[Pdbg] [PATCH v2 0/8] Correct HTM machine checking

[Cyril Bur]

This is v2 because it turns out v1 was subtly broken:

Alistair:
Oh I think I see what's going on, and you're kind of getting lucky. If you pass
-p/-c/-t only that part of the tree will be probed which includes the parent
core, however all other cores will be "disabled" rather than marked
"nonexistant" because we don't even attempt to probe unintersting things
(otherwise pdbg startup is prohibitivly slow as it tries to figure out all the
HW that exists).

In order to get around this problem we need to probe lazily. This
appears to work. Unfortunately I haven't had the time I would have
liked but I really wanted to get this on the list.

All the best everyone.

Cyril

Cyril Bur (8):
  pdbg: Fix using FSI backend on palmetto
  libpdbg/host: Actually check the return value of open() in
    host_pib_probe()
  libpdbg/p8chip: Return thread SMT state
  pdbg/htm: Fix error checking condition for htm dump command
  htm: Fix HTM naming
  libpdbg: Lazy probing
  libpdbg/htm: Enable NCU and set single dispatch/instruction mode
  pdbg/htm: Enforce powersave=off

 libpdbg/host.c       |  13 +++---
 libpdbg/htm.c        |  43 +++++++++++++++++-
 libpdbg/libpdbg.c    |   2 +
 libpdbg/libpdbg.h    |  10 ++++-
 libpdbg/operations.h |   8 ++++
 libpdbg/p8chip.c     |   3 +-
 libpdbg/target.c     |  30 +++++++++----
 libpdbg/target.h     |   1 +
 src/htm.c            | 122 ++++++++++++++++++++++++++++++++++-----------------
 src/main.c           |  41 +++++++++--------
 src/main.h           |   3 +-
 src/options_arm.c    |  26 +++++++----
 src/thread.c         |   7 +--
 13 files changed, 217 insertions(+), 92 deletions(-)

-- 
2.16.3