10986 Commits

Author SHA1 Message Date
twinaphex
d7aa48aa33 (Menu) Don't pass rgui pointer to menu_build_scroll_indices 2014-05-30 22:06:54 +02:00
twinaphex
f49ea47040 (RMenu) Implement shader parameter representation logic in
RMenu display layer
2014-05-30 22:04:53 +02:00
twinaphex
6307202a6d (Menu) render_messagebox/render/frame functions from menu_ctx no longer
get a rgui handle passed to them
2014-05-30 21:51:12 +02:00
twinaphex
a59c3cba6e (D3D PC) Add quitting variable 2014-05-30 21:17:34 +02:00
twinaphex
a282a10ba9 (Lakka) Build fix 2014-05-30 21:15:39 +02:00
twinaphex
0c6b167e4d (Menu) Don't invoke rarch_set_fullscreen for 'defaulting' SoftFilter 2014-05-30 21:12:09 +02:00
twinaphex
68c9a6003c (Menu) Prevent another few segfaults when trying to 'default'
softfilter - have to continually grab pointer back to driver.menu
after some function occurred that could have changed driver.menu
2014-05-30 21:10:15 +02:00
twinaphex
e313c9d181 (Filters) Fixes filter applying - don't do rarch_set_fullscreen
for applying filters - instead do rarch_deinit_filter and rarch_init_filter - faster and less bug-prone since rarch_set_fullscreen will also bring
down menu
2014-05-30 21:01:17 +02:00
twinaphex
776d4599e3 (Menu) Fixes crash when toggling fullscreen mode in menu mode 2014-05-30 20:55:59 +02:00
twinaphex
8c1b4a3835 (Menu) Get rid of spammy log 2014-05-30 20:53:49 +02:00
twinaphex
b2d9c60613 (Menu) Don't pass menu data to menu_iterate
(Menu) Don't pass menu data to backend - postprocess
2014-05-30 20:53:10 +02:00
twinaphex
04c84dc36c (RGUI) Return null if calloc of rgui struct fails in rgui_init 2014-05-30 20:41:31 +02:00
twinaphex
34ed0b3e04 (Menu) Get rid of hacky menu_reinit_context - NOTE - looks like the
RGUI display driver does not clean up all the state since continually
invoking rarch_set_fullscreen will eventually cause segfaults
2014-05-30 20:37:47 +02:00
twinaphex
779c467191 (Menu) Get rid of passing rgui pointer around to most functions in
menu backend. NOTE: Applying filters seems to be broken on PC
right now, investigate
2014-05-30 20:23:20 +02:00
twinaphex
c57f780d9f (Menu) Start undoing passing rgui pointer to every menu-related
function - the only functions where it should actually pass a
pointer instead of just grabbing a pointer from menu.driver should
be in the menu init/deinit functions
2014-05-30 19:34:01 +02:00
Themaister
0ca79f964f Make dumping EQ filter coefficients a config option. 2014-05-30 19:04:40 +02:00
twinaphex
0488dca77a (Menu) Some cleanups to prevent crashing when applying softfilter
several times - should do something about this menu pointer being
passed around from function to function
2014-05-30 18:52:03 +02:00
twinaphex
d267f27130 (Style nit) Rename 'init' to 'initialize' in comments
(Filter) Get rid of questionable ifdef for HAVE_FILTERS_BUILTIN for
filter init
2014-05-30 18:23:32 +02:00
Themaister
35363148b2 Don't let HAVE_OVERLAY depend on OpenGL. 2014-05-30 18:18:06 +02:00
Themaister
c5ef4330d1 Remove reset interface from SW SDL and Xvideo. 2014-05-30 18:13:34 +02:00
twinaphex
c020f362b6 Remove hacky restart function in video interface 2014-05-30 17:57:21 +02:00
twinaphex
af440bf545 (Menu) Menu now gets inited from driver.c instead of frontend.c
(Console) Better way to 'restart' instead of using the function
'restart' - call rarch_set_fullscreen instead. 'Restart' function
in driver interface can go now
(Menu) Menu now gets properly deinited/reinited when - say -
rarch_set_fullscreen is toggled or some other state change like this.
If init_assets function of menu_ctx_driver is implemented, this should
ensure all GL/D3D assets get 'cleaned up' prior to teardown of the video
driver, and properly reinited after the video driver is brought up again
2014-05-30 17:49:04 +02:00
twinaphex
6f62101180 (360) Build fix 2014-05-30 09:24:52 +02:00
twinaphex
6f2927cbac (Filter) rarch_init_failter - Filter init failure was wrongly reported
for HAVE_FILTERS_BUILTIN targets
2014-05-30 09:11:56 +02:00
twinaphex
9319fb7c5b (Menu) When applying softfilter, use rarch_set_fullscreen
for all targets
2014-05-30 09:02:56 +02:00
twinaphex
3a9e468ac6 (XDK) Reinit entire renderchain instead of 'reinit texture' 2014-05-30 08:51:34 +02:00
twinaphex
0217bcfbce (PC D3D) Update PC D3D driver to make it more inline with XDK D3D
driver
2014-05-30 08:22:56 +02:00
twinaphex
1f06f610e2 (XDK D3D) Make drdiver more feature-complete and inline with PC
D3D driver
2014-05-30 08:23:29 +02:00
twinaphex
1eef295450 (XDK) Fixes FPS count 2014-05-30 06:31:53 +02:00
twinaphex
6e48692bf5 (XDK D3D) Cleanup - call to make_d3d_d3dpp unnecessary 2014-05-30 06:17:10 +02:00
twinaphex
61fdcf4836 (XDK D3D) More cleanups - calling SetViewport from init_texture
is also not necessary
2014-05-30 06:08:45 +02:00
twinaphex
bb2bf81f00 (XDK D3D) Cleanup init_textures further - memsetting first pass
texture to 0 is not necessary
2014-05-30 05:56:45 +02:00
twinaphex
b38caa3cd3 (XDK D3D) Cleanups - don't set vertex buffer immediately at
init texture time - wait until set_vertices gets triggered
2014-05-30 05:38:36 +02:00
twinaphex
c8f44ba94d (XDK) set_vertices changes 2014-05-30 05:35:10 +02:00
twinaphex
288cf25319 (360 HLSL) Build fix 2014-05-30 03:43:24 +02:00
twinaphex
31c731705a (XDK1) Compile with HAVE_FILTERS_BUILTIN defined 2014-05-30 03:26:53 +02:00
twinaphex
985a5d9779 Remove unnecessary uninit_video_misc 2014-05-30 03:14:10 +02:00
twinaphex
a97b53f9f1 Reimplement driver.*_data_own variables so that they actually
work for a usecase now -

