When gathering the attributes of each hwmon chip, libsensors uses a
temporary structure in memory to order and group all the attributes
into features. This temporary structure used to be a single array with
room for every possible attribute/subfeature. While simple, this
approach required to predefine a maximum number of per-type sensor
that could be handled.
In order to get rid of this arbitrary limit, which we hit and had to
raise three times already, I changed the temporary structure to an
array of dynamically allocated per-type subattribute arrays. This lets
us not allocate any memory for types which aren't implemented by a
given chip, and more importantly, this lets us reallocate room for
more attributes of a given type as needed.
I decided to allocate chunks of 8 attributes at a time, as this seemed
a good compromise between two frequent reallocations and
over-provisioning. It could be tweaked if needed.
Icing on the cake, I benchmarked this change on two different systems
and it results in performance gains. The total heap usage as reported
by valgrind is down by 50% on average, with peak memory consumption
(as reported by valgrind's massif) also down by 43% on average. The
total instructions count (as reported by valgrind's callgrind) is down
by 11% on average, with measured execution time also down by a few
percents. Valgrind rocks, BTW.
I have some ideas to optimize the memory allocations further, but I do
not expect such a huge gain from them. They may not even improve peak
memory consumption as massif shows the peak is somewhere else now at
least in some cases.
attributes. These are defined in the standard sysfs interface for quite
some time, and at least three drivers (max6650, lm63 and applesmc)
implement them so we should support them.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@6029 7894878c-1315-0410-8ee3-d5d059ff63e0
representation of chip names.
sensord, sensors: Fix a memory leak when one or more chip names are
provided on the command line.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/trunk@5740 7894878c-1315-0410-8ee3-d5d059ff63e0
use it to retrieve a specific subfeature by type. While it is slighly
less efficient than looping over sensors_get_all_subfeatures(), it
often makes the application code much more elegant.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/branches/lm-sensors-3.0.0@4846 7894878c-1315-0410-8ee3-d5d059ff63e0
for faster main features lookup. One side effect of this change is
that subfeatures can no longer have labels nor be ignored. I do not
think that this is a problem in practice, and actually this makes a
lot of sense.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/branches/lm-sensors-3.0.0@4834 7894878c-1315-0410-8ee3-d5d059ff63e0
get the list of all main features, and one to get the list of all the
subfeatures of a given main feature. This is a more logical interface for
applications to use. The current implementation is admittedly less than
optimal, because the storage structures weren't meant for it, but this
issue can (and will) be addressed later.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/branches/lm-sensors-3.0.0@4831 7894878c-1315-0410-8ee3-d5d059ff63e0
standard sysfs interface. I don't think it makes much sense to
display the offset values as part of the output of "sensors" (it
would even probably confuse people) but having support for these
in libsensors makes it possible to adjust the offsets in
sensors.conf, which is convenient.
This closes ticket #2248.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/branches/lm-sensors-3.0.0@4769 7894878c-1315-0410-8ee3-d5d059ff63e0
to add more "standard" feature types such as current or power,
we want to give them numbers that immediately follow the fan, voltage
and temperature features, so that the code is as simple as possible.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/branches/lm-sensors-3.0.0@4767 7894878c-1315-0410-8ee3-d5d059ff63e0
feature number as the logical mapping. This means that the compute
mapping can be turned into a simple boolean flag. Doing so makes
struct sensors_feature_data smaller, which saves some memory (about
17 kB in my tests.)
git-svn-id: http://lm-sensors.org/svn/lm-sensors/branches/lm-sensors-3.0.0@4758 7894878c-1315-0410-8ee3-d5d059ff63e0
sensors_bus_id. This prevents wasting space with padding on 64-bit
systems, saving some memory (128 bytes for sensors_bus_id, nothing
yet for sensors_feature_data but it will come later).
git-svn-id: http://lm-sensors.org/svn/lm-sensors/branches/lm-sensors-3.0.0@4733 7894878c-1315-0410-8ee3-d5d059ff63e0
MAX_SUB_FEATURES in the process.) This value could change in the
future so it's not safe to make it part of the libsensors API.
And it isn't needed either, applications shouldn't need to use
this value.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/branches/lm-sensors-3.0.0@4732 7894878c-1315-0410-8ee3-d5d059ff63e0
caller select which subset of chips it wants. This is slightly better
size-wise than letting all applications do the filtering by themselves.
This will change the way the command line parameters of "sensors" are
interpreted. Beforehand, the chips were always returned in the order
in which they were listed by the library. Also, each chip could be listed
only once. From now on, the chips will be listed in the order in which
they are passed on the command line, which I think makes more sense. A
side effect is that chips can be listed more than once, if that's what
the user asks for. Not very useful though.
This change makes it possible to make sensors_match_chip() internal
to the library. Filtering the list of chips returned by
sensors_get_detected_chips() was the last known external use for this
function.
This patch looks much bigger than it really is, but the largest part is
really only code reindentation.
git-svn-id: http://lm-sensors.org/svn/lm-sensors/branches/lm-sensors-3.0.0@4701 7894878c-1315-0410-8ee3-d5d059ff63e0
