add a helper function to allow plugins to parse the parameter string
passed into plugins
support included for bool, char, string and numbers+decimals
Change-Id: I39f35c8bd3c21b83097a538c19f46d362c468fa4
share all the core icons with plugins
these are all small mono icons like usb plug icon or
play, fast forward, rewind icons
--include the icon_helper function
Change-Id: I385028815a4dd368515f491a9e19dee3d500252d
talk_value is just talk_value_decimal with 0 decimals
lets add the extended function instead
static inline int talk_val(long n, int unit, bool enqueue)
{
#define NODECIMALS 0
return rb->talk_value_decimal(n, unit, NODECIMALS, enqueue);
}
Change-Id: Iaba3d2f95785f2e1855e294ccf099a977bb6cb20
One of these was an unintentional fallthrough -- if you said no to
'quit without saving' it would save and quit.
Change-Id: I33828b58962a05adab65ac5f71dd0e2b1e07001b
I'm not sure this would happen but I suspect its possible to get malformed
data back from lineinfo as this is how the error system tracks code flow
and this is a hacked add-on to allow RLE (run length encoding) of
said code flow data
Change-Id: I4b914c632935f8b1a9fa960212c58e810b2db738
Since battery_bench is a TSR (terminate & stay resident) plugin it
cannot allocate the plugin buffer -- various parts of the Rockbox
core make use of it as temporary storage space. The buffer was used
for stack space so this generated false 'Stkov' panics when one of
those in-core users overwrote the buffer.
The default stack size of 1kb is a bit small, so use 4x the default
size to be on the safe side.
This also fixes a minor issue where trying to view OS stack usage
while battery_bench was running would cause audio dropouts, due to
scanning the possibly huge stack with IRQs disabled.
Change-Id: I2e6b76f9946db9090c1af61f7aa0a0dac0698c20
Since this is a TSR plugin, it is not safe to use the plugin buffer.
Convert to using static buffers instead.
Change-Id: Ic5b297468a99d77d56f5b75e52dafabeb80d2f78
- Make the argument const since it's not actually mutated
- Actually return the size of the buffer used since this is what it
was supposed to do (although no existing callers cared anyway)
Change-Id: I0802071cf63d4af419f427ff54adca50b14918ad
What works:
- LCD: 16-bit RGB565
- all buttons, including scrollwheel
- SD Card
- Battery level and charging/not charging status
- USB
- audio
- sample rate switching
- HP / LO detect, with "safe" fixed LO volume -
LO volume will only be put to user-defined max volume
if headphones are not present.
- rtc
- Plugins build, tried a couple and they seem OK
- Bootloader, installable to nand via usbboot
What doesn't work:
- Dual Boot
- power on/off has intermittent, low volume audio click
(sometimes it's completely silent, sometimes there's
a click)
- Audio uses 16-bit volume scaling, so clicking/popping
is pretty bad at lower volumes - need 32 bit volume
scaling, 24 bit I2S data
- USB HID keys not yet defined
- no jztool support
Unknowns:
- Stereo Switch pins: Direction select, AC_DC
(probably not even hooked up)
- What is the actual purpose of the Stereo Swtich?
- How does the bluetooth module connect?
"Someday" stuff:
- get LCD working at higher bit depth
- Bluetooth
Change-Id: I70dda8fc092c6e3f4352f2245e4164193f803c33
* direct use of memcpy() instead of ci->memcpy() in flac and mod
* uninitialized variable in mpegplayer
Change-Id: I2d08682d5f66c319780e69e3ff63d600c61d8f5a
- Audio playback works
- Touchscreen and buttons work
- Bootloader works and is capable of dual boot
- Plugins are working
- Cabbiev2 theme has been ported
- Stable for general usage
Thanks to Marc Aarts for porting Cabbiev2 and plugin bitmaps.
There's a few minor known issues:
- Bootloader must be installed manually using 'usbboot' as there is
no support in jztool yet.
- Keymaps may be lacking, need further testing and feedback.
- Some plugins may not be fully adapted to the screen size and could
benefit from further tweaking.
- LCD shows abnormal effects under some circumstances: for example,
after viewing a mostly black screen an afterimage appears briefly
when going back to a brightly-lit screen. Sudden power-off without
proper shutdown of the backlight causes a "dissolving" effect.
- CW2015 battery reporting driver is buggy, and disabled for now.
