management: refactor bytecount routines
PATCH 1/2 management: refactor bytecount routines
From: Lev Stipakov <[email protected]...>
In preparation of DCO stats support, simplify call chains of bytecount routines. No functional changes.
--- src/openvpn/forward.c | 4 +-- src/openvpn/manage.c | 64 ++++++++++++++++++++++++++++------------- src/openvpn/manage.h | 66 ++++--------------------------------------- 3 files changed, 52 insertions(+), 82 deletions(-)
src/openvpn/forward.c +2 -2
@@ -948,7 +948,7 @@ process_incoming_link_part1(struct context *c, struct link_socket_info *lsi, boo#ifdef ENABLE_MANAGEMENT if (management) { -management_bytes_in (management, c->c2.buf.len); +management_bytes_client (management, c->c2.buf.len,0 );management_bytes_server (management, &c->c2.link_read_bytes, &c->c2.link_write_bytes, &c->c2.mda_context); }#endif
@@ -1788,7 +1788,7 @@ process_outgoing_link(struct context *c)#ifdef ENABLE_MANAGEMENT if (management) { -management_bytes_out (management, size); +management_bytes_client (management,0 , size);management_bytes_server (management, &c->c2.link_read_bytes, &c->c2.link_write_bytes, &c->c2.mda_context); }#endif
src/openvpn/manage.c +44 -20
75 lines not on this page · raw message
src/openvpn/manage.h +6 -60
76 lines not on this page · raw message
Signature · 1 line
2.38.1.windows.1
Am 12.12.22 um 10:58 schrieb Lev Stipakov:
Quoted 4 lines
From: Lev Stipakov <[email protected]...>
In preparation of DCO stats support, simplify call chains of bytecount routines. No functional changes.
It would be nice to be a bit more verbose what you are actually simplyfing in the commit message. E.g. inlining all the various function that are used only once.
src/openvpn/manage.c man_bytecount(struct management *man, const int update_seconds) · line 503
Quoted 13 lines
src/openvpn/manage.c +12
-0+ +void +management_bytes_client (struct management *man, +const int size_in, +const int size_out) +{ +if (!(man->persist.callback.flags &MCF_SERVER )) + { + man->persist.bytes_in += size_in; + man->persist.bytes_out += size_out; +man_bytecount_output_client (man); + } }
I don't like this part of the change. We are here moving a inline function from the critical path of data channel packets that is called on every single packet into an non-inline function.
The old code was taking care that everything that was in that critical path can be inlined. The new code breaks that promise.
Arne