gcc/fortran/ChangeLog:
PR fortran/50549
* resolve.cc (resolve_structure_cons): Reject pointer assignments
of character with different lengths in structure constructor.
gcc/testsuite/ChangeLog:
PR fortran/50549
* gfortran.dg/char_pointer_assign_7.f90: New test.
This patch fixes a wrong -Wimplicit-fallthrough warning for
case 0:
if (1) // wrong may fallthrough
return 0;
case 1:
which in .gimple looks like
<D.1981>: // case 0
if (1 != 0) goto <D.1985>; else goto <D.1986>;
<D.1985>:
D.1987 = 0;
// predicted unlikely by early return (on trees) predictor.
return D.1987;
<D.1986>: // dead
<D.1982>: // case 1
and the warning thinks that <D.1986>: falls through to <D.1982>:. It
does not know that <D.1986> is effectively a dead label, only reachable
through fallthrough from previous instructions, never jumped to. To
that effect, Jakub introduced UNUSED_LABEL_P, which is set on such dead
labels.
collect_fallthrough_labels has code to deal with cases like
case 2:
if (e != 10)
i++; // this may fallthru, warn
else
return 44;
case 3:
which collects labels that may fall through. Here it sees the "goto <D.1990>;"
at the end of the then branch and so when the warning reaches
...
<D.1990>: // from if-then
<D.1984>: // case 3
it knows it should warn about the possible fallthrough. But an UNUSED_LABEL_P
is not a label that can fallthrough like that, so it should ignore those.
However, we still want to warn about this:
case 0:
if (1)
n++; // falls through
case 1:
so collect_fallthrough_labels needs to return the "n = n + 1;" statement, rather
than the dead label.
Co-authored-by: Jakub Jelinek <jakub@redhat.com>
PR middle-end/103597
gcc/ChangeLog:
* gimplify.cc (collect_fallthrough_labels): Don't push UNUSED_LABEL_Ps
into labels. Maybe set prev to the statement preceding UNUSED_LABEL_P.
(gimplify_cond_expr): Set UNUSED_LABEL_P.
* tree.h (UNUSED_LABEL_P): New.
gcc/testsuite/ChangeLog:
* c-c++-common/Wimplicit-fallthrough-39.c: New test.
We no longer emit a bogus warning for the below testcase after
r11-3266-g4839de55e2c986.
PR c++/71637
gcc/testsuite/ChangeLog:
* c-c++-common/Wmisleading-indentation-6.c: New test.
I noticed that the vsx_extract_<mode> pattern for V2DImode and V2DFmode
only allowed traditional floating point registers, and it did not allow
Altivec registers. The original code was written a few years ago when we
used the old register allocator, and support for scalar floating point in
Altivec registers was just being added to GCC.
I have built the spec 2017 benchmark suite With all 4 patches in this
series applied, and compared it to the build with the previous 3 patches
applied. In addition to the changes from the previous 3 patches, this
patch now changes the code for the following 3 benchmarks (2 floating
point, 1 integer):
bwaves_r, fotonik3d_r, xalancbmk_r
I have built bootstrap versions on the following systems. There were no
regressions in the runs:
Power9 little endian, --with-cpu=power9
Power10 little endian, --with-cpu=power10
Power8 big endian, --with-cpu=power8 (both 32-bit & 64-bit tests)
2022-03-29 Michael Meissner <meissner@linux.ibm.com>
gcc/
* config/rs6000/vsx.md (vsx_extract_<mode>): Allow destination to
be any VSX register.
On aarch64 the AAPCS64 states that an HFA is determined by the 'shape' of
the object after layout has been completed, so anything that adds no
members and does not cause the layout to be modified should be ignored
for the purposes of determining which registers are used for parameter
passing.
A zero-sized bit-field falls into this category. This was not handled
correctly for C structs and in G++-11 only handled correctly because
such fields were eliminated early by the front end.
gcc/ChangeLog:
PR target/102024
* config/aarch64/aarch64.cc (aapcs_vfp_sub_candidate): Handle
zero-sized bit-fields. Detect cases where a warning may be needed.
(aarch64_vfp_is_call_or_return_candidate): Emit a note if a
zero-sized bit-field has caused parameter passing to change.
gcc/testsuite/ChangeLog:
* gcc.target/aarch64/aapcs64/test_28.c: New test.