Battery reporting is currently voltage-based using the AXP192.
Change-Id: I635e83f02a880192c5a82cb0861ad3a61c137c3a
Do not introduce any change for M:Robe 500 as it uses the two LCD frames
in non-obvious way.
Sansa Connect and Creative ZVM use only single front framebuffer.
Place TTB at DRAM end to minimize memory loss due to alignment.
Reserve as little as possible memory for the LCD frames.
On Sansa Connect this change extends audio buffer by 858 KiB.
Change-Id: I21bdeec4cfba86d71803a39acd651a87e73767e6
This fixes a mysterious, long-standing crash that's been bothering me on
ipod6g for ages: a silent stack overflow in the sound mixing thread (which
is triggered upon loading a new sound, apparently) will thrash the memory
which is located directly before it in the address space.
In this case, it was the SDL_ButtonState variable which stores the mouse
button state that was being trashed. This was manifesting itself by making
the player always run forward, since MOUSE2 is mapped to +forward by
default.
Fix this by quadrupling the stack size of SDL-spawned threads (not the main
thread) from 1 KB to 4 KB.
Change-Id: I2d7901b7cee1e3ceb1ccdebb38d4ac5b7ea730e1
- Add support for skip and seek while the player is locked.
(Thanks to @bahus for the suggestion)
- Fix touch zones because the down button zone ended up being
a lot smaller than expected due to the touchpad's wonkiness.
Also added a visual display of the touchpad to its debug menu.
- Fixes the pictureflow keymap because it was mostly unusable.
Change-Id: Ic0da4f8df3613ff7828ce1cb940ed9e77ada3281
rather than dumping all the settings allow a filter function
to choose desired settings
in menucoresettings you can see an how to do exact text matches
or wilcard matches you can even use luas version of regex
Change-Id: I4c7f7592498ea194e06e9a556b77ffd57f5d4223
return the nextfunction and nil instead of pairs it allows a faster
return to lua rather than calling the lua function
pcall(fnpairs) from c and returning the result back
into lua to kick off the search
yeah, no clue why I didn't realize that before..
testing in x86 and ARM..
its more RAM efficient to do the initial creation
of the stack in lua code for the __pairs functon
its not faster but being that its a one time hit per
iter creation the reduced churn alone should be worth it
along with a reduced peak RAM usage
fix bug where a failed module can not be reloaded
optimize filetol
fix potential bug in splash scroller when no break character is found
Change-Id: I42c922e07039a19138b97c0d0e80cf3cf2426471
Settings code tries to cast the bool pointer to an int pointer
and then write through it, which makes AddressSanitizer whine.
Change-Id: I1d6162fc98144cd6b40444016b3b48ef02967324
temp loader allows some lua requires to be loaded and
later garbage collected unfortunately the module needs to be formatted
in such a way to pass back a call table in order to keep the functions
within from being garbage collected too early
BE AWARE this bypasses the module loader which would allow code reuse
so if you aren't careful this memory saving tool could spell disaster
for free RAM if you load the same code multiple times
Change-Id: I0b6f81e481b8c779edbd620c8403794f8353926f
bring custom keyboard layouts to lua
conversion to the proper format requires create_kbd_layout.lua
just pass a lua string with your desired layout
Change-Id: I14a392410846311a4f3cf8dda0e88d39834d0418
updated example script and renamed some functions as well
fixed bug in printtable cursor position if greater than maxlines for the
screen would reset to item 1
now we move the list start and select it
(and try to center it on the screen)
fixed a few bugs in the add_menu code
Change-Id: I01dead0481ef2e925af8b4cc6c14e36c2859dbba
allows menus + submenus + context menus all with simple tables
menu_t which is a table of strings
func_t which are the corresponding functions to go with those strings
see lua_scripts/submenu_demo.lua
Change-Id: I907b74b4abef0ecbe49f181d0ced6e6d20e94de5
The abiflags data is only used to communicate ABI information to a
program loader -- you can see what info is stored with readelf -A.
Dropping it shaves 24 bytes off of every binary (including plugins).
Change-Id: Iae78eeffe5c840ff67717707fb94821d24aac8ec
It never worked, and hasn't compiled in something like a decade, Given
the HW capabilities (limited onboard flash, no expandability) there's
really no point in trying to fix/complete it.
