Allow building on systems with elf.h that includes AMDGPU definitions,
partially or completely.
gcc/ChangeLog:
* config/gcn/mkoffload.c (EM_AMDGPU): Undefine before defining.
(ELFOSABI_AMDGPU_HSA): Likewise.
(ELFABIVERSION_AMDGPU_HSA): Likewise.
(EF_AMDGPU_MACH_AMDGCN_GFX803): Likewise.
(EF_AMDGPU_MACH_AMDGCN_GFX900): Likewise.
(EF_AMDGPU_MACH_AMDGCN_GFX906): Likewise.
(reserved): Delete.
The following patch enables vector permutations optimization by trying
to use ins instruction instead of slow and generic tbl.
example:
vector float f0(vector float a, vector float b)
{
return __builtin_shuffle (a, a, (vector int){3, 1, 2, 3});
}
was compiled into:
...
adrp x0, .LC0
ldr q1, [x0, #:lo12:.LC0]
tbl v0.16b, {v0.16b}, v1.16b
...
and after patch:
...
ins v0.s[0], v0.s[3]
...
bootstrapped and tested on aarch64-linux-gnu with no regressions
gcc/ChangeLog:
2020-07-17 Andrew Pinski <apinksi@marvell.com>
PR target/93720
* config/aarch64/aarch64.c (aarch64_evpc_ins): New function.
(aarch64_expand_vec_perm_const_1): Call it.
* config/aarch64/aarch64-simd.md (aarch64_simd_vec_copy_lane): Make
public, and add a "@" prefix.
gcc/testsuite/ChangeLog:
2020-07-17 Andrew Pinski <apinksi@marvell.com>
PR target/93720
* gcc.target/aarch64/vins-1.c: New test.
* gcc.target/aarch64/vins-2.c: New test.
* gcc.target/aarch64/vins-3.c: New test.
Co-Authored-By: Dmitrij Pochepko <dmitrij.pochepko@bell-sw.com>
The following patch enables vector permutations optimization by using
another vector element size when applicable. It allows usage of simpler
instructions in applicable cases.
example:
vector float f(vector float a, vector float b)
{
return __builtin_shuffle (a, b, (vector int){0, 1, 4,5});
}
was compiled into:
...
adrp x0, .LC0
ldr q2, [x0, #:lo12:.LC0]
tbl v0.16b, {v0.16b - v1.16b}, v2.16b
...
and after patch:
...
zip1 v0.2d, v0.2d, v1.2d
...
bootstrapped and tested on aarch64-linux-gnu with no regressions
gcc/ChangeLog:
2020-07-17 Andrew Pinski <apinksi@marvell.com>
PR target/82199
* config/aarch64/aarch64.c (aarch64_evpc_reencode): New function.
(aarch64_expand_vec_perm_const_1): Call it.
gcc/testsuite/ChangeLog:
2020-07-17 Andrew Pinski <apinksi@marvell.com>
PR target/82199
* gcc.target/aarch64/vdup_n_3.c: New test.
* gcc.target/aarch64/vzip_1.c: New test.
* gcc.target/aarch64/vzip_2.c: New test.
* gcc.target/aarch64/vzip_3.c: New test.
* gcc.target/aarch64/vzip_4.c: New test.
Co-Authored-By: Dmitrij Pochepko <dmitrij.pochepko@bell-sw.com>
In s390_expand_insv the movstrict patterns are always generated with a
CC clobber although only movstricthi actually needs one. The patch
invokes the expanders instead of constructing the pattern by hand.
Bootstrapped and regression tested on s390x.
gcc/ChangeLog:
PR target/96127
* config/s390/s390.c (s390_expand_insv): Invoke the movstrict
expanders to generate the pattern.
* config/s390/s390.md ("*movstricthi", "*movstrictqi"): Remove the
'*' to have callable expanders.
gcc/testsuite/ChangeLog:
PR target/96127
* gcc.target/s390/pr96127.c: New test.
In my testing with cost tweaking for vector with length, I found
two cases below didn't get the expected output. Since the expected
instructions reply on the vectorization occurrence, we don't expect
vectorization gets disabled by cost model.
To make it not fragile, the fix is to force it without vect cost model.
gcc/testsuite/ChangeLog:
* gcc.target/powerpc/conv-vectorize-1.c: Add option
-fno-vect-cost-model.
