Re: Portable binary drivers (was Re: Linux headed for disaster?)

Kendall Bennett (KendallB@scitechsoft.com)
Sun, 5 Dec 1999 15:25:44 -0800


Jeff Garzik <jgarzik@mandrakesoft.com> wrote:

> Supporting portable binary modules adds a lot of work, and slows
> down your driver. If you strongly believe in your case: show us
> the code.

This is a design issue, and not something where 'showing you the
code' is going to help. The fact is that the way Linux is designed,
device driver compatibility problems are ripe and I expect them to
get worse. Binary portable drivers is how they are solved on other
OS'es, and Linux needs this for future compatibility.

I am complaning about this because I have seen these problems with my
own eyes on our own hardware. It is not a myth. Device driver
compatibility in Linux is not as good as it could be, and the device
driver model is the problem.

> I am confident that a solution to support binary loadable modules
> will be accepted by Linus and Alan -- if it does not add overhead
> to the kernel.

So the mandate of the Linux kernel developers is to produce the
fastest and lowest overhead kernel regardless of whether that affects
future compatibility or not? Surely it is important that existing
device drivers will work with new kernel revisions?

> Further, I don't think you'll get very far arguing against open
> source device drivers.

I haven't been arguing that. Please re-read my posts, and you will
see that I am arguing for binary portable modules. Whether those
modules are built from Open Source code or not is a different, and
completely separate issue. I do believe Open Source drivers are
important, and vendors will make their information open and support
Open Source drivers if it makes commercial sense for them to do so.

This is happening in XFree86, and XFree86 allows for binary portable
modules. Hence supporting binary portable modules does not mean that
people are going to suddenly stop writing Open Source device drivers.

> You imply it is bad that the current nature of the kernel
> practically forces vendors to open source their drivers. Most of
> us believe just the opposite... open sourcing the driver means that
> you can fix problems that the vendor can't or won't take care of in
> a reasonable amount of time.

I have nothing against Open Source device drivers, and as I said
before it makes a lot of sense for hardware vendors to Open Source
their device drivers and make technical information available.

I do *not* however feel that hardware vendors should be forced to go
Open Source because there is no other way that they can support
Linux. In the Linux applications area, you can just as easily do an
Open Source app as a closed source one. People do Open Source apps
because it makes sense, not because Linux forces them to.

Not that my company is *not* a hardware company, so I am not talking
about the product my company develops here.

> It means a volunteer can update the driver for a new kernel
> version, without having to wait for the vendor.

Definately. All good reasons for Open Source. But not at all good
reasons for not having binary portable modules.

> There are so many reasons why we should strongly encourage open
> source drivers.

Sure. But that is not what I am arguing!

Regards,

+---------------------------------------------------------------+
| SciTech Software - Building Truly Plug'n'Play Software! |
+---------------------------------------------------------------+
| Kendall Bennett | Email: KendallB@scitechsoft.com |
| Director of Engineering | Phone: (530) 894 8400 |
| SciTech Software, Inc. | Fax : (530) 894 9069 |
| 505 Wall Street | ftp : ftp.scitechsoft.com |
| Chico, CA 95928, USA | www : http://www.scitechsoft.com |
+---------------------------------------------------------------+

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/