Drop MODTCL_CONFIG and MODTK_CONFIG.
Replace with equivalents in the only ports affected:
math/graphviz and math/R.
Tcl/Tk-dependent ports normally need the location,
not the full path to the config files.
ok sthen@
Drop MODTCL_CONFIG and MODTK_CONFIG.
Replace with equivalents in the only ports affected:
math/graphviz and math/R.
Tcl/Tk-dependent ports normally need the location,
not the full path to the config files.
ok sthen@
Drop MODTCL_CONFIG and MODTK_CONFIG.
Replace with equivalents in the only ports affected:
math/graphviz and math/R.
Tcl/Tk-dependent ports normally need the location,
not the full path to the config files.
ok sthen@
Drop MODTCL_CONFIG and MODTK_CONFIG.
Replace with equivalents in the only ports affected:
math/graphviz and math/R.
Tcl/Tk-dependent ports normally need the location,
not the full path to the config files.
ok sthen@
Drop MODTCL_CONFIG and MODTK_CONFIG.
Replace with equivalents in the only ports affected:
math/graphviz and math/R.
Tcl/Tk-dependent ports normally need the location,
not the full path to the config files.
ok sthen@
Mechanical rename of prefix_adjout to adjout_prefix.
All the things for the Adj-RIB-Out handling should use adjout_ as prefix.
So both the struct and the various functions are renamed. This makes it
clearer what belongs to the adjout code.
OK tb@
Introduce an adjout_attr object that decouples some of the common data
from the prefix_adjout. This helps to reduce the memory footprint by
around 20% in large IXP setups.
The size of struct prefix_adjout is critical since we allocate very many
of those. e.g. 250k prefixes * 750 peer = 187mio objects. Every byte saved
makes a big difference.
OK tb@
espeak: work around build failure on !lld archs, ok sthen
audio/espeak doesn't build on sparc64 due to what looks like a bug
in either base libtool or ld.bfd. libtool eats -L options for the
ld.so search directories in the ldconfig -r output, which then causes
linking failures. Work around this by switching to GNU libtool on the
affected architectures.
/usr/bin/libtool --tag=CXX --mode=link c++ -O2 -pipe -static -lm -lpcaudio -L/usr/local/lib -L/usr/X11R6/lib -o src/speak-ng src/speak-ng.o src/libespeak-ng.la
...
libtool: link: c++ -o src/.libs/speak-ng -pthread -O2 -pipe src/speak-ng.o -Lsrc/.libs -lm -lpcaudio -lpulse-simple -lpulse -logg -lvorbis -lpthread -liconv /usr/obj/ports/espeak-1.52.0/espeak-ng-1.52.0/src/.libs/libespeak-ng.a -Wl,-rpath-link,/usr/local/lib
[...]
/usr/bin/ld: warning: libxcb.so.4.1, needed by src/.libs/libpcaudio.so.0.1, not found (try using -rpath or -rpath-link)
/usr/bin/ld: warning: libXau.so.10.0, needed by src/.libs/libpcaudio.so.0.1, not found (try using -rpath or -rpath-link)
/usr/bin/ld: warning: libXdmcp.so.11.0, needed by src/.libs/libpcaudio.so.0.1, not found (try using -rpath or -rpath-link)
/usr/local/lib/pulseaudio/libpulsecommon-17.0.so: undefined reference to `xcb_intern_atom_reply'
/usr/local/lib/pulseaudio/libpulsecommon-17.0.so: undefined reference to `xcb_disconnect'
...
Error while executing c++ -o src/.libs/speak-ng -pthread -O2 -pipe src/speak-ng.o -Lsrc/.libs -lm -lpcaudio -lpulse-simple -lpulse -logg -lvorbis -lpthread -liconv /usr/obj/ports/espeak-1.52.0/espeak-ng-1.52.0/src/.libs/libespeak-ng.a -Wl,-rpath-link,/usr/local/lib
[2 lines not shown]
Add a get-clipboard option which when enabled (the default is off) uses
the same mechanism as palette requests to request clipboard from the
terminal and forward to the requesting pane. Remove the now-redundant
forward-to-pane ability from "refresh-client -l". GitHub issue 4275.
rpki-client: convert mft.c to opaque ASN1_STRING
Mostly more of the same. In addition the comparison functions can use
ASN1_STRING_cmp() instead of doing most of it by hand. The additional
type check this introduces is fine since the ASN.1 template ensures
that. Annotate a spot where we fail to ensure that the unused bits
octet is zero, thus libcrypto's misdesign introduces a few bits of
malleability for every hash that ends in zero bits. This will be easier
to fix later.
ok claudio
use per cpu refs in the input path instead of one refcnt per port.
passing packets into tpmr relies on taking a ref from an smr critical
section so the tprm_port struct can be used while interacting with
the larger network stack because you can't sleep while in an smr
critical section and the network stack has sleeping points.
this is fine, but it's atomic ops against the same refcnt in the
one tpmr_port struct that can be running in parallel on all the
softnet threads. this means the cacheline under this refcnt gets
pulled around a lot, which in turn affects performance.
to compensate for this loss in performance, this diff adds per cpu
refcnts that act as a proxy to the tpmr_port refcnt. this effectively
hashes the softnet threads into individual refcnts on separate cache
lines.
unfortunately, softnet threads can move between cpus because the
network stack has sleeping points. this prevents us from using the
[12 lines not shown]