* gcc.target/powerpc/conv-vectorize-2.c: Likewise.
Since we now only call is_just_move on the original instructions, we
always have an rtx_insn* (not just a pattern), so we can use single_set
on it. This makes no detectable difference at all on all thirty Linux
targets I test, but it does help cris, and it is simpler, cleaner code
anyway.
2020-07-16 Hans-Peter Nilsson <hp@axis.com>
Segher Boessenkool <segher@kernel.crashing.org>
PR target/93372
* combine.c (is_just_move): Take an rtx_insn* as argument. Use
single_set on it.
This patch makes it so that an "attach" operation for a Fortran pointer
with an array descriptor copies that array descriptor to the target,
and similarly that detach operations release the array descriptor.
2020-07-16 Julian Brown <julian@codesourcery.com>
Thomas Schwinge <thomas@codesourcery.com>
gcc/fortran/
* trans-openmp.c (gfc_trans_omp_clauses): Rework OpenACC
attach/detach handling for arrays with descriptors.
gcc/testsuite/
* gfortran.dg/goacc/attach-descriptor.f90: New test.
libgomp/
* testsuite/libgomp.oacc-fortran/attach-descriptor-1.f90: New test.
* testsuite/libgomp.oacc-fortran/attach-descriptor-2.f90: New test.
Co-Authored-By: Thomas Schwinge <thomas@codesourcery.com>
When a full expression contains a co_await (or co_yield), this means
that it might be suspended; which would destroy temporary variables
(on a stack for example). However the language guarantees that such
temporaries are live until the end of the expression.
In order to preserve this, we 'promote' temporaries where necessary
so that they are saved in the coroutine frame (which allows them to
remain live potentially until the frame is destroyed). In addition,
sub-expressions that produce control flow (such as TRUTH_AND/OR_IF
or COND_EXPR) must be handled specifically to ensure that await
expressions are properly expanded.
This patch corrects two mistakes in which we were (a) failing to
promote some temporaries and (b) we we failing to sequence DTORs for
the captures properly. This manifests in a number of related (but not
exact duplicate) PRs.
The revised code collects the actions into one place and maps all the
control flow into one form - a benefit of this is that co_returns are
now expanded earlier (which provides an opportunity to address PR95517
in some future patch).
We replace a statement that contains await expression(s) with a bind
scope that has a variable list describing the temporaries that have
been 'promoted' and a statement list that contains a series of cleanup
expressions for each of those. Where we encounter nested conditional
expressions, these are wrapped in a try-finally block with a guard var
for each sub-expression variable that needs a DTOR. The guards are all
declared and initialized to false before the first conditional sub-
expression. The 'finally' block contains a series of if blocks (one
per guard variable) enclosing the relevant DTOR.
Variables listed in a bind scope in this manner are automatically moved
to a coroutine frame version by existing code (so we re-use that rather
than having a separate mechanism).
gcc/cp/ChangeLog:
PR c++/95591
PR c++/95599
PR c++/95823
PR c++/95824
PR c++/95895
* coroutines.cc (struct coro_ret_data): Delete.
(coro_maybe_expand_co_return): Delete.
(co_return_expander): Delete.
(expand_co_returns): Delete.
(co_await_find_in_subtree): Remove unused name.
(build_actor_fn): Remove unused parm, remove handling
for co_return expansion.
(register_await_info): Demote duplicate info message to a
warning.
(coro_make_frame_entry): Move closer to use site.
(struct susp_frame_data): Add fields for final suspend label
and a flag to indicate await expressions with initializers.
(captures_temporary): Delete.
(register_awaits): Remove unused code, update comments.
(find_any_await): New.
(tmp_target_expr_p): New.
(struct interesting): New.
(find_interesting_subtree): New.
(struct var_nest_node): New.
(flatten_await_stmt): New.
(handle_nested_conditionals): New.
(process_conditional): New.
(replace_statement_captures): Rename to...
(maybe_promote_temps): ... this.
(maybe_promote_captured_temps): Delete.
(analyze_expression_awaits): Check for await expressions with
initializers. Simplify handling for truth-and/or-if.
(expand_one_truth_if): Simplify (map cases that need expansion
to COND_EXPR).
(await_statement_walker): Handle CO_RETURN_EXPR. Simplify the
handling for truth-and/or-if expressions.
