[Bug rtl-optimization/115182] New: [15 Regression] gcc.target/cris/pr93372-47.c at r15-518-g99b1daae18c095

[Bug rtl-optimization/115182] New: [15 Regression] gcc.target/cris/pr93372-47.c at r15-518-g99b1daae18c095

[hp at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115182

            Bug ID: 115182
           Summary: [15 Regression] gcc.target/cris/pr93372-47.c at
                    r15-518-g99b1daae18c095
           Product: gcc
           Version: 15.0
            Status: UNCONFIRMED
          Keywords: missed-optimization, testsuite-fail
          Severity: normal
          Priority: P3
         Component: rtl-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: hp at gcc dot gnu.org
                CC: law at gcc dot gnu.org, pinskia at gcc dot gnu.org,
                    rguenth at gcc dot gnu.org, unassigned at gcc dot gnu.org
        Depends on: 115144
  Target Milestone: ---
            Target: cris-elf

+++ This bug was initially created as a clone of Bug #115144 +++

This bug is only about the unfilled delay-slot that caused
gcc.target/cris/pr93372-47.c to fail starting at r15-518-g99b1daae18c095; not
about the incidental (possibly generic but at least unrelated to
delay-slot-filling) performance regression I noticed.  The bug seems to be due
to a bug in reorg.cc or actually, in resource.cc.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115144
[Bug 115144] [15 Regression] 2% performance regression for some codes with
r15-518-g99b1daae18c095

[Bug rtl-optimization/115182] [15 Regression] gcc.target/cris/pr93372-47.c at r15-518-g99b1daae18c095

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115182

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |15.0

[Bug rtl-optimization/115182] [15 Regression] gcc.target/cris/pr93372-47.c at r15-518-g99b1daae18c095

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115182
Bug 115182 depends on bug 115144, which changed state.

Bug 115144 Summary: [15 Regression] 2% performance regression for some codes with r15-518-g99b1daae18c095
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115144

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|---                         |FIXED

[Bug rtl-optimization/115182] [15 Regression] gcc.target/cris/pr93372-47.c at r15-518-g99b1daae18c095

[hp at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115182

Hans-Peter Nilsson <hp at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2024-05-25
             Status|UNCONFIRMED                 |ASSIGNED
     Ever confirmed|0                           |1

[Bug rtl-optimization/115182] [15 Regression] gcc.target/cris/pr93372-47.c at r15-518-g99b1daae18c095

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115182

--- Comment #1 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Hans-Peter Nilsson <hp@gcc.gnu.org>:

https://gcc.gnu.org/g:84b4ed45ea81ed5c4fb656a17846b26071c23e7d

commit r15-877-g84b4ed45ea81ed5c4fb656a17846b26071c23e7d
Author: Hans-Peter Nilsson <hp@axis.com>
Date:   Tue May 28 23:15:57 2024 +0200

    resource.cc (mark_target_live_regs): Don't look past target insn, PR115182

    The PR115182 regression is that a delay-slot for a conditional branch,
    is no longer filled with an insn that has been "sunk" because of
    r15-518-g99b1daae18c095, for cris-elf w. -O2 -march=v10.

    There are still sufficient "nearby" dependency-less insns that the
    delay-slot shouldn't be empty.  In particular there's one candidate in
    the loop, right after an off-ramp branch, off the loop: a move from
    $r9 to $r3.

            beq .L2
            nop
            move.d $r9,$r3

    But, the resource.cc data-flow-analysis incorrectly says it collides
    with registers "live" at that .L2 off-ramp.  The off-ramp insns
    (inlined from simple_rand) look like this (left-to-right direction):

    .L2:
            move.d $r12,[_seed.0]
            move.d $r13,[_seed.0+4]
            ret
            movem [$sp+],$r8

    So, a store of a long long to _seed, a return instruction and a
    restoring multi-register-load of r0..r8 (all callee-saved registers)
    in the delay-slot of the return insn.  The return-value is kept in
    $r10,$r11 so in total $r10..$r13 live plus the stack-pointer and
    return-address registers.  But, mark_target_live_regs says that
    $r0..$r8 are also live because it *includes the registers live for the
    return instruction*!  While they "come alive" after the movem, they
    certainly aren't live at the "off-ramp" .L2 label.

    The problem is in mark_target_live_regs: it consults a hash-table
    indexed by insn uid, where it tracks the currently live registers with
    a "generation" count to handle when it moves around insn, filling
    delay-slots.  As a fall-back, it starts with registers live at the
    start of each basic block, calculated by the comparatively modern df
    machinery (except that it can fail finding out which basic block an
    insn belongs to, at which times it includes all registers film at 11),
    and tracks the semantics of insns up to each insn.

    You'd think that's all that should be done, but then for some reason
    it *also* looks at insns *after the target insn* up to a few branches,
    and includes that in the set of live registers!  This is the code in
    mark_target_live_regs that starts with the call to
    find_dead_or_set_registers.  I couldn't make sense of it, so I looked
    at its history, and I think I found the cause; it's a thinko or
    possibly two thinkos.  The original implementation, gcc-git-described
    as r0-97-g9c7e297806a27f, later moved from reorg.c to resource.c in
    r0-20470-gca545bb569b756.

    I believe the "extra" lookup was intended to counter flaws in the
    reorg.c/resource.c register liveness analysis; to inspect insns along
    the execution paths to exclude registers that, when looking at
    subsequent insns, weren't live.  That guess is backed by a sentence in
    the updated (i.e. deleted) part of the function head comment for
    mark_target_live_regs: "Next, scan forward from TARGET looking for
    things set or clobbered before they are used.  These are not live."
    To me that sounds like flawed register-liveness data.

    An epilogue expanded as RTX (i.e. not just assembly code emitted as
    text) is introduced in basepoints/gcc-0-1334-gbdac5f5848fb, so before
    that time, nobody would notice that saved registers were included as
    live registers in delay-slots in "next-to-last" basic blocks.

    Then in r0-24783-g96e9c98d59cc40, the intersection ("and") was changed
    to a union ("or"), i.e. it added to the set of live registers instead
    of thinning it out.  In the gcc-patches archives, I see the patch
    submission doesn't offer a C test-case and only has RTX snippets
    (apparently for SPARC).  The message does admit that the change goes
    "against what the comments in the code say":
    https://gcc.gnu.org/pipermail/gcc-patches/1999-November/021836.html
    It looks like this was related to a bug with register liveness info
    messed up when moving a "delay-slotted" insn from one slot to another.
    But, I can't help but thinking it's just papering over a register
    liveness bug elsewhere.

    I think, with a reliable "DF_LR_IN", the whole thing *after* tracking
    from start-of-bb up to the target insn should be removed; thus.

    This patch also removes the now-unused find_dead_or_set_registers
    function.

    At r15-518, it fixes the issue for CRIS and improves coremark scores
    at -O2 -march=v10 a tiny bit (about 0.05%).

            PR rtl-optimization/115182
            * resource.cc (mark_target_live_regs): Don't look for
            unconditional branches after the target to improve on the
            register liveness.
            (find_dead_or_set_registers): Remove unused function.

[Bug rtl-optimization/115182] [15 Regression] gcc.target/cris/pr93372-47.c at r15-518-g99b1daae18c095

[hp at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115182

Hans-Peter Nilsson <hp at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|ASSIGNED                    |RESOLVED

--- Comment #2 from Hans-Peter Nilsson <hp at gcc dot gnu.org> ---
Final verification: pr93372-47.c again passed in my autotester, at
r15-884-ge5fc5d42d25c86.