Change-Id: I7d175089840396f8891645bd10010d730dd5bfdc
They were never finished, never saw any release ever, and haven't
compiled for the better part of a decade. Given their HW capabilities [1],
they are not worth trying to fix.
[1] 1-2MB RAM, ~256MB onboard flash, no expandability
Change-Id: I7b2a5806d687114c22156bb0458d4a10a9734190
Code in cpu.h is correct, so just disable -Wmisleading-indentation for
the specific sections that matter.
Change-Id: I378f3a6fef117ac73edb0d8ce998ef3a818be22e
add stylized lines to lua
the exported do_menu has a severe limitation of 64 items
it also requires double the memory
put_line is the way rockbox builds menus
update printtable
user config from core -- done
code cleanup
fixed for 1-bit screens
changed button behavior
fixed for 2-bit screens
Change-Id: I4de55e42685aa1d2f53a33bc8e980827864e810b
It turns out that aa_cache.buf, used to store decoded album art during
background scanning, was not correctly allocated and overlapped with
memory allocated for buflib. This was what caused all the segfaults.
Also fixed a logic error in read_pfraw(), which returns a buflib handle
on success, but also returned 0 on failure -- since 0 is a valid buflib
handle, it should return -1 on failure instead.
Change-Id: Ifaa1c02ec19b0859e43c40c0462ed7738d07fec3
Image viewer compiles decoders as a plugin and removes CRT0
A resuklt static vars do not get initialized in decoders
Change-Id: I72da326967c7b404b297897000f82852661a7ebb
Added an entry 'Add' to the main screen to lessen confusion
side note:
its actually a shortcut back to this plugin with flag '-add'
passed as a parameter
Change-Id: Ifeb0a1e5be99764b1d25d5a3b0e00f1449ec3d78
Open plugin needs the context menu to allow the user to add an item
on the first run there are no items and therefore nothing to pop the
context menu from
Change-Id: I7a78454ff29b20c7b60db44349dd23d500ed887c
This adds a resync.sh script that handles most of the resyncing process:
copying new sources, regenerating help, and staging for commit. I also took
this opportunity to make the process more efficient and consistent by only
selecting the types of files that we care about and ignoring any
autogenerated temporary files that may be in the upstream tree.
Change-Id: Ib0803462684b8ae3b40eec3742944e5bdb28508e
This updates the upstream sources to include a modified get_cursor_location
patch that I anticipate Simon will merge shortly. Also, I've streamlined
the resync process to only copy the exact files we need to reduce clutter.
Change-Id: I6a5ac60821fce346c500b101c363ae0c63c2ee95
Before this was just implemented inline wherever it was needed. Now
it is provided as 2 inline functions in a header called checksum.h.
There should be no differences in actual behavior.
Change-Id: I5d756cc01dc6225f5cc8b6af90911a3fc7b57cd5
This removes all the pre-release bootloader of version 7 and
replaces them with production releases of version 8. The
pre-releases have their own problems and should be left
behind as they were never officially released.
Change-Id: If96531ba63cf4401d04133bda94735fe94ae4de1
The table is now gated by the FLASH_SIZE macro as any given
target will not have a known rom chip of a differing size
than the FLASH_SIZE. This will reduce the resulting code
a bit as well.
Change-Id: I06a283f9f44118080a106e1bcd410e81e0a48d92
I discovered it was doing this on the H300 during longer operations
and I freaked out at first. This disables this feature until the
program is finished so as to not give anyone else an unnecessary
fright while using it. It will also cut down on frivilous reports.
Change-Id: I0db406912c7f71835576203a540aba8910afa849
* pcm_get_bytes_remaining()
* pcm_calculate_peaks()
* pcm_get_peak_buffer()
Nothing in-tree uses these at all (except for the lua plugin wrapper)
Change-Id: I971b7beed6760250c8b1ce58f401a601e1e2d585
This overhauls most of the code to be easier to understand in terms
of the interactions with the flash. I found the original to be rather
confusing with how it kept switching between byte and word offsets.
My solution was to make all external access to the flash in terms of
sectors and bytes. Whatever the flash uses internally is now handled
by the subroutines for performing the erase, program, and verify
operations.
This helps make it far more consistent for the code that actually uses
these operations as they do not need to concern themselves with word
sizes and offsets anymore.
As a side effect of this change the flash operations are now done
entirely by subroutines; even the batch operations that used to use
custom loops.
