LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
* KASAN: use-after-free Read in tomoyo_realpath_from_path
@ 2019-06-05 18:42 syzbot
2019-06-05 22:09 ` Tetsuo Handa
` (2 more replies)
0 siblings, 3 replies; 13+ messages in thread
From: syzbot @ 2019-06-05 18:42 UTC (permalink / raw)
To: jmorris, linux-kernel, linux-security-module, penguin-kernel,
serge, syzkaller-bugs, takedakn
Hello,
syzbot found the following crash on:
HEAD commit: 788a0249 Merge tag 'arc-5.2-rc4' of git://git.kernel.org/p..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=179848d4a00000
kernel config: https://syzkaller.appspot.com/x/.config?x=60564cb52ab29d5b
dashboard link: https://syzkaller.appspot.com/bug?extid=0341f6a4d729d4e0acf1
compiler: gcc (GCC) 9.0.0 20181231 (experimental)
syz repro: https://syzkaller.appspot.com/x/repro.syz?x=13ac35baa00000
IMPORTANT: if you fix the bug, please add the following tag to the commit:
Reported-by: syzbot+0341f6a4d729d4e0acf1@syzkaller.appspotmail.com
==================================================================
BUG: KASAN: use-after-free in tomoyo_get_socket_name
security/tomoyo/realpath.c:238 [inline]
BUG: KASAN: use-after-free in tomoyo_realpath_from_path+0x722/0x7a0
security/tomoyo/realpath.c:284
Read of size 2 at addr ffff8880a91276d0 by task syz-executor.3/17397
CPU: 0 PID: 17397 Comm: syz-executor.3 Not tainted 5.2.0-rc3+ #12
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
Google 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:77 [inline]
dump_stack+0x172/0x1f0 lib/dump_stack.c:113
print_address_description.cold+0x7c/0x20d mm/kasan/report.c:188
__kasan_report.cold+0x1b/0x40 mm/kasan/report.c:317
kasan_report+0x12/0x20 mm/kasan/common.c:614
__asan_report_load2_noabort+0x14/0x20 mm/kasan/generic_report.c:130
tomoyo_get_socket_name security/tomoyo/realpath.c:238 [inline]
tomoyo_realpath_from_path+0x722/0x7a0 security/tomoyo/realpath.c:284
tomoyo_get_realpath security/tomoyo/file.c:151 [inline]
tomoyo_check_open_permission+0x2a8/0x3f0 security/tomoyo/file.c:771
tomoyo_file_open security/tomoyo/tomoyo.c:319 [inline]
tomoyo_file_open+0xa9/0xd0 security/tomoyo/tomoyo.c:314
security_file_open+0x71/0x300 security/security.c:1454
do_dentry_open+0x373/0x1250 fs/open.c:765
vfs_open+0xa0/0xd0 fs/open.c:887
do_last fs/namei.c:3416 [inline]
path_openat+0x10e9/0x46d0 fs/namei.c:3533
do_filp_open+0x1a1/0x280 fs/namei.c:3563
do_sys_open+0x3fe/0x5d0 fs/open.c:1070
__do_sys_open fs/open.c:1088 [inline]
__se_sys_open fs/open.c:1083 [inline]
__x64_sys_open+0x7e/0xc0 fs/open.c:1083
do_syscall_64+0xfd/0x680 arch/x86/entry/common.c:301
entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x413161
Code: 75 14 b8 02 00 00 00 0f 05 48 3d 01 f0 ff ff 0f 83 04 19 00 00 c3 48
83 ec 08 e8 0a fa ff ff 48 89 04 24 b8 02 00 00 00 0f 05 <48> 8b 3c 24 48
89 c2 e8 53 fa ff ff 48 89 d0 48 83 c4 08 48 3d 01
RSP: 002b:00007f65230f8bb0 EFLAGS: 00000293 ORIG_RAX: 0000000000000002
RAX: ffffffffffffffda RBX: 0000000000000002 RCX: 0000000000413161
RDX: fffffffffffffffa RSI: 0000000000000000 RDI: 00007f65230f8bd0
RBP: 000000000075c060 R08: 0000000000000050 R09: 000000000000000f
R10: 0000000000000004 R11: 0000000000000293 R12: 00007f65230f96d4
R13: 00000000004c83f6 R14: 00000000004dea40 R15: 00000000ffffffff
Allocated by task 17373:
save_stack+0x23/0x90 mm/kasan/common.c:71
set_track mm/kasan/common.c:79 [inline]
__kasan_kmalloc mm/kasan/common.c:489 [inline]
__kasan_kmalloc.constprop.0+0xcf/0xe0 mm/kasan/common.c:462
kasan_kmalloc+0x9/0x10 mm/kasan/common.c:503
__do_kmalloc mm/slab.c:3660 [inline]
__kmalloc+0x15c/0x740 mm/slab.c:3669
kmalloc include/linux/slab.h:552 [inline]
sk_prot_alloc+0x19c/0x2e0 net/core/sock.c:1602
sk_alloc+0x39/0xf70 net/core/sock.c:1656
base_sock_create drivers/isdn/mISDN/socket.c:758 [inline]
mISDN_sock_create+0xb4/0x3a0 drivers/isdn/mISDN/socket.c:780
__sock_create+0x3d8/0x730 net/socket.c:1424
sock_create net/socket.c:1475 [inline]
__sys_socket+0x103/0x220 net/socket.c:1517
__do_sys_socket net/socket.c:1526 [inline]
__se_sys_socket net/socket.c:1524 [inline]
__x64_sys_socket+0x73/0xb0 net/socket.c:1524
do_syscall_64+0xfd/0x680 arch/x86/entry/common.c:301
entry_SYSCALL_64_after_hwframe+0x49/0xbe
Freed by task 17371:
save_stack+0x23/0x90 mm/kasan/common.c:71
set_track mm/kasan/common.c:79 [inline]
__kasan_slab_free+0x102/0x150 mm/kasan/common.c:451
kasan_slab_free+0xe/0x10 mm/kasan/common.c:459
__cache_free mm/slab.c:3432 [inline]
kfree+0xcf/0x220 mm/slab.c:3755
sk_prot_free net/core/sock.c:1639 [inline]
__sk_destruct+0x4f7/0x6e0 net/core/sock.c:1725
sk_destruct+0x7b/0x90 net/core/sock.c:1733
__sk_free+0xce/0x300 net/core/sock.c:1744
sk_free+0x42/0x50 net/core/sock.c:1755
sock_put include/net/sock.h:1723 [inline]
base_sock_release+0x269/0x279 drivers/isdn/mISDN/socket.c:628
__sock_release+0xce/0x2a0 net/socket.c:601
sock_close+0x1b/0x30 net/socket.c:1273
__fput+0x2ff/0x890 fs/file_table.c:280
____fput+0x16/0x20 fs/file_table.c:313
task_work_run+0x145/0x1c0 kernel/task_work.c:113
tracehook_notify_resume include/linux/tracehook.h:185 [inline]
exit_to_usermode_loop+0x273/0x2c0 arch/x86/entry/common.c:168
prepare_exit_to_usermode arch/x86/entry/common.c:199 [inline]
syscall_return_slowpath arch/x86/entry/common.c:279 [inline]
do_syscall_64+0x58e/0x680 arch/x86/entry/common.c:304
entry_SYSCALL_64_after_hwframe+0x49/0xbe
The buggy address belongs to the object at ffff8880a91276c0
which belongs to the cache kmalloc-2k of size 2048
The buggy address is located 16 bytes inside of
2048-byte region [ffff8880a91276c0, ffff8880a9127ec0)
The buggy address belongs to the page:
page:ffffea0002a44980 refcount:1 mapcount:0 mapping:ffff8880aa400c40
index:0x0 compound_mapcount: 0
flags: 0x1fffc0000010200(slab|head)
raw: 01fffc0000010200 ffffea00022c9f88 ffffea0002234408 ffff8880aa400c40
raw: 0000000000000000 ffff8880a91265c0 0000000100000003 0000000000000000
page dumped because: kasan: bad access detected
Memory state around the buggy address:
ffff8880a9127580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
ffff8880a9127600: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
> ffff8880a9127680: fc fc fc fc fc fc fc fc fb fb fb fb fb fb fb fb
^
ffff8880a9127700: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
ffff8880a9127780: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================
---
This bug is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzkaller@googlegroups.com.
