open’ makes pointer from integer without a cast [-Wint-conversion]
file = dentry_open(&event->f_path, flags, current_cred());
^~~~~~~~~~~~~~
In file included from ./include/linux/poll.h:10,
from /home/user/Desktop/redirfs/redirfs/src/avflt/avflt.h:29,
from /home/user/Desktop/redirfs/redirfs/src/avflt/avflt_check.c:24:
./include/linux/fs.h:2469:22: note: expected ‘const struct cred *’ but argument is of type ‘int’
extern struct file * dentry_open(const struct path *, int, const struct cred *);
^~~~~~~~~~~
cc1: some warnings being treated as errors

Is this project support cifs filesystem?

HI,thank your reviving this project. And i have some questions to discuss.
There is a tag which say "cifs is not supported (version 0.7)" in file "CHANGELOG", but in file "rfs_path_check_fs.c",function 'rfs_path_check_fs' comment out some code. So,it has supported cifs now?
BTW,i'm not sure a problem with "rfs_cast_to_rdentry(dentry)" in file "rfs_dentry.c"? it maybe double add count of robject when using "rfs_dentry_get(rfs_cast_to_rdentry)"？

After file move, filter unregister failed

Dear @slavaim,

After some mv operations in filtetring area, we can not unregister filter module with msg of "Device or resource busy".

OS:

• CentOS 6 or 7

Test:

1. rfsctl -f mvflt -i /test1
2. mv /test1/file1 /test1/file2
3. rfsctl -f mvflt -c
4. echo -n "1" > /sys/fs/redirfs/filters/mvflt/unregister
   bash: echo: write error: Device or resource busy

Cause:

• counter mismatch error at following line on "redirfs_unregister_filter" function:
  -- if (atomic_read(&rflt->count) != 3) {

Would you please give me some solution?

Thanks.

Getting it work on linux-kernel 5.x?

Just got in need of redirfs trying to recover Comodo Antivirus still using redirfs. Are there any plans for resurrection for linux-kernel 5.x (5.1)? Tried to compile recent version of this repository, but it failed :-(

Where can I find the docs?

The redirFS website is down, Any ideas?

Question about multiple redirfs

I change the Redirfs Exported symbol name and sysfs name things and kmem_cache names things.
And I loaded two redirfs module. After that I registered kernel module into each redirfs filter.
And I added the path each of them.
When I tried to add path of second redirfs-registered kernel module. panic happened
so my question is, is it possible to make two redirfs load on two kernel, and to make operate well?

Linux 4.17 compatibility

You did a great job reviving this kernel module! Could you please also make it compatible with Linux 4.17? I could really make use of it with COMODO antivirus, which needs this to work. I tried to do this myself, but I didn't succeed...

Empty directory of read-only attribute can be deleted

[root@localhost 11]# ls -la
total 4
drwxr-xr-x 2 root root 6 Aug 20 06:06 .
drwxr-xr-x 3 root root 4096 Aug 20 06:06 ..
[root@localhost 11]# pwd
/root/test/test_1/11
[root@localhost 11]# cat /sys/fs/redirfs/filters/roflt/paths
i:0:/root/test/test_1/11[root@localhost 11]# cd ../
[root@localhost test_1]# rm -fr 11
[root@localhost test_1]#

kernel panic during umount

Dear @slavaim,

Could you please give me some help?
There is kernel panic if umount without deleting path.

test step:

1. insmod roflt
2. rfsctl -f roflt -i /test1
3. mount /dev/sdxx /test1
4. umount /dev/sdxx

