https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126950
Bug ID: 126950
Summary: [17 Regression] Wrong code issues with
libgomp.fortran/use_device_addr-1.f90 since
r17-3342-gf3943597388db4
Product: gcc
Version: 17.0
Status: UNCONFIRMED
Keywords: openmp, wrong-code
Severity: normal
Priority: P3
Component: fortran
Assignee: unassigned at gcc dot gnu.org
Reporter: burnus at gcc dot gnu.org
CC: jvdelisle at gcc dot gnu.org
Target Milestone: ---
Since r17-3342-gf3943597388db4 (cf. also PR 126940). the following testcases
fail with OpenMP offloading to an Nvidia GPU [nvptx] (but oddly not to an AMD
GPU [gcn]).
The testcases are believed to be valid.
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-1.f90
-O1 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-1.f90
-O2 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-1.f90
-O3 -fomit-frame-pointer -funroll-loops -fpeel-loops -ftracer
-finline-functions execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-1.f90
-O3 -g execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-1.f90
-Os execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-2.f90
-O0 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-2.f90
-O1 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-2.f90
-O2 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-2.f90
-O3 -fomit-frame-pointer -funroll-loops -fpeel-loops -ftracer
-finline-functions execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-2.f90
-O3 -g execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-2.f90
-Os execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-3.f90
-O0 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-3.f90
-O1 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-3.f90
-O2 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-3.f90
-O3 -fomit-frame-pointer -funroll-loops -fpeel-loops -ftracer
-finline-functions execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-3.f90
-O3 -g execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-3.f90
-Os execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-4.f90
-O0 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-4.f90
-O1 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-4.f90
-O2 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-4.f90
-O3 -fomit-frame-pointer -funroll-loops -fpeel-loops -ftracer
-finline-functions execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-4.f90
-O3 -g execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_addr-4.f90
-Os execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-1.f90
-O0 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-1.f90
-O1 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-1.f90
-O2 execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-1.f90
-O3 -fomit-frame-pointer -funroll-loops -fpeel-loops -ftracer
-finline-functions execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-1.f90
-O3 -g execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-1.f90
-Os execution test
PASS -> FAIL: qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-3.f90
-O execution test
PASS -> FAIL:
qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-optional-3.f90 -O0
execution test
PASS -> FAIL:
qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-optional-3.f90 -O1
execution test
PASS -> FAIL:
qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-optional-3.f90 -O2
execution test
PASS -> FAIL:
qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-optional-3.f90 -O3
-fomit-frame-pointer -funroll-loops -fpeel-loops -ftracer -finline-functions
execution test
PASS -> FAIL:
qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-optional-3.f90 -O3
-g execution test
PASS -> FAIL:
qemu-rtx4080-02/libgomp.sum:libgomp.fortran/use_device_ptr-optional-3.f90 -Os
execution test
* * *
It fails as follows:
$ gfortran -fopenmp -Wno-deprecated-openmp
libgomp/testsuite/libgomp.fortran/use_device_addr-1.f90 -g
$ LD_LIBRARY_PATH=$HOME/projects-local/gcc-trunk-offload/lib64 ltrace ./a.out
GOMP_offload_register_ver(0x40001, 0x40c780, 5, 0x502840)
= 1
GOMP_offload_register_ver(0x40003, 0x40c780, 8, 0x502980)
= 1
_gfortran_set_args(1, 0x7ffc7cab6a78, 0x7ffc7cab6a78, 0x504db0)
= 1
_gfortran_set_options(7, 0x40c670, 0x7ffc7cab6a78, 0x504db0)
= 0
malloc(8)
= 0x2e021e40
malloc(8)
= 0x2e021e60
malloc(8)
= 0x2e021e80
malloc(8)
= 0x2e021ea0
GOMP_target_data_ext(0xffffffff, 4, 0x7ffc7cab6890, 0x505e00)
= 0x2ee0b440
GOMP_target_ext(0xffffffff, 0x40b28d, 16, 0x7ffc7cab2860 <no return ...>
--- SIGSEGV (Segmentation fault) ---
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Backtrace for this error:
#0 0x7fe8def61def in ???
