platform/upstream/efl.git
7 years agotests: eina_file: check if the fd is really 0 when we set close file to true
Stefan Schmidt [Mon, 13 Feb 2017 13:20:17 +0000 (14:20 +0100)]
tests: eina_file: check if the fd is really 0 when we set close file to true

The return value from create_file_not_empty was saved but never checked if it
really is 0. Whcih should be the case if we call the function with file_close
true.

7 years agotests: eina_file: fix coding style in file_unlink test case
Stefan Schmidt [Mon, 13 Feb 2017 13:12:38 +0000 (14:12 +0100)]
tests: eina_file: fix coding style in file_unlink test case

7 years agotests: eina_file: remove unused variable
Stefan Schmidt [Mon, 13 Feb 2017 13:10:03 +0000 (14:10 +0100)]
tests: eina_file: remove unused variable

7 years agoelm_index: Skip box auto fill when index size is invalid.
Woochan Lee [Mon, 13 Feb 2017 11:21:52 +0000 (20:21 +0900)]
elm_index: Skip box auto fill when index size is invalid.

Summary:
_index_box_auto_fill will be called when user calling elm_index_level_go(), elm_index_omit_enabled_set().

item size re-calculated and each item style reset even index size is invalid.
It can make performance issue during index creation time.

Test Plan:
elementary_test
index sample

Reviewers: taxi2se, jpeg, cedric

Subscribers: cedric, jpeg

Differential Revision: https://phab.enlightenment.org/D4653

7 years agoevas gl: Partially fix MSAA for GLES 3 drivers
Jean-Philippe Andre [Mon, 13 Feb 2017 08:42:10 +0000 (17:42 +0900)]
evas gl: Partially fix MSAA for GLES 3 drivers

MSAA (multisampled_render_to_texture) support was implemented with
GLES 2 in mind, but for GLES 3 it comes as a core feature, not as
an extension. Also it relies on renderbuffers, not textures. GL
is awesome.

Note: MSAA still doesn't work!

7 years agowin: Make ELM_ACCEL work with msaa, depth & stencil
Jean-Philippe Andre [Mon, 13 Feb 2017 08:11:56 +0000 (17:11 +0900)]
win: Make ELM_ACCEL work with msaa, depth & stencil

This makes the env var override and the elementary config
global override on accel preference work for not only "gl"
but also advanced configs such as "gl:msaa_high:depth:stencil"

7 years agodoxygen: Fix typos & invalid statements in ecore
Myoungwoon Roy, Kim [Mon, 13 Feb 2017 06:29:11 +0000 (15:29 +0900)]
doxygen: Fix typos & invalid statements in ecore

This covers ecore audio, avahi, buffer and con.

Summary: There are some typos and cacologigue statements in
doxygen of ecore_audio, ecore_avahi, ecore_buffer, and ecore_con.

Test Plan: API Doxygen Revision

Reviewers: stefan, cedric, raster, Jaehyun_Cho

Subscribers: jpeg, conr2d

Differential Revision: https://phab.enlightenment.org/D4652

7 years agodoxygen: Fix typos & invalid statements in Ecore_Getopt and Ecore_Legacy
Myoungwoon Roy, Kim [Mon, 13 Feb 2017 06:29:01 +0000 (15:29 +0900)]
doxygen: Fix typos & invalid statements in Ecore_Getopt and Ecore_Legacy

Summary: There are some typos and calogique statements in doxygen
of Ecore_Getopt and Ecore_Legacy so I had fixed typos and cacologique statements.

Test Plan: Doxygen revision

Reviewers: stefan, cedric, raster, Jaehyun_Cho

Subscribers: jpeg, conr2d

Differential Revision: https://phab.enlightenment.org/D4651

7 years agoEvasGL: For depth&stencil&msaa, use renderbuffer instead of texture
Minkyoung Kim [Mon, 13 Feb 2017 06:11:56 +0000 (15:11 +0900)]
EvasGL: For depth&stencil&msaa, use renderbuffer instead of texture

Summary:
glFramebufferTexture2DMultisampleEXT() supports color attachment only.
so depth&stencil attachment should be bound via renderbuffer not texture.

According to http://www.khronos.org/registry/OpenGL/extensions/EXT/EXT_multisampled_render_to_texture.txt
In order to allow the use of multisampled depth and stencil buffers
when performing multisampled rendering to a texture, the extension
also adds the command RenderbufferStorageMultisampleEXT.
target, textarget, texture, and level correspond to the same
parameters for FramebufferTexture2D and have the same restrictions.
attachment must be COLOR_ATTACHMENT0.

