+LTTng 0.201 fixes omap3 trace clock build: header moved from mach/ to plat/.<br>
+LTTng 0.202 fixes the ARM omap3 build.<br>
+LTTng 0.203 re-introduces the fligh- channel prefix for flight recorder
+channels. This has been lost in the refactoring done around the 2.6.30 Linux
+kernel support series.<br>
+ltt-control 0.80 ltt-armall disables the input subsystem by default. ltt-armall
+-i should be used to activate these markers.<br>
+ltt-control 0.81 divides the lttd daemon in a LGPLv2.1 library and GPLv2 program
+parts. It makes it possible to spawn a daemon instance from a program without
+requiring to fork a new process.<br>
+LTTng 0.204 supports kernel 2.6.33.2<br>
+LTTng 0.205 fixes a kref race in channel teardown vs lttd file open.<br>
+ltt-control 0.82 ensures lttctl waits for lttd to complete when called at trace
+destruction in overwrite mode. Failure to do so can result in unread channels in
+flight recorder mode.<br>
+LTTng 0.206 marks Immediate Values as BROKEN for PowerPC. Will be replaced by
+asm goto from mainline eventually anyway, so don't bother looking for the cause
+of the bug.<br>
+LTTng 0.207 fixes trivial error from 0.206.<br>
+LTTng 0.208 uses msleep() rather than cpu_relax() for better !PREEMPT UP support
+for the kref teardown.<br>
+ltt-control 0.83 does a temporary fix to ensure lttd grabs file handles at the
+end of overwrite mode tracing.<br>
+LTTV 0.12.31 fixes the statedump end state handler to correctly deal with
+threads scheduled out from userspace (waiting for CPU) when the statedump end
+event happen.<br>
+ltt-control 0.84 fixes /debugfs auto-find in ltt-armall and ltt-disarmall
+scripts.<br>
+LTTng 0.209 fixes a module build issue.<br>
+LTTng 0.210 reverts some powerpc-specific instrumentation from mainline that
+clashes with arch-agnostic LTTng instrumentation.<br>
+LTTng 0.211 fixes build menu Kconfig dependencies.<br>
+LTTng 0.212 adds more complete MIPS syscall instrumentation.<br>
+LTTng 0.213 supports kernel 2.6.33.3, moves call_64.h to the correct header
+location and fixes -rt kernel teardown locking issue.<br>
+LTTng 0.214 support kernel 2.6.33.4.<br>
+LTTng 0.215 fixes module circular dependendy.<br>
+ltt-control 0.85 contains an update lttd which uses fadvise and sync_file_range
+to ensure that the page cache is not filled by trace buffers.<br>
+ltt-control 0.86 puts fadvise after sync writeback to use sync writeback rather
+than dirty page writeback. (should improve streaming)<br>
+LTTng 0.216 fixes a trace-clock-32-to-64 teardown bug recently introcuced. Only
+affects ARM and MIPS SMP systems.<br>
+LTTV 0.12.32 includes various cleanups and fixes.<br>
+ltt-control 0.87 fixes argument parsing for periodical buffer switch option.<br>
+LTTng 0.217 supports kernel 2.6.34.<br>
+LTTng 0.218 fixes a missing irq_desc export in kernel/irq/handle.c, affects
+sparse irq configurations.<br>
+LTTV 0.12.33 fixes statistics handling. Use per-cpu rather than per-tracefile
+event tree head. This bug was introduced when LTTng started supporting multiple
+channels.<br>
+LTTng 0.219 fixes an out-of-bound array use in marker.c.<br>
+LTTng 0.220 forces alignment on 128 bytes (larger than structure) to ensure that
+the section alignment in the linker script matches that of overly enthusiastic
+compilers.<br>
+LTTV 0.12.34 fixes a bug in string handling in libltttraceread. Caused incorrect
+string length calculation when a string appeared in many per-cpu tracefiles of
+the same channel.<br>
+LTTng 0.221 now supports writing strings into the buffer when string input is
+being modified concurrently. Such a race should not render the trace
+unreadable.<br>
+LTTng 0.222 supports kernel 2.6.34.4. It includes a lttng splice actor
+adjustment to match a splice.c change in 2.6.34.2 which changed the checks for
+seekable files.<br>
+LTTV 0.12.35 fixes a problem introduced in 0.12.34, fixing the problem initially
+fixed by 0.12.34 in the "correct" way; without unwanted side-effects. LTTV
+0.12.34 had problems calling hooks for some events.<br>