Linux-specific code in top.cc was moved to linux.cc.
Redundant code (e.g. the parts sorting by CPU usage, CPU time or memory
usage) was removed. Sorting etc. happens in top.cc, whilst
platform-dependent code in linux.cc, freebsd.cc or openbsd.cc just
builds up the process table.
In the Linux code, some functions had a return value which was never
evaluated. They return void now.
I tested it on FreeBSD and Linux; The OpenBSD port does not compile
anyway. I changed the OpenBSD parts, too, so that it will be less
effort to get conky working under OpenBSD.
Signed-off-by: Alexander Graf <agraf@znc.in>
- Displaying of CPU time of the processes now works
- ${top_time} (sorting by CPU time) works
- Using memcpy() instead of many sequential assignments (more clear,
less lines, maybe faster, less effort to change something)
Signed-off-by: Alexander Graf <agraf@znc.in>
Signed-off-by: Pavel Labath <pavelo@centrum.sk>
This allows compilation under FreeBSD without errors.
read_tcpip.cc: Avoid using of SOCK_NONBLOCK argument for socket().
SOCK_NONBLOCK is Linux-specific and its use was substituted with a
fcntl() call after socket().
freebsd.cc: Change return type of update_*() to int, as required by
common.h.
Signed-off-by: Alexander Graf <agraf@znc.in>
Signed-off-by: Pavel Labath <pavelo@centrum.sk>
I've added $memwithbuffers to complement $mem, $memeasyfree, etc. This
variable indicates the used system memory, regardless of buffers/caches. If
the "no_buffers" option is set to "no", it will be the same as $mem. The
reason I find this change useful is that I like to display both the total
used system memory and the memory being used by applications.
I've also included a memwithbuffersbar, which is the same as membar except
that it uses $memwithbuffers as the size instead of $mem.
Signed-off-by: Pavel Labath <pavelo@centrum.sk>