Test Plan: Evas GL on GLES 3

Reviewers: jpeg, raster, dkdk, wonsik

Tags: #efl

Differential Revision: https://phab.enlightenment.org/D4648

7 years agoefl net proxy helper - use infinite for and sleeps instead of pause
Carsten Haitzler (Rasterman) [Mon, 13 Feb 2017 10:00:47 +0000 (19:00 +0900)]
efl net proxy helper - use infinite for and sleeps instead of pause

so the process was waiting to be killed (waiting forever) and i used
pause() but this isnt windows-friendly, so since this process expects
to be killed by the parent process, then use a for (;;) with lon long
long sleeps... instead.

this shoud fix T5174

7 years agoconfigure - eolian seemingly still needs evil due to eina
Carsten Haitzler (Rasterman) [Mon, 13 Feb 2017 09:54:43 +0000 (18:54 +0900)]
configure - eolian seemingly still needs evil due to eina

this should fix T5172

7 years agovpath file - add a keey property to know if obj should be kept around
Carsten Haitzler (Rasterman) [Mon, 13 Feb 2017 06:29:09 +0000 (15:29 +0900)]
vpath file - add a keey property to know if obj should be kept around

if its a file downloading (to a tmp location) or a zip/tar/whatever
file being extracted also to a tmp location and that tmp file needs to
be removed after ...you need to keep the obj around to know when to
delete the file. this makes the keeping optional and you query if the
vpath obj is meant to be kept or not. if it's not it's safe to del
immediately.

this cuts down 1 obj per image obj/edje obj when generally unused.
save some mem.

7 years agoevas: fix bugs in gif image loader
Jaeun Choi [Fri, 10 Feb 2017 07:56:50 +0000 (16:56 +0900)]
evas: fix bugs in gif image loader

gif's logical screen size (which is considered the image size)
might be different from the size of each frame.
when decoding a frame, the width and height of the decoded data should be
based on the size of the frame, not on the size of the logical screen size.
if a frame is decoded into a buffer of screen size, this might happen

(frame = 6 X 3, logical screen = 5 X 3)
OOOXXX      OOOXX
OOOXXX  =>  XOOOX
OOOXXX      XXOOO

@fix

7 years agoevas gl common - make render op explicitly signed as we set it to -1
Carsten Haitzler (Rasterman) [Sun, 12 Feb 2017 06:48:51 +0000 (15:48 +0900)]
evas gl common - make render op explicitly signed as we set it to -1

this just is cleaner and now also matches types for render_op in gl
common context.

7 years agoeina inlist - cast container via void ptr to avoid warnings
Carsten Haitzler (Rasterman) [Sun, 12 Feb 2017 05:45:04 +0000 (14:45 +0900)]
eina inlist - cast container via void ptr to avoid warnings

we really can't do much here but our direct casting causes warnings in
apps or anyone using this macro so keep things silent as our pointer
tricks are actually ok but the compiler can't figure it out.

7 years agoevas_table: abort when there is already the option struct
Marcel Hollerbach [Sat, 11 Feb 2017 22:20:22 +0000 (23:20 +0100)]
evas_table: abort when there is already the option struct

otherwise we would just reuse that struct. This results in
freeing/writing/reading the memory twice.

This error message should give the dev a tip on what went wrong, instead
of leaving cryptic valgrind messages and crashes.

This fixes the sysinfo desktop gadget in enlightenment.

ref T5173

7 years agoevas: fix child position when the container is moved
Marcel Hollerbach [Sat, 11 Feb 2017 22:16:25 +0000 (23:16 +0100)]
evas: fix child position when the container is moved

when the position of the container is changed the children should also be
repositioned, so setting the changed flag.

The visual effect where you saw that was in luncher where items of the
bar did not get fully up.

7 years agoevlog/log bt - fix size casting tot stop warning when printing errs
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 15:29:01 +0000 (00:29 +0900)]
evlog/log bt - fix size casting tot stop warning when printing errs

7 years agoevas color copy neon asm - declare tmp as output not input for warning--
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 14:08:28 +0000 (23:08 +0900)]
evas color copy neon asm - declare tmp as output not input for warning--

less warnings - use output not input for tmp var. much better.