On arm the AAPCS states that an HFA is determined by the 'shape' of
the object after layout has been completed, so anything that adds no
members and does not cause the layout to be modified should be ignored
for the purposes of determining which registers are used for parameter
passing.
A zero-sized bit-field falls into this category. This was not handled
correctly for C structs and in G++-11 only handled correctly because
such fields were eliminated early by the front end.
gcc/ChangeLog:
PR target/102024
* config/arm/arm.cc (aapcs_vfp_sub_candidate): Handle zero-sized
bit-fields. Detect cases where a warning may be needed.
(aapcs_vfp_is_call_or_return_candidate): Emit a note if
a zero-sized bit-field has caused parameter passing to change.
gcc/testsuite/ChangeLog:
PR target/102024
* gcc.target/arm/aapcs/vfp26.c: New test.
The arm port has an optimization used during selection of the
function's ABI to permit deviation from the strict ABI when the
function does not escape the current translation unit.
Unfortunately, the ABI selection it makes can be unsafe if it changes
how a result is returned because not enough information is available
via the RETURN_IN_MEMORY hook to determine where the function gets
used. This can result in some parts of the compiler thinking a value
is returned in memory while others think it is returned in registers.
To mitigate this, this patch temporarily disables the optimization and
falls back to using the default ABI for the translation.
gcc/ChangeLog:
PR target/96882
* config/arm/arm.cc (arm_get_pcs_model): Disable selection of
ARM_PCS_AAPCS_LOCAL.
In the docs we have for m64:
...
Ignored, but preserved for backward compatibility. Only 64-bit ABI is
supported.
...
But with --target-help, we have instead:
...
$ gcc --target-help
...
-m64 Generate code for a 64-bit ABI.
...
which could be interpreted as meaning that generating code for a 32-bit ABI is
still possible.
Fix this by instead emitting the same text as in the docs:
...
-m64 Ignored, but preserved for backward compatibility. Only 64-bit
ABI is supported.
...
Tested on nvptx.
gcc/ChangeLog:
2022-03-29 Tom de Vries <tdevries@suse.de>
* config/nvptx/nvptx.opt (m64): Update help text to reflect that it
is ignored.
Say we have an sm_50 board, and we want to run a benchmark using the highest
possible march setting.
Currently there's march=sm_30, march=sm_35, march=sm_53, but no march=sm_50.
So, we'd need to pick march=sm_35.
Likewise, for a test script that handles multiple boards, we'd need a mapping
from native board sm_xx to march, which might have to be updated with newer
gcc releases.
Add an option march-map, such that we can just specify march-map=sm_50, and
let the compiler map this to the appropriate march.
The option is implemented as a list of aliases, such that we have a somewhat
lengthy (17 lines in total):
...
$ gcc --help=target
...
-march-map=sm_30 Same as -misa=sm_30.
-march-map=sm_32 Same as -misa=sm_30.
...
-march-map=sm_87 Same as -misa=sm_80.
-march-map=sm_90 Same as -misa=sm_80.
...
This implementation was chosen in the hope that it'll be easier if
we end up with some misa multilib.
It would be nice to have the mapping list generated from an updated
nvptx-sm.def, but for now it's spelled out in nvptx.opt.
Tested on nvptx.
gcc/ChangeLog:
2022-03-29 Tom de Vries <tdevries@suse.de>
PR target/104714
* config/nvptx/nvptx.opt (march-map=*): Add aliases.
gcc/testsuite/ChangeLog:
2022-03-29 Tom de Vries <tdevries@suse.de>
PR target/104714
* gcc.target/nvptx/march-map.c: New test.
gcc/ChangeLog:
2022-03-28 Jan Hubicka <hubicka@ucw.cz>
* config/i386/i386-builtins.cc (ix86_vectorize_builtin_gather): Test
TARGET_USE_GATHER_2PARTS and TARGET_USE_GATHER_4PARTS.
* config/i386/i386.h (TARGET_USE_GATHER_2PARTS): New macro.
(TARGET_USE_GATHER_4PARTS): New macro.
* config/i386/x86-tune.def (X86_TUNE_USE_GATHER_2PARTS): New tune
(X86_TUNE_USE_GATHER_4PARTS): New tune
The target option misa has the following description:
...
$ gcc --target-help 2>&1 | grep misa
-misa= Specify the PTX ISA target architecture to use.
...
