linux-kernel - Re: linux-next: manual merge of the userns tree with Linus' tree

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20210628171639.2488fdcc@canb.auug.org.au>
Date:   Mon, 28 Jun 2021 17:16:39 +1000
From:   Stephen Rothwell <sfr@...b.auug.org.au>
To:     "Eric W. Biederman" <ebiederm@...ssion.com>
Cc:     Alexey Gladkov <legion@...nel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Linux Next Mailing List <linux-next@...r.kernel.org>,
        Peter Zijlstra <peterz@...radead.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Linus Torvalds <torvalds@...ux-foundation.org>
Subject: Re: linux-next: manual merge of the userns tree with Linus' tree

Hi all,

On Fri, 25 Jun 2021 18:22:19 +1000 Stephen Rothwell <sfr@...b.auug.org.au> wrote:
>
> Today's linux-next merge of the userns tree got a conflict in:
> 
>   kernel/signal.c
> 
> between commits:
> 
>   69995ebbb9d3 ("signal: Hand SIGQUEUE_PREALLOC flag to __sigqueue_alloc()")
>   4bad58ebc8bc ("signal: Allow tasks to cache one sigqueue struct")
>   399f8dd9a866 ("signal: Prevent sigqueue caching after task got released")
> 
> from Linus' tree and commit:
> 
>   d64696905554 ("Reimplement RLIMIT_SIGPENDING on top of ucounts")
> 
> from the userns tree.

Since commits

  4bad58ebc8bc ("signal: Allow tasks to cache one sigqueue struct")
  399f8dd9a866 ("signal: Prevent sigqueue caching after task got released")

have been reverted in Linus' tree, the conflict resolution is a bit
simpler now:

diff --cc kernel/signal.c
index 20d1d896d5b0,9a6dab712123..de0920353d30
--- a/kernel/signal.c
+++ b/kernel/signal.c
@@@ -408,12 -410,11 +408,12 @@@ void task_join_group_stop(struct task_s
   *   appropriate lock must be held to stop the target task from exiting
   */
  static struct sigqueue *
 -__sigqueue_alloc(int sig, struct task_struct *t, gfp_t flags, int override_rlimit)
 +__sigqueue_alloc(int sig, struct task_struct *t, gfp_t gfp_flags,
 +		 int override_rlimit, const unsigned int sigqueue_flags)
  {
  	struct sigqueue *q = NULL;
- 	struct user_struct *user;
- 	int sigpending;
+ 	struct ucounts *ucounts = NULL;
+ 	long sigpending;
  
  	/*
  	 * Protect access to @t credentials. This can go away when all
@@@ -424,27 -425,26 +424,26 @@@
  	 * changes from/to zero.
  	 */
  	rcu_read_lock();
- 	user = __task_cred(t)->user;
- 	sigpending = atomic_inc_return(&user->sigpending);
+ 	ucounts = task_ucounts(t);
+ 	sigpending = inc_rlimit_ucounts(ucounts, UCOUNT_RLIMIT_SIGPENDING, 1);
  	if (sigpending == 1)
- 		get_uid(user);
+ 		ucounts = get_ucounts(ucounts);
  	rcu_read_unlock();
  
- 	if (override_rlimit || likely(sigpending <= task_rlimit(t, RLIMIT_SIGPENDING))) {
+ 	if (override_rlimit || (sigpending < LONG_MAX && sigpending <= task_rlimit(t, RLIMIT_SIGPENDING))) {
 -		q = kmem_cache_alloc(sigqueue_cachep, flags);
 +		q = kmem_cache_alloc(sigqueue_cachep, gfp_flags);
  	} else {
  		print_dropped_signal(sig);
  	}
  
  	if (unlikely(q == NULL)) {
- 		if (atomic_dec_and_test(&user->sigpending))
- 			free_uid(user);
+ 		if (ucounts && dec_rlimit_ucounts(ucounts, UCOUNT_RLIMIT_SIGPENDING, 1))
+ 			put_ucounts(ucounts);
  	} else {
  		INIT_LIST_HEAD(&q->list);
 -		q->flags = 0;
 +		q->flags = sigqueue_flags;
- 		q->user = user;
+ 		q->ucounts = ucounts;
  	}
- 
  	return q;
  }
  

-- 
Cheers,
Stephen Rothwell

Content of type "application/pgp-signature" skipped