7 years agoeo base - fix warnings for debug logs to get format string types happy
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 13:53:04 +0000 (22:53 +0900)]
eo base - fix warnings for debug logs to get format string types happy

gcc is very unhappy with these log prints - specifically on arm 32bit.
this fixes it so we can focus on real warnings/issues.

7 years agoevas generic loader ps - fix warning about printf format mismatch type
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 13:41:22 +0000 (22:41 +0900)]
evas generic loader ps - fix warning about printf format mismatch type

it's right. types are uint but format exects unsigned long. fix.

7 years agoevas wayland egl engine - fix warning about struct wl_resource not used
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 13:29:26 +0000 (22:29 +0900)]
evas wayland egl engine - fix warning about struct wl_resource not used

this struct is not actually used other than in this func ptr decl and
it's warning-land for gcc, so .. make it void * and do things like
pretty much all other gl extn func ptrs... be very generic

7 years agoevas gl drm engine - fix warning about struct wl_resource not used
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 13:27:21 +0000 (22:27 +0900)]
evas gl drm engine - fix warning about struct wl_resource not used

this struct is not actually used other than in this func ptr decl and
it's warning-land for gcc, so .. make it void * and do things like
pretty much all other gl extn func ptrs... be very generic

7 years agoevas smooth scaler - fix warning where indenting is misleading
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 13:21:48 +0000 (22:21 +0900)]
evas smooth scaler - fix warning where indenting is misleading

indeed the indenting is misleading. fix.

7 years agoevas blit main - put mmx funcs in ifdefs as they are not used on arm
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 04:25:25 +0000 (13:25 +0900)]
evas blit main - put mmx funcs in ifdefs as they are not used on arm

fixes warning

7 years agoevas draw func get - make pixels as unused as it is in neon asm cases
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 04:22:40 +0000 (13:22 +0900)]
evas draw func get - make pixels as unused as it is in neon asm cases

also a little shitespace cleanup and correct ()ing of ops to clean
this function up a bit.

7 years agoevas convert color - neon asm - fix warning by casting - no side effects
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 04:21:30 +0000 (13:21 +0900)]
evas convert color - neon asm - fix warning by casting - no side effects

it's a warning one way or another so reduce noise with a harmless case
as passing in a pit ro a 32bit type is more restrictive than the ptr
it accepts (an 8bit type)

7 years agoefl base class - fix warning about comparing differing sizes
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 04:13:19 +0000 (13:13 +0900)]
efl base class - fix warning about comparing differing sizes

yes - we compare a difference between 2 ptrs and an index which is a
uint. the safe thing here is to promote the unit to the ptrdiff_t
type. reality is we cant have more than 2^32 cb's on an object
anyway... so this should be ok.

7 years agoefl io - fix using wrong type for return values pointed to
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 04:06:16 +0000 (13:06 +0900)]
efl io - fix using wrong type for return values pointed to

this potentially could be a bug on 32bit systems. this fixes that and
addresses the warning that pointed it out

7 years agoecore pipe - fix warning of comparing integers of differing sizes
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 04:05:44 +0000 (13:05 +0900)]
ecore pipe - fix warning of comparing integers of differing sizes

7 years agoelm code - silence possibly uninitialied compiler warnings
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 01:39:42 +0000 (10:39 +0900)]
elm code - silence possibly uninitialied compiler warnings

every time i build efl i see all these warnigns spew by about elm_code
and possibly uninited vars being used. yes - they're filled in via ptr
ref bya  func but gcc definitely seems to think they MIGHT not be
filled in in some cases... and either way tyhe cost of setting to 0 is
minimal and less noise is better so we see real warnings more
easily... :)

7 years agoevas wl shm - fix warning with unused variable
Carsten Haitzler (Rasterman) [Sat, 11 Feb 2017 01:24:42 +0000 (10:24 +0900)]
evas wl shm - fix warning with unused variable

7 years agoevas: use the right structure to not over allocate memory.
Cedric BAIL [Fri, 10 Feb 2017 21:40:44 +0000 (13:40 -0800)]
evas: use the right structure to not over allocate memory.

Eina_Trash is designed for storing cached pointer without any memory
consumption. Please be careful with EFL memory consumption.

