Diff of /ChangeLog [a117f8] .. [092ee5]  Maximize  Restore

Switch to side-by-side view

--- a/ChangeLog
+++ b/ChangeLog
@@ -1,1034 +1,6 @@
-2004-12-25  Will Cohen  <wcohen@redhat.com>
+2005-01-02  Philippe Elie  <phil.el@wanadoo.fr>
 
-	* events/Makefile.am: Corrected type for events_files.
+	* pp/opannotate.cpp: fix #1093162 by simplifying the code, this was
+	  likely to be a compiler problem but the new code is simpler.
 
-2004-12-12  John Levon  <levon@movementarian.org>
-
-	* doc/oprofile.xml: expand list of supported CPU types.
-
-2004-12-12  Ralf Baechle  <ralf@linux-mips.org>
-
-	* daemon/opd_cookie.c: Define __NR_lookup_dcookie for all three
-	  supported MIPS ABIs.
-
-	* events/mips/r10000/events:
-	* events/mips/r10000/unit_masks:
-	* events/mips/r12000/events:
-	* events/mips/r12000/unit_masks:
-	* events/mips/r5432/events:
-	* events/mips/r5432/unit_masks:
-	* events/mips/r5500/events:
-	* events/mips/r5500/unit_masks:
-	* events/mips/rm7000/events:
-	* events/mips/rm7000/unit_masks:
-	* events/mips/rm9000/events:
-	* events/mips/rm9000/unit_masks:
-	* events/mips/sb1/events:
-	* events/mips/sb1/unit_masks:
-	* events/Makefile.am:
-	* libop/op_events.c:
-	* utils/op_help.c:
-	* libop/op_cpu_type.h:
-	* libop/op_cpu_type.c: Add support for the MIPS R10000 and R12000,
-	  QED RM7000 and PMC-Sierra RM9000, NEC VR5432 and VR5500 and Sibyte SB1
-	   processors.
-
-	* README: List myself as the caretaker of the MIPS bits.
-
-2004-11-25  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libop/op_events.c: simplify a bit get_mapping()
-
-2004-11-20  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/opd_stats.c: some statistics are useless, don't print them
-	  if we can't read the oprofilefs files.
-
-2004-11-20  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/opd_sfile.c: sfile_clear_kenel() : clear also callgraph
-	  file depending on module.
-
-2004-11-19  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/opd_sfile.h:
-	* daemon/opd_sfile.c: fix #971487, daemon was unable to retrieve
-	  correctly callgraph sample file for kernel space samples. Fixed by
-	  adding relevant information to cg_hash_entry.
-
-	* libpp/callgraph_container.cpp: more verbosity
-
-2004-11-19  John Levon  <levon@movementarian.org>
-
-	* doc/oprofile.xml: Fix AMD link (noticed by Emmanuel Araman).
-
-2004-11-18  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* pp/opgprof.cpp: ensure we load all callgraph file
-
-2004-11-18  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libutil++/locate_images.cpp: ensure error is always set,
-	  callgraph_container was using an uninitialized value.
-
-2004-11-17  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libutil++/op_bfd.cpp: improve a bit the symbol chosen when
-	  multiple symbol live at same vma
-
-2004-11-17  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* doc/oprofile.xml: some words about UP kernel > 2.6.9 and the need of
-	  lapic at boot.
-
-2004-11-17  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* pp/common_option.h:
-	* pp/common_option.cpp: move threshold option from here ...
-	* pp/opannotate_options.cpp:
-	* pp/opgprof_options.cpp:
-	* pp/opreport_options.cpp:
-	* pp/opstack_options.cpp: to each of these file using it. Side effect
-	  is to remove this option from oparchive which didn't handle it.
-	
-	* doc/oparchive.1.in:
-	* doc/oprofile.xml: there isn't any --threshold option to oparchive
-
-2004-11-07  Kristis Makris  <kristis.makris@asu.edu>
-
-	* doc/internals.xml: fix typo
-
-2004-11-01  Maynard Johnson  <maynardj@us.ibm.com>
-
-	* events/ppc64/970/events: Make min count 10000 for all PM_INST_CMPL;
-	  and remove redundant event in group 2
-	* events/ppc64/power4/event_mappings: Set LSB of MMCRA so kernel kludge
-	  can be removed at some later date
-	* events/ppc64/power4/events: Make min counts consistent with ppc970
-	* events/ppc64/power5/events: Make min counts consistent with ppc970	  
-
-2004-11-01  Greg Banks  <gnb@melbourne.sgi.com>
-
-	* libop/op_events.c: Make the default sample rate for
-	  IA64 sensible (i.e. 10 times smaller).
-
-2004-10-15  Will Cohen  <wcohen@redhat.com>
-
-	* events/ppc64/power4/events: Corrected group 4 counter assignments.
