[PATCH] Re: Linksys PCM200

[soralx at cydem.org]

> > <>I don't see a way how it could break other cards' functionality -
> > should be no concerns here
> D-Link isn't the only 0x00a8; The AboCom FE2500MX bears 0x13d1 0xab08.

Are there any cards that have exactly the same 32-bit PCIID, and
different/modified chipsets? I didn't think that something like that
is possible...
Anyway, the PCM200's PCIID doesn't seem to conflict with any other
card's, so I'm sure it won't break anything.

> >0. More info is _always_ better. In any case, the message will take 2
> > lines on console, so shortening the description will not gain anything
> Yes, it does. It gains readability.

compare:

dc0: <Linksys PCM200 EtherFast 10/100 CardBus NIC, v.03 (ADMtek Centaur-C)> port
 0x1000-0x10ff mem 0x88002000-0x880023ff irq 9 at device 0.0 on cardbus1
		and
dc0: <Linksys PCM200 EtherFast CardBus 10/100> port 0x1000-0x10ff mem 0x88002000
-0x880023ff irq 9 at device 0.0 on cardbus1

Is the latter more readable? I think they're the same. The former is even
more readable for me, because the description is on one line, and all the
I/O info is on the other. Also, if I will need to send dmesg to somebody
(for example, the card could create a problem by conflicting with some other
hardware), they will know _exactly_ what card is it (and even if they're
not familiar with this card, they'll know the chipset).

> Long descriptions should generally
> be reserved for pciconf -lv and driver comments.

for this card, `pciconf -lv` outputs not much useful info:

dc0 at pci2:0:0:   class=0x020000 card=0xab091737 chip=0xab091737 rev=0x11 hdr=0x00
    vendor   = 'Linksys'
    device   = 'PCM200 10/100 CardBus Ethernet Adapter'

Even the chipID is changed. It also does not state card's version (one will
need to figure that out from revID) - and, according to info in the INet,
card's versions differ between each other noticeably.

> > <>2. when PCI IDs for previous card versions will be added, the
> > description will
> > need to be changed anyway to include the version number
>
> Only because D-Link has a 'change everything except the model name'
> fetish. Unless D-Link pulled the same crap they did with the DWL-520 and
> DWL-650, personally I don't see any compelling reason to include chipset
> and revision in the dev's desc.

why not? there can never be too much info :)

> Now, if D-Link pulled the same crap on
> this as they did with the DWL-520, I'd say just slap Rev.D in there;
> there's no need for chipset name

chipset name makes the dmesg message separate nicely into 2 lines :p

Anyway, enough on this. I'd personally prefer to see my long description
commited, but seems like the FreeBSD developers know better what to do
(as they were able to create such a great OS).


I'll test the store-and-fwd patch later today. I thought however, that it
has to swith to that mode because my notebook (and CardBus controller)
isn't fast enough. Form `man 4 dc`:

dc%d: TX underrun -- increasing TX threshold  The device generated a
     transmit underrun error while attempting to DMA and transmit a packet.
     This happens if the host is not able to DMA the packet data into the
     NIC's FIFO fast enough.

Timestamp: 0x417C1321
[SorAlx]  http://cydem.org.ua/
ridin' VN1500-B2