7 years agoecore-wl2: roundtrip during client disconnect
Mike Blumenkrantz [Fri, 10 Feb 2017 22:23:36 +0000 (17:23 -0500)]
ecore-wl2: roundtrip during client disconnect

ensure all pending requests are processed by the compositor and fixes
session recovery destroy

@fix

7 years agoelm_code: Fix syntax crash on trailing newline in multiline macro
Andy Williams [Fri, 10 Feb 2017 20:16:31 +0000 (20:16 +0000)]
elm_code: Fix syntax crash on trailing newline in multiline macro

@fix

7 years agoecore: Don't call _ecore_fd_valid() in release builds
Derek Foreman [Fri, 10 Feb 2017 17:58:24 +0000 (11:58 -0600)]
ecore: Don't call _ecore_fd_valid() in release builds

Not sure _ecore_fd_valid() is all that useful anymore, as the
commit that introduced it said it would be removed "before release"
a long time ago - it's a debug assist that probably doesn't need
to be in release builds.

(I'm counting syscalls on rpi3 - still, calling this an optimization
seems like a bit of a stretch.)

7 years agoSysinfo gadget themes: Account for vertical theme.
Stephen 'Okra' Houston [Fri, 10 Feb 2017 17:58:46 +0000 (11:58 -0600)]
Sysinfo gadget themes: Account for vertical theme.

7 years agoefl_object: document when callbacks are called
Marcel Hollerbach [Fri, 10 Feb 2017 17:49:36 +0000 (18:49 +0100)]
efl_object: document when callbacks are called

7 years agoecore_wl2: Update compositor bind now that damage_buffer is released
Derek Foreman [Fri, 10 Feb 2017 16:25:53 +0000 (10:25 -0600)]
ecore_wl2: Update compositor bind now that damage_buffer is released

We had to do some weird special case stuff when binding wl_compositor
because we implemented damage buffer before depending on a wayland
release that provided it.

Now our wayland dep is newer than damage buffer, so the test can be
more conventional.

Should have no functional change.

7 years agoeolian: silence static analyzer null check (CID 1369018)
Daniel Kolesa [Fri, 10 Feb 2017 15:56:35 +0000 (16:56 +0100)]
eolian: silence static analyzer null check (CID 1369018)

7 years agoeolian: fix memory leak in doc tokens (CID 1367505)
Daniel Kolesa [Fri, 10 Feb 2017 15:54:19 +0000 (16:54 +0100)]
eolian: fix memory leak in doc tokens (CID 1367505)

7 years agovpath usage - simplify to bare minimum to make gustavo happy
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 23:58:29 +0000 (08:58 +0900)]
vpath usage - simplify to bare minimum to make gustavo happy

since these are only local path resolves, the do and wait are
technically not needed. also remove any other tmp strings
and use the vpath string resolving feature to avoid
printfs/strjoins/cats etc. etc. as well.

also remember to remove old name string from evas shm code - it worked
for me. i guess i was lucky and it happened to be NULL thus free was
fine.

7 years agobuild: add mapping between ecore_file and ecore-file for pkgconfig files
Stefan Schmidt [Fri, 10 Feb 2017 09:55:19 +0000 (10:55 +0100)]
build: add mapping between ecore_file and ecore-file for pkgconfig files

elua.pc was depending on a ecore_file pc file it is named ecore-file. We have
mapping for similar cases. This one was just forgotten it seems.

Fixes T5075

7 years agosrc/utils: remove no longer used gitigonre and directories
Stefan Schmidt [Fri, 10 Feb 2017 09:02:23 +0000 (10:02 +0100)]
src/utils: remove no longer used gitigonre and directories

When the XCB support was removed end of last year the .gitignore file left
behind. It was the only file still sitting in the src/utils tree so git removes
the whoile subtree now.

7 years agonet_proxy_helper: fix typo in dependencies
Romain Naour [Thu, 9 Feb 2017 22:42:14 +0000 (23:42 +0100)]
net_proxy_helper: fix typo in dependencies

Fixes:
error: cannot find the library 'lib/eina/libeina.la' or unhandled argument 'lib/eina/libeina.la'

Signed-off-by: Romain Naour <romain.naour@gmail.com>
7 years agoeina_modinfo: add missing dependency on libeina.la
Romain Naour [Thu, 9 Feb 2017 22:14:39 +0000 (23:14 +0100)]
eina_modinfo: add missing dependency on libeina.la

Fixes:
libtool:   error: cannot find the library 'lib/eina/libeina.la' or unhandled
argument 'lib/eina/libeina.la'

Signed-off-by: Romain Naour <romain.naour@gmail.com>
7 years agoelementary-cxx: Make C++ test compile a not used archive so linking can be avoided
Felipe Magno de Almeida [Thu, 9 Feb 2017 23:24:33 +0000 (21:24 -0200)]
elementary-cxx: Make C++ test compile a not used archive so linking can be avoided

Fixes linking errors for some older combos of C++ compiler and linker.

