[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAK8P3a3y3vvgdWXU3x9f1cwYKt3AvLUfN6sMEo0SXFPTCuxjCw@mail.gmail.com>
Date: Sun, 30 May 2021 22:54:15 +0200
From: Arnd Bergmann <arnd@...nel.org>
To: Nikolai Zhubr <zhubr.2@...il.com>
Cc: netdev <netdev@...r.kernel.org>,
tedheadster <tedheadster@...il.com>,
Heiner Kallweit <hkallweit1@...il.com>, whiteheadm@....org,
Jeff Garzik <jgarzik@...ox.com>
Subject: Re: Realtek 8139 problem on 486.
On Sun, May 30, 2021 at 7:19 PM Nikolai Zhubr <zhubr.2@...il.com> wrote:
>
> Hi all,
>
> 30.05.2021 13:36, Nikolai Zhubr:
> > So, 8139too 0.9.18-pre4 as of kernel 2.2.20 is the one that apparently
> > works correctly here. I feel the bisect is going to be massive.
>
> It appears the problem arrived between kernel 2.6.2 and 2.6.3, that is
> respectively 8139too versions 0.9.26 and 0.9.27.
> Unmodified kernel 2.6.2 works fine, unmodified kernel 2.6.3 shows
> reproducable connectivity issues.
>
> The diff is not small, I'm not sure I can dig through.
> Any hints/ideas greatly appreciated.
This is apparently when NAPI was introduced into the driver.
One thing I noticed here was the handling for shared IRQs changing in
the process. Do you happen to have shared IRQs, and if so, can you
change it so this card has an IRQ that is not shared with any other
device?
Arnd
Powered by blists - more mailing lists