-fstack-protector only needs a small amount of runtime
support to work on native builds. It increases code size
by ~1.5% on ARM/MIPS; -fstack-protector-strong adds 3-4%.
This is disabled by default and must be enabled by passing
'--with-stack-protector' to configure.
Change-Id: If952e711d3673c9b469895f08c7bff70b3d95df6
They haven't seen development activity for the better part
of two decades and apparently were never able to even boot
to Rockbox, although the Rockbox bootloader could load the
original firmware.
Change-Id: I5cfa5909c21feaf2825aa685a05e78044b893a13
Both targets were part of the (presumably dead) Lyre project
and no longer build. The Mini2440 was much more complete than
the Lyre and doesn't seem terribly difficult to fix up to the
point where it at least builds, if someone still cares -- but
given it is a dev board in a box, it's unlikely it ever saw
much use.
Change-Id: I09745379d28db69ea9aaf77f0a62b049884260e1
This assembly implementation is marginally faster than
the non-size-optimized C version for large copies, but
is around half the code size.
Unaligned loads/stores will be used on platforms that
support it: though slower than aligned accesses, this
is still faster than copying byte-by-byte and has the
advantage of simplicity and small code size.
Change-Id: Ieee73d7557318d510601583f190ef3aa018c9121
The "mov lr, pc" instruction doesn't link a proper return
address in Thumb mode: bit 0 will be unset, leading to a
UsageFault exception when returning from the thread's main
function.
Cortex-M has the "blx" instruction to automatically branch
and link the correct return address, so use that.
Change-Id: I4c0ca55b1b2204286343f906f0b53be0c0ddc392
The implementations diverge enough that it is too confusing
to support them in the same source file; split them so it
is easier to understand.
Change-Id: Ic2f91c75e8a9bb605241441f2caed841585f5b87
The ARM assembly versions of memcpy and memmove try to
directly modify the PC based on the number of bytes to
be copied as an "optimization"; but this only works if
instructions are guaranteed to be 4 bytes wide.
This trick is completely broken on Cortex-M, since an
instruction can be 2-4 bytes wide: for example a memcpy
of 4 bytes ends up storing 32 bytes worth of garbage to
the target address!
Use the C memcpy/memmove implementations for Cortex-M
instead; fixing the ASM version is more trouble than
it's worth.
Change-Id: I695587fd585ec25ef276f2dbc61e2290b7015e13
asm.make replaces source files listed in asm/SOURCES with
a corresponding .c or .S file under firmware/asm/${ARCH}.
This makes it difficult to handle differences within one
architecture, eg. different instruction sets or versions
of the architecture.
Get rid of asm.make, and instead use the preprocessor to
select the architecture-specific sources directly from
asm/SOURCES. This is slightly more verbose but is more
flexible.
Change-Id: Id2a484794d5d951b7ba7b54a552de6960414fa45
The layout of 'struct regs' is a bit different on Cortex-M
and start_thread() wasn't updated to match; it was loading
'sp' from the wrong offset.
Change-Id: I57dbef26809821d411dc86e2066a2f53e78a3f2d
Turns out I compile-tested stale trees instead of the broken change
Net result is just the simpler MIPS revision (32/64/"classic") lookup rather than
the sub-revision (eg mips32r2 etc).
Change-Id: Ideebe522d29132f00f3769222f3846000b3a89fd
This allows us to easily distinguish between mips32 and mips32r2
(Works at least as far back as gcc 4.9)
Change-Id: I2bcba194fd9cbeedf76cea739252271908bf73d0
GCC cannot compile the existing assembly here on ARMv7-M,
claiming impossible constraints. It is actually possible to
compile if the input arguments (addresses and sizes) are
first moved to a high register so as not to conflict with
the use of r0-r7 in ldm/stm -- this is exactly what GCC does
for ARMv6, but it won't do it on ARMv7-M for some reason.