T2838

7 years agorelease: Update NEWS and bump version for 1.19.0-alpha1 release
Stefan Schmidt [Thu, 9 Feb 2017 15:44:06 +0000 (16:44 +0100)]
release: Update NEWS and bump version for 1.19.0-alpha1 release

7 years agoRevert "vpath usage - simplify to bare minimum to make gustavo happy"
Derek Foreman [Thu, 9 Feb 2017 14:58:03 +0000 (08:58 -0600)]
Revert "vpath usage - simplify to bare minimum to make gustavo happy"

This reverts commit 2037474dc0fd2b360452f2a15abcbe533b57ca37.

This causes the wayland_shm engine to seg fault immediately at
startup when attempting to create shm buffers.

Please make sure when committing changes to the wayland_shm engine to
test on intel, exynos, and at least one platform without dmabuf
capabilities - or using the EVAS_WAYLAND_SHM_DISABLE_DMABUF env var
to disable dmabuf on intel or exynos.

Anyone without the time or hardware to fully test changes to wayland_shm
can submit a patch to phabricator and assign it to me so I can fully
test it before landing.

7 years agodocs: avoid possible duplicates when figuring out all impls of a func
Daniel Kolesa [Thu, 9 Feb 2017 15:04:59 +0000 (16:04 +0100)]
docs: avoid possible duplicates when figuring out all impls of a func

7 years agodocs: list all existing implements of a method/property
Daniel Kolesa [Thu, 9 Feb 2017 14:58:22 +0000 (15:58 +0100)]
docs: list all existing implements of a method/property

7 years agodocs: reverse inheritance hierarchy api in doctree
Daniel Kolesa [Thu, 9 Feb 2017 14:25:10 +0000 (15:25 +0100)]
docs: reverse inheritance hierarchy api in doctree

7 years agoelm: examples: add new layout_model_connect binary to gitignore
Stefan Schmidt [Thu, 9 Feb 2017 13:46:35 +0000 (14:46 +0100)]
elm: examples: add new layout_model_connect binary to gitignore

7 years agodocs: eina: add doc for new eina value type
Stefan Schmidt [Thu, 9 Feb 2017 13:44:51 +0000 (14:44 +0100)]
docs: eina: add doc for new eina value type

7 years agodocs: elm: document new efl UI image factory class
Stefan Schmidt [Thu, 9 Feb 2017 13:44:22 +0000 (14:44 +0100)]
docs: elm: document new efl UI image factory class

7 years agodocs: efl mvc: document all new model view controller related eo files
Stefan Schmidt [Thu, 9 Feb 2017 13:43:37 +0000 (14:43 +0100)]
docs: efl mvc: document all new model view controller related eo files

7 years agodocs: ecore: document newly added efl model composite classes
Stefan Schmidt [Thu, 9 Feb 2017 13:40:50 +0000 (14:40 +0100)]
docs: ecore: document newly added efl model composite classes

7 years agodocs: efl filters: document all new data structures for efl filters
Stefan Schmidt [Thu, 9 Feb 2017 13:38:10 +0000 (14:38 +0100)]
docs: efl filters: document all new data structures for efl filters

7 years agovpath usage - simplify to bare minimum to make gustavo happy
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 13:06:16 +0000 (22:06 +0900)]
vpath usage - simplify to bare minimum to make gustavo happy

since these are only local path resolves, the do and wait are
technically not needed. also remove any other tmp strings and use the
vpath string resolving feature to avoid printfs/strjoins/cats etc.
etc. as well.

7 years agoefl vpath - for runtime dir dont fall back to tmp but instead abort
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 10:52:53 +0000 (19:52 +0900)]
efl vpath - for runtime dir dont fall back to tmp but instead abort

if we can't create a runtimedir maybe its best to abort. also ensure
it is created as the effective user id.