Additionally some functions were merged with other functions in order
to reduce the amount of functions as well as consolidating common
code fragments.
Change-Id: I4698e920a226a3bbe8070004a14e5848abdd70ec
Nothing in the core has used it for some time. It's exported to the
plugin API but the last plugins to use it were switched to the mixer API
back in 2011.
This allows us to get rid of pcm_play_dma_pause() from all audio drivers
Change-Id: Ic3fa02592316f84963e41d792d1cabb436d1ff6b
First the argument should be const since the original parameter is.
Second the pointer arithmetic for detecting whether rockbox is running
from ROM or not is incorrect. It ends up being at a location twice as
far as intended since the arithmetic does not account for the pointer
type's underlying size. It should also be dependent on the target's
FLASH_SIZE.
Third the LCD setup is moved to the entry point since it is the best
place to setup and restore the LCD changes.
Change-Id: If9ddaf2cd937f1edf61c82a8a27f48d01807068a
The main change is revising how the checksums are guarded by macros.
But both are also converted to static linkage so they can be better
optimized by GCC. I also change the types around to reflect how the
space the data types actually need. Furthermore I make use of C99
changes to how variables can be declared to move them closer to where
they are used.
Change-Id: I0b21b655f3f4a7c4bbd4365a384a551e75451159
This moves the checksum into the local stack and turns the second
parameter into an optional argument. This also reads the model
segment that was previously unused so it can also be checked as
an extra safeguard in the event the checksum somehow matches
yet the model is incorrect.
Change-Id: I9a8c2d731e4f1818e6e4aee3c3978777c16ccf19
First neither of these functions can fail on supported targets
so they have become void functions. Their return values were
not being used anyway.
Second support for other flash chips not even used on the
supported targets has been removed. It appears they were
only ever used on the discontinued Arch devices.
Third cfi_read_id was restructured to remove obsolete code
for error checking that is not necessary at all. The datasheets
appear to indicate that the commands used cannot fail.
Fourth cfi_get_flash_info was restructured to use a new approach
to initializing the flash_info struct. It no longer initializes
the structure twice.
Fifth the relevant code has been updated to use the full 16 bits
that are exposed by the flash rom ID interface.
Change-Id: I25b1ada3d4621e2d80ac66d3d9a964964268cb3b
This uses the toggle bit method referenced in the datasheets for the
supported ROM chips and moves the code to a reusable subroutine. It
also introduces a short delay to give the bus a chance to recover. The
older ROM datasheet doesn't mention this delay as necessary but the
newer one does so it now does this for both.
Change-Id: Ie9dc8833bbd3ee545072b0c724ab220d13208d3d
I'm currently running up against the limitations of the lcd_draw functions
I want these functions to be able to be used on any size buffer not
just buffers with a stride matching the underlying device
[DONE] allow the framebuffer to be decoupled from the device framebuffer
[DONE need examples] allow for some simple blit like transformations
[DONE] remove the device framebuffer from the plugin api
[DONE}ditto remote framebuffer
[DONE] remove _viewport_get_framebuffer you can call struct *vp = lcd_set_viewport(NULL) and vp->buffer->fb_ptr
while remote lcds may compile (and work in the sim) its not been tested on targets
[FIXED] backdrops need work to be screen agnostic
[FIXED] screen statusbar is not being combined into the main viewport correctly yet
[FIXED] screen elements are displayed incorrectly after switch to void*
[FIXED] core didn't restore proper viewport on splash etc.
[NEEDS TESTING] remote lcd garbled data
[FIXED] osd lib garbled screen on bmp_part
[FIXED] grey_set_vp needs to return old viewport like lcd_set_viewport
[FIXED] Viewport update now handles viewports with differing buffers/strides by copying to the main buffer
[FIXED] splash on top of WPS leaves old framebuffer data (doesn't redraw)
[UPDATE] refined this a bit more to have clear_viewport set the clean bit and have skin_render do its own screen clear
scrolling viewports no longer trigger wps refresh
also fixed a bug where guisyncyesno was displaying and then disappearing
[ADDED!] New LCD macros that allow you to create properly size frame buffers in you desired size without wasting bytes
(LCD_ and LCD_REMOTE_)
LCD_STRIDE(w, h) same as STRIDE_MAIN
LCD_FBSTRIDE(w, h) returns target specific stride for a buffer W x H
LCD_NBELEMS(w, h) returns the number of fb_data sized elemenst needed for a buffer W x H
LCD_NATIVE_STRIDE(s) conversion between rockbox native vertical and lcd native stride (2bitH)
test_viewports.c has an example of usage
[FIXED!!] 2bit targets don't respect non-native strides
[FIXED] Few define snags
Change-Id: I0d04c3834e464eca84a5a715743a297a0cefd0af
It was originally hard-coded at 0x200 which is 512 sectors. This
only works for the H100 and H120. The larger ROM of the H300 is 1024
sectors in size. In either case the bootloader starts 16 sectors before
the end of the ROM so rely on this fact to correctly calculate where to
stop the bootloader erasure.
