pool/gperftools
SHA256
4
0
Fork 1
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
gperftools/gperftools.changes
Michal Vyskocil 15145287ff Accepting request 280951 from home:ganglia:perftools-2.4 
- gperftools 2.4, available 10 Jan 2015.
  Skipped version 2.3, release 7 Dec 2014.
  Changes for version 2.4
  * enabled aggressive decommit option by default. It was found to 
    significantly improve memory fragmentation with negligible impact 
    on performance. (Thanks to investigation work performed by Adhemerval 
    Zanella) 
  * added ./configure flags for tcmalloc pagesize and tcmalloc allocation 
    alignment. Larger page sizes have been reported to improve performance 
    occasionally. (Patch by Raphael Moreira Zinsly) 
  * sped-up hot-path of malloc/free. By about 5% on static library and 
    about 10% on shared library. Mainly due to more efficient checking 
    of malloc hooks. 
  * improved accuracy of stacktrace capturing in cpu profiler (due to 
    issue found by Arun Sharma). As part of that issue pprof's handling 
    of cpu profiles was also improved. 
- Removed missing_includes.patch; incorporated upstream.

old: devel:libraries:c_c++/gperftools
new: home:ganglia:perftools-2.4/gperftools rev 3
Index: gperftools.changes
===================================================================
--- gperftools.changes (revision 26)
+++ gperftools.changes (revision 3)
@@ -1,4 +1,60 @@
 -------------------------------------------------------------------
+Mon Jan 10 16:30:24 UTC 2015 - cgardner@suse.com
+
+- gperftools 2.4, available 10 Jan 2015.
+  Skipped version 2.3, release 7 Dec 2014. 2.3 changes included:
+  * ( issue 631 ) fixed debugallocation miscompilation on mmap-less 
+    platforms (courtesy of user iamxujian) 
+  * ( issue 630 ) reference to wrong PROFILE (vs. correct CPUPROFILE) 
+    environment variable was fixed (courtesy of WenSheng He) 
+  * pprof now has option to display stack traces in output for heap 
+    checker (courtesy of Michael Pasieka) 
+  * ( issue 636 ) pprof web command now works on mingw 
+  * ( issue 635 ) pprof now handles library paths that contain spaces 
+    (courtesy of user mich...@sebesbefut.com) 
+  * ( issue 637 ) pprof now has an option to not strip template 
+    arguments (patch by jiakai) 
+  * ( issue 644 ) possible out-of-bounds access in GetenvBeforeMain 
+    was fixed (thanks to user abyss.7) 
+  * ( issue 641 ) pprof now has an option --show_addresses (thanks to 
+    user yurivict). New option prints instruction address in addition 
+    to function name in stack traces 
+  * ( issue 646 ) pprof now works around some issues of addr2line 
+    reportedly when DWARF v4 format is used (patch by Adam McNeeney) 
+  * ( issue 645 ) heap profiler exit message now includes remaining 
+    memory allocated info (patch by user yurivict) 
+  * pprof code that finds location of /proc/pid/maps in cpu profile 
+    files is now fixed (patch by Ricardo M. Correia) 
+  * ( issue 654 ) pprof now handles "split text segments" feature of 
+    Chromium for Android (patch by simonb) 
+  * ( issue 655 ) potential deadlock on windows caused by early call 
+    to getenv in malloc initialization code was fixed (bug reported 
+    and fix proposed by user zndmitry) 
+  * incorrect detection of arm 6zk instruction set support 
+    (-mcpu=arm1176jzf-s) was fixed. (Reported by pedronavf on old issue-493) 
+  * new cpu profiling mode on Linux is now implemented. It sets up separate 
+    profiling timers for separate threads. Which improves accuracy of 
+    profiling on Linux a lot. It is off by default. And is enabled if both 
+    librt.f is loaded and CPUPROFILE_PER_THREAD_TIMERS environment 
+    variable is set. But note that all threads need to be registered 
+    via ProfilerRegisterThread. 
+  Changes for version 2.4~rc
+  * enabled aggressive decommit option by default. It was found to 
+    significantly improve memory fragmentation with negligible impact 
+    on performance. (Thanks to investigation work performed by Adhemerval 
+    Zanella) 
+  * added ./configure flags for tcmalloc pagesize and tcmalloc allocation 
+    alignment. Larger page sizes have been reported to improve performance 
+    occasionally. (Patch by Raphael Moreira Zinsly) 
+  * sped-up hot-path of malloc/free. By about 5% on static library and 
+    about 10% on shared library. Mainly due to more efficient checking 
+    of malloc hooks. 
+  * improved accuracy of stacktrace capturing in cpu profiler (due to 
+    issue found by Arun Sharma). As part of that issue pprof's handling 
+    of cpu profiles was also improved. 
+- Removed missing_includes.patch; incorporated upstream.
+
+-------------------------------------------------------------------
 Sun Sep 28 18:44:03 UTC 2014 - jengelh@inai.de
 
 - Implement shlib packaging guidelines: split libraries
Index: gperftools.spec
===================================================================
--- gperftools.spec (revision 26)
+++ gperftools.spec (revision 3)
@@ -15,9 +15,8 @@
 # Please submit bugfixes or comments via http://bugs.opensuse.org/
 #
 
-
 Name:           gperftools
-Version:        2.2
+Version:        2.4
 Release:        0
 Summary:        Performance Tools for C++
 License:        BSD-3-Clause
@@ -28,8 +27,6 @@
 Patch4:         %{name}-fix_docdir.patch
 Patch12:        %{name}_fix_unassigned_malloc_in_unittest.patch
 Patch14:        %{name}_gcc46.patch
-#PATCH-FIX-UPSTREAM https://code.google.com/p/gperftools/issues/detail?id=268
-Patch15:        missing_includes.patch
 BuildRequires:  autoconf >= 2.59
 BuildRequires:  automake
 BuildRequires:  gcc-c++
@@ -82,7 +79,6 @@
 %patch4 -p1
 %patch12 -p1
 %patch14 -p1
-%patch15 -p1
 
 %build
 autoreconf -fi
@@ -109,7 +105,6 @@
 %files -n libprofiler0
 %defattr(-,root,root)
 %_libdir/libprofiler.so.0*
-# %_libdir/libstacktrace.so.0*   # libstacktrace.so got removed in 0.96
 
 %files -n libtcmalloc4
 %defattr(-,root,root)
@@ -124,8 +119,6 @@
 %_includedir/google
 %_includedir/gperftools
 %_libdir/libprofiler.so
-# %_libdir/libstacktrace.*a    # libstacktrace.so got removed in 0.96
-# %_libdir/libstacktrace.so    # libstacktrace.so got removed in 0.96
 %_libdir/libtcmalloc.so
 %_libdir/libtcmalloc_debug.so
 %_libdir/libtcmalloc_minimal.so
Index: gperftools-2.4.tar.gz
===================================================================
Binary files gperftools-2.4.tar.gz (revision 3) added
Index: gperftools-2.2.tar.gz
===================================================================
Binary files gperftools-2.2.tar.gz (revision 26) deleted
Index: missing_includes.patch
===================================================================
--- missing_includes.patch (revision 26)
+++ missing_includes.patch (deleted)
@@ -1,415 +0,0 @@
-commit 75b65f96b0bb44d70e0a461a03a3f8c928390283
-Author: Raphael Moreira Zinsly <rzinsly@linux.vnet.ibm.com>
-Date:   Tue Apr 8 17:45:13 2014 -0300
-
-    PowerPC: stacktrace function refactor and fixes
-    
-    This patch fixes the stacktrace creating when the function is
-    interrupted by a signal. For Linux, the vDSO signal trampoline symbol is
-    compared against LR from stack backchain and handled different in that
-    case (since the signal trampoline layout a different stack frame).
-    
-    Because of this extensive change the PowerPC stacktrace code has now
-    been refactored to split in Linux and Darwin specific codes.
-
-diff --git a/src/stacktrace_powerpc-darwin-inl.h b/src/stacktrace_powerpc-darwin-inl.h
-new file mode 100644
-index 0000000..c4c2edb
---- /dev/null
-+++ b/src/stacktrace_powerpc-darwin-inl.h
-@@ -0,0 +1,158 @@
-+// Copyright (c) 2007, Google Inc.
-+// All rights reserved.
-+//
-+// Redistribution and use in source and binary forms, with or without
-+// modification, are permitted provided that the following conditions are
-+// met:
-+//
-+//     * Redistributions of source code must retain the above copyright
-+// notice, this list of conditions and the following disclaimer.
-+//     * Redistributions in binary form must reproduce the above
-+// copyright notice, this list of conditions and the following disclaimer
-+// in the documentation and/or other materials provided with the
-+// distribution.
-+//     * Neither the name of Google Inc. nor the names of its
-+// contributors may be used to endorse or promote products derived from
-+// this software without specific prior written permission.
-+//
-+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
-+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-+
-+// ---
-+// Produce stack trace.  ABI documentation reference can be found at:
-+// * PowerPC32 ABI: https://www.power.org/documentation/
-+// power-architecture-32-bit-abi-supplement-1-0-embeddedlinuxunified/
-+// * PowerPC64 ABI:
-+// http://www.linux-foundation.org/spec/ELF/ppc64/PPC-elf64abi-1.9.html#STACK
-+
-+#ifndef BASE_STACKTRACE_POWERPC_INL_H_
-+#define BASE_STACKTRACE_POWERPC_INL_H_
-+// Note: this file is included into stacktrace.cc more than once.
