Hi, :)
Currently perf-annotate have several problems.
- Wrong line numbers on perf-annotate (both stdio and TUI)
- Wrong sum of overhead(percent) matching source lines
- It's so confusing that the output is mixed with
both source code and assembly code. (related to depending on 'objdump -S')
So fix them regarding wrong line numbers
and Introduce new source_code collecting actual code,
not depending on 'objdump -S'.
In near future, I have a plan for new reable annotate view
base on source code (per function(sym)) using the 'struct source_code'. :)
I'd appreciate some feedback.
(Current v1 don't consider improvement of performance of annotate
and don't make code of annotate more compact so I'll consider them in v2)
Thanks,
Taeung
Taeung Song (4):
perf annotate: Remove needless regular expression for filename:linenr
perf annotate: Align filename:linenr and more correct summary
perf annotate: Change the method counting line numbers
perf annotate: Introduce source_code to collect actual code
tools/perf/util/annotate.c | 174 +++++++++++++++++++++++++++++++++++----------
tools/perf/util/annotate.h | 7 ++
2 files changed, 144 insertions(+), 37 deletions(-)
--
2.7.4
