On Tue, May 29, 2012 at 7:02 PM, R. Diez <[email protected]> wrote: > Hallo Julius: > >> [...] > >> I created a bug and applied these fixes: >> http://bugzilla.opencores.org/bugzilla4/show_bug.cgi?id=91 > > > > I have re-run my modified test suite once more against the patched ORPSoC > V2, and a different test fails now. That test used to succeed before the > patch. It's the second test in asm-sub.S : > > // Subtract two small positive numbers. Sets the carry, but never the > overflow if the result is negative. > > OK -> TEST_INST_FF_I32_I32 0, SPR_SR_CY | SPR_SR_OV, l.sub, 0x00000003, > 0x00000002 > > fails -> TEST_INST_FF_I32_I32 0, SPR_SR_CY | SPR_SR_OV, l.sub, 0x00000001, > 0x00000002 > > > According to the test case (and to or1ksim's results), the overflow flag > should _not_ be set in the case, but it is with the new ORPSoC V2.
OK, I've looked into this. I certainly think that last fix was not sufficient. I have expanded the ORPSoC overflow flag testsuite and got a better fix. I've marked the bug as open again, too. If I submit an RTL patch, are you happy to test it against your suite before I check this in? Thanks, Julius _______________________________________________ OpenRISC mailing list [email protected] http://lists.openrisc.net/listinfo/openrisc