syzbot will keep track of this bug report. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
syzbot can test patches for this bug, for details see:
https://goo.gl/tpsmEJ#testing-patches
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: KASAN: use-after-free Read in tomoyo_realpath_from_path
2019-06-05 18:42 KASAN: use-after-free Read in tomoyo_realpath_from_path syzbot
@ 2019-06-05 22:09 ` Tetsuo Handa
2019-06-06 2:08 ` Tetsuo Handa
2019-06-06 5:20 ` Tetsuo Handa
2 siblings, 0 replies; 13+ messages in thread
From: Tetsuo Handa @ 2019-06-05 22:09 UTC (permalink / raw)
To: Alexander Viro
Cc: syzbot, jmorris, linux-kernel, linux-security-module, serge,
syzkaller-bugs, takedakn, David S. Miller
Hello, Al.
syzbot found that SOCKET_I(d_backing_inode("struct path"->dentry))->sk
was already kfree()d when trying to calculate pathname for open().
"struct path"->dentry should remain valid but portion of memory
reachable via inode already became invalid. What should we do?
On 2019/06/06 3:42, syzbot wrote:
> Hello,
>
> syzbot found the following crash on:
>
> HEAD commit: 788a0249 Merge tag 'arc-5.2-rc4' of git://git.kernel.org/p..
> git tree: upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=179848d4a00000
> kernel config: https://syzkaller.appspot.com/x/.config?x=60564cb52ab29d5b
> dashboard link: https://syzkaller.appspot.com/bug?extid=0341f6a4d729d4e0acf1
> compiler: gcc (GCC) 9.0.0 20181231 (experimental)
> syz repro: https://syzkaller.appspot.com/x/repro.syz?x=13ac35baa00000
>
> IMPORTANT: if you fix the bug, please add the following tag to the commit:
> Reported-by: syzbot+0341f6a4d729d4e0acf1@syzkaller.appspotmail.com
>
> ==================================================================
> BUG: KASAN: use-after-free in tomoyo_get_socket_name security/tomoyo/realpath.c:238 [inline]
> BUG: KASAN: use-after-free in tomoyo_realpath_from_path+0x722/0x7a0 security/tomoyo/realpath.c:284
> Read of size 2 at addr ffff8880a91276d0 by task syz-executor.3/17397
>
> CPU: 0 PID: 17397 Comm: syz-executor.3 Not tainted 5.2.0-rc3+ #12
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> Call Trace:
> __dump_stack lib/dump_stack.c:77 [inline]
> dump_stack+0x172/0x1f0 lib/dump_stack.c:113
> print_address_description.cold+0x7c/0x20d mm/kasan/report.c:188
> __kasan_report.cold+0x1b/0x40 mm/kasan/report.c:317
> kasan_report+0x12/0x20 mm/kasan/common.c:614
> __asan_report_load2_noabort+0x14/0x20 mm/kasan/generic_report.c:130
> tomoyo_get_socket_name security/tomoyo/realpath.c:238 [inline]
> tomoyo_realpath_from_path+0x722/0x7a0 security/tomoyo/realpath.c:284
> tomoyo_get_realpath security/tomoyo/file.c:151 [inline]
> tomoyo_check_open_permission+0x2a8/0x3f0 security/tomoyo/file.c:771
> tomoyo_file_open security/tomoyo/tomoyo.c:319 [inline]
> tomoyo_file_open+0xa9/0xd0 security/tomoyo/tomoyo.c:314
> security_file_open+0x71/0x300 security/security.c:1454
> do_dentry_open+0x373/0x1250 fs/open.c:765
> vfs_open+0xa0/0xd0 fs/open.c:887
> do_last fs/namei.c:3416 [inline]
> path_openat+0x10e9/0x46d0 fs/namei.c:3533
> do_filp_open+0x1a1/0x280 fs/namei.c:3563
> do_sys_open+0x3fe/0x5d0 fs/open.c:1070
> __do_sys_open fs/open.c:1088 [inline]
> __se_sys_open fs/open.c:1083 [inline]
> __x64_sys_open+0x7e/0xc0 fs/open.c:1083
> do_syscall_64+0xfd/0x680 arch/x86/entry/common.c:301
> entry_SYSCALL_64_after_hwframe+0x49/0xbe
> RIP: 0033:0x413161
> Code: 75 14 b8 02 00 00 00 0f 05 48 3d 01 f0 ff ff 0f 83 04 19 00 00 c3 48 83 ec 08 e8 0a fa ff ff 48 89 04 24 b8 02 00 00 00 0f 05 <48> 8b 3c 24 48 89 c2 e8 53 fa ff ff 48 89 d0 48 83 c4 08 48 3d 01
> RSP: 002b:00007f65230f8bb0 EFLAGS: 00000293 ORIG_RAX: 0000000000000002
> RAX: ffffffffffffffda RBX: 0000000000000002 RCX: 0000000000413161
> RDX: fffffffffffffffa RSI: 0000000000000000 RDI: 00007f65230f8bd0
> RBP: 000000000075c060 R08: 0000000000000050 R09: 000000000000000f
> R10: 0000000000000004 R11: 0000000000000293 R12: 00007f65230f96d4
> R13: 00000000004c83f6 R14: 00000000004dea40 R15: 00000000ffffffff
>
> Allocated by task 17373:
> save_stack+0x23/0x90 mm/kasan/common.c:71
> set_track mm/kasan/common.c:79 [inline]
> __kasan_kmalloc mm/kasan/common.c:489 [inline]