dump stack:
#0 [ffff98e3f2ccfac0] machine_kexec at ffffffffaf063674
#1 [ffff98e3f2ccfb20] __crash_kexec at ffffffffaf11cf32
#2 [ffff98e3f2ccfbf0] crash_kexec at ffffffffaf11d020
#3 [ffff98e3f2ccfc08] oops_end at ffffffffaf76d758
#4 [ffff98e3f2ccfc30] die at ffffffffaf02f95b
#5 [ffff98e3f2ccfc60] do_trap at ffffffffaf76cea0
#6 [ffff98e3f2ccfcb0] do_invalid_op at ffffffffaf02c2a4
#7 [ffff98e3f2ccfd60] invalid_op at ffffffffaf77912e
[exception RIP: shrink_dcache_for_umount_subtree+460]
RIP: ffffffffaf25a9bc RSP: ffff98e3f2ccfe10 RFLAGS: 00010246
RAX: 000000000000004d RBX: ffff98e354797680 RCX: 0000000000000000
RDX: 0000000000000000 RSI: ffff98e49f013898 RDI: ffff98e49f013898
RBP: ffff98e3f2ccfe30 R8: 6e6544203a475542 R9: 66207972746e6544
R10: 0000000000000833 R11: 3533653839666666 R12: ffff98e354797680
R13: 00000000000003f3 R14: ffffffffb00e65b0 R15: ffff98e3e1a85910
ORIG_RAX: ffffffffffffffff CS: 0010 SS: 0018
#8 [ffff98e3f2ccfe38] shrink_dcache_for_umount at ffffffffaf25c2ef
#9 [ffff98e3f2ccfe50] generic_shutdown_super at ffffffffaf24416f
#10 [ffff98e3f2ccfe70] kill_block_super at ffffffffaf244627
#11 [ffff98e3f2ccfe90] deactivate_locked_super at ffffffffaf24496e
#12 [ffff98e3f2ccfeb0] deactivate_super at ffffffffaf2450f6
#13 [ffff98e3f2ccfec8] cleanup_mnt at ffffffffaf26366f
#14 [ffff98e3f2ccfee0] __cleanup_mnt at ffffffffaf263702
#15 [ffff98e3f2ccfef0] task_work_run at ffffffffaf0be7db
#16 [ffff98e3f2ccff30] do_notify_resume at ffffffffaf02bc65
#17 [ffff98e3f2ccff50] int_signal at ffffffffaf776124

Notes:

1. If i delete path by "rfsctl -f roflt -c" before umount then no problem.
2. If i hold some mounted file like "tail -f /test1/locktest" and umount by "umount -l /dev/sdxx" wirh lazy option then no problem.
3. OS is CentOS 7.

thanks.

system stuck when mv normal file to redirfs dir

for example:
insmod redirfs.ko
insmod dummyflt.ko
echo "c" > /sys/fs/redirfs/filters/dummyflt/paths
echo -n "a:i:/home/work/test" > /sys/fs/redirfs/filters/dummyflt/paths
mv /home/work/normal/hhh > /home/work/test/
//then the system will stuck，and dmesg will print

[ +0.000008] [] schedule_preempt_disabled+0x29/0x70
[ +0.000003] [] __mutex_lock_slowpath+0xc7/0x1d0
[ +0.000003] [] mutex_lock+0x1f/0x2f
[ +0.000008] [] rfs_inode_mutex_lock+0x1e/0x20 [redirfs]
[ +0.000006] [] rfs_dcache_get_subs_mutex+0x47/0x77 [redirfs]
[ +0.000045] [] rfs_dcache_walk+0x10d/0x1fd [redirfs]
[ +0.000056] [] ? rfs_dcache_add_dir+0x62/0x62 [redirfs]
[ +0.000004] [] ? kzalloc+0x33/0x35 [redirfs]
[ +0.000006] [] rfs_info_add+0x79/0xa9 [redirfs]
[ +0.000004] [] rfs_fsrename_add_dentry+0x166/0x1c8 [redirfs]
[ +0.000002] [] rfs_fsrename_add+0xda/0xee [redirfs]
[ +0.000002] [] rfs_fsrename+0x14e/0x192 [redirfs]
[ +0.000004] [] rfs_rename+0x291/0x3d2 [redirfs]
[ +0.000003] [] vfs_rename+0x82d/0x880
[ +0.000002] [] SYSC_renameat2+0x503/0x5a0
[ +0.000002] [] ? __do_page_fault+0x1bc/0x4f0
[ +0.000002] [] SyS_renameat2+0xe/0x10
[ +0.000002] [] system_call_fastpath+0x1c/0x21

win like path

Can redirfs redict win path to file on Linux? for exmaple,
One process read "c:\test\test.txt"， the read file it read located "/mnt/test/test.txt"

With mvflt.ko module, move-in operation failed on CentOS 7

Dear @slavaim,

Threre is OS version problem during move-in test with mvflt sample module.
On the CentOS 6.4, rename_in callback captures move-in operation, successfully.
But on the CentOS 7.6, move-in operation failed. mv command be hung.

