netdev - Re: [PATCH 1/1] net: macb: A different way to restart a stuck TX descriptor ring.

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 for Android: free password hash cracker in your pocket

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

Message-ID: <5cd0369c-4bd6-e5bb-c2a2-04be52ebaa72@intel.com>
Date:   Tue, 11 Apr 2023 08:25:08 -0700
From:   Jesse Brandeburg <jesse.brandeburg@...el.com>
To:     Ingo Rohloff <ingo.rohloff@...terbach.com>,
        <robert.hancock@...ian.com>
CC:     <Nicolas.Ferre@...rochip.com>, <claudiu.beznea@...rochip.com>,
        <davem@...emloft.net>, <kuba@...nel.org>, <netdev@...r.kernel.org>,
        <tomas.melin@...sala.com>
Subject: Re: [PATCH 1/1] net: macb: A different way to restart a stuck TX
 descriptor ring.

On 4/7/2023 2:33 PM, Ingo Rohloff wrote:
> This implements a different approach than Commit 4298388574dae6 ("net:
> macb: restart tx after tx used bit read"):
> 
> When reaping TX descriptors in macb_tx_complete(), if there are still
> active descriptors pending (queue is not empty) and the controller
> additionally signals that it is not any longer working on the TX ring,
> then something has to be wrong. Reasoning:
> Each time a descriptor is added to the TX ring (via macb_start_xmit()) the
> controller is triggered to start transmitting (via setting the TSTART
> bit).
> At this point in time, there are two cases:
> 1) The controller already has read an inactive descriptor
>    (with a set TX_USED bit).
> 2) The controller has not yet read an inactive descriptor
>    and is still actively transmitting.
> 
> In case 1) setting the TSTART bit, should restart transmission.
> In case 2) the controller should continue transmitting and at some point
> reach the freshly added descriptors and then process them too.
> 
> This patch checks in macb_tx_complete() if the TX queue is non-empty and
> additionally if the controller indicates that it is not transmitting any
> longer. If this condition is detected, the TSTART bit is set again to
> restart transmission.
> 
> Signed-off-by: Ingo Rohloff <ingo.rohloff@...terbach.com>


I see this series is still under discussion. Next time you send please
use correct subject line:
[PATCH net v1] macb: ...

Also please be sure to cc the correct maintainers.