The name misa is somewhat poorly chosen. It suggests that for a use
-misa=sm_30, sm_30 is the name of a specific Instruction Set Architecture.
Instead, sm_30 is the name of a specific target architecture in the generic
PTX Instruction Set Architecture.
Futhermore, there's mptx, which also has ISA in the description:
...
-mptx= Specify the PTX ISA version to use.
...
Add the more intuitive alias march for misa:
...
$ gcc --target-help 2>&1 | grep march
-march= Alias: Same as -misa=.
...
Tested on nvptx.
gcc/ChangeLog:
2022-03-29 Tom de Vries <tdevries@suse.de>
* config/nvptx/nvptx.opt (march): Add alias of misa.
gcc/testsuite/ChangeLog:
2022-03-29 Tom de Vries <tdevries@suse.de>
* gcc.target/nvptx/main.c: New test.
* gcc.target/nvptx/march.c: New test.
2022-03-29 Chenghua Xu <xuchenghua@loongson.cn>
Lulu Cheng <chenglulu@loongson.cn>
gcc/ChangeLog:
* config/loongarch/constraints.md: New file.
* config/loongarch/generic.md: New file.
* config/loongarch/la464.md: New file.
* config/loongarch/loongarch-ftypes.def: New file.
* config/loongarch/loongarch-modes.def: New file.
* config/loongarch/loongarch.md: New file.
* config/loongarch/predicates.md: New file.
* config/loongarch/sync.md: New file.
Originally introduced almost ten years ago in
r193303/commit 0829c7f7c5
"optc-gen.awk: Factor code out to...".
gcc/
* opt-functions.awk (lang_enabled_by): Fix 'enabledby_negargs'
typo.
Once we no longer care about older compilers without this feature, we
can drop these static data members, so the names don't have to be
visible at class scope.
libstdc++-v3/ChangeLog:
* libsupc++/compare (_Strong_order) [!__cpp_using_enum]: Add
static data members for _Fp_fmt enumerators.
The conditions that guard the feature test macros in <version> should
match the main definitions of the macros in other headers.
This doesn't matter for GCC, because it supports all the conditions
being tested here, but it does matter for non-GCC compilers without the
relevant C++20 features.
libstdc++-v3/ChangeLog:
* include/std/version (__cpp_lib_variant): Fix conditions to
match <variant>.
(__cpp_lib_expected): Fix condition to match <expected>.
Test must check for effective support of fpic.
gcc/testsuite/ChangeLog:
* gcc.target/i386/pr103275.c: Add missing
dg-require-effective-target for checking fpic.
On targets that do not have MXX/SSE enabled by default, pr97521
and pr96713 fail because they emit warnings:
pr97521.c:12:1: warning: MMX vector return without MMX enabled
changes the ABI [-Wpsabi]
pr97521.c:11:1: note: the ABI for passing parameters with
16-byte alignment has changed in GCC 4.6
pr97521.c:11:1: warning: SSE vector argument without SSE enabled
changes the ABI [-Wpsabi]
Add -Wno-psabi to dg-options.
gcc/testsuite/ChangeLog:
* gcc.target/i386/pr97521.c: Add -Wno-psabi to dg-options.
* gcc.dg/analyzer/pr96713.c: Likewise.
When doing format diagnostics at -O0 we should make sure to make
SCEV available to avoid false positives due to ranges we otherwise
can trivially compute.
2022-03-28 Richard Biener <rguenther@suse.de>
PR tree-optimization/105080
* tree-ssa-strlen.cc (printf_strlen_execute): Always init
loops and SCEV.
* gcc.dg/pr105080.c: New testcase.
PR analyzer/105087 describes a false positive from
-Wanalyzer-double-free in which the analyzer erroneously considers two
successive inlined vasprintf calls to have allocated the same pointer.
The root cause is that the result written back from vasprintf is a
conjured_svalue, and that we normally purge state when reusing a
conjured_svalue, but various places in the code were calling
region_model_manager::get_or_create_conjured_svalue but failing to
then call region_model::purge_state_involving on the result.
This patch fixes things by moving responsibility for calling
region_model::purge_state_involving into
region_model_manager::get_or_create_conjured_svalue, so that it is
always called when reusing a conjured_svalue, fixing the false positive.
gcc/analyzer/ChangeLog:
PR analyzer/105087
* analyzer.h (class conjured_purge): New forward decl.