-
-2004-10-15  Maynard Johnson  <maynardj@us.ibm.com>
-
-	* events/Makefile.am:
-	* events/ppc64/970/event_mappings:
-	* events/ppc64/970/events:
-	* events/ppc64/970/unit_masks:
-	* libop/op_cpu_type.c:
-	* libop/op_cpu_type.h:
-	* libop/op_events.c:
-	* utils/op_help.c:
-	* utils/opcontrol: Added ppc64/970 support.
-
-2004-10-14  Greg Banks  <gnb@melbourne.sgi.com>
-
-	* configure.in:
-	* daemon/opd_perfmon.c: autoconf for sched_setaffinity
-	  and perfmonctl, fix sched_setaffinity call for early
-	  2.6 kernels
-
-2004-10-07  Will Cohen  <wcohen@redhat.com>
-
-	* utils/opcontrol: Correct ppc64 check that events are in same group.
-
-2004-09-21  Will Cohen  <wcohen@redhat.com>
-
-	* libutil++/op_bfd.h:
-	* libutil++/op_bfd.cpp: Add logic to use the preferred symbol name.
-
-2004-09-17  John Levon  <levon@movementarian.org>
-
-	* utils/op_help.c: don't deref a NULL pointer
-	  on failure to read /dev/oprofile/cpu_type
-
-2004-09-17  Anton Blanchard  <anton@samba.org>
-
-	* libop/op_events.c: fix compile warnings
-
-2004-09-16  John Levon  <levon@movementarian.org>
-
-	* m4/configmodule.m4: disable x86-64 on 2.4
-
-2004-09-15  Maynard P. Johnson  <maynardj@us.ibm.com>
-
-	* events/Makefile.am:
-	* libop/op_cpu_type.h:
-	* libop/op_events.h:
-	* libop/op_events.c:
-	* utils/op_help.c:
-	* utils/opcontrol: PPC64 support
-	
-2004-09-13  John Levon  <levon@movementarian.org>
-
-	* configure.in: bump to 0.8.2cvs
-
-2004-09-11  John Levon  <levon@movementarian.org>
-
-	* configure.in: bump to 0.8.1
-
-2004-09-11  John Levon  <levon@movementarian.org>
-
-	* libpp/locate_images.h: doxygen for archive_path
-
-2004-08-18  Will Cohen  <wcohen@redhat.com>
-
-	* doc/oprofile.xml: List default events in documentation.
-
-2004-08-12  John Levon  <levon@movementarian.org>
-
-	* libutil++/op_bfd.cpp: we need to verify
-	functionname is non-NULL in *both* places
-
-2004-08-12  John Levon  <levon@movementarian.org>
-
-	* libutil++/op_bfd.cpp: add a hack to support
-	Objective C
-
-2004-07-21  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libdb/db_manage.c: thinko in #931871 bug fix.
-
-2004-07-21  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/arrange_profiles.cpp: fix opreport -m lib, broken in 0.8
-	  when using opcontrol --separate=kernel with message
-	  "add_to_profile_sample_files(): sample file parsed twice"
-
-2004-07-18  John Levon  <levon@movementarian.org>
-
-	* libop/op_mangling.c: dep image always used
-	'{root}' token
-
-2004-07-15  John Levon  <levon@movementarian.org>
-
-	* doc/internals.xml: document some more
-
-2004-07-09  John Levon  <levon@movementarian.org>
-
-	* libpp/populate.cpp: quash timestamp warnings
-	 when using archive: profile specifier
-
-2004-07-09  John Levon  <levon@movementarian.org>
-
-	* pp/opstack_options.cpp: include call-graph
-	  files in archive.
-
-2004-07-06  John Levon  <levon@movementarian.org>
-
-	* doc/oprofile.1.in:
-	* doc/oprofile.xml:
-	* libpp/profile_spec.h:
-	* libpp/profile_spec.cpp: remove sample-file
-	  and binary profile specifiers
-
-	* doc/opcontrol.1.in:
-	* doc/oprofile.xml:
-	* pp/common_option.cpp:
-	* utils/opcontrol: allow either "lib" or "library"
-	 for merge/separate option. Document "lib"
-
-2004-07-06  Will Cohen <wcohen@redhat.com>
-
-	* doc/oprofile.xml:
-	* doc/Makefile.am:
-	* doc/oparchive.1.in:
-	* doc/oprofile.1.in:
-	* libpp/profile_spec.cpp:
-	* libpp/profile_spec.h:
-	* libpp/populate.h:
-	* libpp/populate.cpp:
-	* libpp/callgraph_container.cpp:
-	* libpp/callgraph_container.h:
-	* libpp/arrange_profiles.cpp:
-	* libpp/arrange_profiles.h:
-	* libpp/locate_images.cpp:
-	* libpp/locate_images.h:
-	* libutil++/op_bfd.h:
-	* libutil++/op_bfd.cpp:
-	* libutil++/file_manip.cpp:
-	* libutil++/file_manip.h:
-	* pp/Makefile.am:
-	* pp/oparchive.cpp:
-	* pp/oparchive_options.cpp:
-	* pp/oparchive_options.h:
-	* pp/opreport.cpp:
-	* pp/opreport_options.h:
-	* pp/opreport_options.cpp:
-	* pp/opstack.cpp:
-	* pp/opstack_options.h:
-	* pp/opstack_options.cpp:
-	* pp/opannotate_options.cpp:
-	* pp/opannotate_options.h:
-	* pp/opgprof.cpp:
-	* pp/opgprof_options.cpp:
-	* pp/opgprof_options.h:
-	* pp/opannotate.cpp:
-	* configure.in: Support for oparchive.
-
-2004-07-05  John Levon  <levon@movementarian.org>
-
-	* doc/oprofile.1.in: add opstack(1) to SEE ALSO
-
-2004-06-21  John Levon  <levon@movementarian.org>
-
-	* events/i386/p4/events:
-	* events/i386/p4-ht/events:
-	* module/x86/op_model_athlon.c:
-	* module/x86/op_model_p4.c:
-	* module/x86/op_model_ppro.c:
-	* module/ia64/op_pmu.c: fix bug 964097 (event
-	code of 0x00 doesn't work) by relying consistently
-	on "enabled" not "event"
-
-2004-05-29  John Levon  <levon@movementarian.org>
-
-	* libdb/tests/Makefile.am: fix build
-
-	* daemon/opd_mangling.c:
-	* daemon/opd_mangling.h:
-	* daemon/opd_sfile.c:
-	* daemon/opd_sfile.h:
-	* daemon/liblegacy/opd_sample_files.h:
-	* libabi/abi.cpp:
-	* libabi/abi_test.cpp:
-	* libabi/op_import.cpp:
-	* libdb/Makefile.am:
-	* libdb/db_debug.c:
-	* libdb/db_insert.c:
-	* libdb/db_manage.c:
-	* libdb/db_stat.c:
-	* libdb/db_travel.c:
-	* libdb/odb.h:
-	* libdb/odb_hash.h:
-	* libdb/tests/db_test.c:
-	* libpp/op_header.cpp:
-	* libpp/profile.cpp:
-	* libpp/profile.h: s/samples_odb_t/odb_t/, move
-	odb_hash.h to odb.h
-
-2004-05-28  John Levon  <levon@movementarian.org>
-
-	* daemon/opd_mangling.cpp:
-	* daemon/opd_sfile.cpp:
-	* daemon/liblegacy/opd_proc.c:
-	* daemon/liblegacy/opd_sample_files.c:
-	* libabi/abi_test.cpp:
-	* libabi/op_import.cpp:
-	* libdb/db_manage.cpp:
-	* libdb/odb_hash.h:
-	* libpp/profile.cpp: introduce and use odb_get_data(),
-	odb_open_count()
-
-2004-05-28  John Levon  <levon@movementarian.org>
-
-	* libpp/callgraph_container.cpp:
-	* libpp/arrange_profiles.cpp: trivialities
-
-2004-05-28  Bin Ren  <bin.ren@cl.cam.ac.uk>
-
-	* daemon/opd_mangling.c:
-	* daemon/opd_sfile.c:
-	* daemon/oprofiled.c:
-	* daemon/liblegacy/opd_image.c:
-	* daemon/liblegacy/opd_proc.c:
-	* daemon/liblegacy/opd_sample_files.c:
-	* libdb/db_debug.c:
-	* libdb/db_insert.c:
-	* libdb/db_manage.c:
-	* libdb/db_stat.c:
-	* libdb/odb_hash.h:
-	* libdb/db_travel.c:
-	* libpp/op_header.cpp:
-	* libpp/profile.cpp:
-	* libutil/op_string.c:
-	* libutil/op_string.h: fix dcookie alias bug by
-	  introducing one level of indirection and hash
-	  sample files by path names (bug #931871)
-
-2004-05-11  John Levon  <levon@movementarian.org>
-
-	* libutil++/file_manip.h: clarify behaviour of
-	  op_realpath()
-
-	* libpp/opannotate.cpp: add some checks to avoid
-	  creating files outside of the output directory
-
-2004-05-11  Andi Kleen  <ak@suse.de>
-
-	* libpp/callgraph_container.cpp:
-	* libpp/profile_container.cpp:
-	* libutil++/op_bfd.h:
-	* libutil++/op_bfd.cpp: use unsigned long in various
-	  places - u32 was breaking 64-bit usage
-
-2004-05-11  John Levon  <levon@movementarian.org>
-
-	* configure.in: bump to 0.8.1cvs
-
-2004-05-07  John Levon  <levon@movementarian.org>
-
-	* configure.in: bump to 0.8
-
-2004-05-07  John Levon  <levon@movementarian.org>
-
-	* libpp/filename_spec.h:
-	* libpp/filename_spec.cpp:
-	* libpp/profile_spec.cpp: "dependant" not "dependent"
-
-2004-04-19  Will Cohen  <wcohen@redhat.com>
-
-	* events/i386/p4/unit_masks:
-	* events/i386/p4-ht/unit_masks: Correct machine_clear unit mask.
-
-2004-04-04  Will Cohen  <wcohen@redhat.com>
-
-	* m4/builtinexpect.m4:
-	* m4/compileroption.m4:
-	* m4/configmodule.m4:
-	* m4/copyifchange.m4:
-	* m4/docbook.m4:
-	* m4/extradirs.m4:
-	* m4/findkernel.m4:
-	* m4/kerneloption.m4:
-	* m4/kernelversion.m4:
-	* m4/mallocattribute.m4:
-	* m4/poptconst.m4:
-	* m4/precompiledheader.m4:
-	* m4/qt.m4:
-	* m4/resultyn.m4:
-	* m4/sstream.m4:
-	* m4/typedef.m4: Add quotes as required for automake 1.8. 
-
-2004-04-04  John Levon  <levon@movementarian.org>
-
-	* doc/oprofile.xml: improvements to the call-graph
-	* documentation.
-
-2004-03-28  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libutil++/op_bfd.cpp: remove spurious cerr <<; fix a potential memory
-	  leak
-
-2004-03-28  John Levon  <levon@movementarian.org>
-
-	* libutil++/op_bfd.cpp: fix a BFD leak on format failure. Fix a
-	  small race when the binary changes, and we error fatally
-	  instead of recovering.
-
-2004-03-22  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* module/x86/hammer_op_syscalls.c: use the first map with VM_EXECUTABLE
-	  bit to do the exec notification, this fix a 32 bits application
-	  profiling regression. We dunno exactly what is this second mmap ...
-	  but we must ignore one of these. Fix #921243
-
-2004-03-22  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* module/ia64/op_pmu.c: the way to set pfm_syst_wide and
-	  pfm_dcr_pp changed between 2.4.19 and 2.4.20
-
-2004-03-22  Will Cohen  <wcohen@redhat.com>
-
-	* utils/opcontrol: Correct order of do_init_daemon_vars and
-	decide_orofile_devices.
-	
-2004-03-22  Will Cohen  <wcohen@redhat.com>
-
-	* utils/opcontrol: Tweak error messages and factor out common code.
-
-2004-03-19  Anton Blanchard  <anton@samba.org>
-
-	* daemon/opd_sfile.c: Add function name to some error messages
-
-2004-03-04  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* utils/opcontrol: tweak a bit error message.
-
-2004-03-03  Will Cohen  <wcohen@redhat.com>
-
-	* utils/opcontrol: Support --dump for normal users.
-
-2004-02-29  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/callgraph_container.cpp: make g++ -D_GLIBCXX_DEBUG happy, see
-	  http://gcc.gnu.org/PR14340
-
-	* module/op_dcache.h:
-	* module/op_dname.c: some gcc version are confused by extern
-	  declaration followed by later static.
-
-2004-02-28  Zwane Mwaikambo  <zwane@linuxpower.ca>
-
-	* libutil++/op_bfd.cpp:
-	* utils/opcontrol: This patch fixes the listing of symbols within the
-	  kernel image on ARM i was having. The problem was due to the fact
-	  that ARM/Linux kernel text starts with a section called .init unlike
-	  all the currently supported Linux architectures. So the solution is
-	  to start at section 0
-
-2004-02-28  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* module/x86/op_syscalls.c: warning fix
-
-2004-02-23  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/Makefile.am:
-	* daemon/liblegacy/Makefile.am: force daemon build with frame pointer.
-	  If daemon is built w/o frame pointer we get random arc, some of them
-	  go in the vma used by an existing mmaped sample file, so cookie
-	  lookup return a sample filename which is used as a from or to
-	  component, this break pptools by creating invalid filename.  Note
-	  this doesn't protect agaisnt a make CFLAGS=-fomit-frame-pointer.
-
-	* libpp/callgraph_container.cpp: use the right {cg_}image_name for
-	  verbose output.
-
-2004-02-21  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/format_flags.h:
-	* libpp/format_output.cpp:
-	* libpp/format_output.h:
-	* pp/opstack.cpp: alternate output format.
-
-2004-02-20  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* doc/srcdoc/Doxyfile.in: Fix for doxygen 1.3.6
-	* libpp/profile_spec.h:
-	* libutil++/op_bfd.h: minor Doxygen comments fix
-
-2004-02-20  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libutil++/tests/string_manip_tests.cpp: fix a $ make check failure
-	  from the last format_percent() change
-
-2004-02-18  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/format_output.cpp: change output spacing.
-
-2004-02-17  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* pp/opstack_options.h:
-	* pp/opstack.cpp: pass mege_options::lib to callgraph_container
-	* libpp/callgraph_container.cpp:
-	* libpp/callgraph_container.h: handle merge_options::lib
-
-2004-02-17  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/format_output.cpp: use "self/child" for header
-	* libutil++/string_manip.cpp: format_percent() return "0" if
-	  value == 0.0
-	* libpp/callgraph_container.cpp: remove some dead code.
-
-2004-02-16  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* module/compat24.h:
-	* module/x86/cpu_type.c:
-	* module/x86/op_apic.c:
-	* module/x86/op_model_p4.c:
-	* module/x86/op_nmi.c:
-	* module/x86/op_x86_model.h: backport P4 HyperThreading support from
-	   2.6 to 2.4.
-
-2004-02-16  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* events/i386/p4/events:
-	* events/i386/p4/unit_masks:
-	* events/i386/p4-ht/events:
-	* events/i386/p4-ht/unit_masks: fix MEMORY_COMPLETE unit mask
-
-2004-02-15  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/profile.cpp: samples_range() throw if start > end
-
-2004-02-15  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* events/ia64/itanium2/unit_masks: comments
-	* utils/op_help.c: add --callgraph to sanitize min events count
-	* utils/opcontrol: pass --callgraph to op_help
-
-2004-02-15  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/oprofiled.c: send alarm() after fork not before, behavior
-	  changed between 2.4 and 2.6 and according to posix 2.6 is right.
-
-2004-02-15  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* utils/op_help.c: output the default unit mask. when querying the
-	  default mask output it in decimal not hexadecimal
-	* libpp/op_header.cpp: event number must be output as integer not as
-	  unsigned char type
-	* libop/op_parse_event.c: clarify error message
-
-	* libop/op_events.c: better validation of unit mask
-	* events/ia64/itanium2/unit_masks:
-	* events/x86-64/hammer/unit_masks: fix so make check pass, done through
-	  documentation reading.
-
-
-2004-02-13  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* events/i386/p4-ht/events: multiply all minimum count by two
-
-2004-02-12  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/opd_stats.c: handle new statistics added in callgraph patch
-
-2004-02-11  Will Cohen  <wcohen@redhat.com>
-
-	* utils/opcontrol: fix normalise_events for default event.
-
-2004-02-10  Will Cohen  <wcohen@redhat.com>
-
-	* events/i386/p6_mobile/events:
-	* events/i386/p6_mobile/unit_masks:
-	* events/Makefile.am:
-	* libop/op_cpu_type.c:
-	* libop/op_cpu_type.h:
-	* libop/op_events.c:
-	* utils/op_help.c: Add support for Pentium M (Centrino).
-
-2004-02-10  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* events/i386/p4/events: s/instructions/uops/ when appropriate
-	* events/i386/p4/unit_masks: change many bitmask to mandatory,
-	  fix invalid default value, remove mask only used with HyperThreading
-	* events/i386/p4-ht/unit_masks: s/bitmask/mandatory. Remove unused
-	  unit mask
-
-	* libop/op_events.h:
-	* libop/op_events.c: warn if an unit mask is not used (this add a
-	  warning for Itanium2, added in TODO)
-
-	* libop/op_parse_event.c: don't silently accept invalid value as zero
-	  (e.g FOO:3000:0xx21 was accepted as FOO:3000:0x0)
-
-	* utils/opcontrol: mandatory default value was rejected because we
-	  checked the events before getting the default unit mask.
-
-2004-02-08  Zwane Mwaikambo  <zwane@arm.linux.org.uk>
-
-	* daemon/opd_cookie.c: Fix syscall base number.
-
-2004-02-08  Zwane Mwaikambo  <zwane@arm.linux.org.uk>
-
-	* daemon/opd_cookie.c: arm support for timer interrupt
-
-2004-02-08  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/liblegacy/opd_sample_files.c: op_mangle_filename() presume
-	  mangle_values::dep_name is never NULL, this fix a daemon segfault
-	  with 2.4 kernel
-
-2004-02-07  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* events/i386/p4/unit_masks:
-	* events/i386/p4-ht/unit_masks: global_power_events need mandatory
-	 unit mask
-
-2004-02-07  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/arrange_profiles.cpp: re-order function to fix a build problem
-	  with gcc 3.3.1
-
-2004-02-05  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/callgraph_container.cpp: tweak computation of callee_counts
-
-2004-02-05  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/callgraph_container.cpp: Big Thinko (tm), callee samples offset
-	  are unordered
-
-2004-02-04  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/callgraph_container.cpp:
-	* libpp/format_output.cpp: arc_recorder::get_caller()/get_callee() name
-	  was reversed (caught by Will Cohen)
-
-2004-02-02  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/callgraph_container.cpp:
-	* libpp/callgraph_container.h:
-	* pp/opreport_options.h:
-	* pp/opstack.cpp: implement --threshold
-
-2004-02-02  Will Cohen  <wcohen@redhat.com>
-
-	* libpp/callgraph_container.h:
-	* libpp/format_output.h: missing std:: in header.
