netdev - Re: [PATCH v1 net] lan743x: fix return value for lan743x_tx_napi

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:   Tue, 20 Nov 2018 19:12:05 -0800 (PST)
From:   David Miller <davem@...emloft.net>
To:     Tristram.Ha@...rochip.com
Cc:     Bryan.Whitehead@...rochip.com, netdev@...r.kernel.org,
        UNGLinuxDriver@...rochip.com, andrew@...n.ch, f.fainelli@...il.com
Subject: Re: [PATCH v1 net] lan743x: fix return value for
 lan743x_tx_napi_poll

From: <Tristram.Ha@...rochip.com>
Date: Wed, 21 Nov 2018 02:13:30 +0000

> Slightly out of topic I am not sure why NAPI is used on the transmit side.
> Originally NAPI was designed to fix the receive interrupt happening on each
> receive frame problem, so on transmit side it is to avoid the transmit
> done interrupt on each transmit frame?  Typically hardware has a way
> to trigger transmit done interrupt or not in each transmit frame.

It puts transmit completion, like receive processing, inside of a
software instead of a hardware interrupt.

It is very much intended that all drivers do transmit completion
inside of a NAPI context.

Avoiding TX interrupts by clearing interrupt indication bits in the
TX descriptors or turning TX completion interrupts off compeltely
is a non-starter.

All TX completion events MUST occur in a short finite amount of time,
otherwise you wedge TCP sockets waiting for memory to be free'd up
etc.