so a little perf fun shows malloc/free/realloc/etc. are, combined a
reasonable overhead. this reduced malloc overhead for draw contexts so
whne we duplicate them or create new ones, we re-use a small cache of
8 of them to avoid re-allocation. just take the first one from the
list as it really is that simple. mempool would not have helped more
here and cost more overhead.
@optimize
to date if you use async preload we still load the header
synchronously and this can be horrible especially with generic
loaders. there is no way to farm this off to the preload thread. now
there is. youhave to set it as a skip head load option before doing a
file_set AND you need to issue a preload ... but now it's possible.
@feature
so. on linux signals are delivered to the main process thread/loop.
thats' where signal handlers are set up and always run. this is sane.
it's predicatble. but of course this is not the same in bsd land.
there "just send the signal to any old thread and call the signal
handler there" seems to tbe the order of the day. this explains why
wer are losing sigchld signals in edje_cc - it's heavily threaded and
bsd is just randombly picking a thread to call it on.
this fixes that. in theory. i hope. i can't test, but putting it in to
share
@fix
We have to use void in a function declaration if we want no function
parameters. Using just empty parenthesis means the function takes an
unspecified number of parameters.
We had it correct for most declarations and this series fixes it for
the rest.
This is for now just a small experiment. It was based on the experiment made
with OpenMP. I prefered to only use Eina here as we have already all the infrastructure
to do this nicely and simply. As a result I get a 65% speed improved on average for
the involved scaling operation. The secondary CPU is on my laptop running with a load of
75% percent. I don't have right now the time to do power consumption analysis, but I
think it shouldn't be to bad. I am also not throwing more core at this as we are not able
to use the second core at its max already, so additional core may result in a bigger
energy loss without enough gain.
Change message level from ERR to WRN, when a glyph is not
loadable because FT fails to load it or it contains 0 pixel.
cserve2 used to complain about invalid glyph 3, on a few fonts
An inotify callback is triggered when an image file changes,
and it is supposed to cleanup all references to this image.
Unfortunately, it was doing it in a very unsafe way as pointers
could become invalid, because of hash free callbacks in the
referenced image list. Add some list safety and always assume
the pointer might be dead after free operations.
TBH, this _file_changed_cb function looks very confused to me
(as it tries to bypass eina_hash_del).
this changes the internal encoding of font glyphs in evas to use 4bit
uncompressed if small, or 4bit rle (run length encoded) if larger.
this caves at least 50% of memory on fonts - and more if bigger. with
large fonts (40-80pixel size) we can save in the region of 80% of
memory used for glyphs. this also happesn to allow speedups in
rendering too.
If a slave dies (eg. killed) when it's idle, then cserve2 will crash
miserably at the next request. Indeed, the dead slave's corpse was
removed from the working slaves' list but not from the lazy idle
slaves list.
Also, set read buffer to NULL after free. Just in case. We never know :)
Being annoyed by different types of eina critical macros - CRI, CRIT,
CRITICAL -, I concluded to unify them to one. Discussed on IRC and
finally, CRI was chosen to meet the consistency with other macros -
ERR, WRN, INF, DBG - in terms of the number of characters.
If there is any missing bits, please let me know.
evas_image_load.c's list was updated to match the generic
loaders, in 38dd405712.
The list used by cserve should be the same. Actually, there
should be a common function instead of code duplication here.
Since cserve2 uses inotify to track image file updates,
it will drop its references to a specific file and all
the associated images.
Fix some logic in the deletion code.
Nothing extraordinary here.
Most potential crashes are extremely unlikely.
- Fix CID 1113444
- Fix CID 1113442
- Fix CID 1113441 (Logically dead code, can not be NULL)
- Fix CID 1113440: Explicit null dereferenced
This is actually an impossible situation.
Fixed by checking for nullity and printing out some error
messages instead of just crashing.
- Fix CID 1113439: Dereference after null check
Logically impossible code as both idxpath and datapath
must be either set or null at the same time.
Change the if logic to tell Coverity there's no bug.
- Fix CID 1113438 (Argument cannot be negative)
Fix wrong check of return value from shm_open.
- Fix CID 1113437 (Argument cannot be negative)
Fix wrong check of return value from shm_open.
- Fix CID 1113436 (Dereference null return value)
This case really shouldn't happen.
But the extra check does not hurt.
- Fix CID 1113435 (Dereference before null check)
Check for nullity after map open.
- Fix CID 1113434 (Extra sizeof expression)
Debug buggy debug tool :)
- Fix CID 1113433 (Uninitialized scalar variable)
Insignificant issue: only prints wrong debug logs :)
- Fix CID 1113431 (Uninitialized scalar value)
Check if (!found) only to print out logs. Not a big deal
if found was invalid.
- Fix CID 1039462 (Logically dead code)
Glyphs were previously using 3 shared buffers, now reduce to 2:
- Memory pool (mempool) containing the glyph drawable data
- Index table (Shared_Index / array) containing only the
indexes of the buffers in the mempool
- Glyph_Data table (array) containing the glyphs descriptors
AS WELL as the buffer indexes.
So, we just merge the two index tables into one by using directly
objects of type Glyph_Data for the referencing of the mempool
buffers.