On 2/21/09, Mark Cave-Ayland mark.cave-ayland@siriusit.co.uk wrote:
Stefan Reinauer wrote:
I think you indeed hit a bug. b?branch in openbios only works in word definitions, not in interpretion mode. It needs similar handling as setup-tmp-comp/execute-tmp-comp in the definition of "if"
in bootstrap.fs.
No way! I was actually looking at the code in bootstrap.fs thinking that the code assumed it was working inside a temporary buffer, and wondering if the setup-tmp-comp word would help me. Thanks for the pointer.
In addition to that your example contains non-valid code. A b?branch always needs to point behind a b(>resolve) as in the below example.
here cc c, \ offset16 \ a4 c, \ -1 a5 c, \ 0 14 c, 00 c, 05 c, \ b?branch 02 c, 16 c, \ abort b2 c, \ b(>resolve) 9f c, \ .s 00 c, \ end0 1 byte-load
Indeed, that is my mistake. Please find the attached patch, which when used with your code example above now seems to do the right thing for b?branch:
0 > here ok 1 > cc c, ok 1 > a5 c, ok 1 > 14 c, 00 c, 05 c, ok 1 > 02 c, 16 c, ok 1 > b2 c, ok 1 > 9f c, ok 1 > 00 c, ok 1 > 1 byte-load <e> -1 1 0 -1 0 0 0 ffd79378 0 0 0 0 0 0 ok
0 > here ok 1 > cc c, ok 1 > a4 c, ok 1 > 14 c, 00 c, 05 c, ok 1 > 02 c, 16 c, ok 1 > b2 c, ok 1 > 9f c, ok 1 > 00 c, ok 1 > 1 byte-load byte-load: exception caught! ok
At least now if I find similar errors, I have an idea of how to fix them :)
Thanks, I'll apply this. My marTux and Milax test images get a little further with this patch.