linux-kernel - RE: [PATCH] staging: unisys: visorbus: visorchannel: Refactor locking code to be statically deterministic.

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]

Date:   Wed, 19 Oct 2016 17:00:53 +0000
From:   "Sell, Timothy C" <Timothy.Sell@...sys.com>
To:     'Cathal Mullaney' <chuckleberryfinn@...il.com>
CC:     "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
        *S-Par-Maintainer <SParMaintainer@...sys.com>,
        "devel@...verdev.osuosl.org" <devel@...verdev.osuosl.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "Kershner, David A" <David.Kershner@...sys.com>
Subject: RE: [PATCH] staging: unisys: visorbus: visorchannel: Refactor locking
 code to be statically deterministic.

On Wednesday, October 19, 2016 7:31 AM, Cathal Mullaney wrote:
> This patch makes locking in visorchannel_signalempty statically deterministic.
> As a result this patch fixes the sparse warning:
> Context imbalance in 'visorchannel_signalempty' - different lock contexts for
> basic block.
> 
> The logic of the locking code doesn't change but the layout of the original
> code is "frowned upon"
> according to mails on sparse context checking.
> Refactoring removes the warning and makes the code more readable.
> 
> Signed-off-by: Cathal Mullaney <chuckleberryfinn@...il.com>
> ---
>  drivers/staging/unisys/visorbus/visorchannel.c | 26 +++++++++++++++++---
> ------
>  1 file changed, 17 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/staging/unisys/visorbus/visorchannel.c
> b/drivers/staging/unisys/visorbus/visorchannel.c
> index a1381eb..1eea5d8 100644
> --- a/drivers/staging/unisys/visorbus/visorchannel.c
> +++ b/drivers/staging/unisys/visorbus/visorchannel.c
> @@ -300,22 +300,30 @@
> ---
>  bool
>  visorchannel_signalempty(struct visorchannel *channel, u32 queue)
>  {
>  	unsigned long flags = 0;
> -	struct signal_queue_header sig_hdr;
>  	bool rc = false;

It appears as if you no longer need to initialize 'rc' above.

Although this is NOT caused by your patch, it also looks like 'flags'
is being unnecessarily initialized.  You may want to fix that too
while you're in the neighborhood.

(Kernel folks seem to frown on unnecessary variable initializations.)

> 
> -	if (channel->needs_lock)
> -		spin_lock_irqsave(&channel->remove_lock, flags);
> +	if (!channel->needs_lock)
> +		return queue_empty(channel, queue);
> 
> -	if (sig_read_header(channel, queue, &sig_hdr))
> -		rc = true;
> -	if (sig_hdr.head == sig_hdr.tail)
> -		rc = true;
> -	if (channel->needs_lock)
> -		spin_unlock_irqrestore(&channel->remove_lock, flags);
> +	spin_lock_irqsave(&channel->remove_lock, flags);
> +	rc = queue_empty(channel, queue);
> +	spin_unlock_irqrestore(&channel->remove_lock, flags);
> 
>  	return rc;
>  }
> --
> 2.7.4

Besides that, your patch looks good to me.  Thanks.

- Tim Sell