Fix a much of PWM, serial issues, update doc on PWM

PWM frequencies were not being correctly chosen
#778 #787 #762 #779

Author: SpenceKonde

CODE_OF_CONDUCT.md

@@ -0,0 +1,49 @@
+# Contributor Code of Conduct
+
+As contributors and maintainers of this project, we all pledge to treat
+all people who contribute through reporting issues, posting feature
+requests, updating documentation, submitting pull requests or patches,
+and other activities with common decency at minimum.
+
+To this end, there is to be no harassment, personal attacks, or other
+trashtalking directed at any member of the community, whether over any
+racial, sexual, religious, national identity, over personal grudges
+towards other members from here or elsewhere. In general, most of these
+forms of prohibited content involve either off topic posts, spam, or
+being an asshole or a plagerist - so stay on topic, don't post spam and
+don't be an asshole either to others or in general.
+
+Note that mild trashtalk of organizations or companies as a whole (but
+not individuals working there) - Microchip, Atmel, and Arduino are
+certainly valid targets - but only insofar as it relates to a technical
+matter already being discussed.
+
+Also, obviously, images that are disturbing, sexual, or harassing (or
+otherwise which have nothing to do with the repo) should not be posted.
+No content that is illegal is to be included in any post, commit or PR.
+No intimidating, threatening, advertising (whether goods, services or
+political opinions - this does not prohibit brief mentions of products
+sold by contributors which are that are relevant to an existing
+discussion - but only if it constitutes a small portion of the post,
+and the post is concerned primarily with the matter at hand, not your
+commercial product).
+
+No content that is not open source is to be submitted by anyone other
+than the copyright holder. Open source code with a license incompatible
+with LGPL 2.1 may be permitted, but only if it's particularly valuable,
+as was done for the code from adafruitNeopixel which became tinyNeopixel.
+
+Project maintainers have the right and responsibility to remove, edit,
+or reject comments, commits, code, issues, and other contributions that
+are not aligned to this Code of Conduct. Project maintainers or contributors
+who do not follow the Code of Conduct may be banned temporarally or
+permanently, depending on the severity of the offence. To date we have not
+had any instances of such behavior.
+
+Instances of abusive, harassing, or otherwise unacceptable behavior may
+be reported by emailing `spencekonde@gmail. com>` which only goes to to
+Spence Konde.
+
+This Code of Conduct is adapted from the `Contributor
+Covenant <http://contributor-covenant.org>` version 1.0.0, available at
+http://contributor-covenant.org/version/1/0/0/

avr/cores/tiny/HardwareSerial.cpp

@@ -46,6 +46,11 @@
     volatile uint8_t *udr) {
     _rx_buffer = rx_buffer;
     _tx_buffer = tx_buffer;
+    _ubrrh = ubrrh;
+    _ubrrl = ubrrl;
+    _ucsra = ucsra;
+    _ucsrb = ucsrb;
+    _udr = udr;
   }
   #else
   ) {
@@ -90,66 +95,6 @@
     *_ubrrl = baud_setting;
     *_ucsrb = (_rxen | _txen | _rxcie);
 
