commit 6953471e0d3bfbc2bbccf6ca6a82679f3de5a036
author Ryan Savitski <rsavitski@google.com> Thu Nov 07 17:04:34 2019 +0000
committer Ryan Savitski <rsavitski@google.com> Thu Nov 07 17:04:34 2019 +0000
tree 71f0278c97742ad50e0dd5ee997a0e89151a7798
parent ded3b84bd08b12cfb3d021150aa2aede7e0179cc

compact sched_waking: writer (traced_probes)

Very similar to sched_switch, mostly copy-pasted code.
Two points I want to highlight:
* I've made a simplifying assumption with the "event
  format matches compile-time assumptions" condition
  that needs to be met for us to actually use compact
  encoding at runtime (if requested). I've kept it as
  a single bool that is true only if both sched_switch
  and sched_waking are "valid". This technically stops
  us from e.g. using compact sched_switch on a
  platform that predates sched_waking (~4 yr old
  kernel patch).
* Not happy with how hard to read the default case of
  ParsePagePayload's switch statement has turned out
  to be, but also don't want to hide control flow.
  Not sure how to structure this better, nested switch
  on the outcome of an extracted decision-making logic?

Change-Id: I7caf9a413470f2006d0848ba5c8bdd46ab8d479e