* Limit OPENBLAS to 1 thread in manager
* Restrict openblas threads when running openpilot
* print in CI
* Disable internet for laika, even in CI
* update ref
* Move migration code to new file
* Migrate messages in process replay. Remove old code
* Fix old_logtime
* Keep old logtimes on migrated messages
* Update ref commit
* 1061b1c7-b944-43e3-a940-b56b64d66f54/700
* bump cereal
* bump cereal
* make mypy happy
* write TODO
* read height from written params
* fix certain_if_calib logic and unit tests
* factor moving_avg_with_linear_decay
* remove whitespace
* update model ref commit
* default value for CI
* update process replay ref commit
* update process replay ref commit
* update process replay ref commit
* bump cereal
* refactor radard
* Revert "refactor radard"
This reverts commit 4b3507ff58.
* May work
* No radar for test
* check length
* no accel for now
* First accel
* Cleaner way
* Re-enable radar
* update proc replay
* This might cause oscillation
* Update ref commit
* frequency based messaging
* frequency based messaging
* frequency based messaging
* frequency based messaging
* rename better
* use frame in if statement
* syntax
* remove extra space
* put all behind one if statement
* update refs
---------
Co-authored-by: Shane Smiskol <shane@smiskol.com>
* replay_process_with_fake_sockets implementation
* add missing polled_pubs to configs
* drained_pubs field
* updated cereal
* Remove python/native variations of process replay. Replace with universal one using cereal fake sockets
* Replace old py FakeSocket with DummySocket
* Invalidate and deregister fake sockets after replay is done
* Remove unused import
* Set up new prefix for each replay
* Fixes for radard
* Refactor ReplayContext and ProcessConfig
* Minor fixes
* Reimplement controlsd fingerprinting callback
* time.sleep for sockets to safely reconnect
* Fix fingerprinting for controlsd
* Fixes for regen to work
* Fix replay loop to respect submaster frames
* Fix profiler to use new ProcessConfig fields
* Remove tqdm
* Refactor tests to use new ProcessConfig
* Add FrequencyBasedRcvCallback
* Make tolerance None by default
* Update cereal
* Add get_process_config utility func
* Update cereal. Simplify sync procedure
* Chain context managers
* New sub-socket reconnection procedure
* Fix linter issues
* Revert chaining of context managers
* Init controlsState only when replaying controlsd. Update cereal
* Update cereal
* Update process_replay to use new cereal API
* Update cereal
* Update cereal
* Update cereal
* Simplify radard recv callback
* Update release/files_common
* Drain the sockets until the next cycle in python process replay
* Add waiting mechanism to native process replay, which mimics the syncing
* Sort logs when comparing them. Drain all the sockets instead of only recv_socks.
* Reverse the retrieval order of FakeSocket
* Change pub order for torqued
* Update ref commit
* bump panda
* set safety config/configure experimental longitudinal
* send inactive signals
* bump
* just calculate in function
* already in the variable
* clean that up
* add comment, on ford stock long, prchg seems to be set from the possible future accel
* comment
* this signal didn't have an impact on engaging
* add comment and fix
* worst case while AEB
* and kill ACCDATAs
* this could be dangerous
* need to test this too
* Revert "need to test this too"
This reverts commit 09bba75950.
* Revert "this could be dangerous"
This reverts commit 323143e847.
* Revert "and kill ACCDATAs"
This reverts commit 15f6804bf6.
* Revert "worst case while AEB"
This reverts commit 85e1aaa8e1.
* bump panda
* Update selfdrive/car/ford/fordcan.py
* move around
* was already here
* rm
* use variable
* update docs
* Update ref_commit
---------
Co-authored-by: Shane Smiskol <shane@smiskol.com>
* Add support for whitelist-procs to toggle on/off specific processes during regen
* Use process CONFIGS from process_replay to generate fake processes instead of hard-coded values
* Add rawgpsd to fake_daemons. Remove duplicates from replayable_messages.
* Remove duplicate block of code from setup_env
* Remove rawgpsd comment
* locationd ttff
* log time to first fix
* rename
* max it like laikad
* Update selfdrive/locationd/locationd.cc
* Update ref_commit
* Log when gpsOK first becomes true
* don't forget to update current time!
* stash
* make it deterministic (no proc replay cur time)
* Update ref_commit
* rename to make this clear
* wip
* wip
* wip
* wip
* wip
* wip
* wip
* wip
* wip
* remove compiler flag stuff
* force error for testing
* bump panda and fix mistake
* wip
* bump panda
* wip
* wip
* forgot checksum and stuff
* add to signals
* rename
* test
* test
* bump panda and opendbc
* bump cereal
* add blank filler
* test
* add steer required to infotainment
* fix missing bit
* fix missing bit
* cleanup for PR
* cleanup for PR
* bump submodules
* wip
* wip
* i wonder what 1 is
* seemed to also be obstacle detected?
* bumppanda
* only if global
* only global gen 2
* fix order
* detect and send
* detect in rest of places
* update refs
* bumpopendbc
* copy whole message (camera sometimes sets these signals)
* bumppanda
* update refs!
* bumppanda
* let's not change this
* same as all the other functions
* Revert "same as all the other functions"
This reverts commit e86dbb695e.
---------
Co-authored-by: Shane Smiskol <shane@smiskol.com>
* add erf based ff
* silly bug; diff of nonlinear != nonlinear of diff
* add sigmoid based ff, ensure slope at 0 > 1
* reduce steer down limit and increase driver allowance
* rebase panda
* atry without friction, and with tanh nonlinear
* finalize the nonlinear function
* do not disable friction compensation in the ff
* bump panda
* bump panda
* update refs
* update refs
* resolve comments
* Add type hints
---------
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
* indecisive
* rename to generic FW query
* remove code and update comment
* we need this to start off, unless we set multiplexing immediately
* draft
* draft 2
* try that
* can't do this either, boardd might read Enabled after removing, but before setting new Request param
* this should work
* use one less param
* fix params
* match behavior (set all pandas to safety param of 1, disabling multiplexing for fingerprinting
* clean up (some tests may temp break)
* fix param name and sort
* time it
* yes it does matter
* add to hyundai's bus 5 query
* remove hyundai for now
* this should work
* clean up
* clean up
* flip argument around, clean up
* fix test_startup
* some clean up
* rm line
* comment makes more sense
* required typing
* clean up common type
* comments
* Update selfdrive/car/car_helpers.py
* line
* whoops, need to set before vin!
* fix debug
* annoying
* more debugging
* bug fix (needs both keys always)
* debuGG
debuGG
* Revert "debuGG"
This reverts commit 55b2f42932.
* Revert "more debugging"
This reverts commit 02934c3403.
* Revert "annoying"
This reverts commit 8b4e5e0998.
* clean that up
* bumpback
* bumpback
* every second write param
* flip
* stuff
* move up?
* fix timing out in CI
* rm
Kacper Rączy
Maxime Desroches
Mitchell Goff
Yassine
Harald Schäfer
Shane Smiskol
Dean Lee
YassineYousfi
Justin Newberry
Cameron Clough
Vivek Aithal
Adeeb Shihadeh