-
-2004-02-01  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/callgraph_container.h:
-	* libpp/callgraph_container.cpp: rewrite populate() and ehance
-	  caller_callee_recorder to support multiple profile classes.
-	* pp/opstack_options.cpp: accept multiple profile class
-
-2004-02-01  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/callgraph_container.cpp: get correct source filenname liner
-	* libpp/format_output.cpp: re-factorize a bit by moving anon
-	  namespace from format_output:: to ::
-
-2004-02-01  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* doc/oprofile.xml:
-	* doc/opstack.1.in: document new options
-
-	* libpp/callgraph_container.h:
-	* libpp/callgraph_container.cpp: improve readability, add linenr
-	  debug information recording partially, see TODO
-
-	* libpp/format_output.h: s/formatter/opreport_formatter, define
-	  cg_formatter
-	* libpp/format_output.cpp: implement cg_formatter, factorize some code
-	  in formatter base class
-
-	* pp/opreport.cpp: s/formatter/opreport_formatter
-	* pp/opstack_options.h:
-	* pp/opstack_options.cpp: implement output selection/formatting options
-	* pp/opstack.cpp: use cg_formatter for output.
-
-2004-01-30  Philippe Elie  <phil.el@wanadoo.fr>
-
-	commit not itended but tree is stable so add a ChangeLog ...
-	* libpp/Makefile.am:
-	* libpp/symbol.cpp: new file for output_hint()
-	* libpp/profile_container.cpp: use output_hint()
-	* libpp/callgraph_container.cpp:
-	* libpp/callgraph_container.h:
-	* libpp/symbol.h:
-	* pp/opstack.cpp: get output hint from the callgraph container.
-
-2004-01-29  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/init.c:
-	* daemon/opd_events.c:
-	* daemon/opd_kernel.c:
-	* daemon/opd_mangling.c:
-	* daemon/opd_printf.h:
-	* daemon/opd_sfile.c:
-	* daemon/opd_trans.c:
-	* daemon/oprofiled.c:
-	* daemon/oprofiled.h:
-	* daemon/liblegacy/init.c:
-	* daemon/liblegacy/opd_image.c:
-	* daemon/liblegacy/opd_kernel.c:
-	* daemon/liblegacy/opd_mapping.c:
-	* daemon/liblegacy/opd_parse_proc.c:
-	* daemon/liblegacy/opd_proc.c:
-	* daemon/liblegacy/opd_sample_files.c:
-	* utils/opcontrol: split daemon --verbose, w/o argument it's synonymous
-	  to --verbose=all. Better splitting is welcome
-
-	* daemon/opd_sfile.c: Fix thinko, opcontrol --reset coredump'ed
-
-2004-01-29  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libop/op_events.c:
-	* utils/op_help.c: don't use default in switch (cpu_type) to ensure
-	  adding a cpu type will issue a warning where we forget to handle it.
-
-2004-01-28  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/opd_sfile.c: invert from/to start offset for kernel samples
-
-	* pp/opstack.cpp:
-	* libpp/callgraph_container.h:
-	* libpp/callgraph_container.cpp: if we can't bfd_open a binary and
-	  the sample file show it's a kernel sample file we must ignore
-	  completely the it. (Added to TODO, there is something fragile here)
-
-2004-01-28  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/image_errors.cpp: typo, warn at first error not from second
-
-2004-01-28  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libutil++/tests/utility_tests.cpp:
-	* libopt++/popt_options.cpp: s/template <class/template <typename/
-
-2004-01-28  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/init.c: show buffer read only when --verbose
-
-	* libpp/image_errors.cpp:
-	* libpp/image_errors.h: add needed API for callgraph image error
-	* libpp/callgraph_container.cpp: use it.
-	* pp/opstack.cpp: report_image_errors()
-
-2004-01-27  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libutil++/op_bfd.cpp: Fix bfd_find_nearest_line() and separate
-	  debug info, please test it.
-
-2004-01-27  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* doc/opcontrol.1.in:
-	* doc/oprofile.xml:
-	* utils/opcontrol: add --cpu-buffer-size and document it.
-
-2004-01-27  Will Cohen  <wcohen@redhat.com>
-
-	* events/i386/p4/unit_masks:
-	* events/i386/p4-ht/unit_masks: correct branch_type unit mask.
