Where is Pnut v34n and how do I get it?

Comments

• whicker Posts: 749

  2020-05-30 04:08

  So propeller.parallax.com needs some updating. Scroll down in this same Propeller 2 forum area and you'll find it.
• Cluso99 Posts: 18,069

  2020-05-30 04:52

  You need the latest v34s here
  forums.parallax.com/discussion/171196/pnut-spin2-latest-version-v34s/p1
• evanh Posts: 15,187

  2020-05-30 05:19 edited 2020-05-31 01:47

  v34L is the last release without the hubexec regression in pure Pasm2. Chip has been preoccupied with Spin2 development.
  

  zip

  zip

  PNut_v34L.zip

  281K

  PNut_v34L.zip 281.3K
• Cluso99 Posts: 18,069

  2020-05-30 20:25 edited 2020-05-30 20:26

  evanh wrote: »

  v34L is the last release without the hubexec regression in pure Pasm2. Chip has been preoccupied with Spin2 development.
  

  What do you mean “hubexec regression” please?
  I can still use pnut v34s to compile pasm only programs with or without hubexec.
  Am I missing something?
• evanh Posts: 15,187

  2020-05-30 22:17

  Symbol addresses for long branching (20-bit immediate) from hubexec to hubexec is messed up.
  
  
• evanh Posts: 15,187

  2020-05-30 23:04 edited 2020-05-30 23:09

  Looking at Pnut's lst file, the twelve high bits of the symbol addresses much have meaning. It's notable that all the hub addresses have $fff at the top. Eg: first two here are in cogRAM, the second two are in hubRAM

  ...
  TYPE: DAT_LONG        VALUE: 00800020          NAME: LOCTEST1
  TYPE: DAT_LONG        VALUE: 00900024          NAME: _MAIN
  TYPE: DAT_LONG        VALUE: FFF00110          NAME: LOCTEST2
  TYPE: DAT_BYTE        VALUE: FFF00450          NAME: LOCTESTHUB
  ...
  

  
  EDIT: Seems to be the same for all Pnut versions. Maybe something in the handling of these has changed. :shrug:
  
  
• evanh Posts: 15,187

  2020-05-30 23:34 edited 2020-05-30 23:40

  Not surprisingly, it affects LOCs to hub symbols when in hubexec as well. There is a +748 byte offset added to the target symbol address.
  
  Doesn't affect numerical immediate addresses. Only referencing symbols in hubRAM and only when referenced from hubexec and only for 20-bit immediate encodings. So CALL #, JMP #, and LOC #.
  
  
• Cluso99 Posts: 18,069

  2020-05-31 00:18 edited 2020-05-31 00:19

  Thanks Evan. I had forgotten about those and they are likely to cause problems because my calls/jumps are likely to be large and cause the 20bit addressing to be used. I've had all sorts of issues in the past that i've never been able to pin down, what with all the other addressing problems I've had.
  Anyway, since I cannot live with the spin2 problems in pnut, I'm back to total pasm and fastspin is much better there because of conditional compilation and include files.
• evanh Posts: 15,187

  2020-05-31 01:48

  I've attached the v34L zip file to my first post.