Time for Users to Start Testing 2.5 (Linux Journal)
A lot of people ask me, "When do you think the 2.6 kernel will be released?" My response to that question usually is, "Well, how well is the 2.5 kernel working for you?" Inevitably, during the resulting conversation where I plead with the person to please at least run the kernel once on their hardware, they respond with one of the following reasons why they have not tried 2.5:"
Posted Mar 22, 2003 20:16 UTC (Sat)
by Xanadu (guest, #1215)
[Link] (4 responses)
Posted Mar 22, 2003 23:07 UTC (Sat)
by gregkh (subscriber, #8)
[Link] (1 responses)
Posted Mar 25, 2003 16:00 UTC (Tue)
by freelsjd (guest, #250)
[Link]
Posted Mar 24, 2003 18:28 UTC (Mon)
by plars (guest, #7736)
[Link] (1 responses)
It's well known that there are several drivers that won't even build currently, and those issues ARE being worked though. OSDL has been tracking stats on which drivers have build errors and warnings and they show pretty clearly that progress is being made with every release.
Posted Mar 25, 2003 23:31 UTC (Tue)
by freelsjd (guest, #250)
[Link]
Posted Mar 23, 2003 8:09 UTC (Sun)
by einstein (guest, #2052)
[Link]
It's been getting a lot better lately - I'm currently running 2.5.65-bk2 on a firewall/squid server for a small lan, which also doubles as an occasional X workstation - I'm very impressed with the snappiness of the GUI when the system is under load. Interestingly enough, this box used to crash mysteriously with the original Red Hat 8.0 install - with recent 2.5 kernels it's actually been more stable, as well as having a nicer feel for GUI tasks.I guess the combination of the recent 2.5 kernel and the new glibc really helps - I can't wait for 2.6, I think this new kernel will be a great boon to 3D FPS-playing linux gamers.
Posted Mar 24, 2003 16:34 UTC (Mon)
by freelsjd (guest, #250)
[Link] (8 responses)
Posted Mar 24, 2003 18:30 UTC (Mon)
by wa1hco (subscriber, #3628)
[Link] (7 responses)
Posted Mar 24, 2003 23:01 UTC (Mon)
by freelsjd (guest, #250)
[Link] (6 responses)
Posted Mar 25, 2003 8:09 UTC (Tue)
by Peter (guest, #1127)
[Link] (5 responses)
Install module-init-tools, which is supposed to work with kernels 2.4 (via handoff to modutils) and 2.5.
Posted Mar 25, 2003 15:58 UTC (Tue)
by freelsjd (guest, #250)
[Link] (4 responses)
Posted Mar 25, 2003 18:09 UTC (Tue)
by alexp (guest, #10320)
[Link] (3 responses)
Posted Mar 25, 2003 21:52 UTC (Tue)
by freelsjd (guest, #250)
[Link] (2 responses)
Posted Mar 26, 2003 22:56 UTC (Wed)
by freelsjd (guest, #250)
[Link] (1 responses)
I think the oldconfig should at least get these things right to get the general user like myself to test the new kernel. On the kernel itself...well I am writing this now from the web if that says anything. Seems great so far !
Posted Apr 1, 2003 19:21 UTC (Tue)
by lucaramel (guest, #10432)
[Link]
Why not try one? Time for Users to Start Testing 2.5 (Linux Journal)
How about I've been trying them for the past 15 - 20 releases and have not had a single
one compile yet.
I'd LOVE to try them, I just can't get one to finish compiling.
Gentoo, GCC 2.95.3 *AND* gcc 3.2.2
Any reasons why you haven't posted the problems you have had?Time for Users to Start Testing 2.5 (Linux Journal)
Are you trying to build drivers that you do not have hardware for?
Have you filed a bug in the bugzilla.kernel.org for the problem?
Version 2.5.66 was just released. Maybe what I should do first is a "make Time for Users to Start Testing 2.5 (Linux Journal)
oldconfig" without enabling any new features and see if it will boot.
You might also want to check the kernel errata page at: http://ltp.sourceforge.net/errata which is updated with known fixes/workarounds for blocking problems (can't build, can't boot, etc) for the latest kernel. Also looking at where the error occurred and deciding if you really need (or even have) that driver can help make it through a build.Time for Users to Start Testing 2.5 (Linux Journal)
This errata site enabled me to figure out why I could not boot. It turned out Time for Users to Start Testing 2.5 (Linux Journal)
someone else had the same problem: I needed to enable the VT and VT_CONSOLE
parameters; also enable a vga screen later down the menu. Now it boots !