-	
-2004-01-25  John Levon  <levon@movementarian.org>
-
-	* doc/Makefile.am: fix make chunk
-
-	* doc/buffers.dia: add editable source
-
-	* doc/buffers.png: fix stupid typos
-
-	* doc/internals.xml: add some more text
-
-2004-01-25  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/parse_filename.h:
-	* libpp/parse_filename.cpp: profile_spec_equal() new predicate
-	* libpp/arrange_profiles.cpp: use it to ensure we create a new
-	  profile_samples_file when needed.
-
-2004-01-24  John Levon  <levon@movementarian.org>
-
-	* doc/oprofile.1.in:
-	* doc/opstack.1.in: some cleanups
-
-	* libop/tests/load_events_files_tests.c:
-	* libop/tests/alloc_counter_tests.c:
-	* libop/op_events.c: rename env var to OPROFILE_EVENTS_DIR
-
-2004-01-20  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* daemon/opd_sfile.c: (getfile()) avoid out of bound sf->files[] access
-
-2004-01-20  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/callgraph_container.cpp: tweak comparator to sort arcs.
-	* pp/common_option.cpp: typo.
-
-2004-01-20  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* doc/opannotate.1.in:
-	* doc/opgprof.1.in:
-	* doc/opreport.1.in:
-	* doc/oprofile.xml:
-	* doc/opstack.1.in: document --verbose options a bit
-	* libutil++/cverb.cpp:
-	* libutil++/cverb.h:
-	* pp/common_option.cpp: remove vlevel2 verbose object
-
-2004-01-20  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libutil++/cverb.h:
-	* libutil++/cverb.cpp: new cverb handling. most interresting are
-	  --verbose=sfile,bfd,level1 resp. sample filename handling bfd
-	  handling and more verbosity (only for bfd actually)
-
-	* pp/opgprof.cpp: controlled by --verbose==bfd, add level1 to get
-	  verbose symbols.
-	* libpp/callgraph_container.cpp:
-	* libutil++/op_bfd.cpp:
-	* pp/common_option.cpp:
-	* pp/opannotate_options.cpp:
-	* pp/opgprof_options.cpp:
-	* pp/opreport_options.cpp:
-	* pp/opstack_options.cpp: use it
-
-2004-01-19  John Levon  <levon@movementarian.org>
-
-	* libpp/callgraph_container.h:
-	* libpp/callgraph_container.cpp: rename caller_callee_recorder
-	to arc_recorder.
-
-	* doc/CodingStyle: mention trailing comments
-
-	* utils/opcontrol: fix formatting. Error if --note-table-size
-	is passed on 2.6 kernels.
-
-2004-01-18  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libdb/db_debug.c: fix build on alpha
-	* libpp/callgraph_container.cpp: missing std:: in header
-
-2004-01-18  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libpp/callgraph_container.cpp: Obviously not2(weak_ordering_function)
-	  is not a weak_ordering function leading to mysterious segfault
-	  during sort.
-
-2004-01-18  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* doc/opstack.1.in: new file, describe opstack usage
-	* doc/oprofile.xml: describe opstack
-
-2004-01-18  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libutil++/op_bfd.cpp: Don't fix symbol offset for symbol definition
-	  coming from the original file. A corner case remain broken: symbols
-	  definition coming from debug file and belonging to another section
-	  than .text are incorrecttly offseted.
-
-2004-01-18  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libop/op_sample_file.h: with cg file we need two field is_kernel.
-	* daemon/opd_events.c:
-	* daemon/opd_events.h:
-	* daemon/opd_mangling.c:
-	* daemon/liblegacy/opd_sample_files.c: handle cg_to_is_kernel
-
-	* pp/populate.cpp:
-	* pp/populate.h:
-	* pp/image_errors.cpp:
-	* pp/image_errors.h: move this four files to libpp
-
-	* libpp/populate.cpp: we can get empty filename now, ignore them.
-	* libpp/image_error.h: remove this file by moving its contents ...
-	* libpp/image_errors.h here (this two filenames was confusing)
-
-	* libpp/Makefile.am: update according to files move.
-
-	* libpp/arrange_profiles.h:
-	* libpp/arrange_profiles.cpp: Fix find_profile_sample_files(), I was
-	  creating to many profile_sample_files. Add a bunch of operator<<.
-
-	* libpp/callgraph_container.h:
-	* libpp/callgraph_container.cpp:
-
-	* libpp/filename_spec.h: add cg_filename member
-	* libpp/filename_spec.cpp: handle cg_filename, Fix an error in
-	  is_dependant(), I'm a bit nervous about this predicate...
-
-	* libpp/locate_images.h:
-	* libpp/locate_images.cpp: #include fix
-
-	* libpp/parse_filename.cpp: Fix cg filename handling to not trigger
-	  some invalid_filename error with cg filename.