* region-model-asm.cc (region_model::on_asm_stmt): Add
conjured_purge param to calls binding_cluster::on_asm and
region_model_manager::get_or_create_conjured_svalue.
* region-model-impl-calls.cc
(call_details::get_or_create_conjured_svalue): Likewise for call
to region_model_manager::get_or_create_conjured_svalue.
(region_model::impl_call_fgets): Remove call to
region_model::purge_state_involving, as this is now done
implicitly by call_details::get_or_create_conjured_svalue.
(region_model::impl_call_fread): Likewise.
(region_model::impl_call_strchr): Pass conjured_purge param to
call to region_model_manager::get_or_create_conjured_svalue.
* region-model-manager.cc (conjured_purge::purge): New.
(region_model_manager::get_or_create_conjured_svalue): Add
param "p". Use it to purge state when reusing an existing
conjured_svalue.
* region-model.cc (region_model::on_call_pre): Replace call to
region_model::purge_state_involving with passing conjured_purge
to region_model_manager::get_or_create_conjured_svalue.
(region_model::handle_unrecognized_call): Pass conjured_purge to
store::on_unknown_fncall.
* region-model.h
(region_model_manager::get_or_create_conjured_svalue): Add param
"p".
* store.cc (binding_cluster::on_unknown_fncall): Likewise. Pass
it on to region_model_manager::get_or_create_conjured_svalue.
(binding_cluster::on_asm): Likewise.
(store::on_unknown_fncall): Add param "p" and pass it on to
binding_cluster::on_unknown_fncall.
* store.h (binding_cluster::on_unknown_fncall): Add param p.
(binding_cluster::on_asm): Likewise.
(store::on_unknown_fncall): Likewise.
* svalue.h (class conjured_purge): New.
gcc/testsuite/ChangeLog:
* gcc.dg/analyzer/pr105087-1.c: New test.
* gcc.dg/analyzer/pr105087-2.c: New test.
* gcc.dg/analyzer/vasprintf-1.c: New test.
Signed-off-by: David Malcolm <dmalcolm@redhat.com>
gcc/analyzer/ChangeLog:
PR analyzer/105074
* region.cc (ipa_ref_requires_tracking): Drop "context_fndecl",
instead using the ref->referring to get the cgraph node of the
caller.
(symnode_requires_tracking_p): Likewise.
gcc/testsuite/ChangeLog:
PR analyzer/105074
* gcc.dg/analyzer/pr105074.c: New test.
* gcc.dg/analyzer/untracked-1.c (extern_fn_char_ptr): New decl.
(test_13): New.
Signed-off-by: David Malcolm <dmalcolm@redhat.com>
The top-level configure options --with-gmp-dir and --with-mpfr-dir
were obsoleted and marked as "REMOVED" back in 2006. I think that's
long enough ago for everyone to have updated their scripts, so this
patch removes them entirely. While doing this, I also found one other
leftover that wasn't removed by the earlier patch. This is also
removed here.
2022-03-28 Tom Tromey <tromey@adacore.com>
* configure.ac: Remove --with-mpfr-dir and --with-gmp-dir.
* configure: Rebuild.
These have been misdocumented since C++98 POD was split into C++11 trivial
and standard-layout in r149721.
PR c++/59426
gcc/ChangeLog:
* doc/extend.texi: Refer to __is_trivial instead of __is_pod.
We weren't rejecting a concept declared with multiple template
parameter lists.
PR c++/105067
gcc/cp/ChangeLog:
* pt.cc (finish_concept_definition): Check that a concept is
declared with exactly one template parameter list.
gcc/testsuite/ChangeLog:
* g++.dg/cpp2a/concepts-err4.C: New test.
Here during declaration matching for the two constrained template
friends, we crash from maybe_substitute_reqs_for because the second
friend doesn't yet have DECL_TEMPLATE_INFO set (we're being called
indirectly from push_template_decl).
As far as I can tell, this situation happens only when declaring a
constrained template friend within a non-template class (as in the
testcase), in which case the substitution would be a no-op anyway.
So this patch rearranges maybe_substitute_reqs_for to gracefully
handle missing DECL_TEMPLATE_INFO by just skipping the substitution.
PR c++/105064
gcc/cp/ChangeLog:
* constraint.cc (maybe_substitute_reqs_for): Don't assume
DECL_TEMPLATE_INFO is available.
gcc/testsuite/ChangeLog:
* g++.dg/cpp2a/concepts-friend9.C: New test.