-+// Anything that should only be defined once should be here:
-+
-+#include <stdint.h>   // for uintptr_t
-+#include <stdlib.h>   // for NULL
-+#include <gperftools/stacktrace.h>
-+
-+// Given a pointer to a stack frame, locate and return the calling
-+// stackframe, or return NULL if no stackframe can be found. Perform sanity
-+// checks (the strictness of which is controlled by the boolean parameter
-+// "STRICT_UNWINDING") to reduce the chance that a bad pointer is returned.
-+template<bool STRICT_UNWINDING>
-+static void **NextStackFrame(void **old_sp) {
-+  void **new_sp = (void **) *old_sp;
-+
-+  // Check that the transition from frame pointer old_sp to frame
-+  // pointer new_sp isn't clearly bogus
-+  if (STRICT_UNWINDING) {
-+    // With the stack growing downwards, older stack frame must be
-+    // at a greater address that the current one.
-+    if (new_sp <= old_sp) return NULL;
-+    // Assume stack frames larger than 100,000 bytes are bogus.
-+    if ((uintptr_t)new_sp - (uintptr_t)old_sp > 100000) return NULL;
-+  } else {
-+    // In the non-strict mode, allow discontiguous stack frames.
-+    // (alternate-signal-stacks for example).
-+    if (new_sp == old_sp) return NULL;
-+    // And allow frames upto about 1MB.
-+    if ((new_sp > old_sp)
-+        && ((uintptr_t)new_sp - (uintptr_t)old_sp > 1000000)) return NULL;
-+  }
-+  if ((uintptr_t)new_sp & (sizeof(void *) - 1)) return NULL;
-+  return new_sp;
-+}
-+
-+// This ensures that GetStackTrace stes up the Link Register properly.
-+void StacktracePowerPCDummyFunction() __attribute__((noinline));
-+void StacktracePowerPCDummyFunction() { __asm__ volatile(""); }
-+#endif  // BASE_STACKTRACE_POWERPC_INL_H_
-+
-+// Note: this part of the file is included several times.
-+// Do not put globals below.
-+
-+// The following 4 functions are generated from the code below:
-+//   GetStack{Trace,Frames}()
-+//   GetStack{Trace,Frames}WithContext()
-+//
-+// These functions take the following args:
-+//   void** result: the stack-trace, as an array
-+//   int* sizes: the size of each stack frame, as an array
-+//               (GetStackFrames* only)
-+//   int max_depth: the size of the result (and sizes) array(s)
-+//   int skip_count: how many stack pointers to skip before storing in result
-+//   void* ucp: a ucontext_t* (GetStack{Trace,Frames}WithContext only)
-+int GET_STACK_TRACE_OR_FRAMES {
-+  void **sp;
-+  // Apple OS X uses an old version of gnu as -- both Darwin 7.9.0 (Panther)
-+  // and Darwin 8.8.1 (Tiger) use as 1.38.  This means we have to use a
-+  // different asm syntax.  I don't know quite the best way to discriminate
-+  // systems using the old as from the new one; I've gone with __APPLE__.
-+  // TODO(csilvers): use autoconf instead, to look for 'as --version' == 1 or 2
-+  __asm__ volatile ("mr %0,r1" : "=r" (sp));
-+
-+  // On PowerPC, the "Link Register" or "Link Record" (LR), is a stack
-+  // entry that holds the return address of the subroutine call (what
-+  // instruction we run after our function finishes).  This is the
-+  // same as the stack-pointer of our parent routine, which is what we
-+  // want here.  While the compiler will always(?) set up LR for
-+  // subroutine calls, it may not for leaf functions (such as this one).
-+  // This routine forces the compiler (at least gcc) to push it anyway.
-+  StacktracePowerPCDummyFunction();
-+
-+#if IS_STACK_FRAMES
-+  // Note we do *not* increment skip_count here for the SYSV ABI.  If
-+  // we did, the list of stack frames wouldn't properly match up with
-+  // the list of return addresses.  Note this means the top pc entry
-+  // is probably bogus for linux/ppc (and other SYSV-ABI systems).
-+#else
-+  // The LR save area is used by the callee, so the top entry is bogus.
-+  skip_count++;
-+#endif
-+
-+  int n = 0;
-+  while (sp && n < max_depth) {
-+    // The GetStackFrames routine is called when we are in some
-+    // informational context (the failure signal handler for example).
-+    // Use the non-strict unwinding rules to produce a stack trace
-+    // that is as complete as possible (even if it contains a few
-+    // bogus entries in some rare cases).
-+    void **next_sp = NextStackFrame<!IS_STACK_FRAMES>(sp);
-+
-+    if (skip_count > 0) {
-+      skip_count--;
-+    } else {
-+      // PowerPC has 3 main ABIs, which say where in the stack the
-+      // Link Register is.  For DARWIN and AIX (used by apple and
-+      // linux ppc64), it's in sp[2].  For SYSV (used by linux ppc),
-+      // it's in sp[1].
-+#if defined(__PPC64__)
-+      // This check is in case the compiler doesn't define _CALL_AIX/etc.
-+      result[n] = *(sp+2);
-+#elif defined(__linux)
-+      // This check is in case the compiler doesn't define _CALL_SYSV.
-+      result[n] = *(sp+1);
-+#endif
-+
-+#if IS_STACK_FRAMES
-+      if (next_sp > sp) {
-+        sizes[n] = (uintptr_t)next_sp - (uintptr_t)sp;
-+      } else {
-+        // A frame-size of 0 is used to indicate unknown frame size.
-+        sizes[n] = 0;
-+      }
-+#endif
-+      n++;
-+    }
-+    sp = next_sp;
-+  }
-+  return n;
-+}
-diff --git a/src/stacktrace_powerpc-linux-inl.h b/src/stacktrace_powerpc-linux-inl.h
-new file mode 100644
-index 0000000..5d16fa1
---- /dev/null
-+++ b/src/stacktrace_powerpc-linux-inl.h
-@@ -0,0 +1,231 @@
-+// -*- Mode: C++; c-basic-offset: 2; indent-tabs-mode: nil -*-
-+// Copyright (c) 2007, Google Inc.
-+// All rights reserved.
-+//
-+// Redistribution and use in source and binary forms, with or without
-+// modification, are permitted provided that the following conditions are
-+// met:
-+//
-+//     * Redistributions of source code must retain the above copyright
-+// notice, this list of conditions and the following disclaimer.
-+//     * Redistributions in binary form must reproduce the above
-+// copyright notice, this list of conditions and the following disclaimer
-+// in the documentation and/or other materials provided with the
-+// distribution.
-+//     * Neither the name of Google Inc. nor the names of its
-+// contributors may be used to endorse or promote products derived from
-+// this software without specific prior written permission.
-+//
-+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
-+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-+
-+// ---
-+// Author: Craig Silverstein
-+//
-+// Produce stack trace.  ABI documentation reference can be found at:
-+// * PowerPC32 ABI: https://www.power.org/documentation/
-+// power-architecture-32-bit-abi-supplement-1-0-embeddedlinuxunified/
-+// * PowerPC64 ABI:
-+// http://www.linux-foundation.org/spec/ELF/ppc64/PPC-elf64abi-1.9.html#STACK
-+
-+#ifndef BASE_STACKTRACE_POWERPC_INL_H_
-+#define BASE_STACKTRACE_POWERPC_INL_H_
-+// Note: this file is included into stacktrace.cc more than once.
-+// Anything that should only be defined once should be here:
-+
-+#include <stdint.h>   // for uintptr_t
-+#include <stdlib.h>   // for NULL
-+#include <gperftools/stacktrace.h>
-+#include <base/vdso_support.h>
-+
-+#if defined(HAVE_SYS_UCONTEXT_H)
-+#include <sys/ucontext.h>
-+#elif defined(HAVE_UCONTEXT_H)
-+#include <ucontext.h>  // for ucontext_t
-+#endif
-+typedef ucontext ucontext_t;
-+
-+// PowerPC64 Little Endian follows BE wrt. backchain, condition register,
-+// and LR save area, so no need to adjust the reading struct.
-+struct layout_ppc {
-+  struct layout_ppc *next;
-+#ifdef __PPC64__
-+  long condition_register;
-+#endif
-+  void *return_addr;
-+};
-+
-+// Signal callbacks are handled by the vDSO symbol:
-+//
-+// * PowerPC64 Linux (arch/powerpc/kernel/vdso64/sigtramp.S):
-+//   __kernel_sigtramp_rt64
-+// * PowerPC32 Linux (arch/powerpc/kernel/vdso32/sigtramp.S):
-+//   __kernel_sigtramp32
-+//   __kernel_sigtramp_rt32
-+//
-+// So a backtrace may need to specially handling if the symbol readed is
-+// the signal trampoline.
-+
-+// Given a pointer to a stack frame, locate and return the calling
-+// stackframe, or return NULL if no stackframe can be found. Perform sanity
-+// checks (the strictness of which is controlled by the boolean parameter
-+// "STRICT_UNWINDING") to reduce the chance that a bad pointer is returned.
-+template<bool STRICT_UNWINDING>
-+static layout_ppc *NextStackFrame(layout_ppc *current) {
-+  uintptr_t old_sp = (uintptr_t)(current);
-+  uintptr_t new_sp = (uintptr_t)(current->next);
-+
-+  // Check that the transition from frame pointer old_sp to frame
-+  // pointer new_sp isn't clearly bogus
-+  if (STRICT_UNWINDING) {
-+    // With the stack growing downwards, older stack frame must be
-+    // at a greater address that the current one.
