[Bug tdep/32549] New: [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[Bug tdep/32549] New: [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

            Bug ID: 32549
           Summary: [gdb/tdep, linaro CI] FAIL:
                    gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1,
                    i2)>
           Product: gdb
           Version: HEAD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: tdep
          Assignee: unassigned at sourceware dot org
          Reporter: vries at gcc dot gnu.org
  Target Milestone: ---

After commit 84067a55fcb ("[gdb/tdep] Fix gdb.cp/non-trivial-retval.exp on
riscv64-linux"), I got an email from the linaro CI (
https://linaro.atlassian.net/browse/GNU-1498 ) stating that I introduced
regressions in test-case gdb.cp/non-trivial-retval.exp:
...
FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>
FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f2 (i1, i2)>
FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f22 (i1, i2)>
FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f3 (i1, i2)>
FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f4 (i1, i2)>
FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f5 (i1, i2)>
...

The commit touches only gdb/riscv-tdep.c and the test-case, and in the
test-case just adds an extra test (the FAIL for f5) so it's likely that the
other tests already failed before, and that the CI is incorrectly reporting
those as a regression.

Nevertheless, the tests are failing according to the logs:
...
(gdb) p f1 (i1, i2)
$1 = {a = -136191900}
(gdb) FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>
...
and I have no idea why, and can't reproduce it.  I've tried with target board
unix/-marm and unix/-mthumb, and the test-case passes for me.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug tdep/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |thiago.bauermann at linaro dot org

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug tdep/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[thiago.bauermann at linaro dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

--- Comment #1 from Thiago Jung Bauermann <thiago.bauermann at linaro dot org> ---
(In reply to Tom de Vries from comment #0)
> After commit 84067a55fcb ("[gdb/tdep] Fix gdb.cp/non-trivial-retval.exp on
> riscv64-linux"), I got an email from the linaro CI (
> https://linaro.atlassian.net/browse/GNU-1498 ) stating that I introduced
> regressions in test-case gdb.cp/non-trivial-retval.exp:
> ...
> FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>
> FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f2 (i1, i2)>
> FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f22 (i1, i2)>
> FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f3 (i1, i2)>
> FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f4 (i1, i2)>
> FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f5 (i1, i2)>
> ...
> 
> The commit touches only gdb/riscv-tdep.c and the test-case, and in the
> test-case just adds an extra test (the FAIL for f5) so it's likely that the
> other tests already failed before, and that the CI is incorrectly reporting
> those as a regression.

Yes, of course. Sorry about the wrong report.

> Nevertheless, the tests are failing according to the logs:
> ...
> (gdb) p f1 (i1, i2)
> $1 = {a = -136191900}
> (gdb) FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>
> ...
> and I have no idea why, and can't reproduce it.  I've tried with target
> board unix/-marm and unix/-mthumb, and the test-case passes for me.

I can reproduce the failures on Ubuntu 24.04 armv7l-linux-gnueabihf with
unix/-mthumb:

p f1 (i1, i2)
$1 = {a = -1226727760}
(gdb) FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

But not with unix/-marm. In that case, all tests in
gdb.cp/non-trivial-retval.exp pass.

I'll investigate.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug tdep/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

--- Comment #2 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Thiago Jung Bauermann from comment #1)
> I can reproduce the failures on Ubuntu 24.04 armv7l-linux-gnueabihf with
> unix/-mthumb:
> 
> p f1 (i1, i2)
> $1 = {a = -1226727760}
> (gdb) FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>
> 
> But not with unix/-marm. In that case, all tests in
> gdb.cp/non-trivial-retval.exp pass.
> 

OK, thanks for the concrete data point.  I'll try to get an ubuntu sysroot to
reproduce this.

FWIW, I use an arm32 debian testing chroot on an aarch64-linux machine (also
debian testing).

The test-case output your showing looks like:
- gcc makes f1 return a value using a register containing a pointer to a buffer
- gdb thinks f1 returns a value in a register.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug tdep/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

--- Comment #3 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Tom de Vries from comment #2)
> (In reply to Thiago Jung Bauermann from comment #1)
> > I can reproduce the failures on Ubuntu 24.04 armv7l-linux-gnueabihf with
> > unix/-mthumb:

> OK, thanks for the concrete data point.  I'll try to get an ubuntu sysroot
> to reproduce this.
> 

I managed to reproduce this.  Passes when skipping -fvar-tracking.

The inferior call actually works fine, the problem is in getting the proper
value for i1:
...
$ gdb -q -batch outputs/gdb.cp/non-trivial-retval/non-trivial-retval \
    -ex "b 176" \
    -ex run \
    -ex "p f1 (23, 100)" \
    -ex "p i1" \
    -ex "p i2" \
    -ex 'p /x $pc'
  ...
176       breakpt ();   /* Break here.  */
$1 = {a = 123}
$2 = -136274356
$3 = 100
$3 = 0x400778
...

Variable i1 has a location list at 0x10:
...
 <2><3b4>: Abbrev Number: 13 (DW_TAG_variable)
    <3b5>   DW_AT_name        : i1
    <3ba>   DW_AT_type        : <0x80>
    <3be>   DW_AT_location    : 0x10 (location list)
    <3c2>   DW_AT_GNU_locviews: 0xc
...
which is here:
...
    00000010 00000774 (base address)
    00000015 v0000000 v0000000 views at 0000000c for:
             00000774 00000776 (DW_OP_reg3 (r3))
    0000001a v0000000 v0000000 views at 0000000e for:
             00000776 000007e2 (DW_OP_fbreg: -12)
    00000020 <End of list>
...
and the $pc tells us that i1 resides in (DW_OP_fbreg: -12).