Currently we have:
...
$ gcc --target-help 2>&1 | egrep "misa|mptx"
-misa= Specify the version of the ptx ISA to use.
-mptx= Specify the version of the ptx version to use.
Known PTX ISA versions (for use with the -misa= option):
Known PTX versions (for use with the -mptx= option):
...
As reported in PR104818, the "version of the ptx version" doesn't make much
sense.
Furthermore, the description of misa (and 'Known ISA versions') is misleading
because it does not specify the version of the PTX ISA, but rather the PTX ISA
target architecture.
Fix this by printing instead:
...
$ gcc --target-help 2>&1 | egrep "misa|mptx"
-misa= Specify the PTX ISA target architecture to use.
-mptx= Specify the PTX ISA version to use.
Known PTX ISA target architectures (for use with the -misa= option):
Known PTX ISA versions (for use with the -mptx= option):
...
Tested on nvptx.
gcc/ChangeLog:
2022-03-28 Tom de Vries <tdevries@suse.de>
PR target/104818
* config/nvptx/gen-opt.sh (ptx_isa): Improve help text.
* config/nvptx/nvptx-gen.opt: Regenerate.
* config/nvptx/nvptx.opt (misa, mptx, ptx_version): Improve help text.
* config/nvptx/t-nvptx (s-nvptx-gen-opt): Add missing dependency on
gen-opt.sh.
For PR104008 we thought it might be enough to keep strip_typedefs from
removing this alias template specialization, but this PR demonstrates that
other parts of the compiler also need to know to consider it dependent.
So, this patch changes complex_alias_template_p to no longer consider
template parameters used when their only use appears in a pack expansion,
unless they are the parameter packs being expanded.
To do that I also needed to change it to use cp_walk_tree instead of
for_each_template_parm. It occurs to me that find_template_parameters
should probably also use cp_walk_tree, but I'm not messing with that now.
PR c++/105003
PR c++/104008
PR c++/102869
gcc/cp/ChangeLog:
* pt.cc (complex_alias_template_r): walk_tree callback, replacing
uses_all_template_parms_r, complex_pack_expansion_r.
(complex_alias_template_p): Adjust.
* tree.cc (strip_typedefs): Revert r12-7710 change.
gcc/testsuite/ChangeLog:
* g++.dg/cpp0x/variadic-alias6.C: New test.
* g++.dg/cpp0x/variadic-alias7.C: New test.
PR analyzer/104308 reports that when -Wanalyzer-use-of-uninitialized-value
complains about certain memmove operations where the source is
uninitialized, the diagnostic uses UNKNOWN_LOCATION:
In function 'main':
cc1: warning: use of uninitialized value '*(short unsigned int *)&s + 1' [CWE-457] [-Wanalyzer-use-of-uninitialized-value]
'main': event 1
|
|pr104308.c:5:8:
| 5 | char s[5]; /* { dg-message "region created on stack here" } */
| | ^
| | |
| | (1) region created on stack here
|
'main': event 2
|
|cc1:
| (2): use of uninitialized value '*(short unsigned int *)&s + 1' here
|
The issue is that gimple_fold_builtin_memory_op converts a memmove to:
_3 = MEM <unsigned short> [(char * {ref-all})_1];
MEM <unsigned short> [(char * {ref-all})&s] = _3;
but only sets the location of the 2nd stmt, not the 1st.
