Call For Linux 2.5 Testers

An anonymous reader writes "Linus has put out a call for testers with the release of 2.5.40. IDE appears to be in working condition, and the only really obvious thing that could be a problem anymore is the lack of any working volume manager... (LVM is b0rk, atm) So unless that's a problem, start your kernel compiles."

Comments

if we get this tester

[burns210]

if we actually get people testing this, then we won't have to wait serveral versions down the 2.6.x (or is it 3.0.x) road before it is considered stable.

Re:WRONG!

[leviramsey]

Remember the 2.4 vm fiasco? That was supposedly a finished kernel.

However, this time around, many of the new things have been in development outside the kernel for several years, with heavy testing (ALSA being the best example of this). I may give 2.5 a try...

Re:WRONG!

[spencerogden]

Were you asked to put it into production? Of course there are other problems, thats why its a test kernel. But maybe there is a problem that only shows up on a certain configuration you use, wouldn't it be nice(assuming you have the time) to try it out now, let them know about any problems so that 2.6 is even more stable? Isn't that the whole point of open developement?

2.5.40 on RH8 - decent so far ?

[Zeio]

I just got RedHat 8.0 (psyche) up and working last night (full install over 4GB, no JFS or XFS options during install, only EXT3/2), with their gcc-3.2, all the stuff done just right (that is to say, the RedHat way, I don' not use RedHat or Linux much anymore, I use FreeBSD but like to pay attention to Linux). I've been having problems with getting the 2.5.xx series to compile cleanly of late, lots of broken patches seem to make it into this thing, which is to be expected at this stage in the game. One thing I wish was a requisite before the kernel version is revved is that everything compiles and if it doesn't I gets flagged as such in the configurators so one doesn't spend copious amounts of time figuring stuff like that out empirically. 2.5.40 also crashed out of "make menuconfig" if I went into the ALSA section. I wish that the release process for Linux would get a bit more refined, using a source management tool was step one, now I think its time to build a base system around Linux to ensure me of more things when I get it. You could say, get a standalone kernel, or more desirable, a mini system with a c library and compiler and some tools - so that the kernel guys say, we know it works here, with this compiler, and this library, that's what we use. This is coherency I enjoy in other places. This isn't to say Linux isn't meritorious, quite to the contrary, but I would personally like to see things differently - I'm sure I'm re-hashing something that has been said a billion times already. I just though it ironic that 2.5.40 doesn't compile on RedHat 8.0 release -strange considering a good number of the RedHat people are kernel hackers.

Linux distributions vary wildly in their various eclectic incarnations as to how things are supposed to get done. My favorite system I have seen thus far is Gentoo. I like to see source usage encouraged, base system clearly defined, reference design and methods of extension al la ports (or emerge).

An open question, if I have suggestions or problems, is there a place for people who don't have time to live and die by Linux to "drop it in the suggestion box?" I had some problems here and there in the past and have found that people "don't want to hear it." I don't mind being incorrect, but I don't take correction without explanation. I have yet to year why there is a good reason for things that don't compile being checked into 2.4.STABLE - which I also follow.

So as far as beta testing goes for Linux kernel. Do they want beta testers? The attitude on the mailing lists ranges from super helpful (some code maintainers are very good about dealing with breakage) to this "if you cant write a better implementation, FO, I don't want to hear its broke, don't like it don't use it". In any case, how is it exactly us trying to use this kernel going to help the better it if the method of information ingress is unclear? Is there a procedure? Like Mozilla when it faults, you get to send errors in, stuff like that. Is there a memory dump in the kernel yet or is that still a patch (it's a tradition that kernels dump to swap then copy on boot do you can see what the computer was doing if it panics). One thing about Linux - if you compile it, load the crap out of it to test it, if it doesn't panic in the 1st day it seems to never panic - which is good.

Just an FYI for people getting into kernel stuff with RedHat-ish systems:

Getting Linux via bitkeeper.

First, get BitKeeper:

http://www.bitmover.com/cgi-bin/download.cgi

Follow the instructions and it will tell you how to download and install BitKeeper.

Then, clone the main Linux tree using BitKeeper:
$ cd /usr/src/linux-2.5.40
(or wherever you would like your stuff)
$ bk clone bk://linux.bkbits.net/linux-2.5
$ ln -s linux-2.5.40 linux
$ (optional if needed, ln -s linux-2.5.40 linux-2.4 ; ln -s linux-2.5.40 linux-2.5) - sometimes dists and weird driver SRPMS look for linux/include in all sorts of places
$ cd linux
$ bk -r co

Also don't forget.

- /usr/src/linux , /usr/src/scsi ; /usr/src/asm ; /usr/src/asm-generic should all be re-linked to the right places in /usr/src/linux/include [if this is no longer necessary let me know]
- make install doesn't work with grub, so you have to do your thing manually now
- recommended compiler is gcc-2.9.5.3 [for 2.4 and 2.5 now], I always have extra compilers ready to go just in case. Make sure all the tools are the proper version, and that you have a recent ksymoops (if you need to do any messing around looking for problems ), modutils - etc.

If the build fails, find the offending code and remove from selection, or try to hack it if you need it.

Re:Gotta love how Linux development proceeds...

[Kynde]

Members of the kernel development community don't seem to agree that the development kernel will function on IDE drives without destroying them...

Maybe instead of bloating the featureset, they should be working on getting the I/O working?

I/O? Are you seriously complaining that the IO thruput is lacking? I'm guessing you're talking about the interactive feel.

So and so, but I bet you don't even run 2.5.X. Granted the 2.4 imho is still interactively crap. It's been ever since the vm "improvement" starting with 2.4.10-preX and how last fall those fscking db-benchmarkers showed up on lkml. I bet they got their 1% improvements for their db and on the same they bollocksed my developement desktop usability to a level it felt like win95. Atleast I learned a lot more about the vm and scheduling internals and loads of neat tricks how to improve latency (e.g. more memory and no, Hz up, neg-nicing X, nicing back ground stuff, A.Morton's low latency and lock breaking patches.)

BUT I must say that the 2.5 is a step forward from 2.4 also in this manner. I've been running 2.5s all along and these late 2.5.30s, especially 2.5.39, have been goodish. Not the kind of interactivity as there should be, still.

Although I'm still seriously pissed off how all the main developers seemed to disregard all the cries about interactivity last fall and winter on lkml. There SHOULD be a proc or even a compile time possibility to tweak it into serverish/db or desktop usage. On a desktop developement you couldn't care less if your compile or what ever the fsck you're doing on the back ground takes 5% longer, but if what ever you're actually interacting with lags like hell, it annoys the living crap out of you. I mean, if I press the button while I'm compiling, obviously that button press is more important ffs.