> __kasan_kmalloc.constprop.0+0xcf/0xe0 mm/kasan/common.c:462
> kasan_kmalloc+0x9/0x10 mm/kasan/common.c:503
> __do_kmalloc mm/slab.c:3660 [inline]
> __kmalloc+0x15c/0x740 mm/slab.c:3669
> kmalloc include/linux/slab.h:552 [inline]
> sk_prot_alloc+0x19c/0x2e0 net/core/sock.c:1602
> sk_alloc+0x39/0xf70 net/core/sock.c:1656
> base_sock_create drivers/isdn/mISDN/socket.c:758 [inline]
> mISDN_sock_create+0xb4/0x3a0 drivers/isdn/mISDN/socket.c:780
> __sock_create+0x3d8/0x730 net/socket.c:1424
> sock_create net/socket.c:1475 [inline]
> __sys_socket+0x103/0x220 net/socket.c:1517
> __do_sys_socket net/socket.c:1526 [inline]
> __se_sys_socket net/socket.c:1524 [inline]
> __x64_sys_socket+0x73/0xb0 net/socket.c:1524
> do_syscall_64+0xfd/0x680 arch/x86/entry/common.c:301
> entry_SYSCALL_64_after_hwframe+0x49/0xbe
>
> Freed by task 17371:
> save_stack+0x23/0x90 mm/kasan/common.c:71
> set_track mm/kasan/common.c:79 [inline]
> __kasan_slab_free+0x102/0x150 mm/kasan/common.c:451
> kasan_slab_free+0xe/0x10 mm/kasan/common.c:459
> __cache_free mm/slab.c:3432 [inline]
> kfree+0xcf/0x220 mm/slab.c:3755
> sk_prot_free net/core/sock.c:1639 [inline]
> __sk_destruct+0x4f7/0x6e0 net/core/sock.c:1725
> sk_destruct+0x7b/0x90 net/core/sock.c:1733
> __sk_free+0xce/0x300 net/core/sock.c:1744
> sk_free+0x42/0x50 net/core/sock.c:1755
> sock_put include/net/sock.h:1723 [inline]
> base_sock_release+0x269/0x279 drivers/isdn/mISDN/socket.c:628
> __sock_release+0xce/0x2a0 net/socket.c:601
> sock_close+0x1b/0x30 net/socket.c:1273
> __fput+0x2ff/0x890 fs/file_table.c:280
> ____fput+0x16/0x20 fs/file_table.c:313
> task_work_run+0x145/0x1c0 kernel/task_work.c:113
> tracehook_notify_resume include/linux/tracehook.h:185 [inline]
> exit_to_usermode_loop+0x273/0x2c0 arch/x86/entry/common.c:168
> prepare_exit_to_usermode arch/x86/entry/common.c:199 [inline]
> syscall_return_slowpath arch/x86/entry/common.c:279 [inline]
> do_syscall_64+0x58e/0x680 arch/x86/entry/common.c:304
> entry_SYSCALL_64_after_hwframe+0x49/0xbe
>
> The buggy address belongs to the object at ffff8880a91276c0
> which belongs to the cache kmalloc-2k of size 2048
> The buggy address is located 16 bytes inside of
> 2048-byte region [ffff8880a91276c0, ffff8880a9127ec0)
> The buggy address belongs to the page:
> page:ffffea0002a44980 refcount:1 mapcount:0 mapping:ffff8880aa400c40 index:0x0 compound_mapcount: 0
> flags: 0x1fffc0000010200(slab|head)
> raw: 01fffc0000010200 ffffea00022c9f88 ffffea0002234408 ffff8880aa400c40
> raw: 0000000000000000 ffff8880a91265c0 0000000100000003 0000000000000000
> page dumped because: kasan: bad access detected
>
> Memory state around the buggy address:
> ffff8880a9127580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> ffff8880a9127600: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
>> ffff8880a9127680: fc fc fc fc fc fc fc fc fb fb fb fb fb fb fb fb
> ^
> ffff8880a9127700: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> ffff8880a9127780: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
> ==================================================================
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: KASAN: use-after-free Read in tomoyo_realpath_from_path
2019-06-05 18:42 KASAN: use-after-free Read in tomoyo_realpath_from_path syzbot
2019-06-05 22:09 ` Tetsuo Handa
@ 2019-06-06 2:08 ` Tetsuo Handa
2019-06-06 5:20 ` Tetsuo Handa
2 siblings, 0 replies; 13+ messages in thread
From: Tetsuo Handa @ 2019-06-06 2:08 UTC (permalink / raw)
To: Al Viro, linux-fsdevel
Cc: syzbot, jmorris, linux-kernel, linux-security-module, serge,
syzkaller-bugs, takedakn
Here is a reproducer.
The problem is that TOMOYO is accessing already freed socket from security_file_open()
which later fails with -ENXIO (because we can't get file descriptor of sockets via
/proc/pid/fd/n interface), and the file descriptor is getting released before
security_file_open() completes because we do not raise "struct file"->f_count of
the file which is accessible via /proc/pid/fd/n interface. We can avoid this problem
if we can avoid calling security_file_open() which after all fails with -ENXIO.
How should we handle this race? Let LSM modules check if security_file_open() was
called on a socket?
