|
|
Subscribe / Log in / New account

Development kernel 2.6.0-test9 released

Linus has released 2.6.0-test9 with another set of fixes, along with the "libata" disk driver. "If this works out, then I'll submit -test10 to Andrew Morton, and if he takes it we'll probably have a real 2.6.0 after a final shakedown." See the long-format changelog for the details.

to post comments

Getting close!

Posted Oct 25, 2003 22:43 UTC (Sat) by alspnost (guest, #2763) [Link]

Sounds like we're really close to 2.6 now guys and girls. Can you hear those thunderous drums rolling in the distance? Anyway, I really must get around to booting 2.6 on my machine. I still haven't gotten around to it, partly because of complications with my sound/video drivers....

Development kernel 2.6.0-test9 released

Posted Oct 26, 2003 5:33 UTC (Sun) by yodermk (subscriber, #3803) [Link] (2 responses)

Is there anything specific us non kernel hackers can do to pound on this? It seems to be working very well for me, but I think it may have hung my network card after about 3 days of uptime once under test6-mm2. No idea if I'll be able to reproduce that. It's a 3Com 905B. Now I'm at 4 days, 10 hours under test8-mm1, no problem.

My system is a dual P3 850 with 2 gigs of RAM. If there's something practical i can pound on, let me know!

Development kernel 2.6.0-test9 released

Posted Oct 26, 2003 8:25 UTC (Sun) by TheOneKEA (guest, #615) [Link] (1 responses)

Sure - compile it, boot it and start stressing your system with cpuburn, bonnie++, hdparm, and other benchmarks. If it's flaky or erratic or locks up or starts Oopsing, report it to the list.

SO far I've reported several bugs to linux-kernel with the 2.6.0-test kernels; I'm running 2.6.0-test8 and I'm about to move to 2.6.0-test9.

Development kernel 2.6.0-test9 released

Posted Oct 26, 2003 12:46 UTC (Sun) by TheOneKEA (guest, #615) [Link]

And I just moved to 2.6.0-test9 and got a rude awakening. There are serious issues in the ACPI code which result in a huge overflowing release of debugging messages from the VM code. I've dropped back to 2.6.0-test8-mm1 until I get a response from the list.

Development kernel 2.6.0-test9 released

Posted Oct 26, 2003 23:00 UTC (Sun) by kervel (guest, #9133) [Link] (6 responses)

if you compile SATA support in, the device names for your SATA drives will
change. My SATA drive used to be /dev/hde, but after upgrade to test9 its /dev/
sda1. I had to update lilo.conf and /etc/fstab to get my system boot again ...

Development kernel 2.6.0-test9 released

Posted Oct 26, 2003 23:43 UTC (Sun) by snitm (guest, #4031) [Link] (5 responses)

Its because they moved to libata; which leverages the SCSI subsystem for much of the SATA heavy lifting.

Development kernel 2.6.0-test9 released

Posted Oct 27, 2003 16:58 UTC (Mon) by blindoracle (guest, #15918) [Link] (4 responses)

So what are you supposed to do if you have a SCSI/SATA system? Seems like the SATA devices need to be renamed.

Development kernel 2.6.0-test9 released

Posted Oct 27, 2003 17:33 UTC (Mon) by MathFox (guest, #6104) [Link]

Have two different root partitions on your HD (so that you can have two versions of your /etc directory). You can share /usr, /home and /tmp.

Development kernel 2.6.0-test9 released

Posted Oct 27, 2003 18:11 UTC (Mon) by jwb (guest, #15467) [Link] (2 responses)

Use devfs, where it would be /dev/scsi/hostN/busX/targetY/lunZ/*. It still isn't perfect because your hosts can get renumbered, but at least it makes sense. Other than that my main recommendation is to mount your filesystems by name, instead of by device node.

Development kernel 2.6.0-test9 released

Posted Oct 27, 2003 20:40 UTC (Mon) by blindoracle (guest, #15918) [Link] (1 responses)

I thought devfs was considered depreciated? I thought using it with 2.6 was going to be seriously frowned on?!??

The other option, two roor filesystems, I assume I was supposed to read that with my touge squarely planted in my cheek.

Development kernel 2.6.0-test9 released

Posted Oct 28, 2003 8:36 UTC (Tue) by proski (subscriber, #104) [Link]

I think it was more a political decision to deprecate devfs to stimulate development of alternative solutions (such as udev). udev is still in the initial stages of development. Creating device entries manually is not a better alternative to devfs.


Copyright © 2003, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds