Per manpage of mount(1) in newer util-linux:
The programs mount and umount traditionally maintained a list of
currently mounted filesystems in the file /etc/mtab. This real mtab
file is still supported, but on current Linux systems it is better
to make it a symlink to /proc/mounts instead, because a regular mtab
file maintained in userspace cannot reliably work with namespaces,
containers and other advanced Linux features.
In most new Linux (e.g. Debian) /etc/mtab is already symlinked to
/proc/mounts.
See
e778642a9e
for the gory details.
The code was multiplying the index with the size of the element, and
then adding it to the typed pointer (resulting in a double
multiplication and an OOB access).
Replace the buggy code with a slightly safer c++ alternative.
* eve: fix API, empty queue, conky reload, formatting issues
Since the EVE module was contributed, CCP has changed how the API works and
what the correct API URL is. So the module had to be updated to respect the
changes made on the EVE side.
Additionally, undefined behavior was occurring for characters without active
skill training queue, as is common for folks with multiple characters on a
single account, and was also occurring when Conky was reloaded following a
config file change. Now the `$eve` variable should just contain an empty
string if the character isn't training a skill (to make it detectable by
`$is_empty`), and the module should do The Right Thing (tm) when the
config file is changed while Conky is running.
The time remaining text has been adjusted to be slightly less verbose.
The module should do the right thing for up to 64 distinct characters.
Presently, though, every additional character requires a blocking CURL
request, meaning that querying many characters is presently not ideal. A
rewrite of the module to use asynchronous requests is probably in order, but
will be left for a future PR.
* eve: fix timezone handling; better handling of bad XML
* On Linux `gettimeofday()` doesn't return valid timezone data; we use
`localtime()` and the `tm_gmtoff` instead. Training times are no longer
incorrect as a result.
* Sometimes the EVE API will return spurious non-XML tags as part of the
response. This previously caused the XML parser to return a null document.
We tell libxml to `XML_PARSE_RECOVER` to try to recover from these errors
where they occur on real data.
* Implemented cached versions of get_nvidia_value and get_nvidia_string_balue. Unmutable values now only are queried once.
* Removed the TODO topic related to using static variable as cache + fixed a typo in the sample nvidia config
* Was using mtf variables to store the perf values. Fixed it
* Removed an useless else statement
* llabs() is included in FreeBSD since FreeBSD 5.0.
Oldest supported FreeBSD version is 9.3 now.
* stdio.h needs to be explicitly included here for FreeBSD.
I can wrap it in #ifdefs if needed, but I don't think it can hurt other OSes.
* addr config variable works fine on FreeBSD.
* FreeBSD 5.x is really an ancient version and long unsupported, so I think this check for it can be dropped.
* Fix and simplify code to get battery state and charge.
* Add required include on FreeBSD.
* Add needed include on FreeBSD.
* Also populate basename to avoid printing (null) process names.
Repored by fellow FreeBSD user Szabolcs Grof.
* nvidia: Update the docs and add more arguments
**Updated the docs in variables.xml to reflect the current nvidia object.**
Basically just copied the inline comments into the XML document
**Added the last of the compatible arguments**
The following arguments were added to `get_nvidia_barval`:
* memfree
* ambienttemp
* gpufreq
* memfreq
* mtrfreq
* imagequality
This should leave only incompatible arguments unsupported (e.g. min/max
values).
Decided to base percentages off of zero rather than the specified minimum
values.
Signed-off-by: Matt Sturgeon <matt@sturgeon.me.uk>
* nvidia: Better error/debug printing
* Merged `scan_nvidia_args` and `set_nvidia_type` into new
`set_nvidia_query`
* Added a translation array for `nvidia`, `nvidiabar`, `nvidiagauge` and
`nvidiagraph` commands (helps with error printing)
* Added some NORM_ERR prints to `get_nvidia_barval`, `get_nvidia_value` and
`get_nvidia_string`
* Removed exess newlines in nvidia's core.cc error definitions
Signed-off-by: Matt Sturgeon <matt@sturgeon.me.uk>
* nvidia: Better sanity checks when running queries
Introduced a pair of functions that checks whether the target we are
querying exists and how many of the target there are.
