for client apps that tried to be efficient with preloads to adapt
when the preloaded data is taken away from them. this allows it.
missing callback api bug fix.
SVN revision: 55745
(scalecaches) are flushed which can cause a hang until all data is
"loaded back in" again. it's a bit of a doosey actually and so isn't
fixed here.
SVN revision: 55551
Use EINA_INLIST_CONTAINER_GET instead of cast to
(Evas_Preload_Pthread_Worker*) to get eina_inlist_remove's return.
This works even if the first field of Evas_Preload_Pthread_Worker is no
longer a EINA_INLIST
Patch by: Fabiano Fidêncio <fabianofidencio@gmail.com>
SVN revision: 53826
There is no meaning in negative values for image loading, marking as
dirty or size, so image internals (cache, entry) were changed to
unsigned, reducing possible errors, particularly with overflow.
engines were converted to the new way, but any 3rd party modules will
still work as they should be using values >= 0 only anyway.
please review.
new cases introduced by "comparison between signed and unsigned" were
fixed in the modules that used cache_entry or Image_Entry dimensions.
SVN revision: 52428
As we're heading for a release we better remove as much errors as
possible and as the first step I'm removing warnings due unused
parameters, variables and functions. These tend to pollute real errors
spotted by -Wall and clang/llvm.
This does not fixes all, just the clear that could be set to
__UNUSED__, particularly to do (and I'd like some help from the
authors):
* src/lib/engines/common/evas_font_{draw,query}.c (tasn):
intl_props is just used while doing BIDI, but also used in other
#ifdef blocks :-/
* evas_map_* (raster):
huge amount of warnings, code is quite confusing and thus I'm not
touching it. I have no idea whenever the commented blocks or extra
parameters are intended to be used or no.
* src/modules/engines/fbevas_fb_main.c (raster?):
is fb_setvt() to be used? If not do you mind removing it?
* src/modules/engines/gl_{common,x11} (raster):
huge amount of warnings, code is quite nested and full of #ifdefs
that does not help to give a clear picture of what's going on.
* src/bin/evas_cserve_main.c (raster):
I could have ignored most of the errors, but is the code correct? I
mean, there is no unload of images being applied. If you confirm
none of those warnings are harmful I can flag them as unused.
* src/lib/engines/common_8 (dottedmag):
lots of unused functions that were acquired from common_16, they
are unused and if they will not, then they should be removed.
SVN revision: 52421
Assert cache->usage never becomes negative as was occurring before the
fix in r52149.
Just in time, the fix in r52149 was made by Ulisses, not me.
SVN revision: 52190
Memory usage was not accounted right because
cache->func.mem_size_get(ie) returns 0 when called after
cache->func.destructor(ie). Thus the total memory used, kept on
cache->usage, is never decremented in _evas_cache_image_remove_activ()
This implies that cache->usage will keep growing and eventually it will
overflow, becoming negative. So evas_cache_image_flush() will not do its
job because cache->limit (assumed to be positive) will not be less than
cache->usage anymore. So the total memory allocated will start to grow
and the limit won't be respected.
Strictly speaking, it's not a leak, since all the memory will be
eventually freed when evas shutdown is called, but the program might be
killed by over allocating memory. This is caught by valgrind with the
massif tool. The graphic below shows that in the end a huge memory amount
is allocated. This is the moment when cache->usage became negative.
MB
26.04^ #
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| #::::
| :::::::::::::::::@@:::::::::@:::@::::::@::::::::::::::::::::::::::#::::
| : ::: ::: ::: :::@ :: ::: : @:: @:: :::@: :: ::: : : :: :: : ::: :#::::
0 +----------------------------------------------------------------------->Gi
0 54.83
This patch is a one line fix, which swaps the calls to
_evas_cache_image_remove_activ() and cache->func.destructor() making
cache->limit to be respected.
SVN revision: 52149
Lowering priority was wrong. Some bugs:
1) You don't lower the priority by setting the scheduler policy to some
of the real-time ones (SCHED_RR or SCHER_FIFO). If you do so, you are
actually increasing the priority of the workers and your main thread
you be preempted and stalled until the workers complete their job.
