linux-kernel - Re: [syzbot] [mm?] general protection fault in find_mergeable_anon

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

Message-ID: <20241210150528.GA31266@redhat.com>
Date: Tue, 10 Dec 2024 16:05:28 +0100
From: Oleg Nesterov <oleg@...hat.com>
To: Lorenzo Stoakes <lorenzo.stoakes@...cle.com>
Cc: "Liam R. Howlett" <Liam.Howlett@...cle.com>, mhiramat@...nel.org,
	peterz@...radead.org, Jann Horn <jannh@...gle.com>,
	syzbot <syzbot+2d788f4f7cb660dac4b7@...kaller.appspotmail.com>,
	akpm@...ux-foundation.org, linux-kernel@...r.kernel.org,
	linux-mm@...ck.org, syzkaller-bugs@...glegroups.com, vbabka@...e.cz
Subject: Re: [syzbot] [mm?] general protection fault in
 find_mergeable_anon_vma

On 12/09, Lorenzo Stoakes wrote:
>
> (As discussed on IRC) how about moving up the dup_mmap_sem lock one level, we
> can put the mm before the rmap lookup in build_map_info() is able to find it,
> which should avoid the whole issue?

Not sure I fully understand the problem, but so far I see nothing wrong in
this idea. However,

> @@ -1692,9 +1690,11 @@ static struct mm_struct *dup_mm(struct task_struct *tsk,
>  	if (!mm_init(mm, tsk, mm->user_ns))
>  		goto fail_nomem;
> 
> +	uprobe_start_dup_mmap();
>  	err = dup_mmap(mm, oldmm);
>  	if (err)
>  		goto free_pt;
> +	uprobe_end_dup_mmap();

If try_module_get(mm->binfmt->module)) fails after that, dup_mm() does
"goto free_pt;" and in this case ...

> @@ -1709,6 +1709,7 @@ static struct mm_struct *dup_mm(struct task_struct *tsk,
>  	mm->binfmt = NULL;
>  	mm_init_owner(mm, NULL);
>  	mmput(mm);
> +	uprobe_end_dup_mmap();

... we have the unbalanced uprobe_end_dup_mmap().

Also. Perhaps we can change dup_mmap() to set MMF_XXX before uprobe_end_dup_mmap(),

	fail_uprobe_end:
+		if (retval)
+			set_bit(mm->flags, MMF_XXX);
		uprobe_end_dup_mmap();
		return retval;

Then build_map_info() can check this flag. I guess we can reuse some of
MMF_OOM_ bits? May be MMF_UNSTABLE...

Oleg.