-+    if (new_sp <= old_sp)
-+      return NULL;
-+    // Assume stack frames larger than 100,000 bytes are bogus.
-+    if (new_sp - old_sp > 100000)
-+      return NULL;
-+  } else {
-+    // In the non-strict mode, allow discontiguous stack frames.
-+    // (alternate-signal-stacks for example).
-+    if (new_sp == old_sp)
-+      return NULL;
-+    // And allow frames upto about 1MB.
-+    if ((new_sp > old_sp) && (new_sp - old_sp > 1000000))
-+      return NULL;
-+  }
-+  if (new_sp & (sizeof(void *) - 1))
-+    return NULL;
-+  return current->next;
-+}
-+
-+// This ensures that GetStackTrace stes up the Link Register properly.
-+void StacktracePowerPCDummyFunction() __attribute__((noinline));
-+void StacktracePowerPCDummyFunction() { __asm__ volatile(""); }
-+#endif  // BASE_STACKTRACE_POWERPC_INL_H_
-+
-+// Note: this part of the file is included several times.
-+// Do not put globals below.
-+
-+// Load instruction used on top-of-stack get.
-+#if defined(__PPC64__) || defined(__LP64__)
-+# define LOAD "ld"
-+#else
-+# define LOAD "lwz"
-+#endif
-+
-+// The following 4 functions are generated from the code below:
-+//   GetStack{Trace,Frames}()
-+//   GetStack{Trace,Frames}WithContext()
-+//
-+// These functions take the following args:
-+//   void** result: the stack-trace, as an array
-+//   int* sizes: the size of each stack frame, as an array
-+//               (GetStackFrames* only)
-+//   int max_depth: the size of the result (and sizes) array(s)
-+//   int skip_count: how many stack pointers to skip before storing in result
-+//   void* ucp: a ucontext_t* (GetStack{Trace,Frames}WithContext only)
-+static int GET_STACK_TRACE_OR_FRAMES {
-+  layout_ppc *current;
-+  int n;
-+
-+  // Get the address on top-of-stack
-+  current = reinterpret_cast<layout_ppc*> (__builtin_frame_address (0));
-+  // And ignore the current symbol
-+  current = current->next;
-+
-+  StacktracePowerPCDummyFunction();
-+
-+  n = 0;
-+  skip_count++; // skip parent's frame due to indirection in
-+                // stacktrace.cc
-+
-+  base::VDSOSupport vdso;
-+  base::ElfMemImage::SymbolInfo rt_sigreturn_symbol_info;
-+#ifdef __PPC64__
-+  const void *sigtramp64_vdso = 0;
-+  if (vdso.LookupSymbol("__kernel_sigtramp_rt64", "LINUX_2.6.15", STT_NOTYPE,
-+                        &rt_sigreturn_symbol_info))
-+    sigtramp64_vdso = rt_sigreturn_symbol_info.address;
-+#else
-+  const void *sigtramp32_vdso = 0;
-+  if (vdso.LookupSymbol("__kernel_sigtramp32", "LINUX_2.6.15", STT_NOTYPE,
-+                        &rt_sigreturn_symbol_info))
-+    sigtramp32_vdso = rt_sigreturn_symbol_info.address;
-+  const void *sigtramp32_rt_vdso = 0;
-+  if (vdso.LookupSymbol("__kernel_sigtramp_rt32", "LINUX_2.6.15", STT_NOTYPE,
-+                        &rt_sigreturn_symbol_info))
-+    sigtramp32_rt_vdso = rt_sigreturn_symbol_info.address;
-+#endif
-+
-+  while (current && n < max_depth) {
-+
-+    // The GetStackFrames routine is called when we are in some
-+    // informational context (the failure signal handler for example).
-+    // Use the non-strict unwinding rules to produce a stack trace
-+    // that is as complete as possible (even if it contains a few
-+    // bogus entries in some rare cases).
-+    layout_ppc *next = NextStackFrame<!IS_STACK_FRAMES>(current);
-+    if (skip_count > 0) {
-+      skip_count--;
-+    } else {
-+      result[n] = current->return_addr;
-+#ifdef __PPC64__
-+      if (sigtramp64_vdso && (sigtramp64_vdso == current->return_addr)) {
-+        struct signal_frame_64 {
-+          char dummy[128];
-+          ucontext_t uc;
-+        // We don't care about the rest, since the IP value is at 'uc' field.