----------------------------------------
diff --git a/fs/open.c b/fs/open.c
index b5b80469b93d..995ffcb37128 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -765,6 +765,12 @@ static int do_dentry_open(struct file *f,
error = security_file_open(f);
if (error)
goto cleanup_all;
+ if (!strcmp(current->comm, "a.out") &&
+ f->f_path.dentry->d_sb->s_magic == SOCKFS_MAGIC) {
+ printk("Start open(socket) delay\n");
+ schedule_timeout_killable(HZ * 5);
+ printk("End open(socket) delay\n");
+ }
error = break_lease(locks_inode(f), f->f_flags);
if (error)
----------------------------------------
----------------------------------------
#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
#include <sys/socket.h>
int main(int argc, char *argv[])
{
pid_t pid = getpid();
int fd = socket(AF_ISDN, SOCK_RAW, 0);
char buffer[128] = { };
if (fork() == 0) {
close(fd);
snprintf(buffer, sizeof(buffer) - 1, "/proc/%u/fd/%u", pid, fd);
open(buffer, 3);
_exit(0);
}
sleep(2);
close(fd);
return 0;
}
----------------------------------------
----------------------------------------
getpid() = 32504
socket(AF_ISDN, SOCK_RAW, 0) = 3
clone(strace: Process 32505 attached
child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x7efea30dda10) = 32505
[pid 32504] rt_sigprocmask(SIG_BLOCK, [CHLD], <unfinished ...>
[pid 32505] close(3 <unfinished ...>
[pid 32504] <... rt_sigprocmask resumed> [], 8) = 0
[pid 32505] <... close resumed> ) = 0
[pid 32504] rt_sigaction(SIGCHLD, NULL, {SIG_DFL, [], 0}, 8) = 0
[pid 32505] open("/proc/32504/fd/3", O_ACCMODE <unfinished ...>
[pid 32504] rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
[pid 32504] nanosleep({2, 0}, 0x7ffd3c608150) = 0
[pid 32504] close(3) = 0
[pid 32504] exit_group(0) = ?
[pid 32504] +++ exited with 0 +++
<... open resumed> ) = -1 ENXIO (No such device or address)
exit_group(0) = ?
----------------------------------------
----------------------------------------
[ 95.109628] Start open(socket) delay
[ 97.113150] base_sock_release(00000000506a3239) sk=00000000016d0ceb
[ 100.142235] End open(socket) delay
----------------------------------------
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: KASAN: use-after-free Read in tomoyo_realpath_from_path
2019-06-05 18:42 KASAN: use-after-free Read in tomoyo_realpath_from_path syzbot
2019-06-05 22:09 ` Tetsuo Handa
2019-06-06 2:08 ` Tetsuo Handa
@ 2019-06-06 5:20 ` Tetsuo Handa
2019-06-09 6:41 ` [PATCH] tomoyo: Don't check open/getattr permission on sockets Tetsuo Handa
2 siblings, 1 reply; 13+ messages in thread
From: Tetsuo Handa @ 2019-06-06 5:20 UTC (permalink / raw)
To: Al Viro, linux-fsdevel
Cc: syzbot, jmorris, linux-kernel, linux-security-module, serge,
syzkaller-bugs, takedakn
Tetsuo Handa wrote:
> The problem is that TOMOYO is accessing already freed socket from security_file_open()
> which later fails with -ENXIO (because we can't get file descriptor of sockets via
> /proc/pid/fd/n interface), and the file descriptor is getting released before
> security_file_open() completes because we do not raise "struct file"->f_count of
> the file which is accessible via /proc/pid/fd/n interface. We can avoid this problem
> if we can avoid calling security_file_open() which after all fails with -ENXIO.
> How should we handle this race? Let LSM modules check if security_file_open() was
> called on a socket?
Well, just refusing security_file_open() is not sufficient, for open(O_PATH) allows installing
file descriptor where SOCKET_I(inode)->sk can change at any moment, and TOMOYO cannot tell
whether it is safe to access SOCKET_I(inode)->sk from security_inode_getattr().
But refusing open(O_PATH) as well might break userspace programs. Oh, no...
----------------------------------------
diff --git a/fs/open.c b/fs/open.c
index b5b80469b93d..ea69668e2cd8 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -728,6 +728,16 @@ static int do_dentry_open(struct file *f,
/* Ensure that we skip any errors that predate opening of the file */
f->f_wb_err = filemap_sample_wb_err(f->f_mapping);
+ /*
+ * Sockets must not be opened via /proc/pid/fd/n, even with O_PATH,
+ * for SOCKET_I(inode)->sk can be kfree()d at any moment after a file
+ * descriptor obtained by opening /proc/pid/fd/n was installed.
+ */
+ if (unlikely(S_ISSOCK(inode->i_mode))) {
+ error = (f->f_flags & O_PATH) ? -ENOENT : -ENXIO;
+ goto cleanup_file;
+ }
+
if (unlikely(f->f_flags & O_PATH)) {
f->f_mode = FMODE_PATH | FMODE_OPENED;
f->f_op = &empty_fops;
----------------------------------------
----------------------------------------
#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
#include <sys/socket.h>
int main(int argc, char *argv[])
{
pid_t pid = getpid();
int fd = socket(AF_INET, SOCK_STREAM, 0);
char buffer[128] = { };
if (fork() == 0) {
struct stat buf = { };
close(fd);
snprintf(buffer, sizeof(buffer) - 1, "/proc/%u/fd/%u", pid, fd);
fd = open(buffer, __O_PATH);
sleep(5);
fstat(fd, &buf);
_exit(0);
}
sleep(2);
close(fd);
return 0;
}
----------------------------------------
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH] tomoyo: Don't check open/getattr permission on sockets.
2019-06-06 5:20 ` Tetsuo Handa
@ 2019-06-09 6:41 ` Tetsuo Handa
2019-06-16 6:49 ` Tetsuo Handa
0 siblings, 1 reply; 13+ messages in thread
From: Tetsuo Handa @ 2019-06-09 6:41 UTC (permalink / raw)
To: Al Viro, linux-fsdevel
Cc: syzbot, jmorris, linux-kernel, linux-security-module, serge,
syzkaller-bugs, takedakn
syzbot is reporting that use of SOCKET_I()->sk from open() can result in
use after free problem [1], for socket's inode is still reachable via
/proc/pid/fd/n despite destruction of SOCKET_I()->sk already completed.
But there is no point with calling security_file_open() on sockets
because open("/proc/pid/fd/n", !O_PATH) on sockets fails with -ENXIO.
There is some point with calling security_inode_getattr() on sockets
because stat("/proc/pid/fd/n") and fstat(open("/proc/pid/fd/n", O_PATH))
are valid. If we want to access "struct sock"->sk_{family,type,protocol}
fields, we will need to use security_socket_post_create() hook and
security_inode_free() hook in order to remember these fields because
security_sk_free() hook is called before the inode is destructed. But
since information which can be protected by checking
security_inode_getattr() on sockets is trivial, let's not be bothered by
"struct inode"->i_security management.
