The compile time issue was discovered in SPEC 2017 wrf:
Use time and -ftime-report to analyze the profile data of SPEC 2017 wrf compilation .
Before this patch (Lazy vsetvl):
scheduling : 121.89 ( 15%) 0.53 ( 11%) 122.72 ( 15%) 13M ( 1%)
machine dep reorg : 424.61 ( 53%) 1.84 ( 37%) 427.44 ( 53%) 5290k ( 0%)
real 13m27.074s
user 13m19.539s
sys 0m5.180s
Simple vsetvl:
machine dep reorg : 0.10 ( 0%) 0.00 ( 0%) 0.11 ( 0%) 4138k ( 0%)
real 6m5.780s
user 6m2.396s
sys 0m2.373s
The machine dep reorg is the compile time of VSETVL PASS (424 seconds) which counts 53% of
the compilation time, spends much more time than scheduling.
After investigation, the critical patch of VSETVL pass is compute_lcm_local_properties which
is called every iteration of phase 2 (earliest fusion) and phase 3 (global lcm).
This patch optimized the codes of compute_lcm_local_properties to reduce the compilation time.
After this patch:
scheduling : 117.51 ( 27%) 0.21 ( 6%) 118.04 ( 27%) 13M ( 1%)
machine dep reorg : 80.13 ( 18%) 0.91 ( 26%) 81.26 ( 18%) 5290k ( 0%)
real 7m25.374s
user 7m20.116s
sys 0m3.795s
The optimization of this patch is very obvious, lazy VSETVL PASS: 424s (53%) -> 80s (18%) which
spend less time than scheduling.
Tested on both RV32 and RV64 no regression. Ok for trunk ?
PR target/113495
gcc/ChangeLog:
* config/riscv/riscv-vsetvl.cc (extract_single_source): Remove.
(pre_vsetvl::compute_vsetvl_def_data): Fix compile time issue.
(pre_vsetvl::compute_transparent): New function.
(pre_vsetvl::compute_lcm_local_properties): Fix compile time time issue.
Printing the raw symbol is useful in inline asm (e.g. in C++ to get the
mangled name). Similar constraints are available in other targets (e.g.
"S" for aarch64/riscv, "Cs" for m68k).
There isn't a good way for x86 yet, e.g. "i" doesn't work for
PIC/-mcmodel=large. This patch adds "Ws". Here are possible use cases:
```
namespace ns { extern int var; }
asm (".pushsection .xxx,\"aw\"; .dc.a %0; .popsection" :: "Ws"(&var));
asm (".reloc ., BFD_RELOC_NONE, %0" :: "Ws"(&var));
```
gcc/ChangeLog:
PR target/105576
* config/i386/constraints.md: Define constraint "Ws".
* doc/md.texi: Document it.
gcc/testsuite/ChangeLog:
PR target/105576
* gcc.target/i386/asm-raw-symbol.c: New testcase.
Real-world experience shows that -Wdangling-reference triggers for
user-defined std::span-like classes a lot. We can easily avoid that
by considering classes like
template<typename T>
struct Span {
T* data_;
std::size len_;
};
to be std::span-like, and not warning for them. Unlike the previous
patch, this one considers a non-union class template that has a pointer
data member and a trivial destructor as std::span-like.
PR c++/110358
PR c++/109640
gcc/cp/ChangeLog:
* call.cc (reference_like_class_p): Don't warn for std::span-like
classes.
gcc/ChangeLog:
* doc/invoke.texi: Update -Wdangling-reference description.
gcc/testsuite/ChangeLog:
* g++.dg/warn/Wdangling-reference18.C: New test.
* g++.dg/warn/Wdangling-reference19.C: New test.
* g++.dg/warn/Wdangling-reference20.C: New test.
gcc/ChangeLog:
* config/xtensa/constraints.md (R, T, U):
Change define_constraint to define_memory_constraint.
* config/xtensa/predicates.md (move_operand): Don't check that a
constant pool operand size is a multiple of UNITS_PER_WORD.
* config/xtensa/xtensa.cc
(xtensa_lra_p, TARGET_LRA_P): Remove.
(xtensa_emit_move_sequence): Remove "if (reload_in_progress)"
clause as it can no longer be true.
(fixup_subreg_mem): Drop function.
(xtensa_output_integer_literal_parts): Consider 16-bit wide
constants.
(xtensa_legitimate_constant_p): Add short-circuit path for
integer load instructions. Don't check that mode size is
at least UNITS_PER_WORD.
* config/xtensa/xtensa.md (movsf): Use can_create_pseudo_p()
rather reload_in_progress and reload_completed.
(doloop_end): Drop operand 2.
(movhi_internal): Add alternative loading constant from a
literal pool.
(define_split for DI register_operand): Don't limit to
!TARGET_AUTO_LITPOOLS.
* config/xtensa/xtensa.opt (mlra): Change to no effect.
The original testcase from this PR (fixed by r14-8291) seems rather
different from the others, so let's add it to the testsuite.