We can get a result similar to the ARMv6 code by manually
moving the inputs into temporaries, but the generated code
is a actually a bit smaller on ARMv7-M if the r0-r7 block is
shifted up to r3-r10. This only works since ARMv7-M supports
the 32-bit Thumb encoding -- 16-bit Thumb can't represent an
ldm/stm instruction of this type.
It's worth #ifdef'ing the code because although the ARMv7-M
version works on ARMv6 too, it spills a lot of registers on
the stack even though register use is mostly similar.
Change-Id: I9bc8b5c76e198aecfd0a0e7a2158b1c00f82c4df
On ARMv7-M, stm/ldm instructions can't include SP, so we must
load and store that separately. This changes the order of
registers in the context struct, but it doesn't seem to be
accessed anywhere else so this shouldn't cause any problems.
Change-Id: Ie1cd23272f23384e030f51f0b76739624fa7332b
This makes these files compileable, or in some cases less
broken, on Cortex-M targets.
In lcd-16bit.c, newer versions of GAS complain about the
infix condition codes so we use the suffix form instead,
which requires unified syntax to compile on GCC 4.9.
Change-Id: If45166d3fc83d64c692cbb331096a966397aa9e9
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
Annoyingly, this makes all of the '.S' files we compile get treated as
divided syntax, so we need to make the syntax in them explicit.
Change-Id: I56a3916b7b24c84a1214a5d6bc4ed4d651f002cf
Causes a warning with GCC8 as the protoypes are not the same.
Only affects targets that lack an asm-optimized version (eg mips)
Change-Id: I22e4657f3fb71ebbb915e4f290bf3670b1b87636
GCC 4.9 always emits assembly with divided syntax. Setting unified
syntax in inline assembly causes the assembler to complain about
GCC's generated code, because the directive extends past the scope
of the inline asm. Fix this by setting divided mode at the end of
the inline assembly block.
The assembler directives are hidden behind macros because later
versions of GCC won't need this workaround: they can be told to
use the unified syntax with -masm-syntax-unified.
Change-Id: Ic09e729e5bbb6fd44d08dac348daf6f55c75d7d8
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 appears to solve _some_ of the crashes experienced when using
gcc494 on the multicore PP targets (eg most older ipods).
(With this change, the asm vs plain-C versions behave identically)
corelock_lock(), corelock_unlock(), and corelock_trylock() were declared
with the 'naked' attribute. However, naked functions are only allowed
to have 'Basic Asm' components, and we used some extended asm, but
without declaring clobbered registers, making assumptions about register
arguments, and also directly returned to the caller via asm code.
This is what the GCC docs have to say about this stuff:
"While using extended asm or a mixture of basic asm and C code may
appear to work, they cannot be depended upon to work reliably and are
not supported."
Change-Id: I79a9c4895584f9af365e6c2387595e9c45d89c7d
Provided by Roman Stolyarov
Integration, Refactoring, and Upstreaming by Solomon Peachy
X3II confirmed working by forum tester, X20 is nearly identical.
This includes bootloader, main firmware, and the flash image patcher.
Eventual Todo:
* Further refactor AGPTek Rocker & xduoo hiby bootloaders
* Further refactor AGPTek Rocker & xduoo hosted platform code
Change-Id: I34a674051d368efcc75d1d18c725971fe46c3eee
swr/swl instructions used for word aligning were wrong. This
made memset() terribly broken. I can't imagine how it went
uncaught for soooo long. Spotted by Solomon Peachy.
I run unit tests for alignments 0,1,2,3
size 1, 2, 3, 4, 5, 63, 64, 65, 127, 128, 129;
and fill pattern 0x00 and other (since 0 is special case in this
implementation).
Change-Id: I513a10734335fe97734c10ab5a6c3e3fb3f4687a
The fix is to actually remove unused variables. Those were unused from day 1,
which I find slightly suspicious so either there is some problem or the code was
copy-pasted and modified, making some variables useless.