Fortunately this will only happen if your programming is running as
root, as normal users (without CAP_SYS_NICE) are unable to set
priority to real-time values.
2) setpriority() and getpriority() are not part of pthread and you can't
use the id returned by pthread. Manpage explicitly says so on
pthread_self(3):
"The thread ID returned by pthread_self() is not the same thing as the
kernel thread ID returned by a call to gettid(2)."
Since glibc does not have a gettid, here we are using
syscall(SYS_gettid)
This patch was tested with the program below. Compile and run:
$ gcc p_hello2.c -o p_hello2 -lpthread
$ ./p_hello2 10
You'll see that the main thread remains with its priority and threads
created by the main thread change their own niceness.
#include <errno.h>
#include <pthread.h>
#include <sched.h>
#include <stdio.h>
#include <stdlib.h>
#include <sys/resource.h>
#include <sys/syscall.h>
#include <sys/time.h>
#include <sys/types.h>
/* Lower priority of current thread.
*
* It's used by worker threads so they use up "bg cpu" as it was really intended
* to work. If current thread is running with real-time priority, we decrease
* our priority by 5. This is done in a portable way. Otherwise we are
* running with SCHED_OTHER policy and there's no portable way to set the nice
* level on current thread. In Linux, it does work and it's the only one that is
* implemented.
*/
static void
_ecore_thread_pri_drop(void)
{
struct sched_param param;
int pol, prio, ret;
pid_t tid;
pthread_t pthread_id;
pthread_id = pthread_self();
ret = pthread_getschedparam(pthread_id, &pol, ¶m);
if (ret)
{
fprintf(stderr, "Unable to query sched parameters\n");
return;
}
if (pol == SCHED_RR || pol == SCHED_FIFO)
{
prio = sched_get_priority_max(pol);
param.sched_priority += 5;
if (prio > 0 && param.sched_priority > prio)
param.sched_priority = prio;
pthread_setschedparam(pthread_id, pol, ¶m);
}
#ifdef __linux__
else
{
tid = syscall(SYS_gettid);
errno = 0;
prio = getpriority(PRIO_PROCESS, tid);
if (errno == 0)
{
prio += 5;
if (prio > 19)
prio = 19;
setpriority(PRIO_PROCESS, tid, prio);
}
}
#endif
}
/*
* p_hello.c -- a hello program (in pthread)
*/
#define MAX_THREAD 1000
typedef struct {
int id;
} parm;
void *hello(void *arg)
{
parm *p=(parm *)arg;
pid_t tid;
int prio;
tid = syscall(SYS_gettid);
printf("[%d] Hello from node %d\n", tid, p->id);
pthread_yield();
printf("[%d] HELLO!\n", tid);
_ecore_thread_pri_drop();
prio = getpriority(PRIO_PROCESS, tid);
printf("[%d] New nice value: %d\n", tid, prio);
return (NULL);
}
int main(int argc, char* argv[]) {
int n,i;
pthread_t *threads;
pthread_attr_t pthread_custom_attr;
parm *p;
pid_t tid;
int prio;
if (argc != 2)
{
printf ("Usage: %s n\n where n is no. of threads\n",argv[0]);
exit(1);
}
n=atoi(argv[1]);
if ((n < 1) || (n > MAX_THREAD)) {
printf ("The no of thread should between 1 and %d.\n",MAX_THREAD);
exit(1);
}
threads = (pthread_t *)malloc(n * sizeof(*threads));
pthread_attr_init(&pthread_custom_attr);
p = (parm *)malloc(n * sizeof(parm));
/* Start up thread */
tid = syscall(SYS_gettid);
for (i=0; i<n; i++) {
prio = getpriority(PRIO_PROCESS, tid);
printf("[%d] root thread nice value: %d\n", tid, prio);
p[i].id=i;
pthread_create(&threads[i], &pthread_custom_attr, hello, (void *)(p+i));
}
/* Synchronize the completion of each thread. */
for (i=0; i<n; i++) {
pthread_join(threads[i],NULL);
}
free(p);
return 0;
}
SVN revision: 52040
The offending projects were:
E16/e/src/backgrounds.c | 10 ++++------
PROTO/eon/src/lib/layout/eon_stack.c | 4 +---
ecore/src/lib/ecore_win32/ecore_win32.c | 3 +--
ecore/src/lib/ecore_wince/ecore_wince.c | 3 +--
edje/src/lib/edje_edit.c | 3 +--
evas/src/lib/cache/evas_cache_image.c | 2 +-
exalt/src/lib/libexalt_private.c | 2 +-
This patch assumes code in these places were insane and the fix is to remove
one condition check. Most likely this is not true, but there's no automatic fix
for that.