Looking at the insns, that sort of makes sense, first the value is in r3, then
on stack:
...
 770:   2317            movs    r3, #23
 772:   62fb            str     r3, [r7, #44]   @ 0x2c
 774:   2364            movs    r3, #100        @ 0x64
 776:   62bb            str     r3, [r7, #40]   @ 0x28
 778:   f7ff fff0       bl      75c <_Z7breakptv>
...

Without -fvar-tracking though, we have:
...
 <2><3b4>: Abbrev Number: 3 (DW_TAG_variable)
    <3b5>   DW_AT_name        : i1
    <3ba>   DW_AT_type        : <0x80>
    <3be>   DW_AT_location    : 2 byte block: 91 6c     (DW_OP_fbreg: -20)
...
which suggests that (DW_OP_fbreg: -12) is wrong.

This is with gcc 13.3.0 in an ubuntu chroot.

My usual debian chroot uses gcc 14.2.0, and produces the same code:
...
 76e:   2317            movs    r3, #23
 770:   62fb            str     r3, [r7, #44]   @ 0x2c
 772:   2364            movs    r3, #100        @ 0x64
 774:   62bb            str     r3, [r7, #40]   @ 0x28
 776:   f7ff fff0       bl      75a <_Z7breakptv>
...
but uses (DW_OP_fbreg: -20) instea:
...
    00000010 00000772 (base address)
    00000015 v0000000 v0000000 views at 0000000c for:
             00000772 00000774 (DW_OP_reg3 (r3))
    0000001a v0000000 v0000000 views at 0000000e for:
             00000774 000007e0 (DW_OP_fbreg: -20)
    00000020 <End of list>
...

Looks like some fix happened in gcc.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug tdep/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

--- Comment #4 from Tom de Vries <vries at gcc dot gnu.org> ---
> but uses (DW_OP_fbreg: -20) instea:

but uses (DW_OP_fbreg: -20) instead of (DW_OP_fbreg: -12):

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug tdep/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

--- Comment #5 from Tom de Vries <vries at gcc dot gnu.org> ---
I have a testsuite patch, I'll try to submit today.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug testsuite/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|tdep                        |testsuite

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug testsuite/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

--- Comment #6 from Tom de Vries <vries at gcc dot gnu.org> ---
(In reply to Tom de Vries from comment #5)
> I have a testsuite patch, I'll try to submit today.

https://sourceware.org/pipermail/gdb-patches/2025-January/214824.html

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug testsuite/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[cvs-commit at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

--- Comment #7 from Sourceware Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Tom de Vries <vries@sourceware.org>:

https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=c99345db064e1bd645bc08db65174bae7ac20b0e

commit c99345db064e1bd645bc08db65174bae7ac20b0e
Author: Tom de Vries <tdevries@suse.de>
Date:   Mon Jan 20 05:41:01 2025 +0100

    [gdb/testsuite] Fix gdb.cp/non-trivial-retval.exp on arm-linux with gcc 13

    On arm-linux, with target board unix/-mthumb, we get:
    ...
    (gdb) PASS: gdb.cp/non-trivial-retval.exp: continue to breakpoint: Break
here
    p f1 (i1, i2)^M
    $1 = {a = -136274256}^M
    (gdb) FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>
    ...

    This is not a problem with the inferior call, which works fine:
    ...
    (gdb) p f1 (23, 100)
    $3 = {a = 123}
    ...
    but instead it's a problem with the location information:
    ...
    (gdb) p i1
    $1 = -136274356
    (gdb) p i2
    $2 = 100
    ...
    which tells us to find the value of i1 in (DW_OP_fbreg: -12).

    The test-case passes if we drop -fvar-tracking, in which case the debug
info
    tells us to find the value of i1 in (DW_OP_fbreg: -20).

    This is with gcc 13.3.0 on Ubuntu 24.04.  With gcc 14.2.0 on Debian
testing,
    the code is the same, but -fvar-tracking does use the correct
    '(DW_OP_fbreg: -20)'.

    There seems to be some bugfix in -fvar-tracking for gcc 14.

    Workaround the bug by using constants 23 and 100 instead of i1 and i2 when
    using -fvar-tracking and gcc < 14.

    Tested on arm-linux.

    PR testsuite/32549
    Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=32549

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug testsuite/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[vries at gcc dot gnu.org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

Tom de Vries <vries at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED
   Target Milestone|---                         |17.1

--- Comment #8 from Tom de Vries <vries at gcc dot gnu.org> ---
Fixed.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug testsuite/32549] [gdb/tdep, linaro CI] FAIL: gdb.cp/non-trivial-retval.exp: gdb-command<p f1 (i1, i2)>

[thiago.bauermann at linaro dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=32549

--- Comment #9 from Thiago Jung Bauermann <thiago.bauermann at linaro dot org> ---
(In reply to Tom de Vries from comment #8)
> Fixed.

Thank you!

-- 
You are receiving this mail because:
You are on the CC list for the bug.