re PR tree-optimization/52870 (ICE during SLP pattern matching)
gcc/
PR tree-optimization/52870
* tree-vect-patterns.c (vect_recog_widen_mult_pattern): Verify that
presumed pattern statement is within the same loop or basic block.
gcc/testsuite/
PR tree-optimization/52870
* gcc.dg/vect/pr52870.c: New test.
Richard Guenther [Tue, 10 Apr 2012 08:19:21 +0000 (08:19 +0000)]
tree-pass.h (tree_lowering_passes): Remove.
2012-04-10 Richard Guenther <rguenther@suse.de>
* tree-pass.h (tree_lowering_passes): Remove.
* tree-optimize.c (tree_lowering_passes): Remove.
* cgraph.c (cgraph_add_new_function): Inline relevant parts
of tree_lowering_passes, avoid redundant call of early local
passes.
* cgraphunit.c (cgraph_lower_function): Fold into ...
(cgraph_analyze_function): ... its single caller. Inline
relevant parts of tree_lowering_passes.
Jan Hubicka [Mon, 9 Apr 2012 23:39:11 +0000 (01:39 +0200)]
re PR lto/52722 (ICE in lto_output_varpool_node)
PR lto/52722
PR lto/51765
PR lto/52634
* lto-cgraph.c (compute_ltrans_boundary): When alias is in the boundary,
add its target too.
* lto.c (add_references_to_partition): Add also aliased nodes.
(add_cgraph_node_to_partition,
add_varpool_node_to_partition): Work on nodes, not functions/variables;
when adding alias, add also the aliased object.
* gcc.dg/lto/pr52634_1.c: New testcase.
* gcc.dg/lto/pr52634_0.c: New testcase.
gcc/
* doc/invoke.texi: Clean up Texinfo markup throughout the file.
Use @option markup on command-line options. Use @samp markup on
literal keywords to options. Use @code markup on code fragments.
Use other markup in preference to quotation marks in the text.
Add markup to some passages without any.
sh-protos.h (sh_expand_t_scc): Change return type from int to bool.
* config/sh/sh-protos.h (sh_expand_t_scc): Change return type from int
to bool.
* config/sh/sh.c (sh_expand_t_scc): Likewise.
* config/sh/sh.md (cstoresi4, cstoredi4): Remove GET_CODE checks before
calling sh_expand_t_scc.
Teresa Johnson [Fri, 6 Apr 2012 05:03:49 +0000 (05:03 +0000)]
Avoid instructions that incur expensive length-changing prefix (LCP) stalls on some x86-64 implementations...
Avoid instructions that incur expensive length-changing prefix (LCP) stalls
on some x86-64 implementations, notably Core2 and Corei7. Specifically, a move of
a 16-bit constant into memory requires a length-changing prefix and can incur significant
penalties. Modified an old patch written by H.J to split such instructions
during peephole2.
2012-04-05 Teresa Johnson <tejohnson@google.com>
H.J. Lu <hongjiu.lu@intel.com>
* config/i386/i386.h (ix86_tune_indices): Add
X86_TUNE_LCP_STALL.
* config/i386/i386.md (move immediate to memory peephole2):
Add cases for HImode move when LCP stall avoidance is needed.
* config/i386/i386.c (initial_ix86_tune_features): Initialize
X86_TUNE_LCP_STALL entry.
Co-Authored-By: H.J. Lu <hongjiu.lu@intel.com>
From-SVN: r186176
re PR target/50751 (SH Target: Displacement addressing does not work for QImode and HImode)
PR target/50751
* config/sh/sh.c (sh_find_mov_disp_adjust): Take machine_mode as the
first argument instead of mode size. Move displacement calculations
to ...
(mov_insn_size, max_mov_insn_displacement, mov_insn_alignment_mask):
... these new functions.
(disp_adjust): Remove max_mov_disp field.
(sh_legitimate_index_p): Use max_mov_insn_displacement and
mov_insn_alignment_mask.
(sh_address_cost): Use max_mov_insn_displacement.
gimple.c (walk_gimple_op): Compute val_only for the LHS of an assigment in the canonical way...
2012-04-05 Richard Guenther <rguenther@suse.de>
* gimple.c (walk_gimple_op): Compute val_only for the LHS
of an assigment in the canonical way, avoiding is_gimple_mem_rhs.
