Linked by Elv13 on Fri 20th May 2011 21:06 UTC
Linux Version 2.6.39 once again took Linus Torvalds and his fellow developers less than 70 days to complete. This is further indication of a slight, though ever more apparent, increase in the kernel's development speed, as about 80 to 90 days still passed between the release of two versions one or two years ago. With 2.6.39, this also meant that there was a slight decrease in the number of advancements which are worth mentioning in the Kernel Log; however, there are still plenty of changes that will make Linux faster and better.
Thread beginning with comment 474163
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[2]: Big Kernel Lock
by lucas_maximus on Sun 22nd May 2011 01:12 UTC in reply to "RE: Big Kernel Lock"
lucas_maximus
Member since:
2009-08-18

And the change in wisdom will start showing itself


Will this mean a stable driver ABI?

Reply Parent Score: 2

RE[3]: Big Kernel Lock
by bannor99 on Sun 22nd May 2011 05:14 in reply to "RE[2]: Big Kernel Lock"
bannor99 Member since:
2005-09-15

I think a stable ABI isn't favored by some important kernel devs, Greg Kroah-Hartman chief among them

Reply Parent Score: 3

RE[4]: Big Kernel Lock
by lucas_maximus on Sun 22nd May 2011 10:08 in reply to "RE[3]: Big Kernel Lock"
lucas_maximus Member since:
2009-08-18

I read the reasoning behind it and every single point they make is actually the case for a Stable ABI.

It is a shame ... because my hardware works better on OpenBSD which is a niche Operating system than it does on Linux and I don't have pretty generic kit.

Reply Parent Score: 2

RE[3]: Big Kernel Lock
by lemur2 on Mon 23rd May 2011 01:36 in reply to "RE[2]: Big Kernel Lock"
lemur2 Member since:
2007-02-17

"And the change in wisdom will start showing itself
Will this mean a stable driver ABI? "

A stable driver ABI is only required where you have a binary driver without source code.

To accomodate a stable driver ABI would mandate a lack of flexibility to change that ABI.

It is possible to hide anti-user features within a closed binary. It turns out that users don't want anti-user features. Linux is written by its users.

So, for at least these couple of reasons, a stable driver ABI is the exact opposite of what is wanted.

I read the reasoning behind it and every single point they make is actually the case for a Stable ABI.


So what part of "Linux is written by its users, and not by hardware OEMs or by big media wanting to include DRM" did you fail to understand?

my hardware works better on OpenBSD which is a niche Operating system than it does on Linux and I don't have pretty generic kit


This personal anecdote doesn't change the fact that Linux has working drivers for more hardware than any other OS.

Edited 2011-05-23 01:44 UTC

Reply Parent Score: 3

RE[4]: Big Kernel Lock
by lemur2 on Mon 23rd May 2011 07:03 in reply to "RE[3]: Big Kernel Lock"
lemur2 Member since:
2007-02-17

It is possible to hide anti-user features within a closed binary. It turns out that users don't want anti-user features. Linux is written by its users. So, for at least these couple of reasons, a stable driver ABI is the exact opposite of what is wanted.


I keep making this error ... it is not known as "anti-user features" but rather "anti-features". My apologies for getting the wrong term. Again.

A very good example of an anti-feature is the restrictions on skipping adverts and trailers in DVD players.

http://en.wikipedia.org/wiki/Damaged_good#Anti-features

"Whether a feature is an "anti-feature" often depends on the point of view of the observer: for example, the restrictions on skipping adverts and trailers in DVD players are anti-features from the consumer's viewpoint, but useful features from the viewpoint of the movie studio making the DVD which contains the codes that specify the restrictions".

Spot on. Anyway, since Linux is written by its users, it is written from the point of view that beimg unable to skip adverts and trailers when playing DVDs, or being unable to play a DVD bought in another country ... these kinds of features are bad.

Astute people will also notice that putting this kind of anti-feature into software requires that the software is closed source, and that the end user of such software is therefore unable to remove the anti-feature.

It is also useful to point out that there are vastly many more people who would NOT want this kind of anti-feature included in widely-used desktop software compared to the numbers of those who would want it included. Anyone who would argue ... want to put this to a vote?

Hence having a stable driver ABI is, indirectly, a bad thing, because a stable driver ABI allows for binary-only closed source drivers, and binary-only closed source drivers, in turn, allow anti-features.

A closed source binary-only driver is a "damaged good".

Edited 2011-05-23 07:12 UTC

Reply Parent Score: 2

RE[4]: Big Kernel Lock
by lucas_maximus on Mon 23rd May 2011 14:48 in reply to "RE[3]: Big Kernel Lock"
lucas_maximus Member since:
2009-08-18

To accomodate a stable driver ABI would mandate a lack of flexibility to change that ABI.


The whole point of an interface is that it does not change and the implementation below it changes ... this is standard software engineering practice.

It is possible to hide anti-user features within a closed binary. It turns out that users don't want anti-user features. Linux is written by its users.


Linux is not written by it's users, it is written by large corporations such IBM, Oracle, Redhat and Novell for their own personal needs

So, for at least these couple of reasons, a stable driver ABI is the exact opposite of what is wanted.


It is a common complaint that wireless drivers and video frequently break, this hasn't changed in almost 10 years (since I first used Linux).

So what part of "Linux is written by its users, and not by hardware OEMs or by big media wanting to include DRM" did you fail to understand?


Back to my previous point, much of the GNU/Linux development is done by large corporations, which develop it for their uses, not for the good of you or me ... this is why Desktop Linux has sucked for at least the last 10 years.

This idealogical argument becomes pretty moot, when there is no basement army of hackers improving the operating system, it is done by large corps for their benefit.

This personal anecdote doesn't change the fact that Linux has working drivers for more hardware than any other OS.


My point of it was obvious ... it is pathetic that an Operating system that has less than a 0.01% desktop install base support my hardware than is orders of magnitude far more popular OS and has far more developers.

Dismissing it as far as I am concerned means you can't concoct an explanation.

It is a common complaint that drivers frequently break on Linux, However it they don't frequently break on OpenBSD, Windows and Solaris, I know the latter two have a stable abi and I expect OpenBSD is the same.

Reply Parent Score: 2

RE[3]: Big Kernel Lock
by nej_simon on Mon 23rd May 2011 09:49 in reply to "RE[2]: Big Kernel Lock"
nej_simon Member since:
2011-02-11

"And the change in wisdom will start showing itself


Will this mean a stable driver ABI?
"

That's no longer a large issue anyway, now when we have DKMS and similar systems.

Reply Parent Score: 2

RE[4]: Big Kernel Lock
by lucas_maximus on Mon 23rd May 2011 14:49 in reply to "RE[3]: Big Kernel Lock"
lucas_maximus Member since:
2009-08-18

My wireless still breaks after a kernel upgrade ... So it is an issue.

Reply Parent Score: 2