7 years agoefreetd - remove unused var
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 10:07:09 +0000 (19:07 +0900)]
efreetd - remove unused var

7 years agoeina utils and vpath - use euid not uid if setuid for correct user
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 09:50:15 +0000 (18:50 +0900)]
eina utils and vpath - use euid not uid if setuid for correct user

was using uid not euid for "if setuid use safe way to get location" -
ie the user that effectively owns the process, not nominally.

@fix

7 years agoelm config - handle setuid apps correctly and dont create junk dirs in .
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 09:24:43 +0000 (18:24 +0900)]
elm config - handle setuid apps correctly and dont create junk dirs in .

this fixes a bug where elm_cofnig would firstly use
~/.config/elementary which isnt what we have used before and so would
break older config. also it'd create junk if it was setuid. fix this.

@fix

7 years agoevas fb dev env var - allow in setuid processes with sanitizing
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 08:03:49 +0000 (17:03 +0900)]
evas fb dev env var - allow in setuid processes with sanitizing

this allows only /dev/fb[0-0] or /dev/fb/something where somthing does
not begin with a . - thus no way to break out of the fb subdir... so
it should be ok... this keeps setuid safety and allows this env var to
work now as intended in this situation.

7 years agoefreetd cache - use eina home env get to centralize homedir fetching
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 07:16:40 +0000 (16:16 +0900)]
efreetd cache - use eina home env get to centralize homedir fetching

this cleans up and improves a getenv that should really use eina

7 years agoefreetd - dont use XDG_RUNTIME_DIR or EFREET_LOG env vars for setuid
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 06:48:11 +0000 (15:48 +0900)]
efreetd - dont use XDG_RUNTIME_DIR or EFREET_LOG env vars for setuid

if a setuid app needs efreet - it will not be able to contact the
users' efreetd and thus may spawn it's own... and thus we cant have
this now spawned efreetd using env vars inherited from the
unpriveleged etc. user, so ignore them.

7 years agoefreet - use vpath to get XDG_RUNTIME_DIR so we have a single impl
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 06:38:34 +0000 (15:38 +0900)]
efreet - use vpath to get XDG_RUNTIME_DIR so we have a single impl

having too many places that get XDG_RUNTIME_DIR makes it harder to
keep secure etc, so make it a single location for efreet too.

7 years agoevas wayland shm engine - use vpath for getting xdg runtime dir
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 02:27:48 +0000 (11:27 +0900)]
evas wayland shm engine - use vpath for getting xdg runtime dir

this should make the wayland shm engine safe to use from setuid
processes. it shouldnt depend on getenv of specific file paths if
setuid. vpatrh takes care of this in one central place

7 years agoecore buffer module - use vpath for xdg runtime for setuid safety
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 02:24:52 +0000 (11:24 +0900)]
ecore buffer module - use vpath for xdg runtime for setuid safety

this should make ecore buffer module safe to use in setuid processes

7 years agoevas cserve - remove unused if 0'd out code that is misleading my greps
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 01:23:20 +0000 (10:23 +0900)]
evas cserve - remove unused if 0'd out code that is misleading my greps

looking for XDG_RUNTIME_DIR usage to lock down for setuid apps... this
msled me... remove it as it's unused.

7 years agoelm widget item: add guide about when to use item_signal_emit
Amitesh Singh [Thu, 9 Feb 2017 07:36:42 +0000 (13:06 +0530)]
elm widget item: add guide about when to use item_signal_emit

genlist/gengrid now uses cache for view objects of items and
re-associate edje objects to different obj items whenever needed.
problem here is if app user changes the state of view object
(by sending edje signal, elm_object_item_signal_emit()) and
forget to reset it in unrealized, the old edje object state will link to other
object item, so its application user responsibility to reset
the edje state to normal in "unrealized" callback.

Signed-off-by: Amitesh Singh <amitesh.sh@samsung.com>
7 years agoevas: Fix invalid current state (invalid bool value)
Jean-Philippe Andre [Thu, 9 Feb 2017 02:34:29 +0000 (11:34 +0900)]
evas: Fix invalid current state (invalid bool value)

newframe() resets some of the gl context properties, so match them
to our shadow copy in the gc state.

target_surface_set() also resets some of those properties but only
in our shadow copy, not in the actual GL context. As a consequence
we can't rely on setting a bool to 0 or 1 unless we also call the
equivalent GL function. Expand bitfields to 2 bits allows us to
set a bool to the invalid value of -1 (yes, that's not a real bool).

Also there is no need to reset the target surface to NULL during
newframe. It will be reset during target_surface_set.