-    // For fucks sake, the USART registers aren't in low I/O space!
-    // Even if they WERE, SBI and CBI only work when both arguments are compiletimne known
-    // here they *would* be except that classes completely stymie LTO. the 4 lines of SBI and CBI that were here got turned into this:
-    /*
-        sbi(*_ucsrb, _rxen);
-     376: a0 91 4a 01   lds r26, 0x014A ; 0x80014a <Serial+0x16>
-     37a: b0 91 4b 01   lds r27, 0x014B ; 0x80014b <Serial+0x17>
-     37e: 8c 91         ld  r24, X
-     380: 90 91 4e 01   lds r25, 0x014E ; 0x80014e <Serial+0x1a>
-     384: 9f 01         movw  r18, r30
-     386: 01 c0         rjmp  .+2       ; 0x38a <__stack+0x8b>
-     388: 22 0f         add r18, r18
-     38a: 9a 95         dec r25         ; this is a LOOP. Why is there a loop here? to lefthift the bit position into a bitmask.
-                                        ; CLASSES ARE KYRPYONITE TO LTO. Even though we know that all the places where the bit positions
-                                        ; get passed to the constructor, it takes the same value.... the compiler isn't smart enough to see that.
-     38c: ea f7         brpl  .-6       ; 0x388 <__stack+0x89>
-     38e: 82 2b         or  r24, r18
-     390: 8c 93         st  X, r24
-    C:\Users\Spence\Documents\Arduino\hardware\ATTinyCore\avr\cores\tiny/HardwareSerial.cpp:121
-        sbi(*_ucsrb, _txen);
-     392: a0 91 4a 01   lds r26, 0x014A ; 0x80014a <Serial+0x16>
-     396: b0 91 4b 01   lds r27, 0x014B ; 0x80014b <Serial+0x17>
-     39a: 8c 91         ld  r24, X
-     39c: 90 91 4f 01   lds r25, 0x014F ; 0x80014f <Serial+0x1b>
-     3a0: 9f 01         movw  r18, r30
-     3a2: 01 c0         rjmp  .+2       ; 0x3a6 <__stack+0xa7>
-     3a4: 22 0f         add r18, r18
-     3a6: 9a 95         dec r25         ; THIS IS ANOTHER BLOODY LOOP TO ACHIEVE A LEFTSHIFT!
-     3a8: ea f7         brpl  .-6       ; 0x3a4 <__stack+0xa5>
-     3aa: 82 2b         or  r24, r18
-     3ac: 8c 93         st  X, r24
-    C:\Users\Spence\Documents\Arduino\hardware\ATTinyCore\avr\cores\tiny/HardwareSerial.cpp:122
-        sbi(*_ucsrb, _rxcie);
-     3ae: a0 91 4a 01   lds r26, 0x014A ; 0x80014a <Serial+0x16>
-     3b2: b0 91 4b 01   lds r27, 0x014B ; 0x80014b <Serial+0x17>
-     3b6: 8c 91         ld  r24, X
-     3b8: 90 91 50 01   lds r25, 0x0150 ; 0x800150 <Serial+0x1c>
-     3bc: 9f 01         movw  r18, r30
-     3be: 01 c0         rjmp  .+2       ; 0x3c2 <__stack+0xc3>
-     3c0: 22 0f         add r18, r18
-     3c2: 9a 95         dec r25         ; THIS IS ANOTHER BLOODY LOOP TO ACHIEVE A LEFTSHIFT!
-     3c4: ea f7         brpl  .-6       ; 0x3c0 <__stack+0xc1>
-     3c6: 82 2b         or  r24, r18
-     3c8: 8c 93         st  X, r24
-    C:\Users\Spence\Documents\Arduino\hardware\ATTinyCore\avr\cores\tiny/HardwareSerial.cpp:123
-        cbi(*_ucsrb, _udrie);
-     3ca: a0 91 4a 01   lds r26, 0x014A ; 0x80014a <Serial+0x16>
-     3ce: b0 91 4b 01   lds r27, 0x014B ; 0x80014b <Serial+0x17>
-     3d2: 8c 91         ld  r24, X
-     3d4: 90 91 51 01   lds r25, 0x0151 ; 0x800151 <Serial+0x1d>
-     3d8: 01 c0         rjmp  .+2       ; 0x3dc <__stack+0xdd>
-     3da: ee 0f         add r30, r30
-     3dc: 9a 95         dec r25         ; THIS IS ANOTHER BLOODY LOOP TO ACHIEVE A LEFTSHIFT!
-     3de: ea f7         brpl  .-6       ; 0x3da <__stack+0xdb>
-     3e0: e0 95         com r30
-     3e2: e8 23         and r30, r24
-     3e4: ec 93         st  X, r30
-     3e6: ff cf         rjmp  .-2       ; 0x3e6 <__stack+0xe7>
-     So in total that flagraqnt SBI/CBI abuse swallowed 112 bytes flash and had execution time of around 150 clocks
-    */
   #else
     LINCR = (1 << LSWRES);
     LINBRR = (((F_CPU * 10L / 16L / baud) + 5L) / 10L) - 1;