(is_gimple_mem_rhs, is_gimple_reg_rhs, is_gimple_stmt): Move ...
* gimplify.c (is_gimple_mem_rhs, is_gimple_reg_rhs, is_gimple_stmt):
... here and make static.
* gimple.h (is_gimple_mem_rhs, is_gimple_reg_rhs, is_gimple_stmt):
Remove.
Mike Stump [Wed, 4 Apr 2012 20:27:17 +0000 (20:27 +0000)]
rtl.texi (const_double): Document as sign-extending.
* doc/rtl.texi (const_double): Document as sign-extending.
* expmed.c (expand_mult): Ensure we don't use shift
incorrectly.
* emit-rtl.c (immed_double_int_const): Refine to state the
value is signed.
* simplify-rtx.c (mode_signbit_p): Add a fixme for wider than
CONST_DOUBLE integers.
(simplify_const_unary_operation, UNSIGNED_FLOAT): Ensure no
negative values are converted. Fix conversions bigger than
HOST_BITS_PER_WIDE_INT.
(simplify_binary_operation_1): Ensure we don't use shift
incorrectly.
(simplify_immed_subreg): Sign-extend CONST_DOUBLEs.
* explow.c (plus_constant_mode): Add.
(plus_constant): Implement with plus_constant_mode.
* rtl.h (plus_constant_mode): Add.
Eric Botcazou [Tue, 3 Apr 2012 17:59:07 +0000 (17:59 +0000)]
expr.c (get_bit_range): Add OFFSET parameter and adjust BITPOS.
* expr.c (get_bit_range): Add OFFSET parameter and adjust BITPOS.
Change type of BITOFFSET to signed. Make sure the lower bound of
the computed range is non-negative by adjusting OFFSET and BITPOS.
(expand_assignment): Adjust call to get_bit_range.
h8300.c (h8300_current_function_monitor_function_p): New function.
* h8300/h8300.c (h8300_current_function_monitor_function_p):
New function. Added to check monitor functions.
(h8300_option_override): Modified to generate error/warning
messages for invalid combinations of different command line
options.
* h8300/h8300.md: Generate 'rte' for monitor functions. Do not
save EXR on stack for monitor function in case of H8S target
when "-mno-exr" is passed.
* h8300/h8300-protos.h
(h8300_current_function_monitor_function_p): Add prototype.
* doc/invoke.texi: Document H8S options.
re PR bootstrap/52808 (LTO bootstrap failed with bootstrap-profiled)
2012-04-03 Richard Guenther <rguenther@suse.de>
PR tree-optimization/52808
* tracer.c (tail_duplicate): Return whether we have duplicated
any block.
(tracer): If we have duplicated any block, cleanup the CFG.
* cfghooks.c (duplicate_block): If we duplicated a loop
header but not its loop, destroy the loop because it now has
multiple entries.
* tree-ssa-threadupdate.c (thread_through_loop_header): Tell
the cfg manipulation routines we are not creating a multiple
entry loop.
Eric Botcazou [Tue, 3 Apr 2012 07:11:25 +0000 (07:11 +0000)]
varasm.c (initializer_constant_valid_for_bitfield_p): Return true for REAL_CST as well.
* varasm.c (initializer_constant_valid_for_bitfield_p): Return true
for REAL_CST as well.
(output_constructor): Use RECORD_OR_UNION_TYPE_P predicate.
In the bitfield case, if the value is a REAL_CST, convert it first to
an INTEGER_CST.
H.J. Lu [Tue, 3 Apr 2012 02:56:34 +0000 (02:56 +0000)]
Add --with-abi= support to x86_64-*-*
2012-04-02 H.J. Lu <hongjiu.lu@intel.com>
* config.gcc: Use i386/biarchx32.h instead of i386/biarch64.h
for --with-abi={x32|mx32} or --with-multilib-list=mx32.
(supported_defaults): Add abi for i[34567]86-*-* and x86_64-*-*.
re PR middle-end/52756 (255.vortex in SPEC CPU 2000 failed to build)
2012-04-02 Richard Guenther <rguenther@suse.de>
PR tree-optimization/52756
* tree-ssa-threadupdate.c (def_split_header_continue_p): New function.
(thread_through_loop_header): After threading through the loop latch
remove the split part from the loop and clear further threading
opportunities that would create a multiple entry loop.
