On Mar 7, 2019, "Maciej W. Rozycki" <ma...@linux-mips.org> wrote:
> So this seems backwards to me, port I/O is supposed to be strongly > ordered, so if removing the ordering guarantee "fixes" your problem, then > there must be a second bottom here. Well, it partially restores the earlier state, so the actual bug goes back to latent. We just have to find out what it is. My plan, in the next time slot I can devote to this investigation, is to try to bisect source files that use these interfaces to identify a minimal set that can switch to the new barriers without breaking anything, then report back after failing to make sense of it myself ;-) So, yeah, several more bottoms to dig through. > Does your platform use `war_io_reorder_wmb'? Err... I'm not sure I understand your question. It uses it in __BUILD_IOPORT_SINGLE within the expanded out function, given !barrier, but you already knew that. Did you mean to ask what war_io_reorder_wmb expand to, or whether there are other uses of war_io_reorder_wmb, or what? -- Alexandre Oliva, freedom fighter https://FSFLA.org/blogs/lxo Be the change, be Free! FSF Latin America board member GNU Toolchain Engineer Free Software Evangelist Hay que enGNUrecerse, pero sin perder la terGNUra jamás-GNUChe