CPU page crossing?

NEStress now gives me an error in absolute with X/Y regarding wrapping.
In other words, it's all about page crossing.
When a page cross occurs, an extra cycle is generated, but what's the address for (dummy) reading?

256 bytes less. (i.e. before the carry to the upper byte could happen)

Still gives me the same error. Strange. Am I missing something?

NESStress doesn't seem to pay attention to the dummy reads at all ... at least not in these tests. Are you handling overflow of addresses from $FFxx down to $00xx?

Bingo. Thanks for the help.