Looking at the patch, it seems that some places should use "x" and "y" vars but
used just one of them and therefore they were caught by coccinelle.
SVN revision: 51666
Revert previous patch generated by badnull.cocci script, and apply the new one.
The main difference is that assert and assert-like functions are not touched
anymore.
SVN revision: 51650
Apply badzero.cocci, badnull.coci and badnull2.cocci
This should convert all cases where there's a comparison to NULL to simpler
forms. This patch applies the following transformations:
code before patch ||code after patch
===============================================================
return a == NULL; return !a;
return a != NULL; return !!a;
func(a == NULL); func(!a);
func(a != NULL); func(!!a);
b = a == NULL; b = !a;
b = a != NULL; b = !!a;
b = a == NULL ? c : d; b = !a ? c : d;
b = a != NULL ? c : d; b = a ? c : d;
other cases:
a == NULL !a
a != NULL a
SVN revision: 51487
* Remove vim modelines:
find . -name '*.[chx]' -exec sed -i '/\/\*$/ {N;N;/ \* vim:ts/d}' \{\} \;
find . -name '*.[chx]' -exec sed -i '/\/[\*\/] *vim:/d' \{\} \;
* Remove leading blank lines:
find . -name '*.[cxh]' -exec sed -i '/./,$!d'
If you use vim, use this in your .vimrc:
set ts=8 sw=3 sts=8 expandtab cino=>5n-3f0^-2{2(0W1st0
SVN revision: 50816
Ouch, that was nasty! src/lib/cache/evas_cache_image.c was assuming
all Image_Entry were RGBA_Image (why?!?!?), thus doing the cast and
having mutexes and other to operate on garbage (possibly crashing due
segv). This happened to be the case with Soft16_Image used by
software_16 engines.
I'm not sure, but this may fix problems that people noticed with
async-render hanging their systems even if not explicitly enabled
during runtime.
I also found it quite strange the number of locks required by this
code! Clearly we could use macros to simplify and avoid bugs, but
maybe some atomic_inc/dec code should be used to remove half of the
mutexes in that code?
/me wonders what kinds of bug more we can expect from this code :-/
SVN revision: 50300
The notnull.cocci script from Coccinelle finds places where you check if a
variable is NULL, but it's known not to be NULL. The check can be safely
removed. For example, this code would be caught by notnull:
if (!var) return;
if (var && var->fld) { ... }
It's needless to check again if var is not NULL because if it's in fact NULL,
it would have returned on the previous "if". This commit removes all the
trivial places where this pattern happens. Another patch will be generated for
the more complex cases.
SVN revision: 50241
rendering. to turn on:
1.
configure with --enable-async-render
2.
export EVAS_RENDER_MODE=non-blocking
presto. necessitates some api swizzling (thus the expedite. ecore etc. changes)
the kind of results you get on a desktop:
http://www.rasterman.com/files/evas-async-vs-none.html
SVN revision: 49087
Evas image load was always reporint "generic" error, since it was
disconnected from actual loader modules.
This commit will break the module loader API (as it's restricted to
inside Evas, this should be no problem). The return was turned into
"Eina_Bool" for clarity, while an extra "int *error" is responsible to
report errors. This approach was choosen to force compiler warnings
and to try avoid mistakes as EINA_FALSE == EVAS_LOAD_ERROR_NONE and
thus we'd get opposite behavior if something slips.
Most loaders play well, except by eet that does not provide means to
know if the file open failed due missing file, incorrect format or
corrupted file :-(
Please report any issues. I added eina_log debugging to loader
functions, just run your Evas application as:
EINA_LOG_LEVELS=evas_main:4 your_app
SVN revision: 44666