Date: Thu, 20 Nov 2014 14:07:52 -0800 From: John-Mark Gurney <jmg@funkthat.com> To: John Baldwin <jhb@freebsd.org> Cc: arch@freebsd.org Subject: Re: I'd like to axe some drivers Message-ID: <20141120220752.GI24601@funkthat.com> In-Reply-To: <201411201631.27556.jhb@freebsd.org> References: <201411201631.27556.jhb@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
John Baldwin wrote this message on Thu, Nov 20, 2014 at 16:31 -0500:
> I'm >< close to removing timeout/untimeout from the tree. As part of this I
> have updated several older drivers to use callout(9), but most of those
> patches were untested. Keeping old code around that no one uses does add
> future work as tree-wide API changes are made as well as things like locking
> (note that several of these drivers weren't locked until I recently changed
> them). To that end, here is my short list of things that I think we can bid
> farewell to in 11. Note that many of these are for ISA devices.
>
> asr(4): This is a driver for a set of older Adaptec PCI RAID adapters. This
> driver is _really_ crufty and is the only thing I didn't convert to
> callout(9) because it has no notion of software state for a given
> request. It is also 32-bit only since it stuff kernel pointers into
> 32 bit fields in hardware-defined structures.
>
> mcd(4): This is a driver for a pre-ATAPI ISA CD-ROM adapter. As noted in
> the manpage, this driver is only useful as a backend to cdcontrol to
> play audio CDs since it doesn't use DMA, so its data performance is
> "abysmal" (and that was true in the mid 90's).
>
> scd(4): Similar to mcd(4), this is a pre-ATAPI ISA CD-ROM adapter. (Note
> that the more-popular matcd(4) driver that was used for the CD-ROM
> controller on certain SoundBlaster cards was removed in 2002.)
>
> si(4): This is a driver for an older ISA/EISA/PCI multiport serial card.
> It doesn't use bus_space. It was hacked up to use new tty, but
> still uses Giant. I have a partial set of outstanding patches to
> this to fix it to use bus_space, but when I sent them out for
> testing on current and stable, no one replied.
>
> wds(4): This is an ISA SCSI HBA that does not use DMA (only PIO). I
> actually had one of these a long time ago to use with a SCSI
> ZIP drive. Even if I still have it in a box somewhere, I'm not
> digging it out.
>
> wl(4): This is a driver for an ancient pre-802.11 wireless adapter. It
> also includes wlconfig(8). Warner promises he won't test any
> patches for this. It's older and slower than wi(4) and that driver
> hasn't really worked in years. (One could make the case for axing
> an(4) and wi(4) as well, but I'm just pushing for wl(4).)
>
> spic(4): At one point the Sony VAIO was "the" cool laptop, and this driver
> controlled the "jogdial" found on it and presented it as a mouse.
> This is a tiny driver and is less invasive in terms of future
> maintenance than others perhaps, but my recent calls for testing
> on current@ and stable@ found no takers. It's a fairly obscure
> device and not one that exists on any recently shipped hardware.
>
> ie(4): Unfortunately, someone actually found one of these and tested it
> several years ago when I added locking to it. It is the only ISA
> NIC driver that doesn't have a pccard attachment (you can in theory
> still use a pccard NIC in a cardbus slot (though not ExpressCard)).
> This also only does 10Mb using PIO (no DMA). It doesn't use
> bus_space.
I'm fine w/ removing these... Should we do some house cleaning on
amd64's GENERIC too?
amd64's GENERIC has a lot of ISA or 100Mbit ethernet cards that are
clearly not going to be used on these machines...
My recommended list to remove:
ae, bfe, dc, fxp, hme?, pcn, rl, tx, vr, wb, xl, cs, ed, ex, ep, fe,
sn, xe
All of these are modules, so if someone really needs them, they can
load the module...
--
John-Mark Gurney Voice: +1 415 225 5579
"All that I will do, has been done, All that I have, has not."
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20141120220752.GI24601>