linux-kernel - Re: [RFC PATCH 1/2] serial: sh-sci: start hrtimer after setting up DMA

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: <20240502180150.r3pb2izsjlpqzszz@ninjato>
Date: Thu, 2 May 2024 20:01:50 +0200
From: Wolfram Sang <wsa+renesas@...g-engineering.com>
To: linux-renesas-soc@...r.kernel.org
Cc: Dirk Behme <dirk.behme@...bosch.com>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	Jiri Slaby <jirislaby@...nel.org>,
	Geert Uytterhoeven <geert+renesas@...der.be>,
	linux-kernel@...r.kernel.org, linux-serial@...r.kernel.org
Subject: Re: [RFC PATCH 1/2] serial: sh-sci: start hrtimer after setting up
 DMA

On Tue, Apr 16, 2024 at 02:35:47PM +0200, Wolfram Sang wrote:
> In the RX DMA completion handler, the hrtimer was restarted before DMA
> was set up. If DMA failed, for some reason, it would clean up and the
> hrtimer would run into a NULL-pointer. Restart the timer after DMA was
> successfully set up.

Further investigations, please review:

Originally, I thought this was the race condition Dirk encountered. But
I didn't take locking into account. sci_dma_rx_timer_fn() is protected
by the uart_port_lock. sci_dma_rx_complete() is also protected by the
uart_port_lock. So, the position of restarting the hrtimer should not
matter.

However, I still think it is good coding practice (and easier to
understand) if we cancel the hrtimer at the begin of
sci_dma_rx_complete() and reenable it only if setting DMA was
successful. Because that basically means the timer only runs when DMA
was scheduled and has not finished yet.

There is some unnecessary unlock/lock in the error handling of
sci_dma_rx_complete(). I'll simplify this by moving the dev_err
downwards.

Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)