security,selinux,smack: kill security_task_wait hook
cover security,selinux,smack: kill security_task_wait hook
The following (backported) patch removes a hook which has already been removed upstream. [1] Reason is that a permission denial can lead to soft lockups and zombies.
A reproducer can be found in the initial report. [2]
I hence consider this a bugfix which is allowed for stable branches.
Background:
To reduce divergence of 4.9 to upstream before proposing to apply upstream commit 3dfc9b02864bt (LSM: Initialize security_hook_heads upon registration.) I'm checking which changes to the LSM hooks may be applicable to 4.9 as doing so after backporting 33dfc9b02864bt will lead to conflicts for each such commit.
[1] https://patchwork.kernel.org/project/linux-security-module/patch/[email protected]/
[2] https://patchwork.kernel.org/project/selinux/patch/[email protected]
Stephen Smalley (1):
security,selinux,smack: kill security_task_wait hook
include/linux/lsm_hooks.h | 7 ------- include/linux/security.h | 6 ------ kernel/exit.c | 19 ++----------------- security/security.c | 6 ------ security/selinux/hooks.c | 6 ------ security/smack/smack_lsm.c | 20 -------------------- 6 files changed, 2 insertions(+), 62 deletions(-)
Signature · 1 line
2.25.1
PATCH 1/1 security,selinux,smack: kill security_task_wait hook
From: Stephen Smalley <[email protected]>
commit 3a2f5a59a695a73e0cde9a61e0feae5fa730e936 upstream.
As reported by yangshukui, a permission denial from security_task_wait() can lead to a soft lockup in zap_pid_ns_processes() since it only expects sys_wait4() to return 0 or -ECHILD. Further, security_task_wait() can in general lead to zombies; in the absence of some way to automatically reparent a child process upon a denial, the hook is not useful. Remove the security hook and its implementations in SELinux and Smack. Smack already removed its check from its hook.
--- include/linux/lsm_hooks.h | 7 ------- include/linux/security.h | 6 ------ kernel/exit.c | 19 ++----------------- security/security.c | 6 ------ security/selinux/hooks.c | 6 ------ security/smack/smack_lsm.c | 20 -------------------- 6 files changed, 2 insertions(+), 62 deletions(-)
include/linux/lsm_hooks.h +0 -7
@@ -666,11 +666,6 @@* @sig contains the signal value. * @secid contains the sid of the process where the signal originated * Return 0 if permission is granted. -* @task_wait: -* Check permission before allowing a process to reap a child process @p -* and collect its status information. -* @p contains the task_struct for process. -* Return 0 if permission is granted. * @task_prctl: * Check permission before performing a process control operation on the * current process.
@@ -1507,7 +1502,6 @@ union security_list_options {int (*task_movememory)(struct task_struct *p);int (*task_kill)(struct task_struct *p,struct siginfo *info,int sig,u32 secid); -int (*task_wait)(struct task_struct *p);int (*task_prctl)(int option,unsigned long arg2,unsigned long arg3,unsigned long arg4,unsigned long arg5);void (*task_to_inode)(struct task_struct *p,struct inode *inode);
@@ -1768,7 +1762,6 @@ struct security_hook_heads {struct list_head task_getscheduler;struct list_head task_movememory;struct list_head task_kill; -struct list_head task_wait;struct list_head task_prctl;struct list_head task_to_inode;struct list_head ipc_permission;
include/linux/security.h +0 -6
@@ -332,7 +332,6 @@ int security_task_getscheduler(struct task_struct *p);int security_task_movememory (struct task_struct *p);int security_task_kill (struct task_struct *p,struct siginfo *info,int sig,u32 secid); -int security_task_wait (struct task_struct *p);int security_task_prctl (int option,unsigned long arg2,unsigned long arg3,unsigned long arg4,unsigned long arg5);void security_task_to_inode (struct task_struct *p,struct inode *inode);
@@ -980,11 +979,6 @@ static inline int security_task_kill(struct task_struct *p,return 0 ; } -static inline int security_task_wait (struct task_struct *p) -{ -return 0 ; -} - static inline int security_task_prctl(int option, unsigned long arg2, unsigned long arg3, unsigned long arg4,
kernel/exit.c +2 -17
@@ -14,7 +14,6 @@#include <linux/tty.h> #include <linux/iocontext.h> #include <linux/key.h> -#include <linux/security.h> #include <linux/cpu.h> #include <linux/acct.h> #include <linux/tsacct_kern.h>
@@ -1342,7 +1341,7 @@ static int wait_task_continued(struct wait_opts *wo, struct task_struct *p)* Returns nonzero for a final return, when we have unlocked tasklist_lock. * Returns zero if the search for a child should continue; * then ->notask_error is 0 if @p is an eligible child, -* or another error from security_task_wait(), or still -ECHILD. +* or still -ECHILD. */ static int wait_consider_task (struct wait_opts *wo,int ptrace,struct task_struct *p)
@@ -1362,20 +1361,6 @@ static int wait_consider_task(struct wait_opts *wo, int ptrace,if (!ret)return ret; - ret =security_task_wait (p); -if (unlikely (ret <0 )) { -/* -* If we have not yet seen any eligible child, -* then let this error code replace -ECHILD. -* A permission error will give the user a clue -* to look for security policy problems, rather -* than for mysterious wait bugs. -*/ -if (wo->notask_error) - wo->notask_error = ret; -return 0 ; - } -if (unlikely (exit_state ==EXIT_TRACE )) {/* * ptrace == 0 means we are the natural parent. In this case
