[Bug target/109000] New: LoongArch: "unmatched" -mabi and -mfpu setting can break ABI silently

[Bug target/109000] New: LoongArch: "unmatched" -mabi and -mfpu setting can break ABI silently

[xry111 at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109000

            Bug ID: 109000
           Summary: LoongArch: "unmatched" -mabi and -mfpu setting can
                    break ABI silently
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: xry111 at gcc dot gnu.org
  Target Milestone: ---

$ cat t.c
double t(double x)
{
        return 1.0 / x;
}
$ ~/gcc-trunk/bin/gcc t.c -O2 -c -mabi=lp64s
$ readelf -a t.o | grep FLOAT
  Flags:                             0x41, SOFT-FLOAT, OBJ-v1
$ objdump -d t.o

t.o:     file format elf64-loongarch


Disassembly of section .text:

0000000000000000 <t>:
   0:   01145800        frecip.d        $fa0, $fa0
   4:   4c000020        jirl            $zero, $ra, 0

So we now have a SOFT-FLOAT object using FPR for arguments and return values
:(.

[Bug target/109000] LoongArch: "unmatched" -mabi and -mfpu setting can break ABI silently

[xry111 at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109000

Xi Ruoyao <xry111 at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |12.2.0, 13.0
   Last reconfirmed|                            |2023-03-03
             Target|                            |loongarch64-linux-gnu
           Assignee|unassigned at gcc dot gnu.org      |xry111 at gcc dot gnu.org
     Ever confirmed|0                           |1
           Keywords|                            |wrong-code
             Status|UNCONFIRMED                 |ASSIGNED
   Target Milestone|---                         |12.3

--- Comment #1 from Xi Ruoyao <xry111 at gcc dot gnu.org> ---
I've already made the patch.  Just create a PR to track the issue.

[Bug target/109000] LoongArch: "unmatched" -mabi and -mfpu setting can break ABI silently

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109000

--- Comment #2 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Xi Ruoyao <xry111@gcc.gnu.org>:

https://gcc.gnu.org/g:75eccddef5784bc5e262af31f535267a9c4e993e

commit r13-6500-g75eccddef5784bc5e262af31f535267a9c4e993e
Author: Xi Ruoyao <xry111@xry111.site>
Date:   Thu Mar 2 18:05:23 2023 +0800

    LoongArch: Stop -mfpu from silently breaking ABI [PR109000]

    In the toolchain convention, we describe -mfpu= as:

    "Selects the allowed set of basic floating-point instructions and
    registers. This option should not change the FP calling convention
    unless it's necessary."

    Though not explicitly stated, the rationale of this rule is to allow
    combinations like "-mabi=lp64s -mfpu=64".  This will be useful for
    running applications with LP64S/F ABI on a double-float-capable
    LoongArch hardware and using a math library with LP64S/F ABI but native
    double float HW instructions, for a better performance.

    And now a case in Linux kernel has again proven the usefulness of this
    kind of combination.  The AMDGPU DCN kernel driver needs to perform some
    floating-point operation, but the entire kernel uses LP64S ABI.  So the
    translation units of the AMDGPU DCN driver need to be compiled with
    -mfpu=64 (the kernel lacks soft-FP routines in libgcc), but -mabi=lp64s
    (or you can't link it with the other part of the kernel).

    Unfortunately, currently GCC uses TARGET_{HARD,SOFT,DOUBLE}_FLOAT to
    determine the floating calling convention.  This causes "-mfpu=64"
    silently allow using $fa* to pass parameters and return values EVEN IF
    -mabi=lp64s is used.  To make things worse, the generated object file
    has SOFT-FLOAT set in the eflags field so the linker will happily link
    it with other LP64S ABI object files, but obviously this will lead to
    bad results at runtime.  And for now all loongarch64 CPU models (-march
    settings) implies -mfpu=64 on by default, so the issue makes a single
    "-mabi=lp64s" option basically broken (fortunately most projects for eg
    the Linux kernel have used -msoft-float which implies both -mabi=lp64s
    and -mfpu=none as we've recommended in the toolchain convention doc).

    The fix is simple: use TARGET_*_FLOAT_ABI instead.

    I consider this a bug fix: the behavior difference from the toolchain
    convention doc is a bug, and generating object files with SOFT-FLOAT
    flag but parameters/return values passed through FPRs is definitely a
    bug.

    Bootstrapped and regtested on loongarch64-linux-gnu.  Ok for trunk and
    release/gcc-12 branch?

    gcc/ChangeLog:

            PR target/109000
            * config/loongarch/loongarch.h (FP_RETURN): Use
            TARGET_*_FLOAT_ABI instead of TARGET_*_FLOAT.
            (UNITS_PER_FP_ARG): Likewise.

    gcc/testsuite/ChangeLog:

            PR target/109000
            * gcc.target/loongarch/flt-abi-isa-1.c: New test.
            * gcc.target/loongarch/flt-abi-isa-2.c: New test.
            * gcc.target/loongarch/flt-abi-isa-3.c: New test.
            * gcc.target/loongarch/flt-abi-isa-4.c: New test.

[Bug target/109000] LoongArch: "unmatched" -mabi and -mfpu setting can break ABI silently

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109000

--- Comment #3 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-12 branch has been updated by Xi Ruoyao <xry111@gcc.gnu.org>:

https://gcc.gnu.org/g:eb50091457a789e2f2a15c2e5d08e0d79d3195b1

commit r12-9225-geb50091457a789e2f2a15c2e5d08e0d79d3195b1
Author: Xi Ruoyao <xry111@xry111.site>
Date:   Thu Mar 2 18:05:23 2023 +0800

    LoongArch: Stop -mfpu from silently breaking ABI [PR109000]

    In the toolchain convention, we describe -mfpu= as:

    "Selects the allowed set of basic floating-point instructions and
    registers. This option should not change the FP calling convention
    unless it's necessary."

    Though not explicitly stated, the rationale of this rule is to allow
    combinations like "-mabi=lp64s -mfpu=64".  This will be useful for
    running applications with LP64S/F ABI on a double-float-capable
    LoongArch hardware and using a math library with LP64S/F ABI but native
    double float HW instructions, for a better performance.

    And now a case in Linux kernel has again proven the usefulness of this
    kind of combination.  The AMDGPU DCN kernel driver needs to perform some
    floating-point operation, but the entire kernel uses LP64S ABI.  So the
    translation units of the AMDGPU DCN driver need to be compiled with
    -mfpu=64 (the kernel lacks soft-FP routines in libgcc), but -mabi=lp64s
    (or you can't link it with the other part of the kernel).

    Unfortunately, currently GCC uses TARGET_{HARD,SOFT,DOUBLE}_FLOAT to
    determine the floating calling convention.  This causes "-mfpu=64"
    silently allow using $fa* to pass parameters and return values EVEN IF
    -mabi=lp64s is used.  To make things worse, the generated object file
    has SOFT-FLOAT set in the eflags field so the linker will happily link
    it with other LP64S ABI object files, but obviously this will lead to
    bad results at runtime.  And for now all loongarch64 CPU models (-march
    settings) implies -mfpu=64 on by default, so the issue makes a single
    "-mabi=lp64s" option basically broken (fortunately most projects for eg
    the Linux kernel have used -msoft-float which implies both -mabi=lp64s
    and -mfpu=none as we've recommended in the toolchain convention doc).

    The fix is simple: use TARGET_*_FLOAT_ABI instead.

    I consider this a bug fix: the behavior difference from the toolchain
    convention doc is a bug, and generating object files with SOFT-FLOAT
    flag but parameters/return values passed through FPRs is definitely a
    bug.

    Bootstrapped and regtested on loongarch64-linux-gnu.  Ok for trunk and
    release/gcc-12 branch?

    gcc/ChangeLog:

            PR target/109000
            * config/loongarch/loongarch.h (FP_RETURN): Use
            TARGET_*_FLOAT_ABI instead of TARGET_*_FLOAT.
            (UNITS_PER_FP_ARG): Likewise.

    gcc/testsuite/ChangeLog:

            PR target/109000
            * gcc.target/loongarch/flt-abi-isa-1.c: New test.
            * gcc.target/loongarch/flt-abi-isa-2.c: New test.
            * gcc.target/loongarch/flt-abi-isa-3.c: New test.
            * gcc.target/loongarch/flt-abi-isa-4.c: New test.

    (cherry picked from commit 75eccddef5784bc5e262af31f535267a9c4e993e)

[Bug target/109000] LoongArch: "unmatched" -mabi and -mfpu setting can break ABI silently

[xry111 at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109000

Xi Ruoyao <xry111 at gcc dot gnu.org> changed:

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

--- Comment #4 from Xi Ruoyao <xry111 at gcc dot gnu.org> ---
Fixed for trunk and gcc-12.