(register_local_var_uses): Ensure that we create names in the
implementation namespace.
(morph_fn_to_coro): Add final suspend label to suspend frame
callback data and remove it from the build_actor_fn call.
gcc/testsuite/ChangeLog:
PR c++/95591
PR c++/95599
PR c++/95823
PR c++/95824
PR c++/95895
* g++.dg/coroutines/pr95591.C: New test.
* g++.dg/coroutines/pr95599.C: New test.
* g++.dg/coroutines/pr95823.C: New test.
* g++.dg/coroutines/pr95824.C: New test.
gcc/testsuite/ChangeLog:
PR testsuite/96014
* g++.dg/analyzer/pr94028.C: Replace dynamic exception
specification with noexcept-specifier for C++11 and later.
CMPXCHG instruction sets ZF flag if the values in the destination operand
and EAX register are equal; otherwise the ZF flag is cleared and value
from destination operand is loaded to EAX. Following assembly:
xorl %eax, %eax
lock cmpxchgl %edx, (%rdi)
testl %eax, %eax
sete %al
can be optimized by removing the unneeded comparison, since set ZF flag
signals that no update to EAX happened. This patch adds peephole2
pattern to also handle XOR zeroing and load of -1 by OR.
2020-07-16 Uroš Bizjak <ubizjak@gmail.com>
gcc/ChangeLog:
PR target/96189
* config/i386/sync.md
(peephole2 to remove unneded compare after CMPXCHG):
New pattern, also handle XOR zeroing and load of -1 by OR.
gcc/testsuite/ChangeLog:
PR target/96189
* gcc.target/i386/pr96189-1.c: New test.
The stack clash protection mechanism in the x86 back-end was implemented
by largely duplicating the existing stack checking implementation. Now
the only significant difference between them is the probing window, which
is shifted by 1 probing interval (not 2 as documented in explow.c), but we
can certainly do 1 more probe for stack checking even if it is redundant
in almost all cases.
gcc/ChangeLog:
* config/i386/i386.c (ix86_compute_frame_layout): Minor tweak.
(ix86_adjust_stack_and_probe): Delete.
(ix86_adjust_stack_and_probe_stack_clash): Rename to above and add
PROTECTION_AREA parameter. If it is true, probe PROBE_INTERVAL plus
a small dope beyond SIZE bytes.
(ix86_emit_probe_stack_range): Use local variable.
(ix86_expand_prologue): Adjust calls to ix86_adjust_stack_and_probe
and tidy up the stack checking code.
* explow.c (get_stack_check_protect): Fix head comment.
(anti_adjust_stack_and_probe_stack_clash): Likewise.
(allocate_dynamic_stack_space): Add comment.
* tree-nested.c (lookup_field_for_decl): Set the DECL_IGNORED_P and
TREE_NO_WARNING but not TREE_ADDRESSABLE flags on the field.
Forward the early debug information from the input LTO file to the output
HSACO file, in the same way lto-wrapper does. This is a little more
complicated, however, because the ELF file containing the debug needs to be
converted from x86_64 to amdgcn, and because the offloaded code will have less
content than the host program the debug info describes.
gcc/ChangeLog:
* config/gcn/mkoffload.c: Include simple-object.h and elf.h.
(EM_AMDGPU): New macro.
(ELFOSABI_AMDGPU_HSA): New macro.
(ELFABIVERSION_AMDGPU_HSA): New macro.
(EF_AMDGPU_MACH_AMDGCN_GFX803): New macro.
(EF_AMDGPU_MACH_AMDGCN_GFX900): New macro.
(EF_AMDGPU_MACH_AMDGCN_GFX906): New macro.
(R_AMDGPU_NONE): New macro.
(R_AMDGPU_ABS32_LO): New macro.
(R_AMDGPU_ABS32_HI): New macro.
(R_AMDGPU_ABS64): New macro.
(R_AMDGPU_REL32): New macro.
(R_AMDGPU_REL64): New macro.
(R_AMDGPU_ABS32): New macro.
(R_AMDGPU_GOTPCREL): New macro.
(R_AMDGPU_GOTPCREL32_LO): New macro.
(R_AMDGPU_GOTPCREL32_HI): New macro.
(R_AMDGPU_REL32_LO): New macro.
(R_AMDGPU_REL32_HI): New macro.