@@ -1468,7 +1453,7 @@ static int wait_consider_task(struct wait_opts *wo, int ptrace,* Returns nonzero for a final return, when we have unlocked tasklist_lock. * Returns zero if the search for a child should continue; then * ->notask_error is 0 if there were any eligible children, -* or another error from security_task_wait(), or still -ECHILD. +* or still -ECHILD. */ static int do_wait_thread (struct wait_opts *wo,struct task_struct *tsk) {
security/security.c +0 -6
@@ -1032,11 +1032,6 @@ int security_task_kill(struct task_struct *p, struct siginfo *info,return call_int_hook (task_kill,0 , p, info, sig, secid); } -int security_task_wait (struct task_struct *p) -{ -return call_int_hook (task_wait,0 , p); -} -int security_task_prctl (int option,unsigned long arg2,unsigned long arg3,unsigned long arg4,unsigned long arg5) {
@@ -1776,7 +1771,6 @@ struct security_hook_heads security_hook_heads = { .task_movememory = LIST_HEAD_INIT(security_hook_heads.task_movememory), .task_kill = LIST_HEAD_INIT(security_hook_heads.task_kill), - .task_wait = LIST_HEAD_INIT(security_hook_heads.task_wait), .task_prctl = LIST_HEAD_INIT(security_hook_heads.task_prctl), .task_to_inode = LIST_HEAD_INIT(security_hook_heads.task_to_inode),
security/selinux/hooks.c +0 -6
@@ -3951,11 +3951,6 @@ static int selinux_task_kill(struct task_struct *p, struct siginfo *info,return rc; } -static int selinux_task_wait (struct task_struct *p) -{ -return task_has_perm (p, current,PROCESS__SIGCHLD ); -} -static void selinux_task_to_inode (struct task_struct *p,struct inode *inode) {
@@ -6220,7 +6215,6 @@ static struct security_hook_list selinux_hooks[] = {LSM_HOOK_INIT (task_getscheduler, selinux_task_getscheduler),LSM_HOOK_INIT (task_movememory, selinux_task_movememory),LSM_HOOK_INIT (task_kill, selinux_task_kill), -LSM_HOOK_INIT (task_wait, selinux_task_wait),LSM_HOOK_INIT (task_to_inode, selinux_task_to_inode), LSM_HOOK_INIT(ipc_permission, selinux_ipc_permission),
security/smack/smack_lsm.c +0 -20
@@ -2276,25 +2276,6 @@ static int smack_task_kill(struct task_struct *p, struct siginfo *info,return rc; } -/** -* smack_task_wait - Smack access check for waiting -* @p: task to wait for -* -* Returns 0 -*/ -static int smack_task_wait (struct task_struct *p) -{ -/* -* Allow the operation to succeed. -* Zombies are bad. -* In userless environments (e.g. phones) programs -* get marked with SMACK64EXEC and even if the parent -* and child shouldn't be talking the parent still -* may expect to know when the child exits. -*/ -return 0 ; -} -/** * smack_task_to_inode - copy task smack into the inode blob * @p: task to copy from
@@ -4686,7 +4667,6 @@ static struct security_hook_list smack_hooks[] = {LSM_HOOK_INIT (task_getscheduler, smack_task_getscheduler),LSM_HOOK_INIT (task_movememory, smack_task_movememory),LSM_HOOK_INIT (task_kill, smack_task_kill), -LSM_HOOK_INIT (task_wait, smack_task_wait),LSM_HOOK_INIT (task_to_inode, smack_task_to_inode), LSM_HOOK_INIT(ipc_permission, smack_ipc_permission),
Signature · 1 line
2.25.1
Hi Greg,
after the previous discussion about what kind of patches are acceptable for stable
and your hints on how to send them to the ML in https://lore.kernel.org/all/[email protected]/
I'd like to know if this patch meets the requirements and if it can be considered.
I do have a few more similar ones which I think meet the stable requirements and finally the init-cleanup patch
(upstream 3dfc9b02864bt "LSM: Initialize security_hook_heads upon registration.") which I'd like to backport to 4.9. But first I want to know whether I now got the formal requirements right before sending further patches.
Thanks,
Alex
Quoted 13 lines · Alexander Grund
From: Stephen Smalley <[email protected]>
commit
3a2f5a59a695a73e0cde9a61e0feae5fa730e936upstream.As reported by yangshukui, a permission denial from
security_task_wait()can lead to a soft lockup inzap_pid_ns_processes()since it only expectssys_wait4()to return 0 or-ECHILD. Further,security_task_wait()can in general lead to zombies; in the absence of some way to automatically reparent a child process upon a denial, the hook is not useful. Remove the security hook and its implementations in SELinux and Smack. Smack already removed its check from its hook.<snip>
Greg KH
maintainerA: http://en.wikipedia.org/wiki/Top_post
Q: Were do I find info about this thing called top-posting?
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
A: No.
Q: Should I include quotations after my reply?
http://daringfireball.net/2007/07/on_top
Quoted 5 lines · Alexander Grund
Hi Greg,
after the previous discussion about what kind of patches are acceptable for stable
and your hints on how to send them to the ML in https://lore.kernel.org/all/[email protected]/
I'd like to know if this patch meets the requirements and if it can be considered.
Yes, it is now queued up. It was delayed due to the RETBLEED mess.
Quoted 5 lines
I do have a few more similar ones which I think meet the stable requirements and finally the init-cleanup patch
(upstream 3dfc9b02864bt "LSM: Initializesecurity_hook_headsupon registration.") which I'd like to backport to 4.9. But first I want to know whether I now got the formal requirements right before sending further patches.
Let's see what the other patches look like, I can't guarantee anything without reviewing them first.
thanks,
greg k-h