dmesg log:

2월 13 10:17:12 drbd9-01 kernel: INFO: task mv:44495 blocked for more than 120 seconds.
2월 13 10:17:12 drbd9-01 kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
2월 13 10:17:12 drbd9-01 kernel: mv D ffff8b18d9dcb200 0 44495 21987 0x00000080
2월 13 10:17:12 drbd9-01 kernel: Call Trace:
2월 13 10:17:12 drbd9-01 kernel: [] schedule_preempt_disabled+0x29/0x70
2월 13 10:17:12 drbd9-01 kernel: [] __mutex_lock_slowpath+0xc7/0x1d0
2월 13 10:17:12 drbd9-01 kernel: [] mutex_lock+0x1f/0x2f
2월 13 10:17:12 drbd9-01 kernel: [] rfs_dcache_walk+0xaf/0x2a0 [redirfs]
2월 13 10:17:12 drbd9-01 kernel: [] ? kmem_cache_alloc_trace+0x1d6/0x200
2월 13 10:17:12 drbd9-01 kernel: [] ? rfs_dcache_add_dir+0x40/0x40 [redirfs]
2월 13 10:17:12 drbd9-01 kernel: [] rfs_info_add+0x36/0x80 [redirfs]
2월 13 10:17:12 drbd9-01 kernel: [] rfs_fsrename+0x497/0x670 [redirfs]
2월 13 10:17:12 drbd9-01 kernel: [] rfs_rename+0x27d/0x3e0 [redirfs]
2월 13 10:17:12 drbd9-01 kernel: [] ? legitimize_mnt+0x16/0x40
2월 13 10:17:12 drbd9-01 kernel: [] ? mutex_lock+0x12/0x2f
2월 13 10:17:12 drbd9-01 kernel: [] vfs_rename+0x82d/0x880
2월 13 10:17:12 drbd9-01 kernel: [] SYSC_renameat2+0x503/0x5a0
2월 13 10:17:12 drbd9-01 kernel: [] ? system_call_after_swapgs+0xae/0x146
2월 13 10:17:12 drbd9-01 kernel: [] ? system_call_after_swapgs+0xa2/0x146
2월 13 10:17:12 drbd9-01 kernel: [] ? system_call_after_swapgs+0xae/0x146
2월 13 10:17:12 drbd9-01 kernel: [] ? system_call_after_swapgs+0xa2/0x146
2월 13 10:17:12 drbd9-01 kernel: [] ? system_call_after_swapgs+0xae/0x146
2월 13 10:17:12 drbd9-01 kernel: [] ? system_call_after_swapgs+0xa2/0x146
2월 13 10:17:12 drbd9-01 kernel: [] SyS_renameat2+0xe/0x10
2월 13 10:17:12 drbd9-01 kernel: [] system_call_fastpath+0x22/0x27
2월 13 10:17:12 drbd9-01 kernel: [] ? system_call_after_swapgs+0xae/0x146

Thanks.

kernel panic as soon as reboot

There is kernel panic if reboot without deleting path.

How to rmmod redirfs.ko

HI， I want ro rmmod redirfs.ko, and add the code:
"
static void __exit rfs_exit(void)
{
rfs_sysfs_delete();
rfs_file_cache_destory();
rfs_inode_cache_destroy();
rfs_dentry_cache_destory();
if (rfs_info_none)
rfs_info_put(rfs_info_none);
}
module_exit(rfs_exit);
"

/rfsctl/rfsctl -f roflt -i /home/test
/rfsctl/rfsctl -f roflt -R /home/test
/rfsctl/rfsctl -f roflt -u
rmmod roflt.ko
rmmod redirfs.ko
The system has been restarted， How can i rmmod redirfs.ko

System total freeze when used with COMODO antivirus

I installed kernel 4.14.78 to test the current RedirFS version with the COMODO antivirus. The original RedirFS driver provided with it only compiles on kernel < 2.6, so I wanted to give it a try. Successfully compiled, installed and loaded the module, but 2 seconds after "modprobe avflt" and COMODO starting to use the redirfs and avflt modules, the whole system freezes as if it cannot read any file from the disk. I cannot even login from other tty, the only solution being to restart and boot with another kernel.
Is it not compatible with COMODO for some reason?