Commit graph

120 commits

Author SHA1 Message Date
Alex Stewart
1da996ee10 First batch of opensdk additions
Replacements for:
    libmain/misc.o
    libmain/os_cpu_a.o
    libmain/spi_flash.o
    libmain/timers.o
    libmain/uart.o
    libmain/xtensa_context.o
2015-10-10 22:14:08 -07:00
Angus Gratton
c2e3541c05 Disable default line-buffering of stdout
As discussed in #52. Application can re-enable buffering on stdout with setbuf() or setbuf()
2015-10-11 09:21:30 +11:00
Angus Gratton
0c6a8881a4 Fix sdk_uart_rx_one_char implementation, move into sdk_compat.c 2015-10-06 23:48:28 +11:00
Angus Gratton
be3968abf0 stdout: Bring back auto-CRLF behaviour as discussed 2015-10-06 23:24:40 +11:00
Angus Gratton
7823a28ff8 Change default baud rate back to 74906bps, compatible with boot ROM rate
This reverts a behaviour change from 3ceadfc0a6.
2015-10-06 23:11:17 +11:00
Angus Gratton
cd68622292 Merge branch 'master' into uart_h 2015-10-06 18:37:28 +11:00
Angus Gratton
dc0fa4e207 Merge pull request #50 from SuperHouse/fix/interrupts
Fix/interrupts
2015-10-06 18:36:03 +11:00
Angus Gratton
12399c05ef exception_vectors: Use call0 for sdk__xt_int_exit, in case it needs a literal 2015-10-06 18:25:48 +11:00
Angus Gratton
0d35c0c02d exception_vectors.S: Use .Lnnn syntax for local labels, as noted by @foogod 2015-10-06 18:24:35 +11:00
Angus Gratton
0c0668bfa0 app_main: Refactor to use uart.h functions instead of raw register access
This removes the automatic insertion of CR ahead of LF by the SDK putc
function when called inside the SDK (although we'd already replaced
printf, so most were bypassing this putc function anyhow.)
2015-10-06 18:14:05 +11:00
Angus Gratton
3ceadfc0a6 uart.h: Add uart_get_baud/uart_set_baud functions, change default baud rate from 74906 to 115200 2015-10-06 17:49:00 +11:00
Alex Stewart
707d0ed981 Fix issue #52. Add esp/uart.h 2015-10-05 21:51:57 -07:00
Angus Gratton
afd58bc1ff core headers: Add C++ linker compatibility 2015-10-05 18:33:32 +11:00
Angus Gratton
d39b3f6bb6 NMI Handler: Save 512 bytes of RAM via NMI stack space. Add stack overflow detection. 2015-09-29 17:33:15 +10:00
Angus Gratton
697d5b72bc NMIExceptionHandler: Don't save callee-saved registers, fix stack alignment
NMI routine sdk_wDev_ProcessFiq seems to be written in C, meets ABI
calling conventions for callee-saved registers. Not sure why SDK
handlers saved them.

NMI handler now also meets the ABI requirement that stack is 16 byte
aligned (doesn't seem strictly necessary, but can't hurt.)
2015-09-29 15:05:57 +10:00
Angus Gratton
4448380c5e NMIExceptionHandler: Clean up, refactor to use excsave3 for stack pointer 2015-09-29 14:37:33 +10:00
Angus Gratton
89c481c606 Simplify interrupt and RTOS timer tick handlers
RTOS Timer tick handler is now the same as any other ISR.

This causes a few subtle behaviour changes that seem OK but are worth noting:

* RTOS tick handler sdk__xt_timer_int() is now called from one stack
  frame deeper (inside _xt_isr_handler()), whereas before it was called
  from the level above in UserHandleInterrupt. I can't see any way that
  the extra ~40 bytes of stack use here hurt, though.

* sdk__xt_timer_int() was previous called after all other interrupts
  flagged in the handler, now it's called before the TIMER FRC1 & FRC2
  handlers. The tick handler doesn't appear to do anything particularly
  timing intensive, though.