This fixes some issues I encounter while working on GL filters.

7 years agoelput - use vpath to get xdg runtime to also be setuid safe
Carsten Haitzler (Rasterman) [Thu, 9 Feb 2017 00:27:41 +0000 (09:27 +0900)]
elput - use vpath to get xdg runtime to also be setuid safe

this time around... use a / at the end of the (:run:) virtual dir
path to make it work.

7 years agodoxygen: fix typos, cacologique statements in API reference doxygen of Ecore_Common
Myoungwoon Roy, Kim [Wed, 8 Feb 2017 19:09:15 +0000 (11:09 -0800)]
doxygen: fix typos, cacologique statements in API reference doxygen of Ecore_Common

Summary: There are some typos and calogique statements in doxygen of Ecore_Common so I had fixed typos, cacologique statements.

Test Plan: doxygen revision

Reviewers: stefan, cedric, raster, Jaehyun_Cho

Subscribers: conr2d, jpeg

Differential Revision: https://phab.enlightenment.org/D4650

Signed-off-by: Cedric BAIL <cedric@osg.samsung.com>
7 years agoecore-wl2: Support sending focus for wl_shell
Chris Michael [Wed, 8 Feb 2017 17:30:10 +0000 (12:30 -0500)]
ecore-wl2: Support sending focus for wl_shell

wl_shell does not use a surface state for activation, so we still need
to send the focus in/out for that based on keyboard enter/leave.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
7 years agoecore-wl2: Don't send focus in/out events based on keyboard enter/leave
Chris Michael [Wed, 8 Feb 2017 16:51:21 +0000 (11:51 -0500)]
ecore-wl2: Don't send focus in/out events based on keyboard enter/leave

We should not be sending focus events based on keyboard behaviour, but
rather send them according to xdg shell activate status. This makes
our focus behaviour more "standards" compliant.

@fix

Signed-off-by: Chris Michael <cp.michael@samsung.com>
7 years agoRevert "elput - use vpath to get xdg runtime to also be setuid safe"
Derek Foreman [Wed, 8 Feb 2017 15:37:49 +0000 (09:37 -0600)]
Revert "elput - use vpath to get xdg runtime to also be setuid safe"

This reverts commit 24e34e19a1db84cdcb6241207cb99d14ca83c41b.

This broke keyboard input for the enlightenment wayland compositor,
please test elput changes on at least one of the drm backends,
preferably with enlightenment.

The wayland compositor is hard enough to keep stable due to breakage
from core changes only tested on X - but elput's main user is our
wayland compositor, was this tested anywhere?

7 years agoRevert "elput: Minor formatting fix"
Derek Foreman [Wed, 8 Feb 2017 15:36:46 +0000 (09:36 -0600)]
Revert "elput: Minor formatting fix"

This reverts commit 95d93cb7208857cc89d79f3ffbbd7ee71ba1936f.

Formatting change on top of a totally untested coverity fix that
breaks input for enlightenment's wayland compositor.

 #TeamworkIsDeprecated

7 years agoGive Batman it's own edc.
Stephen 'Okra' Houston [Wed, 8 Feb 2017 14:54:44 +0000 (08:54 -0600)]
Give Batman it's own edc.

7 years agowhen resetting tex_target, set it to -1 so we force a reset
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 13:59:37 +0000 (22:59 +0900)]
when resetting tex_target, set it to -1 so we force a reset

7 years agoevas gl commonnon - context - expand fields to 2 bits for invalid vals
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 13:57:28 +0000 (22:57 +0900)]
evas gl commonnon - context - expand fields to 2 bits for invalid vals

we use invalid values (-1 or 0xff or 0x3 etc. invalid vals to know to reset
some state). expand fields out in size a little bit to allow that to
work again.

