|| ||Steven Rostedt <firstname.lastname@example.org>|
|| ||LKML <email@example.com>,
|| ||Thu, 15 Nov 2007 13:40:30 -0500|
|| ||Ingo Molnar <firstname.lastname@example.org>, Thomas Gleixner <email@example.com>,
Gregory Haskins <firstname.lastname@example.org>,
Peter Zijlstra <email@example.com>,
Daniel Walker <firstname.lastname@example.org>,
"Paul E. McKenney" <email@example.com>|
We are pleased to announce 2.6.24-rc2-rt1.
This is probably one of the toughest ports the -rt patch had to face. It
was definitely the toughest one I had to perform. With a 50 Meg
difference between 2.6.23 and 2.6.24-rc2 it caused a lot of problems
with conflicts. The 188.8.131.52-rt11 patch consisted of 386 patches. Of
them 25 of them were incorporated into mainline (mostly hrtimers for
x86_64 and other archs).
With fixes and such, the 2.6.24-rc2-rt1 has 370 patches. 361 of them
came from 184.108.40.206-rt11 and of those 129 had conflicts. Some conflicts
were trivial, but a lot of them were not. In order to allow others to
audit my changes I have archived the .rej reject files as well as the
quilt .pc files and original patch.
This holds a list of tar balls that contain the necessary files to get
to the state before I preformed the "quilt refresh".
For example: to fix the patch preempt-realtime-mips.patch
rostedt@gandalf:~$ mkdir bah
rostedt@gandalf:~$ cd bah
rostedt@gandalf:~/bah$ ketchup -r -G 2.6.24-rc2
None -> 2.6.24-rc2
Current directory renamed to /home/rostedt/linux-2.6.24-rc2
rostedt@gandalf:~/bah$ cd .
rostedt@gandalf:~/linux-2.6.24-rc2$ tar -xf patch-2.6.24-rc2-rt1-broken-out.tar.bz2
rostedt@gandalf:~/linux-2.6.24-rc2$ quilt push preempt-realtime-mips.patch
rostedt@gandalf:~/linux-2.6.24-rc2$ tar -xf preempt-realtime-mips.patch.tar.bz2
rostedt@gandalf:~/linux-2.6.24-rc2$ find . -name '*.rej'
And now I'm in the position before I did the "quilt refresh" against the
preempt-realtime-mips.patch. Others can do this too, and see where I
might have messed up the conflicts. Note, doing a "quilt pop -f" now
will remove the patch as I resolved it, and doing a "quilt push -f" will
put you to the position before I did the patch.
Note, I may have messed up by running a "quilt refresh" before running
the reject archive script. If that's the case (the saved patch will be
the resolved patch, not the original), simply take the patch from the
patch-220.127.116.11-rt11-broken-out.tar.bz2 and use that as the original.
I've tested this on x86_64 both SMP and UP, as well as SMP on i386. I'm
pretty sure that the other archs may be broken (arm, mips and powerpc).
Because this was such a big change, there is probably a bit of
regression bugs that might have crept in. So please test and report your
Special thanks goes to Paul E. Mckenney (for pointing me to the wrong
bug, that happened to sit right next to the correct one ;-), as well as
Peter Zijlstra and Daniel Walker for helping me debug the initial ports.
Changes since 18.104.22.168-rt11
- Port to 2.6.24-rc2
- latency tracer bootmem allocate (Yanmin Zhang)
- removed Gregory Haskins patch 8 (will probably be back in -rt2)
This was due to some initial porting problems. But the fixes
came too late in the testing period. (he decided cleaning his
garage was more important than working on this kernel.;-)
to build a 2.6.24-rc2-rt1 tree, the following patches should be applied:
And like always, my RT version of Matt Mackall's ketchup will get this
for you nicely:
The broken out patches are also available.
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to firstname.lastname@example.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/