Export the posix signal constants in the rx/perf perl test. Fixes a
perl syntax error on solaris.
Change-Id: Iaad361b8533787f9ad97fa00221e01e687f50723
Reviewed-on: http://gerrit.openafs.org/9836
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@your-file-system.com>
Change rxevent_Put so that it takes a pointer to the event being
put, and NULLs that pointer. This removes a lot of duplicate code
in callers, as well as making it harder to reuse a discarded event.
Change-Id: Ib7a51f01687e08ea3dced5932ec9ec27797a784a
Reviewed-on: http://gerrit.openafs.org/8540
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@your-file-system.com>
Reviewed-by: Jeffrey Altman <jaltman@your-file-system.com>
Many different structures can be passed to the rxevent package as
data. Don't give calls special treatment by making rxevent aware of
how to release their reference counts when an event is cancelled.
Update all of the callers of rxevent_Cancel to use the new arguments,
and where they were cancelling functions with calls as parameters add
the appropriate CALL_RELE directives. In many cases, this has led to
new helper functions to cancel particular call-based events.
Change-Id: Ic02778e48fd950e8850b77bd3c076c235453274d
Reviewed-on: http://gerrit.openafs.org/8538
Reviewed-by: Derrick Brashear <shadow@your-file-system.com>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Jeffrey Altman <jaltman@your-file-system.com>
Build a pthreaded, libtool, version of librx.a called liboafs_rx.la.
librx.a remains for LWP applications to use. With this change, all RX
objects are built in both the LWP and pthread cases, so some #ifdef
guards are required to protect code that isn't relevant in a given
build.
Currently, all of our pthreaded objects use libafsrpc to get RX
functionality, so this change is fairly minimal outside of the RX
directory.
Change-Id: I8e629e2319fb1964058e70c3c0c3ed548b09b22d
Reviewed-on: http://gerrit.openafs.org/8058
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@your-file-system.com>
Convert opr so that it uses libtool. For backwards compatibility we
still build libopr.a, but we do so as a static convenience library.
As libopr.a may, in the future, be converted to an LWP library, change
all of the pthreaded binaries so that they link against the libtool
library liboafs_opr.la
Change-Id: Icee04ff4745334f06ffba16df5bb07fc9dcc0b54
Reviewed-on: http://gerrit.openafs.org/8034
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Tested-by: Derrick Brashear <shadow@dementix.org>
Instead of bundling our own copies of Russ's C TAP Harness, start using
source pulled from his git repository using the src/external import
mechanism. Note that we are not currently building the floating
point (is_double) portion of the harness.
In the process of doing so, we also upgrade our test harness to the latest
upstream version, 1.11. This is somewhat problematic, as there have been
some significant code changes since the version bundled with OpenAFS.
Work around these by
*) Referencing the basic.h header as <tests/tap/basic.h>, rather than
just <tap/basic.h>, to match the new upstream layout
*) Changing the include path so that the tests/ directory can be
found within it.
Change-Id: I63efbb30248165e5729005b0a791e7eb7afb051d
Reviewed-on: http://gerrit.openafs.org/7374
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Russ Allbery <rra@stanford.edu>
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Use the rxperf performance testing tools to add a couple of simple
RX tests. The first moves 1Mbyte of data backwards and forwards 30
times. The second starts 30 threads, which each move 1MByte of data
once.
This is by no means an exhaustive test of RX, but the single and
multi-threaded invocations should provide a useful smoke test if
things get very broken.
Change-Id: I11267be067cf6c05a20aeb90a18ed4031502a1b1
Reviewed-on: http://gerrit.openafs.org/7244
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>
On Linux, the RX library has a dependency on libroken for the rk_socket
function. Add this dependency to the RX tests.
Change-Id: I306e846524232bc136cd969ab1b8664d1c570e2d
Reviewed-on: http://gerrit.openafs.org/7127
Tested-by: Simon Wilkinson <simonxwilkinson@gmail.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>
Instead of the current event stack, which uses a sorted linked
list, use a red/black tree to maintain the timer stack. This
dramatically improves event insertion times, at the expense of
some additional implementation complexity.
This change also adds reference counting to the rxevent
structure. We've always had a race between an event being
fired, and that event being simultaneously cancelled by
the user thread. Reference counting avoids that race resulting
in the structure appearing twice in the free list.
Change-Id: Icbef6e04e01f3eef5b888bc3cb77b7a3d1be26ae
Reviewed-on: http://gerrit.openafs.org/5841
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Jeffrey Altman <jaltman@secure-endpoints.com>
Reviewed-by: Jeffrey Altman <jaltman@secure-endpoints.com>
