Why does the JSR operation not jump?

Question

I am building a computer based on the 65C02S, and when stepping through the code, I see strange things happening.

This is my code:

SEI
CLD
LDX #$ff
TXS

JSR someRoutine
...

This translates to this behaviour (when stepping though and monitoring with an Arduino):

READ  address:   80 0d   1000.0000 0000.1101   data:   78   0111.1000 SEI
READ  address:   80 0e   1000.0000 0000.1110   data:   d8   1101.1000 CLD
READ  address:   80 0e   1000.0000 0000.1110   data:   d8   1101.1000 d8 
READ  address:   80 0f   1000.0000 0000.1111   data:   a2   1010.0010 LDX
READ  address:   80 0f   1000.0000 0000.1111   data:   a2   1010.0010 a2 
READ  address:   80 10   1000.0000 0001.0000   data:   ff   1111.1111 ff 
READ  address:   80 11   1000.0000 0001.0001   data:   9a   1001.1010 TXS
READ  address:   80 12   1000.0000 0001.0010   data:   20   0010.0000 JSR
READ  address:   80 12   1000.0000 0001.0010   data:   20   0010.0000 JSR
READ  address:   80 13   1000.0000 0001.0011   data:   28   0010.1000 28 
READ  address:   01 ff   0000.0001 1111.1111   data:   80   1000.0000 ---
WRITE address:   01 ff   0000.0001 1111.1111   data:   80   1000.0000 ---
WRITE address:   01 fe   0000.0001 1111.1110   data:   ff   1111.1111 ---    ; Why write FF?
READ  address:   80 14   1000.0000 0001.0100   data:   80   1000.0000 80 
READ  address:   ff ff   1111.1111 1111.1111   data:   ff   1111.1111 ---
READ  address:   80 14   1000.0000 0001.0100   data:   80   1000.0000 80     ; Why not jump?
READ  address:   80 15   1000.0000 0001.0101   data:   a9   1010.1001 LDA

As you can see, the WRITEs are already weird (why write 80 ff, when the PC is at 80 12 at that point?), but also, the JSR doesn't seem to jump at all, it just continues at 80 14.

I first thought it might be some noise somewhere, but the behaviour is always the same. I also tried adding the reset code (SEI, etc), but that just gives slightly different, incorrect, results. I am trying to figure out if my code is wrong, or that I should look at other parts somewhere (noise, design ...)

For clarity, this is the binary code that I am running:

00000000  48 65 6c 6c 6f 20 57 6f  72 6c 64 21 00 78 d8 a2  |Hello World!.x..|
00000010  ff 9a 20 28 80 a9 00 85  10 a9 80 85 11 20 5d 80  |.. (......... ].|
00000020  ea 4c 20 80 a2 ff 9a 60  a9 00 8d f8 7f 8d f9 7f  |.L ....`........|
00000030  a9 30 20 45 80 a9 01 20  45 80 a9 0f 20 45 80 60  |.0 E... E... E.`|
00000040  3a 8d f9 7f 60 8d f8 7f  a9 01 8d f9 7f 20 40 80  |:...`........ @.|
00000050  60 8d f8 7f a9 05 8d f9  7f 20 40 80 60 a0 00 b1  |`........ @.`...|
00000060  10 f0 07 20 51 80 c8 4c  5f 80 60                 |... Q..L_.`|

Update I created a very simple program that loops forever, with a simple JMP and that runs fine. Which make me think noise is not a problem here.

Update This is how I attach the Arduino (pin 2 is connected to PHI2 (pin 37) of the CPU):

#define CLOCK 2

pinMode(CLOCK, INPUT_PULLUP);
attachInterrupt(digitalPinToInterrupt(CLOCK), onClock, FALLING);

Update I attached an oscilloscope to the databus and see that at some point the level is 1.5V instead of 0 or 5, so I think the design is failing somewhere...

Update this is my design:

Answer 1

The 6502 cannot be single stepped, and neither can the 65C02, as they use DRAM cells for their internal registers.

If you want single-stepping, you have to buy a 65C02S.