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: <aacdcc85c9f8d3a51a85b6429646c38d4b239449.camel@redhat.com>
Date: Tue, 23 Sep 2025 10:30:18 +0200
From: Gabriele Monaco <gmonaco@...hat.com>
To: Nam Cao <namcao@...utronix.de>, Nathan Chancellor <nathan@...nel.org>
Cc: Steven Rostedt <rostedt@...dmis.org>, Masami Hiramatsu
 <mhiramat@...nel.org>,  Mathieu Desnoyers <mathieu.desnoyers@...icios.com>,
 linux-trace-kernel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] rv: Fix wrong type cast in enabled_monitors_next()

On Tue, 2025-09-23 at 07:28 +0200, Nam Cao wrote:
> Hi Nathan,
> 

Thanks for finding this!

> Nathan Chancellor <nathan@...nel.org> writes:
> > I am seeing a crash when reading from
> > /sys/kernel/tracing/rv/enabled_monitors
> > on a couple of my arm64 boxes running Fedora after this change, which
> > landed in mainline in 6.17-rc7. I can reproduce this in QEMU pretty
> > easily.
> ...
> > With this change reverted, there is no crash. As this change seems to
> > have proper justification, is there some other latent bug here?
> 
> Thanks for the report.
> 
> Yes, this patch is broken, because argument 'p' of
> enabled_monitors_next() *is* a pointer to struct rv_monitor. I'm not
> sure how did I even test this patch...

Damn, I'm wondering the same :facepalm: ..

> Steven is right, we really need something in kselftest for RV, another thing
> in my RV TODO list.

I can work on that, at least a few selftests for the sysfs, I think this gets
the top priority now.

> 
> But reverting is not the real fix, because monitors_show() still expects
> a pointer to list_head. Changing monitors_show() is not an option,
> because it is shared with the 'available_monitors' interface.
> 
> So the real fix is completely changing the iterator to be list_head
> instead of rv_monitor.

Looks reasonable, can you work on the fix?
I see Steve is out for conferences so this won't be too urgent.

Thanks,
Gabriele

> 
> Best regards,
> Nam
> 
> diff --git a/kernel/trace/rv/rv.c b/kernel/trace/rv/rv.c
> index 48338520376f..43e9ea473cda 100644
> --- a/kernel/trace/rv/rv.c
> +++ b/kernel/trace/rv/rv.c
> @@ -501,7 +501,7 @@ static void *enabled_monitors_next(struct seq_file *m,
> void *p, loff_t *pos)
>  
>  	list_for_each_entry_continue(mon, &rv_monitors_list, list) {
>  		if (mon->enabled)
> -			return mon;
> +			return &mon->list;
>  	}
>  
>  	return NULL;
> @@ -509,7 +509,7 @@ static void *enabled_monitors_next(struct seq_file *m,
> void *p, loff_t *pos)
>  
>  static void *enabled_monitors_start(struct seq_file *m, loff_t *pos)
>  {
> -	struct rv_monitor *mon;
> +	struct list_head *head;
>  	loff_t l;
>  
>  	mutex_lock(&rv_interface_lock);
> @@ -517,15 +517,15 @@ static void *enabled_monitors_start(struct seq_file *m,
> loff_t *pos)
>  	if (list_empty(&rv_monitors_list))
>  		return NULL;
>  
> -	mon = list_entry(&rv_monitors_list, struct rv_monitor, list);
> +	head = &rv_monitors_list;
>  
>  	for (l = 0; l <= *pos; ) {
> -		mon = enabled_monitors_next(m, mon, &l);
> -		if (!mon)
> +		head = enabled_monitors_next(m, head, &l);
> +		if (!head)
>  			break;
>  	}
>  
> -	return mon;
> +	return head;
>  }
>  
>  /*


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