dtrace: Fix spec file for 0.6.1-3

diff --git a/dtrace/NEWS b/dtrace/NEWS
index db04b44fe159f1c472ec5d72a37529dd0b20b536..9e2b137b268aa19fbf76639855165e20b7b6f430 100644 (file)
--- a/dtrace/NEWS
+++ b/dtrace/NEWS
@@ -1,9 +1,60 @@
 DTrace Kernel Modules News
 ==========================
 
-Release 0.6.0 (Apr  3rd, 2017)
+Release 0.6.1 (Jul 14th, 2017)
 ------------------------------
-Kernel release: 4.1.12-97.el6uek
+Kernel release: 4.1.12-103.el6uek
+
+New features:
+
+ - The FBT provider now supports loadable kernel modules.  The current
+   implementation does not support creation of probes inside .init sections.
+
+ - FBT supports creation of return probes on SPARC platform.  At the moment the
+   return probe support is limited to functions the optimizer has not turned
+   into a tail call.
+
+ - On all platforms, return probes now provide correct offsets in their arg0 (on
+   both x86 and SPARC).
+
+ - Added link_ntop() subroutine.
+
+ - Added support for new x86 CPUS with SMAP support.
+
+Changes:
+
+ - FBT probes on x86 are now using int3 instead of the LOCK prefix to trigger
+   a probe.
+
+ - The dynamic variable cleanup routine is now self-throttling.  If the
+   opearating system gets too overloaded and is no longer able to do the cleanup
+   with the given period then it no longer tries to run the handler more than
+   once.
+
+Bugfixes:
+
+ - Fixed mutex_* subroutines that were not accessing mutexes correctly, causing
+   incorrect results.
+
+ - The ustackdepth built-in variable was always returning 1 regardless of the
+   depth of the ustack().
+
+ - Fixed various memory accesses to properly check for privileges.
+
+ - When a FBT probe used int3 as its trigger, it was always handled as if it ran
+   in interrupt context, impacting handling of self variables.
+
+ - The deadman timer values have been adjusted to lower values.  DTrace now
+   gives up sooner rather than allowing the watchdog to reboot the host.
+
+ - The deadman routine was not properly checking state of all CPUs, causing it
+   to miss stuck CPUs in some circumstances (though this was hard to notice due
+   to the overly long timer values).
+
+
+Release 0.6.0 (Apr 3rd, 2017)
+------------------------------
+Kernel release: 4.1.12-97.el6uek
 
 New features:
 

diff --git a/dtrace/dtrace-module.spec b/dtrace/dtrace-module.spec
index 55e1625d7d8c4563597a54a6ca13bfe258cb0223..a9a2dc08a9ce2d384fc1f1a4b9acf3a2d5766f7a 100644 (file)
--- a/dtrace/dtrace-module.spec
+++ b/dtrace/dtrace-module.spec
@@ -20,7 +20,7 @@
 Name: dtrace-modules-%{kver}
 Summary: dtrace module
 Version: 0.6.1
-Release: 2%{dist}
+Release: 3%{dist}
 Provides: dtrace-modules
 Requires: at
 License: CDDL
@@ -127,6 +127,11 @@ rm -rf %{buildroot}
 /usr/include/linux/dtrace/types.h
 
 %changelog
+* Fri Jul 14 2017 - <tomas.jedlicka@oracle.com> - 0.6.1-3
+- FBT module support and SPARCs return probes [Orabug: 26384179, 26384765]
+- DTrace state cleanmust use dtrace_sync() [Orabug: 26385102]
+- Make dynamic variable cleanup self-throtling [Orabug: 26385177]
+- Restore deadman original timing values [Orabug: 26385159]
 * Wed Jun 28 2017 - <kris.van.hees@oracle.com> - 0.6.1-2
 - Change FBT entry probes on x86 to use int3 [Orabug: 26324039]
 - Support x86 CPUs with SMAP (Nick Alcock) [Orabug: 26166784]