There is point with calling security_file_ioctl() on sockets. Since
ioctl(open("/proc/pid/fd/n", O_PATH)) is invalid, security_file_ioctl()
on sockets should remain safe.
[1] https://syzkaller.appspot.com/bug?id=73d590010454403d55164cca23bd0565b1eb3b74
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Reported-by: syzbot <syzbot+0341f6a4d729d4e0acf1@syzkaller.appspotmail.com>
---
security/tomoyo/tomoyo.c | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/security/tomoyo/tomoyo.c b/security/tomoyo/tomoyo.c
index 716c92e..9661b86 100644
--- a/security/tomoyo/tomoyo.c
+++ b/security/tomoyo/tomoyo.c
@@ -126,6 +126,9 @@ static int tomoyo_bprm_check_security(struct linux_binprm *bprm)
*/
static int tomoyo_inode_getattr(const struct path *path)
{
+ /* It is not safe to call tomoyo_get_socket_name(). */
+ if (path->dentry->d_inode && S_ISSOCK(path->dentry->d_inode->i_mode))
+ return 0;
return tomoyo_path_perm(TOMOYO_TYPE_GETATTR, path, NULL);
}
@@ -316,6 +319,10 @@ static int tomoyo_file_open(struct file *f)
/* Don't check read permission here if called from do_execve(). */
if (current->in_execve)
return 0;
+ /* Sockets can't be opened by open(). */
+ if (f->f_path.dentry->d_inode &&
+ S_ISSOCK(f->f_path.dentry->d_inode->i_mode))
+ return 0;
return tomoyo_check_open_permission(tomoyo_domain(), &f->f_path,
f->f_flags);
}
--
1.8.3.1
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH] tomoyo: Don't check open/getattr permission on sockets.
2019-06-09 6:41 ` [PATCH] tomoyo: Don't check open/getattr permission on sockets Tetsuo Handa
@ 2019-06-16 6:49 ` Tetsuo Handa
2019-06-18 20:49 ` Al Viro
0 siblings, 1 reply; 13+ messages in thread
From: Tetsuo Handa @ 2019-06-16 6:49 UTC (permalink / raw)
To: Al Viro, linux-fsdevel
Cc: syzbot, jmorris, linux-kernel, linux-security-module, serge,
syzkaller-bugs, takedakn, David S. Miller
Hello, Al.
Q1: Do you agree that we should fix TOMOYO side rather than SOCKET_I()->sk
management.
Q2: Do you see any problem with using f->f_path.dentry->d_inode ?
Do we need to use d_backing_inode() or d_inode() ?
Regards.
On 2019/06/09 15:41, Tetsuo Handa wrote:
> syzbot is reporting that use of SOCKET_I()->sk from open() can result in
> use after free problem [1], for socket's inode is still reachable via
> /proc/pid/fd/n despite destruction of SOCKET_I()->sk already completed.
>
> But there is no point with calling security_file_open() on sockets
> because open("/proc/pid/fd/n", !O_PATH) on sockets fails with -ENXIO.
>
> There is some point with calling security_inode_getattr() on sockets
> because stat("/proc/pid/fd/n") and fstat(open("/proc/pid/fd/n", O_PATH))
> are valid. If we want to access "struct sock"->sk_{family,type,protocol}
> fields, we will need to use security_socket_post_create() hook and
> security_inode_free() hook in order to remember these fields because
> security_sk_free() hook is called before the inode is destructed. But
> since information which can be protected by checking
> security_inode_getattr() on sockets is trivial, let's not be bothered by
> "struct inode"->i_security management.
>
> There is point with calling security_file_ioctl() on sockets. Since
> ioctl(open("/proc/pid/fd/n", O_PATH)) is invalid, security_file_ioctl()
> on sockets should remain safe.
>
> [1] https://syzkaller.appspot.com/bug?id=73d590010454403d55164cca23bd0565b1eb3b74
>
> Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
> Reported-by: syzbot <syzbot+0341f6a4d729d4e0acf1@syzkaller.appspotmail.com>
> ---
> security/tomoyo/tomoyo.c | 7 +++++++
> 1 file changed, 7 insertions(+)
>
> diff --git a/security/tomoyo/tomoyo.c b/security/tomoyo/tomoyo.c
> index 716c92e..9661b86 100644
> --- a/security/tomoyo/tomoyo.c
> +++ b/security/tomoyo/tomoyo.c
> @@ -126,6 +126,9 @@ static int tomoyo_bprm_check_security(struct linux_binprm *bprm)
> */
> static int tomoyo_inode_getattr(const struct path *path)
> {
> + /* It is not safe to call tomoyo_get_socket_name(). */
> + if (path->dentry->d_inode && S_ISSOCK(path->dentry->d_inode->i_mode))
> + return 0;
> return tomoyo_path_perm(TOMOYO_TYPE_GETATTR, path, NULL);
> }
>
> @@ -316,6 +319,10 @@ static int tomoyo_file_open(struct file *f)
> /* Don't check read permission here if called from do_execve(). */
> if (current->in_execve)
> return 0;
> + /* Sockets can't be opened by open(). */
> + if (f->f_path.dentry->d_inode &&
> + S_ISSOCK(f->f_path.dentry->d_inode->i_mode))
> + return 0;
> return tomoyo_check_open_permission(tomoyo_domain(), &f->f_path,
> f->f_flags);
> }
>
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH] tomoyo: Don't check open/getattr permission on sockets.
2019-06-16 6:49 ` Tetsuo Handa
@ 2019-06-18 20:49 ` Al Viro
2019-06-22 4:45 ` [PATCH v2] " Tetsuo Handa
0 siblings, 1 reply; 13+ messages in thread
From: Al Viro @ 2019-06-18 20:49 UTC (permalink / raw)
To: Tetsuo Handa
Cc: linux-fsdevel, syzbot, jmorris, linux-kernel,
linux-security-module, serge, syzkaller-bugs, takedakn,
David S. Miller
On Sun, Jun 16, 2019 at 03:49:00PM +0900, Tetsuo Handa wrote:
> Hello, Al.
>
> Q1: Do you agree that we should fix TOMOYO side rather than SOCKET_I()->sk
> management.
You do realize that sockets are not unique in that respect, right?
All kinds of interesting stuff can be accessed via /proc/*/fd/*, and
it _can_ be closed under you. So I'd suggest checking how your code
copes with similar for pipes, FIFOs, epoll, etc., accessed that way...