Change-Id: Iec4112ebf24379f80a7bf1363035e005c434907e
This also modifies the configuration file to include macros
defined in the H100 / H120 implementation.
Change-Id: Iae845889c98661ec548c04fc57e733dcc346c0f1
at 1 GHZ the intermediate calculation for cycles overflows 32 bits
this makes timer fail even with sensible values
solution divide both sides by 100
Change-Id: I18a4054c2d06fb72531d5496bba562f71b03984f
This resolves a long-standing issue with ATA errors in certain
situations, mostly in CF card modified units.
Change-Id: Iec23fa6de117a6c9657320432ed585170e95a4eb
* HOME_DIR is now either "/" or special "<HOME>"
* target-specific "home dir path" is defined solely by PIVOT_ROOT
* PIVOT_ROOT path is now defined in toplevel config files
* Make Samsung YP-R0/R1 and SONY_NWZ use PIVOT_ROOT too
* Do not prepend PIVOT_ROOT path if the path already has it!
* Do not play these games for __PCTOOL__ builds
Change-Id: I3d51ad902a5f9cafe45ba15ba654f30f1ec6113a
The Q and K have a slightly different case, but the hardware under the
shell is completely identical.
These models are rebadged versions:
* Hifiwalker H2 (== Q)
* AGPTek H3 (== K)
* Surfans F20 (== K)
Other notes:
* Significant improvements in the shared Hiby-platform launcher/loader
* SD card can theoretically be hot-swapped now
* Support external USB mass storage!
* Some consolidation of Hiby-platform targets
* Some consolidation of plugin keymaps
Todo/known issues:
* Keymaps need to be gone over properly
* Convert to HAVE_SCROLLWHEEL?
Change-Id: I5a8a4f22c38a5b69392ca7c0a8ad8c4e07d9523c
Most credit goes to: Roman Skylarov
Additional integration and refactoring by myself.
*** COMPLETELY UNTESTED ***
Change-Id: Ia64c36d92e0214c6b15f7a868df286f8113ea27b
This codec requires floating point.
Original author: Peter Sovietov
Ported to Rockbox: Roman Skylarov
Further integration and bugfixes: Solomon Peachy
Change-Id: I781ecd3592dfcdbbc694063334350342534f1d6c
lua gives you a memory used number that only reflects the current allocations
if fact it doesn't even give you a way to get the amount of ram free
rb.mem_stats() seeks to fill this gap by marking the memory allocated for lua
with a sentinel value which can later be checked to get a high water mark
of the ram used by lua and a pretty good idea of how much ram is available
Also includes an example script
usage:
used, allocd, free = rb.mem_stats()
Change-Id: Ia282869f989848324d7d88c7df4827fdbce4fb4e
checks button_status in rockev
strpbrk_n custom implementation allows setting max search len in source string
add some branch prediction where appropriate
fix formatting in splash_scroller script
Change-Id: Id5d8e9d83f4b3e361ccb67b403af8f9a8a31b8f0
lua currently splashes a stack traceback on error
for deep tracebacks and especially on devices with smaller screens
this leaves out a lot of vital information
in the past I have resorted to splitting the traceback string or
even saving the return to a file
This patch provides a scrollable buffer with rudimentary text reflow
to allow you to read the whole traceback string
Upon traceback if you press nothing the screen will display for 5 seconds
If you press OK or CANCEL it will quit immediately
PREV/NEXT scrolls the list on button press timeout is disabled
lua now provides rb.splash_scroller(timeout, str)
example script provided too
Change-Id: Idbc8ce0c514196f0fae48c43aeaea8b60d6da1a5
check if any buttons are waiting in the queue before triggering
the event thread for action & button events
makes button events quicker and also spend less time interrupting lua
both wins
Change-Id: I38346c084afdd99e4608f40b52053ee39730fb40
cool little lua based audio player
creates dynamic playlist of 10 mp3s found on device if no music loaded
I had to limit the depth of search to 3 levels due to the recursive
nature of the current dirbrowser functions this could be rectified with a
bit more code
fixed a bug in print.lua that kept scrolling text even after screen clear
Change-Id: Ifd285332df41a409ecaeb1ea447ad15537b5d04c
I had previously added the fuctionality to luadir but I didn't update
the examples
also breaks out the file_browser function to be a bit more accessible
Change-Id: I14067256b9d76a757f732840cbee1cf84d775b1b
Remove mp3_is_playing() entirely, in favor of pcm_is_playing()
Remove mp3_play_pause() entirely, as it's a dummy/no-op call
Remoce some archos-specific comments
Change-Id: I4e9ff323490a93add00809efd19e0d4e3f198b2d
Fixes deficiencies with the button system on the X3
The x3 has an interesting button layout.
