netdev - Re: [PATCH net-next] pcnet32: remove VLB support

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: <20260115222954.248e9f79@pumpkin>
Date: Thu, 15 Jan 2026 22:29:54 +0000
From: David Laight <david.laight.linux@...il.com>
To: Ethan Nelson-Moore <enelsonmoore@...il.com>
Cc: netdev@...r.kernel.org
Subject: Re: [PATCH net-next] pcnet32: remove VLB support

On Tue,  6 Jan 2026 23:18:31 -0800
Ethan Nelson-Moore <enelsonmoore@...il.com> wrote:

> This allows the code managing device instances to be simplified
> significantly. The VLB bus is very obsolete and last appeared on
> P5 Pentium-era hardware. Support for it has been removed from
> other drivers, and it is highly unlikely anyone is using it with
> modern Linux kernels.

That device bring back memories, but only the PCnet/ISA and PCnet/PCI
variants.
IIRC both are basically AMD 'lance' ethernet chips with a built-in
bus interface (the ISA one is ISA pnp).
So have a limit of 24 address bits (I don't remember using a different
ring format than any other lance variant - just for the Sun HME).
I don't remember anything about VLB - just ISA, EISA and PCI.

There are two variants of the PCnet/PCI - the '790 and '791.
Each had its own quirks, I later thought that the 791 might have
been intended to support 100M - but didn't work.

The ISA variant needed one of the ISA DMA channels put into 'cascade
mode' (the kernel didn't really want to allow that), but was about
the only ISA card capable of saturating 10M ethernet with smallish
packets (IIRC in a 33MHz 486, but not the similar 386 box).
(And 'fun' to set to fixed IO addresses on a motherboard that supported
ISA pnp for a kernel that didn't.)

One bit you might disable is the probe_vlbus() code.
You really never want that sort of probe code, it can have a disastrous
effect on other hardware at the address being probed.

I didn't 'steal' the green datasheet book :-(

	David