on Xbox, full teardown/re-initing of D3D context seems to be
error-prone - so by flagging driver.video_data_own to true inside
of the video_init function, we signal later on to the uninit_drivers
function that we DO NOT want to call the free function of this driver
and clean up the handle.

Instead, this driver should properly retain the handle by returning
the pre-existing handle when (for example) driver.video_data is
not NULL. You can see an example of this in xdk/xdk_d3d.cpp.

Overall still a quite clean solution and we will only use this in
extraordinary conditions (like this Xbox one I suppose) - full
teardown/setup will be the goal for all other platforms where we
can be certain that the state can be brought down and up entirely
during runtime without any problems.
2014-05-30 02:37:48 +02:00
twinaphex
346701fce8 (Xbox) Make hack less ugly 2014-05-30 02:01:29 +02:00
twinaphex
1e1e95a663 (XDK) Don't put early return in free function since this
function won't be called during game time anyway
2014-05-30 01:53:55 +02:00
twinaphex
ffd68d4878 (Xbox) Seems that Xbox D3D does not free up all resources
for some reason and that they didn't anticipate full teardown/setup.
Not sure if I can confirm, but it has been hell so far trying to get
this to work. So, instead, implement a butt-ugly hack that forces
ownership over video/input lifetime state. Will properly refactor
the driver.*_data_own variables later so that this hack is less
ugly. Ideally we'd solve the teardown/setup problem but given this
is esoteric D3D gibberish - good luck debugging >_>
2014-05-30 01:48:55 +02:00
twinaphex
a747ad29d3 (XDK1) Remove manual usage of XInputClose in input free function 2014-05-30 00:33:04 +02:00
twinaphex
c3f19c13bf (GX) Audio driver - some cleanups 2014-05-30 00:24:09 +02:00
twinaphex
4ac47a9c56 (PS3) Input - free data variable at end of free function 2014-05-30 00:02:06 +02:00
twinaphex
77dd37409f (GX) Properly set up GX video driver for proper teardown/setup 2014-05-29 23:59:59 +02:00
twinaphex
bbf3f17b85 (Android) Properly implement input free function 2014-05-29 23:57:58 +02:00
twinaphex
9894e2db0c (QNX) Properly implement qnx input free function 2014-05-29 23:57:00 +02:00
twinaphex
07cd28d00f (XDK) Properly implement XDK input free function 2014-05-29 23:54:00 +02:00
twinaphex
d1bcf03a3d (GX) Input - properly implement free function 2014-05-29 23:49:33 +02:00
twinaphex
e48123ba30 (XDK) Cleanups 2014-05-29 23:34:26 +02:00