This PR adds the possibility of specifying a memory error callback in
the "ram" element of the "config" section. The new configuration
attribute is called "memory_error_callback" and its value is the name of
a function with the signature "void f(size_t)" that will be called
automatically in the following cases:
- when malloc, calloc, or realloc (with the 'size' argument greater than
0) return NULL. The callback function will be called with the 'size'
argument of the allocation function.
- when the allocator encounters an internal error (only when using the
'multiple' allocator). The callback function will be called with the
argument set to 0.
For now, this feature is only availabsle when using the multiple
allocator or the simple allocator, not the built-in libc allocator.
Usage example (in the board configuration file):
```
config = {
ram = { memory_error_callback = "memory_error" }
}
```
The original implementation of net.accept did not unlisten the port
after returing. So additional clients can connect to the port, but the
connection is not used. The easiest way to solve this, is to make an
unlisten() call before returning from accept. But I think it is better
to separate listen/unlisten from accept and let the system allow
accepting new connections in the background which can be later taken
with a call to accept. This allows for example using a coroutine waiting
for new connections with a non-blocking accept loop and a yield call.
This commit contains everything to implement the new behaviour and also
add the methods net.listen and net.unlisten. To be compatible with the
old semantics net.accept automatically calls listen to the port.
In addtion the error handling of accept was changed in a way that a
timeout returns the new return value net.ERR_WAIT_TIMEDOUT instead of
returning -1. I find this more consistent.
Before this, you could put the console on USB CDC without BUILD_USB_CDC
and eLua compiled OK but nothing worked. This change makes it spit out
an appropriate message if that should happen again.
It was being set to ( ( 1LL << 31 ) - 2 ) which only used 31 bits when 32
are available. By setting it to ( ( 1LL << 32 ) - 2 ) the range of the timer
is doubled to around 4295 seconds. Note that time values greater than
approx 2147 seconds will have the top bit (1 << 31) set and so will appear
to be negative as all numbers are considered to be signed.
Now, can.send() returns a boolean result to indicate whether the message
was successfully sent or not. The various platform functions differ in how
they handled the situation where all CAN transmitters are occupied. In
particular, the STM32x and STR9 implementations ignored the status returns
from the underlying libraries and so if all the CAN transmitters were busy,
the latest message simply got thrown away. Now, can.send() returns true/false
to indicate if the message really did get queued for transmission. All
platforms should behave the same as before.
The material that used to be in platform_conf.h has
been split to several files, including:
platform_ints.h, cpu_xxxx.h, and in the board.lua.
Still more fixes needed in documentation.
This should really be more than one commit, but here it goes anyway:
- added STM32F4 CDC UART support. For some reason, it seems to work only
on the OTG_FS interface, the OTG_HS interface enumerates it, but
doesn't seem to send/receive any data.
- removed the "platform interface" functions for CDC UART. That was
never really a platform interface, just a couple of functions reading
and writing data from/to a CDC UART. Now each backend takes care of this
as a special case in its platform_uart_xxx functions.
- added buffering support for CDC UARTs
- added uart.CDC to the uart module, so the CDC uart can be used directly
from Lua.
- stm32f4discovery now defaults to using the CDC, since it doesn't have
a dedicated UART connector.
Conflicts:
.gitignore
src/common_tmr.c
src/platform/lm3s/platform.c
src/platform/lm3s/platform_conf.h
src/platform/sim/platform_conf.h
src/platform/stm32/platform_conf.h
src/platform/str9/platform_conf.h
utils/build.lua
Also fixed some minor issues in the builder.
There is a new command in the shell ('mv' - move/rename files).
Also, the shell code was refactored into its own directory (src/shell).
The 'help' command in the shell was updated and supports 'help <command>'.
There is a new BUILD_ADVANCED_SHELL build time macro that enables the
advanced shell features (currently 'cp' with recursion (and other improvements),
'mv' and 'rm').
