> From: Paul Koning <paulkon...@comcast.net> > Date: Wed, 26 Apr 2023 21:02:31 -0400
> > On Apr 26, 2023, at 8:05 PM, Hans-Peter Nilsson <h...@axis.com> wrote: > > > > Not many targets define this besides msp430, pdp1, xtensa, > > and arm compared to those that appear to unconditionally > > have a hardware division instruction (also, pdp11 and > > msp430 seem confused and should be empty instead of "1" ... > > How so, "confused"? The documentation says it should be > defined, it doesn't say that it should be defined as > empty. What goes wrong if it's defined as 1 rather than > empty? Only future edits, expecting action to follow as if it was a non-zero expression like many of the target macros. > The documentation is also somewhat misleading, because it > says to define it if the hardware has no divide > instruction. The more accurate statement is that it > should be defined if the hardware has no 64 / 32 bit > divide hardware support. pdp11.h points this out in a > comment, because most pdp11s do have divide instructions > but those are for 32 / 16 bits. That might be true, and I've heard patches are welcome. brgds, H-P