PR c++/67898
gcc/testsuite/ChangeLog:
* g++.dg/cpp0x/temp_default8.C: New test.
As with r14-6796-g2fa122cae50cd8, avoid mangling compatibility aliases in
mangling tests, and test the new mangling.
PR c++/112846
gcc/testsuite/ChangeLog:
* g++.dg/abi/anon6.C: Specify ABI v18.
* g++.dg/abi/anon6a.C: New test for ABI v19.
When I added -fabi-compat-version=8 to avoid mangling aliases it also
suppressed the -Wabi warning.
PR c++/113451
gcc/testsuite/ChangeLog:
* g++.dg/abi/mangle-regparm1a.C: Use -Wabi=0.
We miscompile the below testcase because keep_unused_object_arg thinks
the object argument of an xobj member function is unused, and so it ends
up duplicating the argument's side effects.
PR c++/113640
gcc/cp/ChangeLog:
* call.cc (keep_unused_object_arg): Punt for an xobj member
function.
gcc/testsuite/ChangeLog:
* g++.dg/cpp23/explicit-obj-lambda14.C: New test.
Reviewed-by: Jason Merrill <jason@redhat.com>
Here when trying to unify P=42 A=T::value we ICE due to the latter's
empty type, which same_type_p dislikes.
PR c++/113644
gcc/cp/ChangeLog:
* pt.cc (unify) <case INTEGER_CST>: Handle NULL_TREE type.
gcc/testsuite/ChangeLog:
* g++.dg/template/nontype30.C: New test.
The test_format() function contained an incorrect assertion but wasn't
actually being called from main.
libstdc++-v3/ChangeLog:
* testsuite/std/time/clock/gps/io.cc: Fix expected result in
assertion and call test_format() from main.
According to the issue as below.
https://hub.fgit.cf/riscv-non-isa/riscv-elf-psabi-doc/pull/416
When the mode size of vls integer mode is less than 2 * XLEN, we will
take the gpr for both the args and the return values. Instead of the
reference. For example the below code:
typedef short v8hi __attribute__ ((vector_size (16)));
v8hi __attribute__((noinline))
add (v8hi a, v8hi b)
{
v8hi r = a + b;
return r;
}
Before this patch:
add:
vsetivli zero,8,e16,m1,ta,ma
vle16.v v1,0(a1) <== arg by reference
vle16.v v2,0(a2) <== arg by reference
vadd.vv v1,v1,v2
vse16.v v1,0(a0) <== return by reference
ret
After this patch:
add:
addi sp,sp,-32
sd a0,0(sp) <== arg by register a0 - a3
sd a1,8(sp)
sd a2,16(sp)
sd a3,24(sp)
addi a5,sp,16
vsetivli zero,8,e16,m1,ta,ma
vle16.v v2,0(sp)
vle16.v v1,0(a5)
vadd.vv v1,v1,v2
vse16.v v1,0(sp)
ld a0,0(sp) <== return by a0 - a1.
ld a1,8(sp)
addi sp,sp,32
jr ra
For vls floating point, we take the same rules as integer and passed by
the gpr or reference.
The riscv regression passed for this patch.
gcc/ChangeLog:
* config/riscv/riscv.cc (riscv_v_vls_mode_aggregate_gpr_count): New function to
calculate the gpr count required by vls mode.
(riscv_v_vls_to_gpr_mode): New function convert vls mode to gpr mode.
(riscv_pass_vls_aggregate_in_gpr): New function to return the rtx of gpr
for vls mode.
(riscv_get_arg_info): Add vls mode handling.
(riscv_pass_by_reference): Return false if arg info has no zero gpr count.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/rvv/autovec/vls/def.h: Add new helper macro.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-1.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-10.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-2.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-3.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-4.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-5.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-6.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-7.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-8.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-9.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-run-1.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-run-2.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-run-3.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-run-4.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-run-5.c: New test.
* gcc.target/riscv/rvv/autovec/vls/calling-convention-run-6.c: New test.
Signed-off-by: Pan Li <pan2.li@intel.com>
This is useful for debugging the analyzer.
gcc/analyzer/ChangeLog:
* region-model.cc
(dubious_allocation_size::dubious_allocation_size): Add
"capacity_sval" param. Drop unused ctor.
(dubious_allocation_size::maybe_add_sarif_properties): New.
(dubious_allocation_size::m_capacity_sval): New field.
(region_model::check_region_size): Pass capacity svalue to
dubious_allocation_size ctor.
Signed-off-by: David Malcolm <dmalcolm@redhat.com>
If a struct type with a variant that has fields is initialized with some fields the expression HIR StructExprStructFields is checked that all the fields are assigned. However, if no fields are initialized the HIR StructExprStruct is generated. This doesn't check if the struct is a unit during typechekc and only fails at the compile stage with a ICE.
Add a check at the typecheck stage that makes sure the struct does not have a variant with fields and give an error message based on the rustc one.