We are _not_ going to be checking that in fs/open.c - the stuff found
via /proc/*/fd/* can have the associated file closed by the time
we get to calling ->open() and we won't know that until said call.
> Q2: Do you see any problem with using f->f_path.dentry->d_inode ?
> Do we need to use d_backing_inode() or d_inode() ?
Huh? What's wrong with file_inode(f), in the first place? And
just when can that be NULL, while we are at it?
> > static int tomoyo_inode_getattr(const struct path *path)
> > {
> > + /* It is not safe to call tomoyo_get_socket_name(). */
> > + if (path->dentry->d_inode && S_ISSOCK(path->dentry->d_inode->i_mode))
> > + return 0;
Can that be called for a negative?
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v2] tomoyo: Don't check open/getattr permission on sockets.
2019-06-18 20:49 ` Al Viro
@ 2019-06-22 4:45 ` Tetsuo Handa
2019-08-22 6:30 ` Eric Biggers
0 siblings, 1 reply; 13+ messages in thread
From: Tetsuo Handa @ 2019-06-22 4:45 UTC (permalink / raw)
To: Al Viro
Cc: linux-fsdevel, syzbot, jmorris, linux-kernel,
linux-security-module, serge, syzkaller-bugs, takedakn,
David S. Miller
On 2019/06/19 5:49, Al Viro wrote:
> On Sun, Jun 16, 2019 at 03:49:00PM +0900, Tetsuo Handa wrote:
>> Hello, Al.
>>
>> Q1: Do you agree that we should fix TOMOYO side rather than SOCKET_I()->sk
>> management.
>
> You do realize that sockets are not unique in that respect, right?
> All kinds of interesting stuff can be accessed via /proc/*/fd/*, and
> it _can_ be closed under you. So I'd suggest checking how your code
> copes with similar for pipes, FIFOs, epoll, etc., accessed that way...
I know all kinds of interesting stuff can be accessed via /proc/*/fd/*,
and it _can_ be closed under me.
Regarding sockets, I was accessing "struct socket" memory and
"struct sock" memory which are outside of "struct inode" memory.
But regarding other objects, I am accessing "struct dentry" memory,
"struct super_block" memory and "struct inode" memory. I'm expecting
that these memory can't be kfree()d as long as "struct path" holds
a reference. Is my expectation correct?
>
> We are _not_ going to be checking that in fs/open.c - the stuff found
> via /proc/*/fd/* can have the associated file closed by the time
> we get to calling ->open() and we won't know that until said call.
OK. Then, fixing TOMOYO side is the correct way.
>
>> Q2: Do you see any problem with using f->f_path.dentry->d_inode ?
>> Do we need to use d_backing_inode() or d_inode() ?
>
> Huh? What's wrong with file_inode(f), in the first place? And
> just when can that be NULL, while we are at it?
Oh, I was not aware of file_inode(). Thanks.
>
>>> static int tomoyo_inode_getattr(const struct path *path)
>>> {
>>> + /* It is not safe to call tomoyo_get_socket_name(). */
>>> + if (path->dentry->d_inode && S_ISSOCK(path->dentry->d_inode->i_mode))
>>> + return 0;
>
> Can that be called for a negative?
>
I check for NULL when I'm not sure it is guaranteed to hold a valid pointer.
You meant "we are sure that path->dentry->d_inode is valid", don't you?
By the way, "negative" associates with IS_ERR() range. I guess that
"NULL" is the better name...
Anyway, here is V2 patch.
From c63c4074300921d6d1c33c3b8dc9c84ebfededf5 Mon Sep 17 00:00:00 2001
From: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Date: Sat, 22 Jun 2019 13:14:26 +0900
Subject: [PATCH v2] tomoyo: Don't check open/getattr permission on sockets.
syzbot is reporting that use of SOCKET_I()->sk from open() can result in
use after free problem [1], for socket's inode is still reachable via
/proc/pid/fd/n despite destruction of SOCKET_I()->sk already completed.
But there is no point with calling security_file_open() on sockets
because open("/proc/pid/fd/n", !O_PATH) on sockets fails with -ENXIO.
There is some point with calling security_inode_getattr() on sockets
because stat("/proc/pid/fd/n") and fstat(open("/proc/pid/fd/n", O_PATH))
are valid. If we want to access "struct sock"->sk_{family,type,protocol}
fields, we will need to use security_socket_post_create() hook and
security_inode_free() hook in order to remember these fields because
security_sk_free() hook is called before the inode is destructed. But
since information which can be protected by checking
security_inode_getattr() on sockets is trivial, let's not be bothered by
"struct inode"->i_security management.
There is point with calling security_file_ioctl() on sockets. Since
ioctl(open("/proc/pid/fd/n", O_PATH)) is invalid, security_file_ioctl()
on sockets should remain safe.
[1] https://syzkaller.appspot.com/bug?id=73d590010454403d55164cca23bd0565b1eb3b74
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Reported-by: syzbot <syzbot+0341f6a4d729d4e0acf1@syzkaller.appspotmail.com>
---
security/tomoyo/tomoyo.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/security/tomoyo/tomoyo.c b/security/tomoyo/tomoyo.c
index 716c92e..8ea3f5d 100644
--- a/security/tomoyo/tomoyo.c
+++ b/security/tomoyo/tomoyo.c
@@ -126,6 +126,9 @@ static int tomoyo_bprm_check_security(struct linux_binprm *bprm)
*/
static int tomoyo_inode_getattr(const struct path *path)
{
+ /* It is not safe to call tomoyo_get_socket_name(). */
+ if (S_ISSOCK(d_inode(path->dentry)->i_mode))
+ return 0;
return tomoyo_path_perm(TOMOYO_TYPE_GETATTR, path, NULL);
}
@@ -316,6 +319,9 @@ static int tomoyo_file_open(struct file *f)
/* Don't check read permission here if called from do_execve(). */
if (current->in_execve)
return 0;
+ /* Sockets can't be opened by open(). */
+ if (S_ISSOCK(file_inode(f)->i_mode))
+ return 0;
return tomoyo_check_open_permission(tomoyo_domain(), &f->f_path,
f->f_flags);
}
--
1.8.3.1
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v2] tomoyo: Don't check open/getattr permission on sockets.
