Nick Piggin [interview] released version w25 of his scheduler policy patches [story], primarily targeting improvements for SMP and NUMA servers. Testing his changes on a 16-way NUMAQ at OSDL, Nick notes, "I have done some tweaking of the 'domains', and found some pretty impressive performance improvements." These improvements best show up in his dbench results, included in the following email.
Nick notes that SMT (Symmetric multithreading, aka hyperthreading) [story] is currently broken, suggesting it will remain this way until he finds an SMP P4 HT to use for testing. Read on for the full description of his changes, and numerous benchmarking comparisons.
The 2.6.x kernel seems to have included the i2c/lm_sensors code bases. Things getting clean and proper, except the fact that not all the chips and busses from i2c/lm_sensors 2.8.x have been ported to kernel 2.6..
Does anybody know where to find missing drivers (e.g. the vt1211 used on my EPIA-M MB to probe CPU temp ?).
Does anybody know who is working on this ? Could I help them in any way ?
I recently upgraded to 2.6-test11 due to a problem I was having on test9. I would notice on large (70MB or greater) file transfers via the network (to or from my box) would freeze up the network. My messages file would complain about the following:
Nov 20 09:15:25 senor kernel: ip_local_deliver: bad loopback skb: PRE_ROUTING LOCAL_IN
I Compiled linux-2.6-test 11 but i i can't get my mouse working.
With 2.4.23 there is no problem.
this is my mouse config for 2.6
# cat .config | grep MOUSE
Does anyone has an idea why my mouse ain't working?
Marcelo Tosatti has released the final 2.4.23 stable Linux kernel unchanged from 2.4.23-rc5 [story]. I included a more detailed list of changes since 2.4.22 [story] with the announcement of 2.4.32-rc3 [story]. Among the many fixes and updates, changes include a significant update to the VM including the removal of the OOM killer [story], the addition of "laptop mode" to help increase battery life [story], and DRM support for Xfree86 v4.3.0+.
2.4.23 is available from numerous kernel.org mirrors. Read on for the complete changelog.
Did anyone manage to get 3D acceleration working with the ATI binary drivers for the Radeon 9500pro on an nForce based motherboard? The fglrx driver works fine for me but in the XFree logs I get the following message:
(EE) fglrx(0): [agp] unable to acquire AGP, error "xf86_ENOMEM"
(EE) fglrx(0): cannot init AGP
This is with the built-in agpgart. If I use the ncforce agpgart driver in the kernel starting XFree will result in a black screen and then XFree just terminates.
Linus Torvalds announced 2.6.0-test11, calling it the "Beaver in Detox" kernel and saying it was "for everybody who thought 'stoned beaver' wasn't an appropriate name for a kernel". As to the actual changes, Linus explains:
"This is mainly brought on by the fact that the old aic7xxx driver was broken in -test10, and Ingo [interview] found this really evil test program that showed an error case in do_fork() that we had never handled right. Well, duh! While at it, this also pulls in some firewire fixes and a few potential skbuff leakage points."
Linus went on to note that he'll be too busy eating for the next few days to check his email, and from here on out the 2.6 kernel is in Andrew Morton's [interview] able hands. Whether Andrew will go on to release a 2.6.0-test12, or the official 2.6.0, this is yet to be seen.
first off I hope this forum is the right place. At least I know the interactivity gods post on here, so at least people that should know a answer might read this :).
Im running various 2.6.0-testX kernels (vanilla sources), and I am having situations where I cannot believe how slow my system gets (for example I type text into this window, and under certain load I will not see the characters I type for ~15 seconds, when then suddenly all characters I wrote in that time are displayed in one go.
Andrew Morton [interview], the upcoming 2.6 kernel maintainer, released 2.6.0-test10-mm1 with "various small fixes". He also points out that this release contains some "device mapper and RAID updates which need testing", and that it still includes "the problematic PNP patches". He goes on to add that if 2.6.0-test10-mm1 "crashes mysteriously during boot", you should try reverting the three PNP patches available in the -test10-mm1 'broken-out' directory and see if that helps.
Read on for the complete changelog.
Yet another thing showed up (possible data corruption on x86-64), so here
Summary of changes from v2.4.23-rc4 to v2.4.23-rc5
Con Kolivas [interview] recently posted a "demo" patch to the lkml in response to concern that his interactivity estimator [story] might be resulting in a performance hit. The patch entirely removes his interactivty estimator from the 2.6 kernel, allowing the unbelievers a chance to witness with their own eyes that indeed these ~350 lines of code are not the source of any throughput problems. He explains:
"The estimator should not somehow make your cpu 20% slower unless something is horribly wrong. It simply reorders which tasks go first, and if they do get to go first they round robin more frequently. Overall within a larger timespace the amount of time taken to do the work is the same (or slightly better in some settings). Delaying the cpu bound tasks and then letting them run for a longer timeslice when there is a larger cpu window allows them to benefit more from cpu cache. While the code looks complicated, the overhead is miniscule."
System: Fedora Core 1, i686,Celeron 2.2Ghz Toshiba A10 Laptop
I have used the RPM kernel updates available from arjanv since the test series of kernels have been released. Up until test10 I could literally just rpm -ivh the newest kernel as a complete drop in for any of my machines. Now the kernel oops' in /include/linux/rwsem.h - which I think I can trace back to the inclusion of the 4G/4G split patch. The kernel spits out into messages "Unable to handle kernel paging request at virtual address..." and dumps on me.
Greg Kroah-Hartman released version 007 of udev for the 2.6 kernel [forum], an "implementation of devfs in userspace using sysfs and /sbin/hotplug." He says:
"I think with the ability to capture the output of the CALLOUT rule, combined with the ability to put format modifiers in the CALLOUT program string, we now have everything in place to emulate the existing devfs naming scheme. Anyone want to verify this or not?"
Read on for the complete changelog.