* Added constant time strings comparison to avoid possible time-based attacks
* Fixed data types
* Fixed indentation
* Moved string comnparison in constant time to String class; modified function body to assure constant time comparison despite compiler optimizations
* Removed wrong code
* Fixed error and prevented compiler optimization to delete u1 local variable
* Avoid timing attacks on string comparison
* Minor
* changed counter names, removed else
* Add a workaround-delay in Serial.flush()
In relation to #2536 and #2502
Tested at 80MHz and 160MHz with flash-frequency at both 40MHz and 80MHz, the bug mentioned in the above issues manifests in all cases. The proposed workaround seems to work fine, I tested at 2400bps, 9600bps, 115200bps, 230400bps and 2Mbps and didn't see anomalous output.
* Remove extraneous character
* Update HardwareSerial.cpp
SDK uses final 4 sectors of flash for configuration data.
ESP.eraseConfig would only erase 2 sectors, so in some cases of
corrupted data ("system param error"), users could not fix the issue
using ESP.eraseConfig, and had to use esptool instead.
Thanks @HugoML for reporting this.
realloc() is called with newSize > 0 (at least 16), so newbuffer==0 means the old memory was not deallocated. Therefore, the pointer should still point to the old buffer. This change should resolve issue #3516.
UMM debugging strings are normally placed in RODATA, which uses up scarse
memory. Move them to PROGMEM and use macros to replace printf with a
version that can handle ROM strings.
Every assert() includes a __FILE__ constant string to RODATA which
can be quite large as it includes the complete path as well as
the filename.
Move that string into PMEM, and update the postmortem to work with
either PMEM or RAM strings for dumping abort/assert/exception
information.
- Update SDK header files and libraries to SDK 2.0.0 plus 2.0.0_16_08_09
patch
- Remove mem_manager.o from libmain.a (replaced with umm_malloc)
- Disable switch from DIO to QIO mode for certain flash chips (saves
IRAM space)
- Add user_rf_cal_sector_set; it points to rf_init_data sector.
- Change the way rf_init_data is spoofed.
This is now done by wrapping spi_flash_read and returning the data we
need during startup sequence.
- Place lwip library into flash using linker script instead of section
attributes (saves IRAM space)
* Suppressed -Wunused-parameter and -Wunused-function by casting to void unused identifiers.
* Explicit initialization of all fields to suppress -Wmissing-field-initializers.
* Fixed signed/unsigned integer comparison.
* memset initialization of structs.
* More -Wunused-parameter fixes.
* Fixed libb64 decoder.
Libb64 decoder works when "char" type is signed. In esp8266 xtensa gcc "char" is unsigned, so libb64 decoder does not work. In the implementation file I replaced "char" to "int8_t" and created function wrappers in order to keep existing interface.
* For style consistency.
* Prevent divide by zero error causing code to crash
As per the issue at #2491, there is a divide by error issue resulting from the specification of 0 as the frequency. This does not appear to affect the AVR implementation, but it crashes on ESP8266s. I have merely removed the division if the frequency is zero, which appears to be giving the expected results (no tone), without any code crashes.
To test, simply load the toneMelody sketch included with the Arduino IDE (Examples -> 02. Digital -> toneMelody) and change the piezo to something else if you need to. On the Witty module used to test this, I could also tell by the wifi led blinking every time the code crashed as the ESP8266 immediately rebooted.
* Use noTone when frequency is zero
When a frequency of zero is given to tone(), instead call noTone() and exit. Placed after some of the initialisation stuff to ensure the pin is mapped as a output, etc. Tested as functional against a Node MCU 1.0 board and the toneMelody example sketch, using GPIO5 (pin D1).
* Errant tab in formatting
* Rest of tabs that crept in from web editor
Defaulted to tabs and 8 indent :sigh:
gettimeofday used to return timestamps which had tv_sec 1000 times smaller than the correct value.
s_bootTime is in seconds, so no need to divide it by 1000.