-+        } *sigframe = reinterpret_cast<signal_frame_64*>(current);
-+        result[n] = (void*) sigframe->uc.uc_mcontext.gp_regs[PT_NIP];
-+      }
-+#else
-+      if (sigtramp32_vdso && (sigtramp32_vdso == current->return_addr)) {
-+        struct signal_frame_32 {
-+          char dummy[64];
-+          struct sigcontext sctx;
-+          mcontext_t mctx;
-+          // We don't care about the rest, since IP value is at 'mctx' field.
-+        } *sigframe = reinterpret_cast<signal_frame_32*>(current);
-+        result[n] = (void*) sigframe->mctx.gregs[PT_NIP];
-+      } else if (sigtramp32_rt_vdso && (sigtramp32_rt_vdso == current->return_addr)) {
-+        struct rt_signal_frame_32 {
-+          char dummy[64 + 16];
-+          siginfo_t info;
-+          struct ucontext uc;
-+          // We don't care about the rest, since IP value is at 'uc' field.A
-+        } *sigframe = reinterpret_cast<rt_signal_frame_32*>(current);
-+        result[n] = (void*) sigframe->uc.uc_mcontext.uc_regs->gregs[PT_NIP];
-+      }
-+#endif
-+
-+#if IS_STACK_FRAMES
-+      if (next > current) {
-+        sizes[n] = (uintptr_t)next - (uintptr_t)current;
-+      } else {
-+        // A frame-size of 0 is used to indicate unknown frame size.
-+        sizes[n] = 0;
-+      }
-+#endif
-+      n++;
-+    }
-+    current = next;
-+  }
-+
-+  // It's possible the second-last stack frame can't return
-+  // (that is, it's __libc_start_main), in which case
-+  // the CRT startup code will have set its LR to 'NULL'.
-+  if (n > 0 && result[n-1] == NULL)
-+    n--;
-+
-+  return n;
-+}

OBS-URL: https://build.opensuse.org/request/show/280951
OBS-URL: https://build.opensuse.org/package/show/devel:libraries:c_c++/gperftools?expand=0&rev=27
2015-01-13 07:30:23 +00:00

260 lines
12 KiB
Plaintext
Raw Blame History

-------------------------------------------------------------------
Mon Jan 10 16:30:24 UTC 2015 - cgardner@suse.com
- gperftools 2.4, available 10 Jan 2015.
Skipped version 2.3, release 7 Dec 2014. 2.3 changes included:
* ( issue 631 ) fixed debugallocation miscompilation on mmap-less
platforms (courtesy of user iamxujian)
* ( issue 630 ) reference to wrong PROFILE (vs. correct CPUPROFILE)
environment variable was fixed (courtesy of WenSheng He)
* pprof now has option to display stack traces in output for heap
checker (courtesy of Michael Pasieka)
* ( issue 636 ) pprof web command now works on mingw
* ( issue 635 ) pprof now handles library paths that contain spaces
(courtesy of user mich...@sebesbefut.com)
* ( issue 637 ) pprof now has an option to not strip template
arguments (patch by jiakai)
* ( issue 644 ) possible out-of-bounds access in GetenvBeforeMain
was fixed (thanks to user abyss.7)
* ( issue 641 ) pprof now has an option --show_addresses (thanks to
user yurivict). New option prints instruction address in addition
to function name in stack traces
* ( issue 646 ) pprof now works around some issues of addr2line
reportedly when DWARF v4 format is used (patch by Adam McNeeney)
* ( issue 645 ) heap profiler exit message now includes remaining
memory allocated info (patch by user yurivict)
* pprof code that finds location of /proc/pid/maps in cpu profile
files is now fixed (patch by Ricardo M. Correia)
* ( issue 654 ) pprof now handles "split text segments" feature of
Chromium for Android (patch by simonb)
* ( issue 655 ) potential deadlock on windows caused by early call
to getenv in malloc initialization code was fixed (bug reported
and fix proposed by user zndmitry)
* incorrect detection of arm 6zk instruction set support
(-mcpu=arm1176jzf-s) was fixed. (Reported by pedronavf on old issue-493)
* new cpu profiling mode on Linux is now implemented. It sets up separate
profiling timers for separate threads. Which improves accuracy of
profiling on Linux a lot. It is off by default. And is enabled if both
librt.f is loaded and CPUPROFILE_PER_THREAD_TIMERS environment
variable is set. But note that all threads need to be registered
via ProfilerRegisterThread.
Changes for version 2.4~rc
* enabled aggressive decommit option by default. It was found to
significantly improve memory fragmentation with negligible impact
on performance. (Thanks to investigation work performed by Adhemerval
Zanella)
* added ./configure flags for tcmalloc pagesize and tcmalloc allocation
alignment. Larger page sizes have been reported to improve performance
occasionally. (Patch by Raphael Moreira Zinsly)
* sped-up hot-path of malloc/free. By about 5% on static library and
about 10% on shared library. Mainly due to more efficient checking
of malloc hooks.
