Removed check that was introduced in 54e6bafa. This check doesn't work in WPS as BUTTON_NONE actions come while you still have button pressed. I believe the proper fix for this is already commited in c354e0bd.
Fixes FS#13370
Change-Id: I86e92c296c0bfa42caaed9cb3540e311630daf9c
Now no matter how [de]normalized the input strings are, we will
normalize them to the best of our ability in what we use.
This adds a dependencey for Perl's Unicode::Normalize.
Change-Id: I13e275692ea33a463b19f3a499ea06ce1acbb44a
Fixes ignored button press in yes/no screen.
It depends on keymapping - this issue happens if action before wait_for_release is mapped on release action. Something like:
{ ACTION_STD_OK, BUTTON_SELECT|BUTTON_REL, BUTTON_SELECT }
Change-Id: Idc6a57ef7a7daf75c21c9887e07fbefc814b6ae3
ipod6g was configured with SECTOR_SIZE of 4096, but this ATA driver
unconditionally translated these to 512B operations on the actual
storage device.
Rockbox's storage layer already has robust support for "logical sectors
larger than physical storage sectors" through use of
MAX_LOG_SECTOR_SIZE. So switch to that mechanism, allowing the ipod6g
ATA driver to be simplified.
If we want to support drives with physical sector sizes > 512B, then
we need to port the MAX_PHYS_SECTOR_SIZE logic from the primary ATA driver.
Additional changes:
* Simplify MWDMA/UDMA selection logic
* Report CE-ATA mode in debug menu
* Use LBA48 commands only if drive is over 128GiB.
* Drop default sleep/poweroff time from 20s to 7s (matching main ATA driver)
Finally, the bulk of the changes are the first phase of a badly needed
style cleanup that made reading this driver a lot harder than it should
be. I intend to split this into a separate patch.
Change-Id: I2feca9fd319c8d6cfb3c2610208970428d2fa947
And fix up all the problems I've found so far.
...This will undoubtedly introduce a pile of new warnings.
Change-Id: I868de507a0e9790f289676c198e2977c26755f22
I'm 98% sure that the "byteswapped" value I saw was due to another bug
(bad PIO timings on ipods leading to data corruption) and none of the
growing pile of identify device data files that I have include the
backwards one.
Change-Id: Iea47b7419b120a3c8a282b2e3a8f65b8965356ce
This is actually the root fix for the crash; the fix in
772eff8ca6 didn't restrict it to 2 channels, just sanity
checked the file vs MAX_CHANNELS.
Change-Id: If5a0a9b946a179ad47d6e7955f7c025de3aaaa0e
Since commit e0df995 storage_disk_is_active may be 0
even if there is no disk that needs to be spun up or
device to power on, so it doesn't really make sense
to check for it, anymore, in order to decide whether
to display the loading indicator.
For remaining devices with spinning disks, the noise
or vibration of the disk spinning up may serve as a
sufficient signal when loading a plugin, that the
device isn't frozen. Plus, for the tagtree, search
progress is displayed after 0.5s anyway.
Change-Id: I5791725e2388d7113818a358204968d8a4e6f843
Added jpeg decoder jpegp.c using RAINBOW lib. Currently enabled only for pictures not supported by old decoder (as old decoder more optimized for low mem targets)
Someone TODO:
* Old decoder has optimized downscale logic which new decoder doesn't have (it gives big difference in required memory and time for decoding). This logic should be ported/adapted if possible.
* Add smooth downscaling.
* Grayscale support
Change-Id: Ie96bc62848b51cc6a3942f8e069ec6ab02dc1c56
Added unmodified files from RAINBOW library by Attila Tarpai
Full sources:
https://github.com/Halicery/vc_rainbow
Change-Id: I356486b6a332aa3f610ddcae57f8a2044653b051
It's not clear that we've ever intended to support >2ch files, based on
'#define MAX_CHANNELS 2' and other logic that only seems to care about
mono vs not.
Change-Id: I15e92fb29cceef32e63fc3a821f6e96bbde930b6
This should be a no-op for .lang files, but it can matter when
generating talk clips, as those are based on the raw filename.
Change-Id: Ia449ae7668818e4bf093347fd0fb15fa5b67fa69
The iPod 4g and color always ignored the very first
button press after booting, i.e. you needed to either
press another button or scroll for a bit first.
Not sure whether any other PP iPods exhibit the same
behavior, although it is not reproducible on an iPod
video.
This patch basically copies the code from FS#5230
that fixed unresponsive buttons whenever the HOLD
switch was turned off (committed as a5961c944b).
Change-Id: I8d3444094e5d000e0f8e30e39a23f054abeeb0f5
This is particularly useful for classical music, where track names like
"symphony no.5" are common -- We want to voice this as "symphony number
5" instead.
Change-Id: I3607dfccbafee2174519760b30636e0db0aad819
Basically, there are weak aliases for all possible IRQ handlers,
pointing at a NORETURN function. GCC complains that the aliases
don't have the same attribute, but that's actually what we want.
Change-Id: I7f877e00193ed457589c8442db82e93ddea0d60a
The flag would apparently make hardware accelerated
rendering mandatory. SDL2 will still attempt to use
hardware acceleration, if the flag is missing.
Change-Id: Icfdcad90591d2a5003a1540f2a669c7e54dcfc3b
GMP 4.3.2 was quite old even at the time of the GCC 4.9.4 release, and
it refuses to compile out of the box on modern toolchains (eg GCC 14)
due to issues in the configure script. Instead of trying to backport
fixes, just bump it up to 6.1.2, the final release in the 6.1.x series
that was current when GCC 4.9.4 was released.
Meanwhile, with the update to GMP, the also-quite-old MPFR 3.1.0 fails
to compile; instead bump it up to 3.1.6, which is the final release in
the 3.1.x series.
Change-Id: I356c07282b20629c90fb0de1b8be40d2c55bbefc
The 12K-per-thread necessary on arm64 linux systems caused an
unnecessary 180K increase in RAM usage on the hibyos hosted ports. So
back off to the old size unless we know it was needed.
Change-Id: I1e25417433052027ae02a51903b0f5245819db44
This allows callers to see if it's safe to enqueue something instead
of triggering a panic if the queue turns out to be full.
Change-Id: Idb887e7a47cfbfef998f27d9d85090f3c0ed2230
