From: Thomas Gleixner
Date: Sun Feb 21 2010 - 18:25:24 EST
We are pleased to announce the next update to our new preempt-rt
- Forward to 2.6.33-rc8
Finally I managed to get that beast tamed.
Some folks were already joking, that my dog ate the 33-rt patch. The
truth is that my reject fixup skills were one line off in the mm/*
code which resulted in -rt eating my files :)
When I fixed the wreckage of my own stupidty I found out that -rt
triggered a nasty and hard to reproduce mainline scheduler bug, which
refused to reproduce with kernel tracing enabled. Thanks to Peter
Zijlstra for helping me to decode the initial mainline problem and the
ones which resulted from fixing the mainline bug!
It's all fixed now and I'm pretty happy about the state of this
release with the following exceptions:
1) it is only runtime tested on x86_64 (compiles on x86_32, arm,
I'm going to throw it at my testing infrastructure tomorrow, but I
wanted to release this before .33-final and of course you folks
are heartely invited to help out with fixing the fallout :)
2) I broke the histogram tracer stuff simply because I got lost in
the divergence of -rt and mainline kernel/trace/* changes
I apologize to Carsten, but I'm sure he will fix that in no time
The current state of the patch on x86_64 is: rock stable!
[rt-tests]# chrt -f 1 ./cyclictest -t -a -n -d0 -p80 -i125
policy: fifo: loadavg: 509.19 490.53 486.66 364/1055 31899
T: 0 (27682) P:80 I:125 C:1170997397 Min: 2 Act: 3 Avg: 3 Max: 34
T: 1 (27683) P:79 I:125 C:1170997397 Min: 2 Act: 3 Avg: 3 Max: 34
That's an uptime of ~1.6 days with a load of hackbench, recursive
grep and extensive networking traffic going on.
The merge of the spinlock namespace cleanup and the initial
annotations took a big chunk away from the -rt patch size:
We are zooming in, but there is still a way to go!
http://rt.et.redhat.com/download/ (temporary out of sync)
Git development branch:
Git release branch:
Information on the RT patch can be found at:
To build the 2.6.33-rc8-rt1 tree, the following patches should be
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/