[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnewsense-dev] The Linux kernel, no longer as monolithic as it once

From: Kim Hawtin
Subject: Re: [Gnewsense-dev] The Linux kernel, no longer as monolithic as it once was?
Date: Fri, 06 Aug 2010 13:30:38 +0930
User-agent: Thunderbird (X11/20100317)


Travis wrote:
Kim wrote:
Linux on the other hand, was driven more by being able to support the available hardware, but had one key difference. The X Windows System
was a completely separate project and pushed in different direction
to the kernel team. That changed a while back, but kernel support for
video devices didn't seem to be such big deal, although theres been
plenty of discussion in that space. Its also been the leading example
where many kernel folks use X as the ideal user space service idea
for drivers. Around 2004 there was a push to make lots more hardware
drives user space applications/services. That way the kernel could be
smaller and more general purpose, rather than having many very
different non interoperable interfaces.

How did that effort turn out, if you know?

I don't know, but there is still an ongoing bun fight about what to do with all the libraries built upon libraries built upon libraries built upon libraries ... that nobody uses any more. Read X input, X libs and widgets, etc ...

> It sounds like one of those
deals where the kernel could comply with the GPL while also doing a
run-around as to enable usage of non-free blobs like nVidia's
X11 driver.

Some how I luckily missed all that while doing server land sysadmin. I have a crappy lowend video card that just seems to work with the open drivers ;)

For example in FreeBSD you had all sorts of different low level
devices, char streams, several block devices, different types of
memory devices, etc, etc. At some point they made everything a
stream. How has that worked out? Well, it was generally slow at
first, but its fast now... arguably faster than Linux in some regards.

Did this occur during the major 5.x development?

I think so, I remember I was maintaining FreeBSD 4.[3456] servers at the time. 2001 ish? So thats about right. I believe that OpenBSD lead the way there, standardizing the interfaces to make implementing drivers much easier. Perhaps FreeBSD did it first, can't remember, but I was working on routing with the early WiFi gear at the same time on those two platforms, drivers were hell on Linux at the time.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]