Change-Id: I41caf52d469b48c969ece969540de67d87e77357
SUPPORTED SERIES:
- NWZ-E450
- NWZ-E460
- NWZ-E470
- NWZ-E580
- NWZ-A10
NOTES:
- bootloader makefile convert an extra font to be installed alongside the bootloader
since sysfont is way too small
- the toolsicon bitmap comes from the Oxygen iconset
- touchscreen driver is untested
TODO:
- implement audio routing driver (pcm is handled by pcm-alsa)
- fix playback: it crashes on illegal instruction in DEBUG builds
- find out why the browser starts at / instead of /contents
- implement radio support
- implement return to OF for usb handling
- calibrate battery curve (NB: of can report a battery level on a 0-5 scale but
probabl don't want to use that ?)
- implement simulator build (we need a nice image of the player)
- figure out if we can detect jack removal
POTENTIAL TODOS:
- try to build a usb serial gadget and gdbserver
Change-Id: Ic77d71e0651355d47cc4e423a40fb64a60c69a80
On Windows 64-bit, the size of long is 32-bit, thus any pointer to long cast is
not valid. In any case, one should use intptr_t and ptrdiff_t when casting
to integers. This commit attempts to fix all instances reported by GCC.
When relevant, I replaced code by the macros PTR_ADD, ALIGN_UP from system.h
Change-Id: I2273b0e8465d3c4689824717ed5afa5ed238a2dc
defined in mingw environments.
Renamed defines of UNALIGNED to ROCKBOX_UNALIGNED so that they don't
conflict with definitions in mingw32 cross-compiling environments
(defined in _mingw.h).
Change-Id: I369848c0f507e6bf5ff9ab4a60663bbbda6edc52
The port to for this two targets has been entirely developped by Ilia Sergachev (alias Il or xzcc). His source
can be found at https://bitbucket.org/isergachev/rockbox . The few necesary modifications for the DX90 port
was done by headwhacker form head-fi.org. Unfortunately i could not try out the final state of the DX90 port.
The port is hosted on android (without java) as standalone app. The official Firmware is required to run this port.
Ilia did modify the source files for the "android" target in the rockbox source to make the DX port work. The work I did
was to separate the code for DX50 (&DX90) from the android target.
On this Target Ilia used source from tinyalsa from AOSP. I did not touch that part of the code because I do not understand it.
What else I changed from Ilias sources besides the separation from the target "android":
* removed a dirty hack to keep backlight off
* changed value battery meter to voltage battery meter
* made all plugins compile (named target as "standalone") and added keymaps
* i added the graphics for the manual but did not do anything else for the manual yet
* minor optimizations
known bugs:
* timers are slowed donw when playback is active (tinyalsa related?)
* some minor bugs
Things to do:
* The main prolem will be how to install the app correctly. A guy called DOC2008 added a CWM (by androtab.info) to the
official firmware and Ilia made a CWM installation script and a dualboot selector (rbutils/ibassoboot, build with
ndk-build). We will have to find a way to install rockbox in a proper way without breaking any copyrights.
Maybe ADB is an option but it is not enable with OF by default. Patching the OF is probably the way to go.
* All the wiki and manual
to build:
needed: android ndk installed, android sdk installed with additional build-tools 19.1.0 installed
./tools/configure
select iBasso DX50 or iBasso DX90
make -j apk
the content of rockbox.zip/.rockbox needs to be copied to /system/rockbox/app_rockbox/rockbox/ (rockbox app not needed)
the content of libs/armeabi to /system/rockbox/lib/ (rockbox app needed)
The boot selector is needed as /system/bin/MangoPlayer and the iBasso app as /system/bin/MangoPlayer_original. There
is also the "vold" file. The one from OF does not work with DX50 rockbox (DX90 works!?), the one from Ilia is necessary.
Until we have found a proper way to install it, it can only be installed following the instructions of Ilia on his
bitbucket page, using the CWM-OF and his installation script package.