* sem_util.adb: Minor reformatting, minor code cleanup.
2012-04-02 Ed Schonberg <schonberg@adacore.com>
* lib-xref.adb (Generate_Reference): For a reference to an
operator symbol, set the sloc to point to the first character
of the operator name, and not to the initial quaote.
(Output_References): Ditto for the definition of an operator
symbol.
2012-04-02 Vincent Celier <celier@adacore.com>
* ali.adb (Scan_Ali): Recognize Z lines. Set
Implicit_With_From_Instantiation to True in the With_Record for
Z lines.
* ali.ads (With_Record): New Boolean component
Implicit_With_From_Instantiation, defaulted to False.
* csinfo.adb: Indicate that Implicit_With_From_Instantiation
is special
* lib-writ.adb (Write_ALI): New array Implicit_With.
(Collect_Withs): Set Implicit_With for the unit is it is not Yes.
(Write_With_Lines): Write a Z line instead of a W line if
Implicit_With is Yes for the unit.
* sem_ch12.adb (Inherit_Context): Only add a unit in the context
if it is not there yet.
* sinfo.ads: New flag Implicit_With_From_Instantiation (Flag12)
added.
* osint.adb, osint.ads (Add_Default_Search_Dirs): Add library
search dirs in file specified with option -gnateO.
2012-04-02 Robert Dewar <dewar@adacore.com>
* sem_ch5.adb, exp_util.adb, sem_util.adb, exp_ch4.adb: Minor
reformatting.
2012-04-02 Olivier Hainque <hainque@adacore.com>
* g-sse.ads: Add x86-solaris and x86_64-darwin to the set of
platforms where the use of this spec is supported. Add current
year to the copyright notice.
* gcc-interfaces/Makefile.in: Add g-sse.o and g-ssvety.o to
EXTRA_GNATRTL_NONTASKING_OBJS on x86 32/64 targets that support
it and where they were missing (x86-solaris, x86-freebsd,
x86_64-freebsd, and x86-darwin).
2012-04-02 Gary Dismukes <dismukes@adacore.com>
* bindgen.adb (Gen_Ada_Init): When compiling for the AAMP small
library, where we no longer suppress the Standard_Library,
generate an empty body rather than the usual generation of
assignments to imported globals, since those aren't present in
the small library.
* sem_res.adb (Resolve_Conditional_Expression): Add local variables
Else_Typ and Then_Typ. Add missing type conversions to the "then" and
"else" expressions when their respective types are scalar.
2012-04-02 Vincent Pucci <pucci@adacore.com>
* exp_ch9.adb: Reordering of the local subprograms. New Table
for the lock free implementation that maps each protected
subprograms with the protected component it references.
(Allow_Lock_Free_Implementation): New routine. Check if
the protected body enables the lock free implementation.
(Build_Lock_Free_Protected_Subprogram_Body): New routine.
(Build_Lock_Free_Unprotected_Subprogram_Body): New routine.
(Comp_Of): New routine.
* Makefile.rtl: Add s-atopri.o
* debug.adb: New compiler debug flag -gnatd9 for lock free
implementation.
* rtsfind.ads: RE_Atomic_Compare_Exchange_8,
RE_Atomic_Compare_Exchange_16, RE_Atomic_Compare_Exchange_32,
RE_Atomic_Compare_Exchange_64, RE_Atomic_Load_8,
RE_Atomic_Load_16, RE_Atomic_Load_32, RE_Atomic_Load_64, RE_Uint8,
RE_Uint16, RE_Uint32, RE_Uint64 added.
* s-atropi.ads: New file. Defines atomic primitives used
by the lock free implementation.
* exp_ch4.adb (Expand_N_Quantified_Expression): Reimplemented.
The expansion no longer uses the copy of the original QE created
during analysis.
* sem.adb (Analyze): Add processing for loop parameter specifications.
* sem_ch4.adb (Analyze_Quantified_Expression): Reimplemented. The
routine no longer creates a copy of the original QE. All
constituents of a QE are now preanalyzed and resolved.
* sem_ch5.adb (Analyze_Iteration_Scheme): Remove the guard which
bypasses all processing when the iteration scheme is related to a
QE. Relovate the code which analyzes loop parameter specifications
to a separate routine. (Analyze_Iterator_Specification):
Preanalyze the iterator name. This action was originally
done in Analyze_Iteration_Scheme. Update the check which
detects an iterator specification in the context of a QE.