at ./signal/../sysdeps/unix/sysv/linux/x86_64/libc_sigaction.c:0
#1 0x7fe8df084bc0 in __memmove_avx_unaligned_erms
at ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:364
#2 0x7fe8df179413 in gomp_map_vars_internal
at /net/vbuild-02/srv/data/tburnus/repos/gcc/libgomp/target.c:2063
#3 0x7fe8df179413 in gomp_map_vars
at /net/vbuild-02/srv/data/tburnus/repos/gcc/libgomp/target.c:2266
#4 0x7fe8df17fe00 in GOMP_target_ext
at /net/vbuild-02/srv/data/tburnus/repos/gcc/libgomp/target.c:3691
#5 0x401540 in do_offload_scalar
at libgomp/testsuite/libgomp.fortran/use_device_addr-1.f90:81
#6 0x40173e in __target_procs_MOD_copy3_scalar
at libgomp/testsuite/libgomp.fortran/use_device_addr-1.f90:75
#7 0x40ad98 in __tests_MOD_test_main_1
at libgomp/testsuite/libgomp.fortran/use_device_addr-1.f90:1045
#8 0x40b21c in omp_device_addr
at libgomp/testsuite/libgomp.fortran/use_device_addr-1.f90:1253
#9 0x40b285 in main
at libgomp/testsuite/libgomp.fortran/use_device_addr-1.f90:1245
--- SIGSEGV (Segmentation fault) ---
+++ killed by SIGSEGV +++
* * *
commit f3943597388db4846386b6d5a18d33ebf80ea96c
Author: Jerry DeLisle <[email protected]>
AuthorDate: Mon Aug 10 09:37:01 2026 -0700
Commit: Jerry DeLisle <[email protected]>
CommitDate: Mon Aug 17 13:06:41 2026 -0700
fortran: [PR53800] Wrong copy-in/out with array actual to TARGET dummy
An actual argument whose elements are spaced by more than the element
size - a CLASS array, or a component of a derived-type array - was
copied in and out when passed to a TARGET or POINTER dummy, so pointers
associated with the dummy went stale on return.
Such an actual argument is now passed with a descriptor of its own,
whose element type is that of the subobject and whose span is the
element size of the array, as is already done for a pointer assignment
to a subobject of an array. The dummy addresses its elements through
that span. Passing it on to a dummy that requires contiguous storage
copies it, but only if it turns out not to be contiguous.
PR fortran/53800
gcc/fortran/ChangeLog:
* gfortran.h (gfc_is_span_addressed_dummy): New prototype.
* symbol.cc (gfc_is_span_addressed_dummy): New function.
* expr.cc (is_subref_array): Return true for a span addressed
dummy.
* trans.h (gfc_get_span_descriptor): New prototype.
* trans.cc (gfc_get_span_descriptor): New function returning the
descriptor that carries a pointer array decl's span.
(get_array_span): Use it, including for character types.
* trans-array.cc (is_pointer_array): Note in the comment that the
tree must be a descriptor.
(span_addressed_array): New function returning the decl that
provides the span of an array.
(gfc_get_array_span): Use it.
(gfc_conv_scalarized_array_ref, gfc_conv_array_ref): Likewise, so
that a descriptorless dummy is addressed by its span.
(gfc_get_dataptr_offset): Likewise and dereference a spanned
character element.
(gfc_conv_expr_descriptor): Describe a subobject of the array
elements by a new descriptor when no temporary is made.
* trans-decl.cc (gfc_build_dummy_array_decl): Use
gfc_is_span_addressed_dummy to mark the dummy as a pointer array.
(gfc_get_symbol_decl): Likewise.
* trans-expr.cc (is_subobject_ref): New function.
(copy_in_out_allowed): Take the actual argument. Use
gfc_is_span_addressed_dummy and keep the copy for an actual
argument with a vector subscript.
(gfc_conv_gfc_desc_to_cfi_desc): Remove the now duplicate
offsetting
of the data pointer for a subobject reference.
(gfc_class_array_data_assign): Also copy the span field.
(gfc_conv_procedure_call): Use copy_in_out_allowed to skip
copy-in/copy-out for a class array reference, a class array
function
result and a subref array. Pass a spanned descriptor instead and
make the copy of a span addressed dummy conditional on it not being
contiguous.
* trans-intrinsic.cc (gfc_conv_is_contiguous_expr): Check the span
of a span addressed dummy against the element length.
gcc/testsuite/ChangeLog:
* gfortran.dg/c_loc_test_22.f90: Update dump patterns for span
addressing.
* gfortran.dg/class_to_type_5.f90: New test.
* gfortran.dg/class_to_type_6.f90: New test.
* gfortran.dg/class_to_type_7.f90: New test.
* gfortran.dg/class_to_type_8.f90: New test.
* gfortran.dg/class_to_type_9.f90: New test.
libgomp/ChangeLog:
* testsuite/libgomp.oacc-fortran/host_data-5.F90: Update the
privatization notes for the packing of a TARGET dummy.