We have also updated the message given in the case where one field was present to list the missing fields and match more closely the new message.
gcc/rust/ChangeLog:
* typecheck/rust-hir-type-check-expr.cc (TypeCheckExpr::visit) Add additional check
* typecheck/rust-hir-type-check-struct-field.h: A helper method to make error added
* typecheck/rust-hir-type-check-struct.cc (TypeCheckStructExpr::resolve) Update message
gcc/testsuite/ChangeLog:
* rust/compile/missing_constructor_fields.rs: Added case with no initializers
Signed-off-by: Robert Goss <goss.robert@gmail.com>
gcc/rust/ChangeLog:
* hir/rust-ast-lower-type.cc (ASTLoweringTypeBounds::visit): fix for lifetimes
(ASTLowerWhereClauseItem::visit): fix for lifetimes
Signed-off-by: Jakub Dupak <dev@jakubdupak.com>
This test did not compile with rustc.
gcc/testsuite/ChangeLog:
* rust/compile/torture/utf8_identifiers.rs: add mising lifetime
Signed-off-by: Jakub Dupak <dev@jakubdupak.com>
Fixes#2381
If a struct type is initialized with one of it's fields repeated it will currently issue an error at the use site. However we would like to give the rust error code and (like rustc) show both the specifications for the field to help the user diagnose the issue.
We move the check after the index for the field has been established so we can look up if the field has already been defined and get it's location.
We update the visit method to return if it has handled an error otherwise we then output a second fatal error as not all the field in the specification have been processed.
gcc/rust/ChangeLog:
* typecheck/rust-hir-type-check-struct-field.h: Allow visit to return a bool
* typecheck/rust-hir-type-check-struct.cc: Improve check of repeat fields
gcc/testsuite/ChangeLog:
* rust/compile/repeated_constructor_fields.rs: Added case with constructor field repeated
Signed-off-by: Robert Goss <goss.robert@gmail.com>
gcc/rust/ChangeLog:
* checks/errors/rust-ast-validation.cc (ASTValidation::visit):
Enclose const in single quotes.
gcc/testsuite/ChangeLog:
* rust/compile/const_trait_fn.rs:
Enclose const in single quotes.
Signed-off-by: Kushal Pal <kushalpal109@gmail.com>
If newline strings are encountered while lexing, the lexer now handles
newline characters by incrementing current line number. This provides
correct line number when displaying errors. If the lexer encounters end
of file before string end, then it will use the start of the string as
the location to an report error.
gcc/rust/ChangeLog:
* lex/rust-lex.cc (Lexer::parse_byte_string): Handle newline
while parsing byte strings
(Lexer::parse_string): Handle newline while parsing strings
Signed-off-by: Nirmal Patel <nirmal@nirmal.dev>
gcc/rust/ChangeLog:
* backend/rust-compile-expr.cc (CompileExpr::visit): Use new API.
* typecheck/rust-tyty-call.cc (TypeCheckCallExpr::visit): Use new API.
* typecheck/rust-tyty-cmp.h: Remove old API.
* typecheck/rust-tyty.cc (FnPtr::is_equal): Use new API.
* typecheck/rust-tyty.h: Remove old API.
* typecheck/rust-unify.cc (UnifyRules::expect_fnptr): Use new API.
Signed-off-by: Jakub Dupak <dev@jakubdupak.com>
There was a mismatch whether lifetime 'static is parsed as "static"
or "'static".
gcc/rust/ChangeLog:
* parse/rust-parse-impl.h (Parser::lifetime_from_token): Fix matched pattern.
Signed-off-by: Jakub Dupak <dev@jakubdupak.com>
gcc/rust/ChangeLog:
* hir/tree/rust-hir-item.h: Ad lifetime getter.
* hir/tree/rust-hir-path.h: Make getter const ref.
* hir/tree/rust-hir.h: Const ref and new getter.
Signed-off-by: Jakub Dupak <dev@jakubdupak.com>
Many visit functions in the expand visitor simply visit their components
like the default visitor. Making the expand visitor inherit from the
default visitor allows us to keep all visitor in sync without having to
change every visitor.
gcc/rust/ChangeLog:
* expand/rust-expand-visitor.cc (ExpandVisitor::go): Add call to visit
on the crate.
(ExpandVisitor::visit): Remove some visit functions in favor of their
default visitor counterpart.
* expand/rust-expand-visitor.h (class ExpandVisitor): Inherit from
default visitor and remove now useless function prototypes.
Signed-off-by: Pierre-Emmanuel Patry <pierre-emmanuel.patry@embecosm.com>
The default resolver put some scope in place but mostly has traversal
functions similar to the default ast visitor, making it inherit from the
default visitor allows us to avoid code duplication.
gcc/rust/ChangeLog:
* resolve/rust-default-resolver.cc (DefaultResolver::visit): Remove
duplicated functions.
* resolve/rust-default-resolver.h (class DefaultResolver): Make the
default resolver inherit from the default visitor.
Signed-off-by: Pierre-Emmanuel Patry <pierre-emmanuel.patry@embecosm.com>