Change-Id: Ic4faaf84824c162aabcc08e492cee6e0068719d0
Reviewed-on: http://gerrit.rockbox.org/941
Tested: Chiwen Chang <rock1104.tw@yahoo.com.tw>
Reviewed-by: Michael Giacomelli <giac2000@hotmail.com>
A GNU extension that returns dst + size instead of dst. It's a nice
shortcut when copying strings with a known size or back-to-back blocks
and you have to do it often.
May of course be called directly or alternately through
__builtin_mempcpy in some compiler versions.
For ASM on native targets, it is implemented as an alternate entrypoint
to memcpy which adds minimal code and overhead.
Change-Id: I4cbb3483f6df3c1007247fe0a95fd7078737462b
This patch redoes the filesystem code from the FAT driver up to the
clipboard code in onplay.c.
Not every aspect of this is finished therefore it is still "WIP". I
don't wish to do too much at once (haha!). What is left to do is get
dircache back in the sim and find an implementation for the dircache
indicies in the tagcache and playlist code or do something else that
has the same benefit. Leaving these out for now does not make anything
unusable. All the basics are done.
Phone app code should probably get vetted (and app path handling
just plain rewritten as environment expansions); the SDL app and
Android run well.
Main things addressed:
1) Thread safety: There is none right now in the trunk code. Most of
what currently works is luck when multiple threads are involved or
multiple descriptors to the same file are open.
2) POSIX compliance: Many of the functions behave nothing like their
counterparts on a host system. This leads to inconsistent code or very
different behavior from native to hosted. One huge offender was
rename(). Going point by point would fill a book.
3) Actual running RAM usage: Many targets will use less RAM and less
stack space (some more RAM because I upped the number of cache buffers
for large memory). There's very little memory lying fallow in rarely-used
areas (see 'Key core changes' below). Also, all targets may open the same
number of directory streams whereas before those with less than 8MB RAM
were limited to 8, not 12 implying those targets will save slightly
less.
4) Performance: The test_disk plugin shows markedly improved performance,
particularly in the area of (uncached) directory scanning, due partly to
more optimal directory reading and to a better sector cache algorithm.
Uncached times tend to be better while there is a bit of a slowdown in
dircache due to it being a bit heavier of an implementation. It's not
noticeable by a human as far as I can say.
Key core changes:
1) Files and directories share core code and data structures.
2) The filesystem code knows which descriptors refer to same file.
This ensures that changes from one stream are appropriately reflected
in every open descriptor for that file (fileobj_mgr.c).
3) File and directory cache buffers are borrowed from the main sector
cache. This means that when they are not in use by a file, they are not
wasted, but used for the cache. Most of the time, only a few of them
are needed. It also means that adding more file and directory handles
is less expensive. All one must do in ensure a large enough cache to
borrow from.
4) Relative path components are supported and the namespace is unified.
It does not support full relative paths to an implied current directory;
what is does support is use of "." and "..". Adding the former would
not be very difficult. The namespace is unified in the sense that
volumes may be specified several times along with relative parts, e.g.:
"/<0>/foo/../../<1>/bar" :<=> "/<1>/bar".
5) Stack usage is down due to sharing of data, static allocation and
less duplication of strings on the stack. This requires more
serialization than I would like but since the number of threads is
limited to a low number, the tradoff in favor of the stack seems
reasonable.
6) Separates and heirarchicalizes (sic) the SIM and APP filesystem
code. SIM path and volume handling is just like the target. Some
aspects of the APP file code get more straightforward (e.g. no path
hashing is needed).
Dircache:
Deserves its own section. Dircache is new but pays homage to the old.
The old one was not compatible and so it, since it got redone, does
all the stuff it always should have done such as:
1) It may be update and used at any time during the build process.
No longer has one to wait for it to finish building to do basic file
management (create, remove, rename, etc.).
2) It does not need to be either fully scanned or completely disabled;
it can be incomplete (i.e. overfilled, missing paths), still be
of benefit and be correct.
3) Handles mounting and dismounting of individual volumes which means
a full rebuild is not needed just because you pop a new SD card in the
slot. Now, because it reuses its freed entry data, may rebuild only
that volume.
