wireguard-openbsd/sys/dev/dt, branch jd/queueboostsWireGuard implementation for the OpenBSD kernelhttps://git.zx2c4.com/wireguard-openbsd/atom/sys/dev/dt?h=jd%2Fqueueboosts2020-03-28T15:42:25ZExport the number of arguments exposed by a probe to userland.2020-03-28T15:42:25Zmpimpi@openbsd.org2020-03-28T15:42:25Zurn:sha1:238295763ced58d84d82653d6eb4cda9b1927b27
Also make the syscall provider exports the number of syscall
arguments. This will allow for stricter checks when parsing
`argN' and future `args->name'.
This is an ABI change, don't forget to keep btrace(8) and your
kernel in sync.
Introduce stacktrace_save_at() and make use of it in dt(4).2020-03-25T14:59:23Zmpimpi@openbsd.org2020-03-25T14:59:23Zurn:sha1:ecfa01e02ef35102b8a3242b3b6394ca528f935e
This variant of stacktrace_save() takes an aditionnal argument to skip
an arbitrary number of frame. This allows to skip recording frames used
to execute the profiling code and produces outputs easier to understand.
Inputs from and ok visa@
Replace msleep(9) by sleep_setup/setup_signal/finish() dance.2020-02-04T10:56:15Zmpimpi@openbsd.org2020-02-04T10:56:15Zurn:sha1:1159b64969d2758308ac09b677e99fb50a97ca6a
This handrolled cond_wait() dealing with signals prevents a lock
ordering problem when executing probes inside the scheduler code.
That means we stop relying on a mutex for synchronisation and instead
rely on the sleep_* internals, currently the SCHED_LOCK().
From kettenis@
Improve use of SMR in dt(4) by changing the time when smr_barrier()2020-01-27T17:13:33Zvisavisa@openbsd.org2020-01-27T17:13:33Zurn:sha1:0e146e1b7bfd98a4826341ccc1fd28b5a1f1026c
is invoked. This enables threads to continue gathering events without
interruption when another thread starts or stops recording.
OK mpi@
New `kern.allowdt' button must be set to open(2) /dev/dt.2020-01-24T14:00:31Zmpimpi@openbsd.org2020-01-24T14:00:31Zurn:sha1:c4318b7540a1c4c503732a1e40da9aaf95e3d46b
dt(4) exposes kernel internals, addresses and content of states to
userland. As such its interface shouldn't be available without
enabling it consciously.
ok millert@, deraadt@
Import dt(4) a driver and framework for Dynamic Profiling.2020-01-21T16:16:22Zmpimpi@openbsd.org2020-01-21T16:16:22Zurn:sha1:91b2ecf67e2a667ccefd071f95f44a8068f1bbc4
The design is fairly simple: events, in the form of descriptors on a
ring, are being produced in any kernel context and being consumed by
a userland process reading /dev/dt.
Code and hooks are all guarded under '#if NDT > 0' so this commit
shouldn't introduce any change as long as dt(4) is disable in GENERIC.
ok kettenis@, visa@, jasper@, deraadt@