(reserved): New macro.
(R_AMDGPU_RELATIVE64): New macro.
(gcn_s1_name): Delete global variable.
(gcn_s2_name): Delete global variable.
(gcn_o_name): Delete global variable.
(gcn_cfile_name): Delete global variable.
(files_to_cleanup): New global variable.
(offload_abi): New global variable.
(tool_cleanup): Use files_to_cleanup, not explicit list.
(copy_early_debug_info): New function.
(main): New local variables gcn_s1_name, gcn_s2_name, gcn_o_name,
gcn_cfile_name.
Create files_to_cleanup obstack.
Recognize -march options.
Copy early debug info from input .o files.
The convert_like* macros were introduced in
2000-03-05 Nathan Sidwell <nathan@codesourcery.com>
* call.c (convert_like): Macrofy.
(convert_like_with_context): New macro.
but now we can use overloading so we can do away with the macros.
I've also taken this chance to rename _real to _internal to make it
clear that it should not be called directly.
No functional change intended.
gcc/cp/ChangeLog:
* call.c (convert_like): Remove macro and introduce a new
wrapper instead.
(convert_like_with_context): Likewise.
(convert_like_real): Rename to convert_like.
(convert_like_real_1): Rename to convert_like_internal. Call
convert_like instead of convert_like_real therein.
(perform_direct_initialization_if_possible): Call convert_like
instead of convert_like_real.
2020-07-16 Roger Sayle <roger@nextmovesoftware.com>
gcc/ChangeLog:
* target.def (TARGET_TRULY_NOOP_TRUNCATION): Clarify that
targets that return false, indicating SUBREGs shouldn't be
used, also need to provide a trunc?i?i2 optab that performs this
truncation.
* doc/tm.texi: Regenerate.
gcc/ada/
* exp_ch3.adb (Expand_N_Full_Type_Declaration): Ensure a _master
declaration on limited types that might have tasks.
* exp_ch9.adb (Build_Master_Renaming): For private types, if we
are processing declarations in the private part, ensure that
master is inserted before its full declaration; otherwise the
master renaming may be inserted in the public part of the
package (and hence before the declaration of its _master
variable).
gcc/ada/
* sem_aggr.adb (Resolve_Container_Aggregate): Add semantic
checks for indexed aggregates, including component associations
and iterated component associations.
* exp_aggr.adb (Expand_Iterated_Component): New subprogram,
subsidiary of Expand_Container_Aggreggate, used for positional,
named, and indexed aggregates.
(Aggregate_Size): New subprogram to precompute the size of an
indexed aggregate prior to call to allocate it.
(Expand_Range_Component): New subprogram so generate loop for a
component association given by a range or a subtype name in an
indexed aggregate.
gcc/ada/
* bindo-diagnostics.adb (Output_Invocation_Related_Suggestions):
Use Cumulative_Restrictions.Set, because Restriction_Active only
works at compile time.
gcc/ada/
* gnatbind.adb (Gnatbind): For No_Tasks_Unassigned_To_CPU, check
that CPU has been set on the main subprogram.
(Restriction_Could_Be_Set): Don't print
No_Tasks_Unassigned_To_CPU if it would violate the
above-mentioned rule. Up to now, all restrictions were checked
by the compiler, with the binder just checking for consistency.
But the compiler can't know which subprogram is the main, so
it's impossible to check this one at compile time.
* restrict.ads, restrict.adb: Misc refactoring. Change Warning
to Warn, for consistency, since most already use Warn.
(Set_Restriction): New convenience routine.
* sem_ch13.adb (Attribute_CPU): Check
No_Tasks_Unassigned_To_CPU.
* sem_prag.adb (Pragma_CPU): Check No_Tasks_Unassigned_To_CPU.
Misc refactoring.
* tbuild.ads, tbuild.adb (Sel_Comp): New functions for building
selected components.
gcc/ada/
* sem_ch12.adb (Load_Parent_Of_Generic): If an ancestor is an
instance whose source appears within a formal package of the
current unit, there is no body of the ancestor needed to
complete the current generic compilation.
gcc/ada/
* libgnat/s-thread__ae653.adb (taskVarAdd): Defunct, so remove.
(Current_ATSD): Make it a TLS variable.
(OK): Move to package scope.