* improved accuracy of stacktrace capturing in cpu profiler (due to
issue found by Arun Sharma). As part of that issue pprof's handling
of cpu profiles was also improved.
- Removed missing_includes.patch; incorporated upstream.
-------------------------------------------------------------------
Sun Sep 28 18:44:03 UTC 2014 - jengelh@inai.de
- Implement shlib packaging guidelines: split libraries
- Improve on RPM group classification
-------------------------------------------------------------------
Mon Sep 15 00:33:02 CEST 2014 - ro@suse.de
- set excludearch for s390 and s390x: not ported there
-------------------------------------------------------------------
Mon Jun 16 10:51:46 UTC 2014 - dvaleev@suse.com
- fix issue 628 (gperftools 2.2 fails to build on powerpc)
missing_includes.patch
-------------------------------------------------------------------
Mon May 19 15:39:40 UTC 2014 - cgardner@suse.com
- gperftools 2.2, released 3 May 2014.
* issue 620 (crash on windows when c runtime dll is reloaded) was
fixed
* a number of fixes for a number compilers and platforms. Notably
Visual Studio 2013, recent mingw with c++ threads and some OSX
fixes.
* we now have mips and mips64 support! (courtesy of Jovan Zelincevic,
Jean Lee, user xiaoyur347 and others)
* we now have aarch64 (aka arm64) support! (contributed by Riku
Voipio)
* there's now support for ppc64-le (by Raphael Moreira Zinsly and
Adhemerval Zanella)
* there's now some support of uclibc (contributed by user xiaoyur347)
* google/ headers will now give you deprecation warning. They are
deprecated since 2.0
* there's now new api: tc_malloc_skip_new_handler (ported from chromium
fork)
* issue-557: added support for dumping heap profile via signal (by
Jean Lee)
* issue-567: Petr Hosek contributed SysAllocator support for windows
* Joonsoo Kim contributed several speedups for central freelist code
* TCMALLOC_MAX_TOTAL_THREAD_CACHE_BYTES environment variable now works
* configure scripts are now using AM_MAINTAINER_MODE. It'll only
affect folks who modify source from .tar.gz and want automake to
automatically rebuild Makefile-s. See automake documentation for
that.
* issue-586: detect main executable even if PIE is active (based on
patch by user themastermind1). Notably, it fixes profiler use with
ruby.
* there is now support for switching backtrace capturing method at
runtime (via TCMALLOC_STACKTRACE_METHOD and
TCMALLOC_STACKTRACE_METHOD_VERBOSE environment variables)
* there is new backtrace capturing method using -finstrument-functions
prologues contributed by user xiaoyur347
* few cases of crashes/deadlocks in profiler were addressed. See
(famous) issue-66, issue-547 and issue-579.
* issue-464 (memory corruption in debugalloc's realloc after
memallign) is now fixed
* tcmalloc is now able to release memory back to OS on windows
(issue-489). The code was ported from chromium fork (by a number of
authors).
* Together with issue-489 we ported chromium's "aggressive decommit"
mode. In this mode (settable via malloc extension and via
environment variable TCMALLOC_AGGRESSIVE_DECOMMIT), free pages are
returned back to OS immediately.
* MallocExtension::instance() is now faster (based on patch by
Adhemerval Zanella)
* issue-610 (hangs on windows in multibyte locales) is now fixed
- Removed the following patches. All have either been incorporated
upstream or are no longer applicable in current distros.
* gperftools_fix_multiple_install_headers.patch
* gperftools-glibc216.patch
* spinlock-rand.patch
* gperftools_use_older_autoconf.diff
-------------------------------------------------------------------
Mon Aug 12 15:49:12 UTC 2013 - cgardner@suse.com
- Fixed configure.ac to use older version of autoconf, so that
gperftools can build for SLE10.
-------------------------------------------------------------------
Tue Jul 30 17:44:01 UTC 2013 - cgardner@suse.com
- gperftools 2.1, released 30 July 2013.
* fixes for building on newer platforms. Notably, there's now initial
support for x32 ABI (--enable-minimal only at this time))
* Some fixes for debug allocation on POWER/Linux
* new getNumericProperty stats for cache sizes
* added HEAP_PROFILER_TIME_INTERVAL variable (see documentation)
* added environment variable to control heap size (TCMALLOC_HEAP_LIMIT_MB)
* added environment variable to disable release of memory back to OS
(TCMALLOC_DISABLE_MEMORY_RELEASE)
* cpu profiler can now be switched on and off by sending it a signal
(specified in CPUPROFILESIGNAL)
* (issue 491) fixed race-ful spinlock wake-ups
* (issue 496) added some support for fork-ing of process that is using
tcmalloc
* (issue 368) improved memory fragmentation when large chunks of
memory are allocated/freed
-------------------------------------------------------------------
Thu Dec 27 11:08:09 UTC 2012 - dvaleev@suse.com
- spinlock-rand.patch fixes an issue encountered on PPC with
undefined conversions from 'base::subtle::Atomic64*' to 'const
volatile Atomic32*'
-------------------------------------------------------------------
Sun Jul 22 07:47:04 UTC 2012 - aj@suse.de
- replace struct siginfo with siginfo_t since the former is not
available with glibc 2.16 anymore.