* GPIO interrupt (value 3) is now lower priority than the SPI
  interrupt (value 2), whereas before it would have been called before
  SPI if both interrupts triggered at once.
2015-09-29 13:21:04 +10:00
Angus Gratton
ed8470631f Consolidate interrupt management in core as esp/interrupts.h & esp_interrupts.c 2015-09-28 22:15:40 +10:00
Angus Gratton
fc9d9ff89c Merge branch 'experiments/unaligned_load' into feature/mbedtls_notworking 2015-09-25 16:57:49 +10:00
Alex Stewart
b813489524 Code reformat/cleanup of exception_vectors.S 2015-09-24 22:43:29 -07:00
Alex Stewart
b6c2e4df59 Fix bad memory access in NMI handler
CallNMIExceptionHandler was saving/restoring registers to the wrong offsets,
causing it to potentially corrupt the first two slots of the
LoadStoreErrorHandler's save area.
2015-09-24 20:12:14 -07:00
Angus Gratton
87f77b1021 Refactor Hardware RNG functions to esp/hwrand.h
As suggested by @foogod, thanks!
2015-09-24 20:48:16 +10:00
Angus Gratton
09405a9095 Merge branch 'master' into feature/mbedtls 2015-09-24 20:40:36 +10:00
Angus Gratton
c31b9be9a2 Merge branch 'experiments/unaligned_load' into feature/mbedtls 2015-09-24 20:25:10 +10:00
Alex Stewart
b15d149b09 More LoadStoreErrorHandler performance improvements 2015-09-23 08:51:36 -07:00
Alex Stewart
f1bff97103 Further optimizations for LoadStoreErrorHandler 2015-09-21 22:13:14 -07:00
Alex Stewart
b280f8a725 Fix format strings in app_main.c 2015-09-20 14:09:30 -07:00
Alex Stewart
ba7492756c Merge branch 'master' into open-startup 2015-09-20 14:05:54 -07:00
Angus Gratton
b03f279f74 mbedTLS: Add ESP8266 hardware entropy source discovered by @foogod
Addresses #3, provided turns out to be an effective HWRNG.
2015-09-20 21:15:30 +10:00
Angus Gratton
3aff91c85c Rename GPIO_CONF_PUSH_PULL to GPIO_CONF_OPEN_DRAIN
Seems I got the functionality of this bit inverted when
initially testing.

In testing it also seems open drain mode is ignored on some pins, which
still source current. Needs more investigation though (may be pullups
internal to the ESP modules or set by default in software.)

Relates to #45
2015-09-20 21:13:38 +10:00
Angus Gratton
f31daa7f9a iomux: Fix setting of invalid bits when masking in iomux fields
Fixes regression from 2a939e97

Part of issue #45
2015-09-20 21:13:38 +10:00
Angus Gratton
6c06e5e91e common_macros: Include sys/cdefs.h, defines _Static_assert among others 2015-09-20 21:13:38 +10:00
Angus Gratton
1c8017484c Move c++ operators back to link-time resolutions, part of 'core'
This reverts commit e9b1df5cb5.
2015-09-20 21:13:38 +10:00
Angus Gratton
7be365ef08 gpio.h: Explicit cast to enum type (needed for g++) 2015-09-20 21:13:38 +10:00
Angus Gratton
6416fe7329 Post-merge cleanup of extras/cpp_support 2015-09-20 21:13:38 +10:00
Angus Gratton
94fabc6ceb Add basic C++ new/delete operators, as contributed by @mikejac in #24 2015-09-20 21:13:38 +10:00
Angus Gratton
cc97067fa1 Add C++ support to Makefile, and proof-of-concept simple.cpp example
This is a work in progress based on @mikejac's work.

Missing:
* No 'new' operator.
* I don't think STL is currently supported.
2015-09-20 21:13:38 +10:00
Michael Jacobsen
372827ac42 Mostly updated header files for use in C++ 2015-09-20 21:13:38 +10:00
Angus Gratton
f47aa77982 Moved DoubleLoadStorageHandler to end of functions, improve fast path l8ui performance
This allows a conditional branch that previously required a full jump instruction not to require it.
2015-09-15 15:09:42 +10:00
Angus Gratton
9cd23e08a4 Fix performance regression in 3911a94cc 2015-09-15 15:00:56 +10:00
Angus Gratton
7d779389ab unaligned_load: Fix sign extension, add test case for sign extension
H/T to @foogod again for this.
2015-09-15 12:00:32 +10:00
Angus Gratton
c7a7dfdf82 Remove the unaligned load handler to its own source file as it was getting quite complex
The downside here is needing to use #include so the 'beq' in the
exception vector can land on it directly, save an instruction.

There might be a better way to do this, but it seems hard to "curate"
the order that symbols appear in each section.
2015-09-15 11:34:21 +10:00
Angus Gratton
3911a94cc8 Remove one more branch from l8ui code path
Thanks for @foogod for the heads-up on this.
2015-09-15 11:01:21 +10:00
Angus Gratton
26e1ce325e Fix sign extension as suggested by @foogod
Also removes a branch from l8ui path, saves some cycles
2015-09-14 18:30:11 +10:00
Angus Gratton
259c9b643f Exception vectors: Remove no-op syscallhandler 2015-09-14 18:17:45 +10:00
Angus Gratton
0e0c5d89b8 NMI exception handler: Fix reserving too much RAM for registers 2015-09-14 17:43:29 +10:00
Angus Gratton
9c15b757d3 Exception handlers: remove unused NMIHandlerAddress 2015-09-14 17:22:50 +10:00
Angus Gratton
0432ae3745 Use GPIO2 for led_debug.S 2015-09-14 17:20:00 +10:00
Angus Gratton
45d4e0da2d Exception vectors: Rearrange to use absolute offets instead of .align hacks 2015-09-14 17:20:00 +10:00
Angus Gratton
a5a179beef Extended unaligned load support to work from DoubleExceptionVector
Allows handling bad loads that occur inside level 1 exception handlers.
2015-09-14 17:20:00 +10:00