Fixed thusly, giving:
pr104308.c: In function 'main':
pr104308.c:6:3: warning: use of uninitialized value '*(short unsigned int *)&s + 1' [CWE-457] [-Wanalyzer-use-of-uninitialized-value]
6 | memmove(s, s + 1, 2); /* { dg-warning "use of uninitialized value" } */
| ^~~~~~~~~~~~~~~~~~~~
'main': events 1-2
|
| 5 | char s[5]; /* { dg-message "region created on stack here" } */
| | ^
| | |
| | (1) region created on stack here
| 6 | memmove(s, s + 1, 2); /* { dg-warning "use of uninitialized value" } */
| | ~~~~~~~~~~~~~~~~~~~~
| | |
| | (2) use of uninitialized value '*(short unsigned int *)&s + 1' here
|
One side-effect of this change is a change in part of the output of
gcc.dg/uninit-40.c from:
uninit-40.c:47:3: warning: ‘*(long unsigned int *)(&u[1][0][0])’ is used uninitialized [-Wuninitialized]
47 | __builtin_memcpy (&v[1], &u[1], sizeof (V));
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
uninit-40.c:45:5: note: ‘*(long unsigned int *)(&u[1][0][0])’ was declared here
45 | V u[2], v[2];
| ^
to:
uninit-40.c:47:3: warning: ‘u’ is used uninitialized [-Wuninitialized]
47 | __builtin_memcpy (&v[1], &u[1], sizeof (V));
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
uninit-40.c:45:5: note: ‘u’ declared here
45 | V u[2], v[2];
| ^
What's happening is that pass "early_uninit"(29)'s call to
maybe_warn_operand is guarded by this condition:
1051 else if (gimple_assign_load_p (stmt)
1052 && gimple_has_location (stmt))
Before the patch, the stmt:
_3 = MEM <unsigned long> [(char * {ref-all})&u + 8B];
has no location, and so early_uninit skips this operand at line
1052 above. Later, pass "uninit"(217) tests the var_decl "u$8", and
emits a warning for it.
With the patch, the stmt has a location, and so early_uninit emits a
warning for "u" and sets a NW_UNINIT warning suppression at that
location. Later, pass "uninit"(217)'s test of "u$8" is rejected
due to that per-location suppression of uninit warnings, from the
earlier warning.
gcc/ChangeLog:
PR analyzer/104308
* gimple-fold.cc (gimple_fold_builtin_memory_op): When optimizing
to loads then stores, set the location of the new load stmt.
gcc/testsuite/ChangeLog:
PR analyzer/104308
* gcc.dg/analyzer/pr104308.c: New test.
* gcc.dg/uninit-40.c (foo): Update expression in expected message.
Signed-off-by: David Malcolm <dmalcolm@redhat.com>
This test ICEd after the constexpr new patch (r10-3661) because alloc_call
had a NOP_EXPR around it; fixed by moving the NOP_EXPR to alloc_expr. And
the PR pointed out that the size_t cookie might need more alignment, so I
fix that as well.
PR c++/102071
gcc/cp/ChangeLog:
* init.cc (build_new_1): Include cookie in alignment. Omit
constexpr wrapper from alloc_call.
gcc/testsuite/ChangeLog:
* g++.dg/cpp1z/aligned-new9.C: New test.
When setting up the hidden namespace-scope decl for a local extern, we also
need to set its visibility.
PR c++/103291
gcc/cp/ChangeLog:
* name-lookup.cc (push_local_extern_decl_alias): Call
determine_visibility.
gcc/testsuite/ChangeLog:
* g++.dg/ext/visibility/visibility-local-extern1.C: New test.
When building a deduction guide from the Test constructor, we need to
rewrite the use of _dummy into a dependent reference, i.e. Test<T>::template
_dummy. We were using SCOPE_REF for both type and non-type templates; we
need to use UNBOUND_CLASS_TEMPLATE for type templates.
PR c++/102123
gcc/cp/ChangeLog:
* pt.cc (tsubst_copy): Use make_unbound_class_template for rewriting
a type template reference.
gcc/testsuite/ChangeLog:
* g++.dg/cpp1z/class-deduction110.C: New test.
Here, we were wrongly thinking that (const Options&)Widget<T>::c_options is
not value-dependent because neither the type nor the (value of) c_options
are dependent, but since we're binding it to a reference we also need to
consider that it has a value-dependent address.
PR c++/103968
gcc/cp/ChangeLog:
* pt.cc (value_dependent_expression_p): Check
has_value_dependent_address for conversion to reference.
gcc/testsuite/ChangeLog:
* g++.dg/cpp0x/alias-decl-mem1.C: New test.
More quirks of rewriting member references to dependent references for
CTAD. A reference to a member of dependent scope is definitely dependent.
And since r11-7044, tsubst_baselink builds a SCOPE_REF, so
tsubst_qualified_id should just use it.
PR c++/103943
gcc/cp/ChangeLog:
* pt.cc (tsubst_qualified_id): Handle getting SCOPE_REF from
tsubst_baselink.
(instantiation_dependent_scope_ref_p): Check dependent_scope_p.
gcc/testsuite/ChangeLog:
* g++.dg/cpp1z/class-deduction109.C: New test.
