[Bug middle-end/107601] New: Change SLOW_BYTE_ACCESS into WIDEN_MODE_ACCESS_BITFIELD target hook

[Bug middle-end/107601] New: Change SLOW_BYTE_ACCESS into WIDEN_MODE_ACCESS_BITFIELD target hook

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107601

            Bug ID: 107601
           Summary: Change SLOW_BYTE_ACCESS  into
                    WIDEN_MODE_ACCESS_BITFIELD target hook
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Keywords: internal-improvement, missed-optimization
          Severity: normal
          Priority: P3
         Component: middle-end
          Assignee: unassigned at gcc dot gnu.org
          Reporter: pinskia at gcc dot gnu.org
  Target Milestone: ---

This comes up every few years even. The name SLOW_BYTE_ACCESS and documentation
is confusing even and does not describe what it does.
I found that HJL proposed a decent naming target macro back in 2006 which I
will be using as a basis going forward.

HJL's patch
https://gcc.gnu.org/legacy-ml/gcc-patches/2006-09/msg00555.html

Wilco Dijkstra's patch which just removes SLOW_BYTE_ACCESS 
https://gcc.gnu.org/legacy-ml/gcc-patches/2017-11/msg01547.html

This is part of my bit-field lowering work too.

[Bug middle-end/107601] Change SLOW_BYTE_ACCESS into WIDEN_MODE_ACCESS_BITFIELD target hook

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107601

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |pinskia at gcc dot gnu.org
   Last reconfirmed|                            |2022-11-10
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |ASSIGNED

[Bug middle-end/107601] Change SLOW_BYTE_ACCESS into WIDEN_MODE_ACCESS_BITFIELD target hook

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107601

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Note SLOW_SHORT_ACCESS is unused and has been as far as I can tell it was never
used.
I submitted a patch to remove it:
https://gcc.gnu.org/pipermail/gcc-patches/2022-November/605574.html


Note also at one point SLOW_BYTE_ACCESS was used to control the size of bool
type in C++. This is even why on powerpc-darwin bool is 32bits ...

[Bug middle-end/107601] Change SLOW_BYTE_ACCESS into WIDEN_MODE_ACCESS_BITFIELD target hook

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107601

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Related is probably STRICT_ALIGNMENT vs TARGET_SLOW_UNALIGNED_ACCESS

[Bug middle-end/107601] Change SLOW_BYTE_ACCESS into WIDEN_MODE_ACCESS_BITFIELD target hook

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107601

--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #2)
> Related is probably STRICT_ALIGNMENT vs TARGET_SLOW_UNALIGNED_ACCESS

That one comes up every once in a while though less often though
STRICT_ALIGNMENT and TARGET_SLOW_UNALIGNED_ACCESS SHOULD have a well defined
meaning some folks misunderstand them. RS6000 target has a decent usage of them
really. Definitely Something which should be expanded on in the documentation
and made use it is used correctly in the middle-end but that is for a different
bug (which I am going to file).