7 years agoelput: Minor formatting fix
Chris Michael [Wed, 8 Feb 2017 13:29:24 +0000 (08:29 -0500)]
elput: Minor formatting fix

NB: No functional changes, just some minor formatting adjustments

Signed-off-by: Chris Michael <cp.michael@samsung.com>
7 years agoevas yuv convert - fix unused increments
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 12:07:21 +0000 (21:07 +0900)]
evas yuv convert - fix unused increments

this removes useless yp1 += and yop2 += as they are SET int he loop.
this fixes 1367510 and 1367511

7 years agoevas mask filter - remove unused increment of msk as its set in-loop
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 12:04:46 +0000 (21:04 +0900)]
evas mask filter - remove unused increment of msk as its set in-loop

msk is set inside the loop body, so msk += is pointless.

fix CID 1367512

7 years agoeina debug monitor - dont use XDG_RUNTIME_DIR if setuid
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 11:15:10 +0000 (20:15 +0900)]
eina debug monitor - dont use XDG_RUNTIME_DIR if setuid

avoids possible issues with debug monitor + setuid

7 years agoecore_con - only use XDG_RUNTIME_DIR if not setuid
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 10:15:19 +0000 (19:15 +0900)]
ecore_con - only use XDG_RUNTIME_DIR if not setuid

this should address possible misuse of this env var in setuid apps.
keep using home and tmp to maintain socket "abi" (the filenames that
are used).

7 years agoevas cserver client - remove dead if 0'd code
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 10:12:51 +0000 (19:12 +0900)]
evas cserver client - remove dead if 0'd code

7 years agoelput - use vpath to get xdg runtime to also be setuid safe
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 10:10:44 +0000 (19:10 +0900)]
elput - use vpath to get xdg runtime to also be setuid safe

this fixes CID 1357168 in the case of setuid binaries as vpath handles
setuid cases.

7 years agoevas gl generic - remove dead code else case
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 09:43:29 +0000 (18:43 +0900)]
evas gl generic - remove dead code else case

plane limits are checked at top of function so no need to check in the
middle again. it's dead code. fixes CID 1362729

7 years agoevas gl generic - comment switch fallthrough as intended
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 09:38:42 +0000 (18:38 +0900)]
evas gl generic - comment switch fallthrough as intended

this is intended. document as such.

7 years agoefl vpath - be paranoid about setuid execution and dont use env vars
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 09:11:07 +0000 (18:11 +0900)]
efl vpath - be paranoid about setuid execution and dont use env vars

if setuid, do not trust env vars for any xdg stuff at all.

7 years agoevas callbacks - add comment to note fallthrough is intended
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 08:57:35 +0000 (17:57 +0900)]
evas callbacks - add comment to note fallthrough is intended

swtich fallthrough oftenis not intended but in this case it is, so
documente it as such.

7 years agoelm config - forbid bakcslash in profile name too
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 08:51:52 +0000 (17:51 +0900)]
elm config - forbid bakcslash in profile name too

/ already is forbidden but \ might be valid on windows, so forbid it
too.

7 years agoelm config - handle profile name of ".." as its obviously invalid
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 08:50:47 +0000 (17:50 +0900)]
elm config - handle profile name of ".." as its obviously invalid

this forbids .., as a profile name in addition to the existing "stip /
from profile name".

7 years agoeina utils - home and tmp environ - store statitcally and handle setuid
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 08:35:26 +0000 (17:35 +0900)]
eina utils - home and tmp environ - store statitcally and handle setuid

if setuod we dont want to trust HOME environ at all and get it from
passwd file... also we dont want to keep re-getting too... so store
statically as well as tmp.

this also kind of helps CID 1366469

7 years agoedje_cc - fix potential divide by 0 in info output if file size is 0
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 07:43:50 +0000 (16:43 +0900)]
edje_cc - fix potential divide by 0 in info output if file size is 0

this will realistically never happen in a sane case... but its
theoretically possible. fixes CID 1366923

7 years agoeet data - fix checking return value of eina_value_pset like elsewhere
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 07:36:57 +0000 (16:36 +0900)]
eet data - fix checking return value of eina_value_pset like elsewhere

eina_value_pset() return is checked pretty much everywhere except
here. this addresses that inconsistency. this fixes CID 1367487

7 years agoevas cache image - remove double unlock of image task locks.
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 07:30:54 +0000 (16:30 +0900)]
evas cache image - remove double unlock of image task locks.

this is indeed a bug and fixes coverity CID 1367492

7 years agoeina share - fix spinlock release if magic check fails on share data
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 07:20:17 +0000 (16:20 +0900)]
eina share - fix spinlock release if magic check fails on share data

this fixes a potential double spinlock unrelease if magic check fails
on share data. this fixes CID 1367493

7 years agoevas image obj - remove logically dead code
Carsten Haitzler (Rasterman) [Wed, 8 Feb 2017 06:35:19 +0000 (15:35 +0900)]
evas image obj - remove logically dead code

coverity is right - it's logically dead code. fix CID 1367774