Multiple key presses are NOT supported unless
[BUTTON_POWER] is one of the combined keys
As you can imagine this causes problems as the power button takes
precedence in the button system and initiates a shutdown if the
key is held too long
instead of BUTTON_POWER use BUTTON_PWRALT in combination with other keys
IF using as a prerequsite button then BUTTON_POWER should be used
Multiple buttons are emulated by button_read_device but there are a few
caveats to be aware of:
Button Order Matters!
different keys have different priorities, higher priority keys 'overide'
the lower priority keys
VOLUP[7] VOLDN[6] PREV[5] NEXT[4] PLAY[3] OPTION[2] HOME[1]
There will be no true release or repeat events, the user can let off the
button pressed initially and it will still continue to appear to be
pressed as long as the second key is held
Tree scrolling is PLAY+NEXT or PLAY+PREV
Change-Id: I88dfee1c70a6a99659e8227f5becacc50cc43910
shortcuts can be exported as .opx; now they can import as well
if parameter is a valid file..
plugins with parameters are now hashed on the parameter path
fix bug with empty parameters not overwriting last valid parameter
Change-Id: I149519811f07cb4ba22b7113449e2f89f77f1eee
OP allows you to use Open With.. to call plugins with parameters
called directly it acts as a shortcut list for plugins
open_plugins.rock interfaces with the open_plugin core
When opened directly it acts as a viewer for the plugin.dat file
this allows you to edit the paths and parameters for
core shortcuts as well as your added plugins
If a plugin is supplied to the viewer it is added to the dat file
If instead the plugin has previously been added then it is run
with the parameters you previously supplied
-----------------------------------------------------------------------------
Added export to .opx files
this allows shortcuts to plugins with parameters to be called from
the file browser
Change-Id: Ib8b05a60b049fb1d5881031ca09a07e3307d375a
open_plugin allows arbitrary plugins to be called
in hotkey and start screen
replaces PictureFlow Integration
shortcuts menu plays plugins now too
rather than store paths and parameters in the settings
that reside in memory instead entries in a file are searched by hash.
after all, the plugin has to be loaded from disk anyways
----------------------------------------------------------------------------
shortcut_viewer.rock-- can now call plugins rather than taking you to them
in the browser
-----------------------------------------------------------------------------
Added a new option to menus:
F_CB_ON_SELECT_ONLY
instead of option callback every time a item is accessed
F_CB_ON_SELECT_ONLY fires callback only when item is selected
-----------------------------------------------------------------------------
Added manual entries
-----------------------------------------------------------------------------
Change-Id: I078b57b1d2b4dd633c89212c1082fcbc1b516e6a
* Get rid of the non-functional GT2 loader
* Add the UMX loader
* Add HQ mixer routines (and make it configurable)
* Allow samplerate to be configured at run/playtime
* Support >64KHz mixing/playback
* Correctly restore non-boost status
(The diff to upstream is much smaller now too!)
Change-Id: Iaa4ac901ba9cd4123bb225656976e78271353a72
I'd overlooked this code path earlier. In the case of an unknown GRP, the
game would poll SDL for keyboard input, which it never received. Remove
that, and just warn the user instead.
Change-Id: Ibbabc0f8d43cb1276ed2fcfc3c6138517582e936
