Strange implementation of BRK and JSR

[ldm0]

https://www.masswerk.at/6502/6502_instruction_set.html#BRK https://www.masswerk.at/6502/6502_instruction_set.html#JSR

According to the documentation, BRK and JSR pushes PC + 2 to stack.

But:

• https://github.com/whscullin/apple1js/blob/main/js/cpu6502.js#L458
• https://github.com/whscullin/apple1js/blob/main/js/cpu6502.js#L831

It seems that BRK pushes PC, and JSR pushes PC + 1 to stack.

[whscullin]

I'm not sure what they mean by PC + 2, readBytePC() at the beginning increments the PC once more in addition to the initial increment after reading the opcode. This is in line with what is described in Table 4.1 in Jim Sather's Understanding the Apple II - and although admittedly BRK/RTI might not get exercised a whole lot, JSR/RTS simply wouldn't work any other way.

The 6502 code from sister project apple2js has been tested against the Tom Harte test suite that validates cycle and memory accuracy against a real CPU, the logic there, although re-written in Typescript, is unchanged.