If the target doesn't exist there probably isn't a nvidia card
installed, if the target count query fails then the propietary drivers
probably aren't installed.
Fixes #269
Signed-off-by: Matt Sturgeon <matt@sturgeon.me.uk>
* Add native support for pulseaudio.
This adds ability to get Pulseaudio current volume without calling an
external program. Also it provides a few other parameters like card's
current profile similarly to pavucontrol. Currently it only queries the
default sink (and associated card).
* Switch to pulse async API.
This us by far the most efficient as conky subscribe to a few pulseaudio
notifications. Whenever pulseaudio server changes (volume or default
sink or default card), conky's callback updates the results.
Rewrite the gtksourceview highlighting file to make better use
of available syntax options. This allows us to separate various
contexts, and means we don't have to keep track of all conky
keywords.
Overview of changes:
* Use separate contexts for config block and "lit string".
* (lit string is the lua lit string [[ .. ]], we assume that
this is used to contain text sections or template blocks)
* Use "sub-string" highlighting to highlight parts of identifiers
separately.
* Added conky.text comments (from '#' to EOL).
* Remove lists of keywords and instead rely on syntax implications.
* Uses separate contexts for config 'key=val' and text '${var}'s.
* Match escape sequences more precisely.
* Lua and conky escape sequences matched separately; lua in normal
strings and conky in literal strings. See escapes note.
* Transition from camelCase to hyphen-separated names
* Reword README to reflect changes
Escapes note:
We also match conky template escape sequences when in lua literal
strings as it would be too messy to separate `conky.text` and
`templateN = ` literal strings.
We currently don't match time format escape sequences as this would
require knowing which text variable we are in. This is only possible
if we duplicate the `bracket-var` context (less painful than
duplicating the `lua-literal-string` context).
Signed-off-by: Matt Sturgeon <matt@sturgeon.me.uk>
The compilation fails when the symbols of terminfo are not present in
the libncurses.so. This can happen when the terminfo library libtinfo.so
is compiled separately. Using pkg_config is the correct approach to
handle all cases.
This adds support for reading log entries from the Systemd journal. A
new command, called simply 'journal', is added which takes a number of
lines as a parameter and optionally which journal (user or system) to
read from.
Obviously, libsystemd is required to build journal support. Tested on
Arch Linux (Systemd v229).
Signed-off-by: Toke Høiland-Jørgensen <toke@toke.dk>
When parsing the argument string for graph objects, check if there is a
double-quoted command in the beginning of the string and extract it
before parsing any options. This prevents items in a command from being
improperly recognized as graph options. It should also eliminate the
need to place commands with spaces into a separate script.
Also, fall back on default_graph_{height,width} instead of using
hard-coded values so that the default values set in a user's config will
work.
* Make dev_name resolve symlinks before resolving the name.
- When symlinks are not resolved first, you will get an error on startup when
using path in /dev/disk/by-id, for example, in diskiograph_*.
* Improve diskio device name resolution by always resolving symlinks.
- Any device path will be resolved of any symlinks before getting the usual
checks for labels et device name.
* Check if diskio device is really a block device, not only if it exists.
- Current check would succeed in case the device is a path, like when using
label: with an empty argument.
* First experiment with $nvidiabar
Added nvidiabar, nvidiagraph and nvidiagauge. Using a dummy value of 50% for now until it is integrated with the nvidia value query functions.
* Implement get_nvidia_barval()
`get_nvidia_barval()` no longer returns a dummy "50%" but instead an actual value for supported arguments:
`gpuutil`, `memutil`, `membwutil`, `videoutil`,`pcieutil`, `fanlevel` and `gputemp`
* use CLOCK_MONOTONIC if supported
get_time is used to control the update interval etc. so there's
no need to query the system time
* repaint immediately when the root pixmap changed
Original patch by pweisenburger,
In addition resets the next_update_time to ensure "now"
FIXES: #48