2019-06-22 4:45 ` [PATCH v2] " Tetsuo Handa
@ 2019-08-22 6:30 ` Eric Biggers
2019-08-22 6:55 ` Tetsuo Handa
0 siblings, 1 reply; 13+ messages in thread
From: Eric Biggers @ 2019-08-22 6:30 UTC (permalink / raw)
To: Tetsuo Handa
Cc: Al Viro, linux-fsdevel, syzbot, jmorris, linux-kernel,
linux-security-module, serge, syzkaller-bugs, takedakn,
David S. Miller
Hi Tetsuo,
On Sat, Jun 22, 2019 at 01:45:30PM +0900, Tetsuo Handa wrote:
> On 2019/06/19 5:49, Al Viro wrote:
> > On Sun, Jun 16, 2019 at 03:49:00PM +0900, Tetsuo Handa wrote:
> >> Hello, Al.
> >>
> >> Q1: Do you agree that we should fix TOMOYO side rather than SOCKET_I()->sk
> >> management.
> >
> > You do realize that sockets are not unique in that respect, right?
> > All kinds of interesting stuff can be accessed via /proc/*/fd/*, and
> > it _can_ be closed under you. So I'd suggest checking how your code
> > copes with similar for pipes, FIFOs, epoll, etc., accessed that way...
>
> I know all kinds of interesting stuff can be accessed via /proc/*/fd/*,
> and it _can_ be closed under me.
>
> Regarding sockets, I was accessing "struct socket" memory and
> "struct sock" memory which are outside of "struct inode" memory.
>
> But regarding other objects, I am accessing "struct dentry" memory,
> "struct super_block" memory and "struct inode" memory. I'm expecting
> that these memory can't be kfree()d as long as "struct path" holds
> a reference. Is my expectation correct?
>
> >
> > We are _not_ going to be checking that in fs/open.c - the stuff found
> > via /proc/*/fd/* can have the associated file closed by the time
> > we get to calling ->open() and we won't know that until said call.
>
> OK. Then, fixing TOMOYO side is the correct way.
>
> >
> >> Q2: Do you see any problem with using f->f_path.dentry->d_inode ?
> >> Do we need to use d_backing_inode() or d_inode() ?
> >
> > Huh? What's wrong with file_inode(f), in the first place? And
> > just when can that be NULL, while we are at it?
>
> Oh, I was not aware of file_inode(). Thanks.
>
> >
> >>> static int tomoyo_inode_getattr(const struct path *path)
> >>> {
> >>> + /* It is not safe to call tomoyo_get_socket_name(). */
> >>> + if (path->dentry->d_inode && S_ISSOCK(path->dentry->d_inode->i_mode))
> >>> + return 0;
> >
> > Can that be called for a negative?
> >
>
> I check for NULL when I'm not sure it is guaranteed to hold a valid pointer.
> You meant "we are sure that path->dentry->d_inode is valid", don't you?
>
> By the way, "negative" associates with IS_ERR() range. I guess that
> "NULL" is the better name...
>
> Anyway, here is V2 patch.
>
> From c63c4074300921d6d1c33c3b8dc9c84ebfededf5 Mon Sep 17 00:00:00 2001
> From: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
> Date: Sat, 22 Jun 2019 13:14:26 +0900
> Subject: [PATCH v2] tomoyo: Don't check open/getattr permission on sockets.
>
> syzbot is reporting that use of SOCKET_I()->sk from open() can result in
> use after free problem [1], for socket's inode is still reachable via
> /proc/pid/fd/n despite destruction of SOCKET_I()->sk already completed.
>
> But there is no point with calling security_file_open() on sockets
> because open("/proc/pid/fd/n", !O_PATH) on sockets fails with -ENXIO.
>
> There is some point with calling security_inode_getattr() on sockets
> because stat("/proc/pid/fd/n") and fstat(open("/proc/pid/fd/n", O_PATH))
> are valid. If we want to access "struct sock"->sk_{family,type,protocol}
> fields, we will need to use security_socket_post_create() hook and
> security_inode_free() hook in order to remember these fields because
> security_sk_free() hook is called before the inode is destructed. But
> since information which can be protected by checking
> security_inode_getattr() on sockets is trivial, let's not be bothered by
> "struct inode"->i_security management.
>
> There is point with calling security_file_ioctl() on sockets. Since
> ioctl(open("/proc/pid/fd/n", O_PATH)) is invalid, security_file_ioctl()
> on sockets should remain safe.
>
> [1] https://syzkaller.appspot.com/bug?id=73d590010454403d55164cca23bd0565b1eb3b74
>
> Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
> Reported-by: syzbot <syzbot+0341f6a4d729d4e0acf1@syzkaller.appspotmail.com>
> ---
> security/tomoyo/tomoyo.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/security/tomoyo/tomoyo.c b/security/tomoyo/tomoyo.c
> index 716c92e..8ea3f5d 100644
> --- a/security/tomoyo/tomoyo.c
> +++ b/security/tomoyo/tomoyo.c
> @@ -126,6 +126,9 @@ static int tomoyo_bprm_check_security(struct linux_binprm *bprm)
> */
> static int tomoyo_inode_getattr(const struct path *path)
> {
> + /* It is not safe to call tomoyo_get_socket_name(). */
> + if (S_ISSOCK(d_inode(path->dentry)->i_mode))
> + return 0;
> return tomoyo_path_perm(TOMOYO_TYPE_GETATTR, path, NULL);
> }
>
> @@ -316,6 +319,9 @@ static int tomoyo_file_open(struct file *f)
> /* Don't check read permission here if called from do_execve(). */
> if (current->in_execve)
> return 0;
> + /* Sockets can't be opened by open(). */
> + if (S_ISSOCK(file_inode(f)->i_mode))
> + return 0;
> return tomoyo_check_open_permission(tomoyo_domain(), &f->f_path,
> f->f_flags);
> }
> --
What happened to this patch?
Also, isn't the same bug in other places too?:
- tomoyo_path_chmod()
- tomoyo_path_chown()
- smack_inode_getsecurity()
- smack_inode_setsecurity()
- Eric
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v2] tomoyo: Don't check open/getattr permission on sockets.
2019-08-22 6:30 ` Eric Biggers
@ 2019-08-22 6:55 ` Tetsuo Handa
2019-08-22 7:01 ` Eric Biggers
0 siblings, 1 reply; 13+ messages in thread
From: Tetsuo Handa @ 2019-08-22 6:55 UTC (permalink / raw)
To: Eric Biggers
Cc: Al Viro, linux-fsdevel, syzbot, jmorris, linux-kernel,
linux-security-module, serge, syzkaller-bugs, takedakn,
David S. Miller
Eric Biggers wrote:
> What happened to this patch?
I have to learn how to manage a git tree for sending
pull requests, but I can't find time to try.
