[Bug rtl-optimization/104814] New: [10/11/12 Regression] ifcvt: Deleting live variable in IF-CASE-2

[Bug rtl-optimization/104814] New: [10/11/12 Regression] ifcvt: Deleting live variable in IF-CASE-2

[stefansf at linux dot ibm.com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

            Bug ID: 104814
           Summary: [10/11/12 Regression] ifcvt: Deleting live variable in
                    IF-CASE-2
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: rtl-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: stefansf at linux dot ibm.com
  Target Milestone: ---
            Target: s390x-*-*

Created attachment 52570
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52570&action=edit
dump ce2

short a = 0;
static long b = 0;
int c = 7;
char d = 0;
short *e = &a;
long f = 0;

unsigned long g (unsigned long h, long j) { return j == 0 ? h : h / j; }

int main (void) {
  long k = f;
  for (; c; --c) {
    for (int i = 0; i < 7; ++i)
      ;
    long m = g (f, --b);
    d = ((char)m | *e) <= 43165;
  }
  if (b != -7)
    __builtin_abort ();
  return 0;
}

Variable b should be decremented in each iteration of the outer loop and thus
equal -7 at the end.

After combine a load, decrement, and store insn exists:

(insn 13 12 14 3 (set (reg:DI 62 [ b_lsm.16 ])
        (mem/c:DI (const:DI (plus:DI (symbol_ref:DI ("*.LANCHOR0") [flags
0x182])
                    (const_int 8 [0x8]))) [1 b+0 S8 A64])) 1469 {*movdi_64}
     (nil))


(jump_insn 40 39 46 6 (parallel [
            (set (pc)
                (if_then_else (ne (reg:DI 62 [ b_lsm.16 ])
                        (const_int 1 [0x1]))
                    (label_ref 38)
                    (pc)))
            (set (reg:DI 62 [ b_lsm.16 ])
                (plus:DI (reg:DI 62 [ b_lsm.16 ])
                    (const_int -1 [0xffffffffffffffff])))
            (clobber (scratch:DI))
            (clobber (reg:CC 33 %cc))
        ]) "t.c":8:63 2164 {doloop_di}
     (expr_list:REG_UNUSED (reg:CC 33 %cc)
        (int_list:REG_BR_PROB 536870916 (nil)))
 -> 38)


(insn 48 47 49 7 (set (mem/c:DI (const:DI (plus:DI (symbol_ref:DI
("*.LANCHOR0") [flags 0x182])
                    (const_int 8 [0x8]))) [1 b+0 S8 A64])
        (reg:DI 62 [ b_lsm.16 ])) 1469 {*movdi_64}
     (expr_list:REG_DEAD (reg:DI 62 [ b_lsm.16 ])
        (nil)))

Pass ce2 deletes jump insn 40 including the decrement of variable b:

IF-CASE-2 found, start 6, else 4
deleting insn with uid = 40.
deleting block 4
Conversion succeeded on pass 1.

Thus variable b equals 0 in the end.

[Bug rtl-optimization/104814] [10/11/12 Regression] ifcvt: Deleting live variable in IF-CASE-2

[stefansf at linux dot ibm.com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

--- Comment #1 from Stefan Schulze Frielinghaus <stefansf at linux dot ibm.com> ---
Created attachment 52571
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52571&action=edit
dump combine

[Bug rtl-optimization/104814] [10/11/12 Regression] ifcvt: Deleting live variable in IF-CASE-2

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |10.4

[Bug rtl-optimization/104814] [10/11/12 Regression] ifcvt: Deleting live variable in IF-CASE-2

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
What options is this with?  Which isa/tuning?

[Bug rtl-optimization/104814] [10/11/12 Regression] ifcvt: Deleting live variable in IF-CASE-2

[stefansf at linux dot ibm.com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

--- Comment #3 from Stefan Schulze Frielinghaus <stefansf at linux dot ibm.com> ---
Oh forgot to mention it is just: gcc -O1 t.c

Works fine with -O{0,2,3}

[Bug rtl-optimization/104814] [10/11/12 Regression] ifcvt: Deleting live variable in IF-CASE-2

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2

[Bug rtl-optimization/104814] [10/11/12 Regression] ifcvt: Deleting live variable in IF-CASE-2

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |jakub at gcc dot gnu.org
   Last reconfirmed|                            |2022-03-11
             Status|UNCONFIRMED                 |ASSIGNED
     Ever confirmed|0                           |1

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 52612
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52612&action=edit
gcc12-pr104814.patch

Untested fix.

[Bug rtl-optimization/104814] [10/11/12 Regression] ifcvt: Deleting live variable in IF-CASE-2

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

--- Comment #5 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jakub Jelinek <jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:a2645cd8fb33b36d737b310e26f4c47401305c7b

commit r12-7651-ga2645cd8fb33b36d737b310e26f4c47401305c7b
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Tue Mar 15 09:12:03 2022 +0100

    ifcvt: Punt if not onlyjump_p for find_if_case_{1,2} [PR104814]

    find_if_case_{1,2} implicitly assumes conditional jumps and rewrites them,
    so if they have extra side-effects or are say asm goto, things don't work
    well, either the side-effects are lost or we could ICE.
    In particular, the testcase below on s390x has there a doloop instruction
    that decrements a register in addition to testing it for non-zero and
    conditionally jumping based on that.