However, I cannot get either the keyboard or mouse to work. In the "input devices"
section, I did find a "legacy psaux" parameter, but I'm sure that is not what is
intended. This allowed for X to start having found the mouse. But, neither the
mouse nor keyboard works.
There is an "input device parameter" under "input device drivers" and a "keyboards"
and "mice" parameter. The help for these parameters says something to the effect
that a menu should come up and a list of devices to pick. And it says "This option
doesn't effect the kernel" So, how does one enable the mouse and keyboard ?
Also, is there a better forum for general development kernel help for users like me
who are trying to test ?
I've been using 2.5 on and off since somewhere in the 2.5.3x days -Time for Users to Start Testing 2.5 (Linux Journal)
OK. I did. After running "make oldconfig" and picking which new features I was Time for Users to Start Testing 2.5 (Linux Journal)
interested in (being careful not to be too experimental of course), I also picked the alsa
sound drivers (I use alsa now with 2.4.20). "make modules_install" responded with many
unresolved external references. How do you fix this ? I have not RTFM yet.
Get the new module utils from kernel.org/?/?/?/people/rusty. Note this will break upgrade of Redhat kernels.
Time for Users to Start Testing 2.5 (Linux Journal)
I was successful in using the most recent module utilities from Rusty and it did Time for Users to Start Testing 2.5 (Linux Journal)
allow me to finish the creation of the 2.5.65 modules including the ALSA sound
modules. I then performed a "make install", modified my lilo.conf to add the 2.5.x
kernel as an option, and executed lilo successfully. However, when I tried to boot,
the new kernel, it stopped the boot process just after or at the time of
uncompressing the kernel (would not boot). It was necessary to perform a hard
reset (push the ole reset button).
In addition, the development module utilities were apparently not compatible with
the officially released ones, and I could not use the modules from the stable kernel.
I then had to reinstall the modutils package from Debian/Sid (which required some
manual deletion of a few files under /lib/modules/2.4.20/ and a new "make
modules_install" from the 2.4.20 sources. Basically a real pain to clean up!
So, not much success in using the 2.5.x kernels from this Debian/Sid user.
Time for Users to Start Testing 2.5 (Linux Journal)
In addition, the development module utilities were apparently not compatible with the officially released ones, and I could not use the modules from the stable kernel. I then had to reinstall the modutils package from Debian/Sid
Thanks. I have installed the module-init-tools and they work as expected. Time for Users to Start Testing 2.5 (Linux Journal)
Do you have any idea why the newly compiled kernel will not boot ? Does the boot
process depend on module-init-tools ? It hasn't gotten to the point of loading
modules. It has only uncompressed the bzImage and then locks.
I had similar (not booting) problem with recent 2.5.6x kernels which I was able to track down to an ATA hard drive configuration option - "ATA tagged command queueing (EXPERIMENTAL)". Make sure it's not checked
Time for Users to Start Testing 2.5 (Linux Journal)
I checked my configuration and I did not have this option enabled. But, following Time for Users to Start Testing 2.5 (Linux Journal)
this, I created a new configuration from scratch that was a straight "make
oldconfig" from a working 2.4.20 kernel configuration. This did not boot. Then I
removed any experimental parameters and any parameters that were by default set
to yes as a new option (a couple of places) and verify that I do not need this
option, and then disable any parameter that I absolutely do not need (i.e., create a
bare bones kernel). This did not boot either.
I think there is something fundamental that is not allowing this kernel to boot.
I finally got there ! The "oldconfig" script does not produce consistent configuration at all. There are several parameters that must be enabled in different places just to get the standard ps/2 keyboard and mouse to work. These are the most common by everyone ! Then the console is not enabled by default. This is tricky to get enabled. Most everyone has a console also !Time for Users to Start Testing 2.5 (Linux Journal)
Could you please tell us which options you enabled to make this one boot ?Time for Users to Start Testing 2.5 (Linux Journal)
Because I'm running Debian sid, too, and I can't get 2.5.66 to boot (I've installed module-init-tools BTW).
Thanks in advance.