-------------------------------------------------------------------
Mon May 14 13:25:33 UTC 2012 - cfarrell@suse.com
- license update: BSD-3-Clause
SPDX format
-------------------------------------------------------------------
Fri Feb 17 11:23:45 CET 2012 - hmacht@suse.de
- enable build with libunwind
-------------------------------------------------------------------
Thu Feb 16 22:17:11 UTC 2012 - jengelh@medozas.de
- Remove redundant tags/sections from specfile
- Simply file lists (why would mode 0775 be needed anyway?)
-------------------------------------------------------------------
Thu Feb 16 15:37:36 UTC 2012 - cgardner@suse.com
- Fix make target.
-------------------------------------------------------------------
Thu Feb 16 11:28:14 UTC 2012 - coolo@suse.com
- Remove redudant tags/sections
-------------------------------------------------------------------
Tue Feb 7 18:12:39 UTC 2012 - cgardner@suse.com
- gperftools 2.0, released 3 Feb 2012.
- Primarily renamed from google-perftools to gperftools, with ownership
shifted from Google to the community.
-------------------------------------------------------------------
Sat Dec 24 19:05:13 UTC 2011 - cgardner@suse.com
- Version 1.9.1, released 23 Dec 2011.
* google-perftools: version 1.9 release
* Lightweight check for double-frees (blount)
* BUGFIX: Fix pprof to exit properly if run with no args (dagitses)
* Suggest ASan as a way to diagnose buggy code (ppluzhnikov)
* Get rid of unused CACHELINE_SIZE (csilvers)
* Replace atexit() calls with global dtors; helps freebsd (csilvers)
* Disable heap-checker under AddressSanitizer (kcc)
* Fix bug in powerpc stacktracing (ppluzhnikov)
* PERF: Use exponential backoff waiting for spinlocks (m3b)
* Fix 64-bit nm on 32-bit binaries in pprof (csilvers)
* Add ProfileHandlerDisallowForever (rsc)
* BUGFIX: Shell escape when forking in pprof (csilvers)
* No longer combine overloaded functions in pprof (csilvers)
* Fix address-normalizing bug in pprof (csilvers)
* More consistently call abort() instead of exit() on failure (csilvers)
* Allow NoGlobalLeaks to be safely called more than once (csilvers)
* PORTING/BUGFIX: Fix ARM cycleclock to use volatile asm (dougkwan)
* PORTING: 64-bit atomic ops for ARMv7 (dougkwan)
* PORTING: Implement stacktrace for ARM (dougkwan)
* PORTING: Fix malloc_hook_mmap_linux for ARM (dougkwan)
* PORTING: Update linux_syscall_support.h for ARM/etc (evannier, sanek)
* PORTING: Fix freebsd to work on x86_64 (chapp...@gmail.com)
* PORTING: Added additional SYS_mmap fixes for FreeBSD (chappedm)
* PORTING: Allow us to compile on OS X 10.6 and run on 10.5 (raltherr)
* PORTING: Check for mingw compilers that *do* define timespec
* PORTING: Add "support" for MIPS cycletimer
* PORTING: Fix fallback cycle-timer to work with Now (dougkwan)
* PERF: Move stack trace collecting out of the mutex (taylorc)
* PERF: Get the deallocation stack trace outside the mutex (sean)
* Make PageHeap dynamically allocated for leak checks (maxim)
* BUGFIX: Fix probing of nm -f behavior in pprof (dpeng)
* BUGFIX: Fix a race with the CentralFreeList lock before main (sanjay)
* Support /pprof/censusprofile url arguments (rajatjain)
* Change IgnoreObject to return its argument (nlewycky)
* Update malloc-hook files to support more CPUs
* BUGFIX: write our own strstr to avoid libc problems (csilvers)
* Use simple callgrind compression facility in pprof
* Print an error message when we can't run pprof to symbolize (csilvers)
* Die in configure when g++ is't installed (csilvers)
* DOC: Beef up the documentation a bit about using libunwind (csilvers)
-------------------------------------------------------------------
Sat Oct 15 04:47:04 UTC 2011 - coolo@suse.com
- add libtool as buildrequire to make the spec file more reliable
-------------------------------------------------------------------
Sun Jul 17 21:59:38 UTC 2011 - cgardner@suse.com
- Google perftools 1.8, released 15 July 2011.
Reference in New Issue View Git Blame Copy Permalink