(System.Storage_Elements): Import and Use.
gcc/ada/
* exp_aggr.adb (Max_Aggregate_Size): Use the small size of 64
when copying is needed (for example, for the initialization of a
local variable, and for assignment statements). Use the larger
size when static allocation can be done without copying.
gcc/ada/
* libgnat/s-rident.ads (No_Dynamic_CPU_Assignment): New
restriction. Add it to all relevant profiles.
* sem_ch13.adb (Attribute_CPU): Check No_Dynamic_CPU_Assignment
restriction.
(Attribute_CPU, Attribute_Dispatching_Domain,
Attribute_Interrupt_Priority): Remove error checks -- these are
checked in the parser.
* sem_prag.adb (Pragma_CPU): Check No_Dynamic_CPU_Assignment
restriction. We've got a little violation of DRY here.
* sem.ads, sem_ch3.ads: Minor comment fix.
gcc/ada/
* sem_ch4.adb (Try_Container_Indexing): When the prefix type is
an access type, change it to the designated type, change the
prefix to an explicit dereference, and emit a ?d? warning for
the implicit dereference. Include a ??? comment questioning
whether this is the right context in which to perform the
implicit dereferencing.
gcc/ada/
* sem_ch13.adb (Validate_Literal_Aspect): Ensure that the
parameter is not aliased. Minor reformatting.
* sem_util.adb (Statically_Names_Object): Update comment.
gcc/ada/
* sem_ch6.adb (Null_Exclusions_Match): New function to check
that the null exclusions match, including in the case addressed
by this AI.
(Check_Conformance): Remove calls to Comes_From_Source
when calling Null_Exclusions_Match. These are not
needed, as indicated by an ancient "???" comment.
gcc/ada/
* exp_ch4.adb (Expand_N_Type_Conversion): Remove flawed test for
whether "statically deeper" accessibility rules apply to a given
target type and instead use the new routine
Statically_Deeper_Relation_Applies.
(Statically_Deeper_Relation_Applies): Created to centralize the
calculation of whether a target type within a conversion must
have static accessibility checks.
* sem_ch13.adb (Check_One_Function): Minor comment revision.
gcc/ada/
* einfo.adb, einfo.ads (Is_Named_Access_Type): Created for
readability.
* sem_ch6.adb (Check_Return_Construct_Accessibility): Add
special cases for formals.
* sem_util.adb (Object_Access_Level): Add handling of access
attributes and named access types in the general case.
gcc/ada/
* sem_ch8.adb (Note_Redundant_Use): Add missing warning tag.
Do not check for redundant use clauses in predefined units to avoid
misleading warnings that may occur as part of a rtsfind load.
gcc/ada/
* exp_attr.adb (Has_By_Protected_Procedure_Prefixed_View): New
subprogram.
(Expand_Access_To_Protected_Op): Adding support for prefixed
class-wide view with By_Protected_Procedure convention.
* sem_attr.adb (Get_Convention): New subprogram.
(Get_Kind): Adapted to use Get_Convention.
* sem_ch4.adb (Try_By_Protected_Procedure_Prefixed_View): New
subprogram.
(Analyze_Selected_Component): Invoke
Try_By_Protected_Procedure_Prefixed_View.
* sem_util.ads (Is_By_Protected_Procedure): New subprogram.
* sem_util.adb (Is_By_Protected_Procedure): New subprogram.
gcc/ada/
* libgnat/s-ststop.ads: Fix typo.
* libgnat/s-ststop.adb (Read, Write): Fix block number
computation to avoid overflows in case of large strings.
gcc/ada/
* libgnat/s-genbig.adb ("**"): Remove capacity limit check.
Improve code by using an extended return.
(Normalize): Perform capacity limit check here instead which is
the centralized place where (potentially large) big integers are
allocated.
Andrew Stubbs
Dmitrij Pochepko
xiezhiheng
Jonathan Wakely
Andreas Krebbel
Kewen Lin
GCC Administrator
Segher Boessenkool
Julian Brown
Iain Sandoe
Uros Bizjak
Eric Botcazou
Marek Polacek
Andrea Corallo
Roger Sayle
Javier Miranda
Arnaud Charlet
Ed Schonberg
Bob Duff
Eric Botcazou
Doug Rupp
Gary Dismukes
Ghjuvan Lacambre
Justin Squirek
Piotr Trojanek