    The following patch fixes that by punting for !onlyjump_p case, i.e.
    if there are side-effects in the jump instruction or it isn't a plain PC
    setter.

    Also, it assumes BB_END (test_bb) will be always non-NULL, because basic
    blocks with 2 non-abnormal successor edges should always have some
instruction
    at the end that determines which edge to take.

    2022-03-15  Jakub Jelinek  <jakub@redhat.com>

            PR rtl-optimization/104814
            * ifcvt.cc (find_if_case_1, find_if_case_2): Punt if test_bb
doesn't
            end with onlyjump_p.  Assume BB_END (test_bb) is always non-NULL.

            * gcc.c-torture/execute/pr104814.c: New test.

[Bug rtl-optimization/104814] [10/11 Regression] ifcvt: Deleting live variable in IF-CASE-2

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[10/11/12 Regression]       |[10/11 Regression] ifcvt:
                   |ifcvt: Deleting live        |Deleting live variable in
                   |variable in IF-CASE-2       |IF-CASE-2

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed on the trunk so far.

[Bug rtl-optimization/104814] [10/11 Regression] ifcvt: Deleting live variable in IF-CASE-2

[stefansf at linux dot ibm.com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

--- Comment #7 from Stefan Schulze Frielinghaus <stefansf at linux dot ibm.com> ---
Gave trunk a try and it worked fine for me.  Thanks for the fix!

[Bug rtl-optimization/104814] [10/11 Regression] ifcvt: Deleting live variable in IF-CASE-2

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

--- Comment #8 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:17de662ec26786053caab2cea34c76814352ea4d

commit r11-9727-g17de662ec26786053caab2cea34c76814352ea4d
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Tue Mar 15 09:12:03 2022 +0100

    ifcvt: Punt if not onlyjump_p for find_if_case_{1,2} [PR104814]

    find_if_case_{1,2} implicitly assumes conditional jumps and rewrites them,
    so if they have extra side-effects or are say asm goto, things don't work
    well, either the side-effects are lost or we could ICE.
    In particular, the testcase below on s390x has there a doloop instruction
    that decrements a register in addition to testing it for non-zero and
    conditionally jumping based on that.

    The following patch fixes that by punting for !onlyjump_p case, i.e.
    if there are side-effects in the jump instruction or it isn't a plain PC
    setter.

    Also, it assumes BB_END (test_bb) will be always non-NULL, because basic
    blocks with 2 non-abnormal successor edges should always have some
instruction
    at the end that determines which edge to take.

    2022-03-15  Jakub Jelinek  <jakub@redhat.com>

            PR rtl-optimization/104814
            * ifcvt.c (find_if_case_1, find_if_case_2): Punt if test_bb doesn't
            end with onlyjump_p.  Assume BB_END (test_bb) is always non-NULL.

            * gcc.c-torture/execute/pr104814.c: New test.

    (cherry picked from commit a2645cd8fb33b36d737b310e26f4c47401305c7b)

[Bug rtl-optimization/104814] [10 Regression] ifcvt: Deleting live variable in IF-CASE-2

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[10/11 Regression] ifcvt:   |[10 Regression] ifcvt:
                   |Deleting live variable in   |Deleting live variable in
                   |IF-CASE-2                   |IF-CASE-2

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed for 11.3 too.

[Bug rtl-optimization/104814] [10 Regression] ifcvt: Deleting live variable in IF-CASE-2

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

--- Comment #10 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:1ebc65fde4dd30678da10a17c6cc29bc4e1e041e

commit r10-10694-g1ebc65fde4dd30678da10a17c6cc29bc4e1e041e
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Tue Mar 15 09:12:03 2022 +0100

    ifcvt: Punt if not onlyjump_p for find_if_case_{1,2} [PR104814]

    find_if_case_{1,2} implicitly assumes conditional jumps and rewrites them,
    so if they have extra side-effects or are say asm goto, things don't work
    well, either the side-effects are lost or we could ICE.
    In particular, the testcase below on s390x has there a doloop instruction
    that decrements a register in addition to testing it for non-zero and
    conditionally jumping based on that.

    The following patch fixes that by punting for !onlyjump_p case, i.e.
    if there are side-effects in the jump instruction or it isn't a plain PC
    setter.

    Also, it assumes BB_END (test_bb) will be always non-NULL, because basic
    blocks with 2 non-abnormal successor edges should always have some
instruction
    at the end that determines which edge to take.

    2022-03-15  Jakub Jelinek  <jakub@redhat.com>

            PR rtl-optimization/104814
            * ifcvt.c (find_if_case_1, find_if_case_2): Punt if test_bb doesn't
            end with onlyjump_p.  Assume BB_END (test_bb) is always non-NULL.

            * gcc.c-torture/execute/pr104814.c: New test.

    (cherry picked from commit a2645cd8fb33b36d737b310e26f4c47401305c7b)

[Bug rtl-optimization/104814] [10 Regression] ifcvt: Deleting live variable in IF-CASE-2

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104814

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #11 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed for 10.4 too.