4) Much more fundamental to the file code. When it is built, it is
the keeper of the master file list whether enabled or not ("disabled"
is just a state of the cache). Its must always to ready to be started
and bind all streams opened prior to being enabled.
5) Maintains any short filenames in OEM format which means that it does
not need to be rebuilt when changing the default codepage.
Miscellaneous Compatibility:
1) Update any other code that would otherwise not work such as the
hotswap mounting code in various card drivers.
2) File management: Clipboard needed updating because of the behavioral
changes. Still needs a little more work on some finer points.
3) Remove now-obsolete functionality such as the mutex's "no preempt"
flag (which was only for the prior FAT driver).
4) struct dirinfo uses time_t rather than raw FAT directory entry
time fields. I plan to follow up on genericizing everything there
(i.e. no FAT attributes).
5) unicode.c needed some redoing so that the file code does not try
try to load codepages during a scan, which is actually a problem with
the current code. The default codepage, if any is required, is now
kept in RAM separarately (bufalloced) from codepages specified to
iso_decode() (which must not be bufalloced because the conversion
may be done by playback threads).
Brings with it some additional reusable core code:
1) Revised file functions: Reusable code that does things such as
safe path concatenation and parsing without buffer limitations or
data duplication. Variants that copy or alter the input path may be
based off these.
To do:
1) Put dircache functionality back in the sim. Treating it internally
as a different kind of file system seems the best approach at this
time.
2) Restore use of dircache indexes in the playlist and database or
something effectively the same. Since the cache doesn't have to be
complete in order to be used, not getting a hit on the cache doesn't
unambiguously say if the path exists or not.
Change-Id: Ia30f3082a136253e3a0eae0784e3091d138915c8
Reviewed-on: http://gerrit.rockbox.org/566
Reviewed-by: Michael Sevakis <jethead71@rockbox.org>
Tested: Michael Sevakis <jethead71@rockbox.org>
The changed thread code may not wish to save the old context under
certain circumstances but thread-unix.c assumed it would, cached it
and used it unconditionally.
Also, prevent it from leaking away all the jump buffers (old problem).
Creating and removing threads would eventually run it out of buffers
and then it would crash after that. Plugins, like Pictureflow, which
have worker threads could only be started a few times. Implement a
simple O(1) allocator that will reuse them and stays self-contained
to its own types (as it appears the original author intended).
Change-Id: Icf65413c086b346fb79bf827102b725269e2812c
Abstracts threading from itself a bit, changes the way its queues are
handled and does type hiding for that as well.
Do alot here due to already required major brain surgery.
Threads may now be on a run queue and a wait queue simultaneously so
that the expired timer only has to wake the thread but not remove it
from the wait queue which simplifies the implicit wake handling.
List formats change for wait queues-- doubly-linked, not circular.
Timeout queue is now singly-linked. The run queue is still circular
as before.
Adds a better thread slot allocator that may keep the slot marked as
used regardless of the thread state. Assists in dumping special tasks
that switch_thread was tasked to perform (blocking tasks).
Deletes alot of code yet surprisingly, gets larger than expected.
Well, I'm not not minding that for the time being-- omlettes and break
a few eggs and all that.
Change-Id: I0834d7bb16b2aecb2f63b58886eeda6ae4f29d59
* Seal away private thread and kernel definitions and declarations
into the internal headers in order to better hide internal structure.
* Add a thread-common.c file that keeps shared functions together.
List functions aren't messed with since that's about to be changed to
different ones.
* It is necessary to modify some ARM/PP stuff since GCC was complaining
about constant pool distance and I would rather not force dump it. Just
bl the cache calls in the startup and exit code and let it use veneers
if it must.
* Clean up redundant #includes in relevant areas and reorganize them.
* Expunge useless and dangerous stuff like remove_thread().
Change-Id: I6e22932fad61a9fac30fd1363c071074ee7ab382
