fix stack crash in cx16.push_vera_context()

compiler/res/prog8lib/cx16/syslib.p8

@@ -735,49 +735,53 @@ IRQ_SCRATCH_ZPWORD2	.word  0
 		}}
 }
 
-inline asmsub push_vera_context() clobbers(A) {
+asmsub push_vera_context() clobbers(A) {
     ; -- use this at the start of your IRQ handler if it uses Vera registers, to save the state
     %asm {{
+        ; note cannot store this on cpu hardware stack because this gets called as a subroutine
         lda  cx16.VERA_ADDR_L
-        pha
+        sta  _vera_storage
         lda  cx16.VERA_ADDR_M
-        pha
+        sta  _vera_storage+1
         lda  cx16.VERA_ADDR_H
-        pha
+        sta  _vera_storage+2
         lda  cx16.VERA_CTRL
-        pha
+        sta  _vera_storage+3
         eor  #1
         sta  cx16.VERA_CTRL
         lda  cx16.VERA_ADDR_L
-        pha
+        sta  _vera_storage+4
         lda  cx16.VERA_ADDR_M
-        pha
+        sta  _vera_storage+5
         lda  cx16.VERA_ADDR_H
-        pha
+        sta  _vera_storage+6
         lda  cx16.VERA_CTRL
-        pha
+        sta  _vera_storage+7
+        rts
+_vera_storage:  .byte 0,0,0,0,0,0,0,0
     }}
 }
 
-inline asmsub pop_vera_context() clobbers(A) {
+asmsub pop_vera_context() clobbers(A) {
     ; -- use this at the end of your IRQ handler if it uses Vera registers, to restore the state
     %asm {{
-        pla
+        lda  cx16.push_vera_context._vera_storage+7
         sta  cx16.VERA_CTRL
-        pla
+        lda  cx16.push_vera_context._vera_storage+6
         sta  cx16.VERA_ADDR_H
-        pla
+        lda  cx16.push_vera_context._vera_storage+5
         sta  cx16.VERA_ADDR_M
-        pla
+        lda  cx16.push_vera_context._vera_storage+4
         sta  cx16.VERA_ADDR_L
-        pla
+        lda  cx16.push_vera_context._vera_storage+3
         sta  cx16.VERA_CTRL
-        pla
+        lda  cx16.push_vera_context._vera_storage+2
         sta  cx16.VERA_ADDR_H
-        pla
+        lda  cx16.push_vera_context._vera_storage+1
         sta  cx16.VERA_ADDR_M
-        pla
+        lda  cx16.push_vera_context._vera_storage+0
         sta  cx16.VERA_ADDR_L
+        rts
     }}
 }
 

compiler/res/version.txt

@@ -1 +1 @@
-8.4-dev
+8.3.1

compiler/test/TestTypecasts.kt

@@ -706,7 +706,7 @@ main  {
             }
         """
         val result = compileText(C64Target(), false, text, writeAssembly = true)!!
-        result.program.entrypoint.statements.size shouldBe 13
+        result.program.entrypoint.statements.size shouldBe 15
     }
 
     test("invalid typecasts of numbers") {

docs/source/programming.rst

@@ -495,7 +495,9 @@ Breaking out of a loop prematurely is possible with the ``break`` statement.
     languages if this is *not* the case! A for loop from ubyte 10 to ubyte 2, for example, will iterate through
     all values 10, 11, 12, 13, .... 254, 255, 0  (wrapped), 1, 2. In other languages the entire loop will
     be skipped in such cases. But prog8 omits the overhead of an extra loop range check and/or branch for every for loop
-    by assuming the normal ranges.
+    because the most common case is that it is not needed.
+    You should add an explicit range check yourself if the ending value can be less than the start value and
+    a full wrap-around loop is not what you want!
 
 
 Conditional Execution

docs/source/todo.rst

@@ -3,6 +3,10 @@ TODO
 
 For next release
 ^^^^^^^^^^^^^^^^
+- cx16: sys.reset_system() doesn't silence the vera psg (kernel bug in reset vector?)  use reset vera bit?
+- see if we can let for loops skip the loop if end<start, without adding a lot of code size/duplicating the loop condition
+  this is documented behiavor to now loop around but it's too easy to forget about
+
 ...