(Analyze_Loop_Parameter_Specification): New routine. This
procedure allows for a stand-alone analysis of a loop parameter
specification without the need of a parent iteration scheme. Add
code to update the type of the loop variable when the range
generates an itype and the context is a QE.
(Pre_Analyze_Range): Renamed to Preanalyze_Range. Update all references
to the routine.
* sem_ch5.ads: Code reformatting.
(Analyze_Loop_Parameter_Specification): New routine.
* sem_ch6.adb (Fully_Conformant_Expressions): Detect a case
when establishing conformance between two QEs utilizing different
specifications.
* sem_res.adb (Proper_Current_Scope): New routine.
(Resolve): Do not resolve a QE as there is nothing to be done now.
Ignore any loop scopes generated for QEs when detecting an expression
function as the scopes are cosmetic and do not appear in the tree.
(Resolve_Quantified_Expression): Removed. All resolution of
QE constituents is now performed during analysis. This ensures
that loop variables appearing in array aggregates are properly
resolved.
2012-04-02 Ed Schonberg <schonberg@adacore.com>
* sem_util.adb (Build_Default_Subtype): If the base type is
private and its full view is available, use the full view in
the subtype declaration.
* gnat_ugn.texi: Add some minimal documentation about how to
use GNATtest for cross platforms.
2012-04-02 Vincent Celier <celier@adacore.com>
* opt.ads (Object_Path_File_Name): New variable.
* prj-attr.adb: New Compiler attribute Object_Path_Switches.
* prj-nmsc.adb (Process_Compiler): Recognize new attribute
Object_Path_Switches.
* snames.ads-tmpl: New standard name Object_Path_Switches.
* switch-c.adb (Scan_Front_End_Switches): Recognize new switch
-gnateO= and put its value in Opt.Object_Path_File_Name.
* exp_ch7.adb (Process_Declarations): Detect a case where
a source object was initialized by another source object,
but the expression was rewritten as a class-wide conversion
of Ada.Tags.Displace.
* exp_util.adb (Initialized_By_Ctrl_Function): Removed.
(Is_Controlled_Function_Call): New routine.
(Is_Displacement_Of_Ctrl_Function_Result): Removed.
(Is_Displacement_Of_Object_Or_Function_Result): New routine.
(Is_Source_Object): New routine.
(Requires_Cleanup_Actions): Detect a case where a source object was
initialized by another source object, but the expression was rewritten
as a class-wide conversion of Ada.Tags.Displace.
* exp_util.ads (Is_Displacement_Of_Ctrl_Function_Result): Removed.
(Is_Displacement_Of_Object_Or_Function_Result): New routine.
2012-04-02 Ed Schonberg <schonberg@adacore.com>
* sem_res.adb (Resolve_Call): A call to an expression function
does not freeze if it appears in a different scope from the
expression function itself. Such calls appear in the generated
bodies of other expression functions, or in pre/postconditions
of subsequent subprograms.
* sem_ch12.adb (Analyze_Subprogram_Instantiation): Do not suppress
style checks, because the subprogram instance itself may contain
violations of syle rules.
* style.adb (Missing_Overriding): Check for missing overriding
indicator on a subprogram instance.
* sem_ch6.adb (Last_Implicit_Declaration): New routine.
(Process_PPCs): Insert the body of _postconditions after the
last internally generated declaration. This ensures that actual
subtypes created for formal parameters are visible and properly
frozen as _postconditions may reference them.
Robert Dewar [Mon, 2 Apr 2012 09:14:47 +0000 (09:14 +0000)]
2012-04-02 Robert Dewar <dewar@adacore.com>
* einfo.adb (First_Component_Or_Discriminant) Now applies to
all types with discriminants, not just records.
* exp_attr.adb (Expand_N_Attribute): Add Scalar_Values handling
for arrays, scalars and non-variant records.
* sem_attr.adb (Analyze_Attribute): Handle Valid_Scalars
* sem_attr.ads (Valid_Scalars): Update description
* sem_util.ads, sem_util.adb (No_Scalar_Parts): New function.
projects / thirdparty / gcc.git / log