- For now only supported for MMCFS
- 'dir' function in shell augmented to show directories and
to traverse a path recursively if requested
- new command 'mkdir' in shell
This should really be more than one commit, but I wrote everything in one
shot and I don't feel like arranging the changes logically into different
commits. So, these are the changes:
- added WOFS (Write Once File System). This is a writeable file system that
exists in the MCU's internal Flash memory and allows files to be written,
but only once, in a single shot. More details to follow.
- the platform interface has a new MCU flash access interface.
- added WOFS "reference implementations" for two CPUs: LM3S8962 and
STM32F103RE. They are easily extendable to other CPUs in the same platform
and can be taken as a model for other platforms.
- the ROMFS file layout in memory was slightly changed.
- the simulator (src/platform/sim) got a new function (lseek).
- shell: now each shell command receives its arguments in a C-main-style
(argc, argv) pair. This was originally Marcelo's idea and it finally
made it to the master (although this particular implementation is mine),
after I got fed up with all the argument parsing in the shell functions.
- new shell command: wofmt ("formats" a WOFS, effectively clearing it).
- a couple of small fixes in the shell code
All the functions that implement a FS receive the instance data
of the FS (given at registration time to dm_register) as their
last argument. ROMFS was changed to take advantage of this.
Now it's possible to have more than one instance of a given file
system. For example, one could use more that one ROM file system
in different physical locations (a possible configuration is
internal Flash and external serial memories). This mechanism is
currently implemented only in the device manager (devman.c),
actual instance implementation require per-FS support (to be
implemented later).
This patch adds more RAM optimizations to eLua:
- direct file memory mapping: files in ROMFS will be read directly from Flash,
without allocating any additional buffers. This doesn't help with RAM
consumption in itself, but enables the set of optimizations below.
- pseudo read-only strings. These are still TStrings, but the actual string
content can point directly to Flash. Original Lua strings are kept in
TStrings structures (lobject.h):
typedef union TString {
L_Umaxalign dummy; /* ensures maximum alignment for strings */
struct {
CommonHeader;
lu_byte reserved;
unsigned int hash;
size_t len;
} tsv;
} TString;
The actual string content comes right after the union TString above.
Pseudo RO strings have the same header, but instead of having the string
content after TString, they have a pointer that points to the actual
string content (which should exist in a RO memory (Flash) that is directly
accesbile from the MCU bus (like its internal Flash memory)). lua_newlstr
detects automatically if it should create a regular string or a pseudo RO
string by checking if the string pointer comes from the Flash region of the
MCU. This optimization works for both precompiled (.lc) files that exist in
ROMFS and for internal Lua strings (C code).
- functions in Flash: for precompiled (.lc) files that exist in ROMFS, the code
of the functions and a part of the debug information will be read directly
from Flash.
- ROMFS was changed to support files that are larger than 2**16 bytes and it
aligns all its files to an offset which is a multiple of 4 in order to prevent
data alignment issues with precompiled Lua code.
- the Lua bytecode dumper was changed to align all the instructions in a Lua
function and a part of the debug information to an offset which is a multiple
of 4. This might slightly increase the size of the precompiled Lua file.
These changes were succesfully checked against the Lua 5.1 test suite.
These changes were tested in eLua on LM3S and AVR32.
This patch undoes the change in the order of timer parameters for:
net.accept() net.recv() tmr.delay() tmr.setclock() tmr.set_match_int()
tmr.gettimediff() and also changes the new tmr.getdiffnow() to have the
same parameter ordering as the others.
The default timer ID (the systimer), which previously was obtained with an
optional last parameter, is now obtained by supplying nil as the timer ID.
tmr.getmaxdelay() used to return -1 for the system timer and -2 for
virtual timers in the integer build due to Lua integers being signed.
This makes them return 2147483647 and 2147483646 resectively.
Fixed an error which made the getmindelay/getmaxdelay functions
invalid (thanks to Martin for spotting this). Also refactored the
code that computes min/max delay from platform files to common code.
