On 06/12/2022 16:23, Richard Henderson wrote:
On 12/6/22 10:02, Peter Maydell wrote:
On Tue, 6 Dec 2022 at 15:56, Philippe Mathieu-Daudé <phi...@linaro.org> wrote:
On 6/12/22 13:30, Dr. David Alan Gilbert wrote:
I don't know that bit of qemu well enough to know whether the cpu part
of qemu should be splitting the unaligned accesses or not.
All I/O accesses are gated thru access_with_adjusted_size() in
softmmu/memory.c.
There is an old access_with_adjusted_size_unaligned() version [1] from
Andrew and a more recent series [2] from Richard. Maybe the latter fixes
some long-standing bug [3] we have here?
There definitely are some unaddressed bugs there -- maybe this
is the time to work through what semantics we want that
softmmu code to provide and fix the bugs...
Yes, indeed. Let's not forget Mark C-A's m68k bug[1] which so far has no
resolution.
r~
[1] https://gitlab.com/qemu-project/qemu/-/issues/360
That would definitely be useful: since Richard worked on this series, I managed to
develop a hack that allows me to work around the issue for my particular use-case
which is why I haven't been focusing on this.
The main concerns are listed in the above issue at
https://gitlab.com/qemu-project/qemu/-/issues/360#note_597130838. Defining the
behaviour doesn't seem too bad, but it is likely some things that unintentionally
depend upon the existing behaviour will break.
ATB,
Mark.
