Re: [Simulavr-devel] RJMP/RCALL/BRBC/BRBS bug & configure error in simulavr-0.1.1, test-case and fix provided

[Theodore A. Roth]

This has already been fixed in cvs. I just ran you test case in the latest 
cvs version and got to the third rjmp insn no problem.

Here's a reference to the fix:

http://savannah.nongnu.org/cgi-bin/viewcvs/simulavr/simulavr/src/avrcore.h.diff?r1=1.6&r2=1.7

For your configure fixes, I just committed these:

http://savannah.nongnu.org/cgi-bin/viewcvs/simulavr/simulavr/config/avr_doc_pdf.m4.diff?r1=1.2&r2=1.3
http://savannah.nongnu.org/cgi-bin/viewcvs/simulavr/simulavr/config/avr_doc_ps.m4.diff?r1=1.1&r2=1.2
http://savannah.nongnu.org/cgi-bin/viewcvs/simulavr/simulavr/config/enable_curses.m4.diff?r1=1.2&r2=1.3

Thanks for the bug report.

Ted Roth

On Tue, 8 Oct 2002, Thibaud Gaillard wrote:

:)Hi,
:)
:)Playing a bit with GCC-generated AVR code, I found simulavr-0.1.1
:)to fail when it comes to handle negative relative jumps that cross
:)the memory low boundary (still works fine crossing the high boundary).
:)This bug impacts RJMP/RCALL/BRBC/BRBS opcodes (jump instructions with
:)relative signed offsets).
:)
:)There is also a "configure" script problem, at least on a sparc-sun-solaris2.x
:)host (/bin/sh does not handle "==" in "test" constructs, only "=").
:)
:)Test-case (rjump_bug.s) and fix (simulavr-0.1.1.patch) both for "configure"
:)and "src/avrcore.c" provided.
:)
:)Thibaud
:)
:)PS: Theodore, I chose to resend this mail to the official mailing list,
:)    please forgive me for the double posting inconvenience :-)
:)
:)