-
-	* libpp/profile.h:
-	* libpp/profile.cpp: we can no longer assume we use u32, so use
-	  odb_key_t in public interface. Add operator== acting on iterator
-
-	* libpp/symbol_container.cpp:
-	* libpp/symbol_container.h: add find(symbol const &);
-	* libpp/profile_container.h:
-	* libpp/profile_container.cpp: add a public interface to iterate over
-	  all smybols and to find a specific symbol
-
-	* libpp/symbol.h: typo, missing include
-
-	* pp/Makefile.am: handle opstack various new files and files move to
-	  libpp.
-
-	* pp/opreport.cpp:
-	* pp/opannotate.cpp: pass symbol_filter as params to populate()
-
-	* pp/opannotate_options.cpp: add verbose output of profile_classes
-	* pp/opgprof_options.cpp:
-	* pp/opreport_options.cpp:
-
-	* pp/opstack.cpp:
-	* pp/opstack_options.cpp:
-	* pp/opstack_options.h: new callgraph utility far to be complete but
-	  enough to start to play with callgraph.
-
-2004-01-04  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libutil/op_fileio.c: op_write_file(): allow to write zero byte.
-	* libutil/op_fileio.h: add fatal error for opd_read_int_from_file()
-	* daemon/init.c: add fatal error for opd_read_fs_int
-	* daemon/oprofiled.c:
-	* daemon/oprofiled.h:
-	* daemon/liblegacy/init.c:
-	* daemon/liblegacy/opd_24_stats.c: use this new fatal parameter
-	* daemon/opd_stats.c: dump /dev/oprofile/stats/*
-
-	* daemon/opd_sfile.c:
-	* daemon/opd_sfile.h: augment opd_sfile with an hash table of related
-	  call graph file
-	
-	* daemon/opd_mangling.c:
-	* daemon/opd_mangling.h:
-	* daemon/liblegacy/opd_sample_files.c:
-	* libop/op_mangle.c:
-	* libop/op_mangle.h: use new mangling name scheme, in short word:
-	  {dep} part is no longer optionnal, a {cg}/binary_name can optionnally
-	  follow the {dep} part
-	* libop/tests/mangle_tests.c: update test according to sample filename
-	  change
-
-	* libpp/filename_spec.cpp:
-	* libpp/filename_spec.h: filename_spec::is_dependant() new function
-	* libpp/parse_filename.h: add a cg_image member
-	* libpp/parse_filename.cpp: handle callgraph filename
-
-	* libpp/arrange_profiles.cpp:
-	* libpp/arrange_profiles.h: rather to use string for sample files, use
-	  a { string sample_filename, list<string> cg_files }, if {dep} part
-	  != {cg} part it's a cross-call binary.
-	  Note the sample_filename string can be empty (callgraph file w/o any
-	  samples into a binary)
-
-	* libpp/profile_spec.h:
-	* libpp/profile_spec.cpp: fix filtering of callgraph filename (this fix
-	  Carlo problem where opreport die with an invalid_filename error)
-
-	* pp/opgprof.cpp: use the new arrange_profiles.h API to acquire
-	  callgraph filename associated with a binary
-	* pp/opgprof_options.cpp: don't filter callgraph filename
-
-	* pp/opreport.cpp:
-	* pp/populate.cpp: update according arrange_profile API change
-
-	* utils/opcontrol: add --callgraph=#depth parameter
-	* doc/opcontrol.1.in:
-	* doc/oprofile.xml: document opcontrol --callgraph=#depth
-
-2004-01-16  Will Cohen  <wcohen@redhat.com>
-
-	* libutil++/op_bfd.cpp: Always use filepos of original file.
-
-2004-01-16  Will Cohen  <wcohen@redhat.com>
-
-	* libop/op_config.h (DEBUGDIR): New.
-	* libutil/op_fileio.h:
-	* libutil/op_fileio.c: New function calc_crc32
-	* libutil++/Makefile.am: Add libop to include directory.
-	* libutil++/op_bfd.h:
-	* libutil++/op_bfd.cpp: New functions separate_debug_file_exists,
-	  get_debug_link_info, find_separate_debug_file, open_bfd, and
-	  get_symbols_from_file to support separate debug files. 
-
-2004-01-02  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* pp/opgprof_options.cpp: use the proper type for options::demangle
-
-2004-01-01  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libregex/op_regex.cpp:
-	* libregex/op_regex.h: tidy a bit private interface
-
-2004-01-01  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* gui/oprof_start_config.h:
-	* gui/oprof_start_util.h:
-	* libpp/symbol.h:
-	* libutil++/cverb.cpp:
-	* libutil++/cverb.h:
-	* pp/opannotate_options.cpp:
-	* pp/opgprof_options.cpp: move some #include <> from .h to .cpp
-
-2004-01-01  Philippe Elie  <phil.el@wanadoo.fr>
-
-	* libregex/op_regex.cpp: don't assume int('a') > int('0')
-
-See ChangeLog-2003 for earlier changelogs.
+See ChangeLog-2004 for earlier changelogs.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:





No, thanks