| Age | Commit message (Collapse) | Author |
|---|
|
Refactored the trace processing script to use argument parsing, etc.,
and improve the structure. Then added several command line options to
filter and provide different views.
Bug: none
Test: run the script on a trace, view the flame graph
Change-Id: I66805829c9e7212b2f69b3556134036f8dd21bc2
|
|
Add the ability to weight the counts by the total size allocated at that
stack trace. Also puts the thread/type/size at the top of the stack
trace and fixes a bug that erroneously inserted an additional stack
frame at the base. Some small code cleanups in the agent also.
Bug: none
Test: attach, run app, pull, process, enjoy the flame graph.
Change-Id: I6256a99c2651696b973c25a5955e3bb9251b86a1
|
|
Change the text encoding to be easier to parse off-line. Also fixes the
per-thread sampling bias Hans pointed out yesterday. Adds the python
script I use to uncompress the text encoding.
Bug: none
Test: attach the agent; run the startop test; uncompress; make flame graph; view
Change-Id: I4a882642e659d8512a02d4980c82fb52470ac7f9
|