>
> Also, isn't the same bug in other places too?:
>
> - tomoyo_path_chmod()
> - tomoyo_path_chown()
> - smack_inode_getsecurity()
> - smack_inode_setsecurity()
What's the bug? The file descriptor returned by open(O_PATH) cannot be
passed to read(2), write(2), fchmod(2), fchown(2), fgetxattr(2), mmap(2) etc.
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v2] tomoyo: Don't check open/getattr permission on sockets.
2019-08-22 6:55 ` Tetsuo Handa
@ 2019-08-22 7:01 ` Eric Biggers
2019-08-22 7:42 ` Tetsuo Handa
0 siblings, 1 reply; 13+ messages in thread
From: Eric Biggers @ 2019-08-22 7:01 UTC (permalink / raw)
To: Tetsuo Handa
Cc: Al Viro, linux-fsdevel, syzbot, jmorris, linux-kernel,
linux-security-module, serge, syzkaller-bugs, takedakn,
David S. Miller
On Thu, Aug 22, 2019 at 03:55:31PM +0900, Tetsuo Handa wrote:
> Eric Biggers wrote:
> > What happened to this patch?
>
> I have to learn how to manage a git tree for sending
> pull requests, but I can't find time to try.
>
> >
> > Also, isn't the same bug in other places too?:
> >
> > - tomoyo_path_chmod()
> > - tomoyo_path_chown()
> > - smack_inode_getsecurity()
> > - smack_inode_setsecurity()
>
> What's the bug? The file descriptor returned by open(O_PATH) cannot be
> passed to read(2), write(2), fchmod(2), fchown(2), fgetxattr(2), mmap(2) etc.
>
chmod(2), chown(2), getxattr(2), and setxattr(2) take a path, not a fd.
- Eric
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v2] tomoyo: Don't check open/getattr permission on sockets.
2019-08-22 7:01 ` Eric Biggers
@ 2019-08-22 7:42 ` Tetsuo Handa
2019-08-22 15:47 ` Eric Biggers
0 siblings, 1 reply; 13+ messages in thread
From: Tetsuo Handa @ 2019-08-22 7:42 UTC (permalink / raw)
To: Eric Biggers
Cc: Al Viro, linux-fsdevel, syzbot, jmorris, linux-kernel,
linux-security-module, serge, syzkaller-bugs, takedakn,
David S. Miller
Eric Biggers wrote:
> On Thu, Aug 22, 2019 at 03:55:31PM +0900, Tetsuo Handa wrote:
> > > Also, isn't the same bug in other places too?:
> > >
> > > - tomoyo_path_chmod()
> > > - tomoyo_path_chown()
> > > - smack_inode_getsecurity()
> > > - smack_inode_setsecurity()
> >
> > What's the bug? The file descriptor returned by open(O_PATH) cannot be
> > passed to read(2), write(2), fchmod(2), fchown(2), fgetxattr(2), mmap(2) etc.
> >
>
> chmod(2), chown(2), getxattr(2), and setxattr(2) take a path, not a fd.
>
OK. Then, is the correct fix
inode_lock(inode);
if (SOCKET_I(inode)->sk) {
// Can access SOCKET_I(sock)->sk->*
} else {
// Already close()d. Don't touch.
}
inode_unlock(inode);
thanks to
commit 6d8c50dcb029872b ("socket: close race condition between sock_close() and sockfs_setattr()")
commit ff7b11aa481f682e ("net: socket: set sock->sk to NULL after calling proto_ops::release()")
changes?
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v2] tomoyo: Don't check open/getattr permission on sockets.
2019-08-22 7:42 ` Tetsuo Handa
@ 2019-08-22 15:47 ` Eric Biggers
0 siblings, 0 replies; 13+ messages in thread
From: Eric Biggers @ 2019-08-22 15:47 UTC (permalink / raw)
To: Tetsuo Handa
Cc: Al Viro, linux-fsdevel, syzbot, jmorris, linux-kernel,
linux-security-module, serge, syzkaller-bugs, takedakn,
David S. Miller
On Thu, Aug 22, 2019 at 04:42:26PM +0900, Tetsuo Handa wrote:
> Eric Biggers wrote:
> > On Thu, Aug 22, 2019 at 03:55:31PM +0900, Tetsuo Handa wrote:
> > > > Also, isn't the same bug in other places too?:
> > > >
> > > > - tomoyo_path_chmod()
> > > > - tomoyo_path_chown()
> > > > - smack_inode_getsecurity()
> > > > - smack_inode_setsecurity()
> > >
> > > What's the bug? The file descriptor returned by open(O_PATH) cannot be
> > > passed to read(2), write(2), fchmod(2), fchown(2), fgetxattr(2), mmap(2) etc.
> > >
> >
> > chmod(2), chown(2), getxattr(2), and setxattr(2) take a path, not a fd.
> >
>
> OK. Then, is the correct fix
>
> inode_lock(inode);
> if (SOCKET_I(inode)->sk) {
> // Can access SOCKET_I(sock)->sk->*
> } else {
> // Already close()d. Don't touch.
> }
> inode_unlock(inode);
>
> thanks to
>
> commit 6d8c50dcb029872b ("socket: close race condition between sock_close() and sockfs_setattr()")
> commit ff7b11aa481f682e ("net: socket: set sock->sk to NULL after calling proto_ops::release()")
>
> changes?
inode_lock() is already held during security_path_chmod(),
security_path_chown(), and security_inode_setxattr().
So you can't just take it again.
- Eric
^ permalink raw reply [flat|nested] 13+ messages in thread
end of thread, other threads:[~2019-08-22 15:48 UTC | newest]
Thread overview: 13+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-06-05 18:42 KASAN: use-after-free Read in tomoyo_realpath_from_path syzbot
2019-06-05 22:09 ` Tetsuo Handa
2019-06-06 2:08 ` Tetsuo Handa
2019-06-06 5:20 ` Tetsuo Handa
2019-06-09 6:41 ` [PATCH] tomoyo: Don't check open/getattr permission on sockets Tetsuo Handa
2019-06-16 6:49 ` Tetsuo Handa
2019-06-18 20:49 ` Al Viro
2019-06-22 4:45 ` [PATCH v2] " Tetsuo Handa
2019-08-22 6:30 ` Eric Biggers
2019-08-22 6:55 ` Tetsuo Handa
2019-08-22 7:01 ` Eric Biggers
2019-08-22 7:42 ` Tetsuo Handa
2019-08-22 15:47 ` Eric Biggers
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).