Linaro GCC

Linaro GCC 4.7-2013.01

Milestone information

Project:
Linaro GCC
Series:
4.7
Version:
4.7-2013.01
Released:
 
Registrant:
Matthew Gretton-Dann
Release registered:
Active:
No. Drivers cannot target bugs and blueprints to this milestone.  

Download RDF metadata

Activities

Assigned to you:
No blueprints or bugs assigned to you.
Assignees:
2 Brice Dobry, 1 Christophe Lyon, 3 Matthew Gretton-Dann, 1 Michael Hope, 1 Ulrich Weigand, 1 Yvan Roux, 1 Zhenqiang Chen
Blueprints:
7 Implemented
Bugs:
3 Fix Released

Download files for this release

After you've downloaded a file, you can verify its authenticity using its MD5 sum or signature. (How do I verify a download?)

File Description Downloads
download icon gcc-linaro-4.7-2013.01.tar.bz2 (md5, sig) Source release 22,264
last downloaded 2 weeks ago
Total downloads: 22,264

Release notes 

The Linaro Toolchain Working Group is pleased to announce the 2013.01
release of Linaro GCC 4.7.

Linaro GCC 4.7 2013.01 is the tenth release in the 4.7 series. Based
off the latest GCC 4.7.2+svn194772 release, it includes ARM-focused
performance improvements and bug fixes.

Interesting changes include:
 * Updates to GCC 4.7.2+svn194772
 * Includes arm/aarch64-4.7-branch up to svn revision 194808
 * Support for the rev16 and revsh instructions
 * A15 Neon pipeline backported from trunk
 * FMA intrinsic backported from trunk
 * Better extending core to NEON transfers
 * Fused multiply-add support

Fixes:
 * LP #1088898 regression of x86 gcc bootstrap with Linaro sourcebase
 * LP #1067766 Backport support for arm-linux-gnueabihf to GCC Linaro
 * LP #1084010 __atomic_load doesn't match ACQUIRE memory model

The source tarballs are available from:
 https://launchpad.net/gcc-linaro/+milestone/4.7-2013.01

Downloads are available from the Linaro GCC page on Launchpad:
 https://launchpad.net/gcc-linaro

More information on the features and issues are available from the
release pages:
 https://launchpad.net/gcc-linaro/4.7/4.7-2013.01

Mailing list: http://lists.linaro.org/mailman/listinfo/linaro-toolchain

Bugs: https://bugs.launchpad.net/gcc-linaro/

Questions? https://ask.linaro.org/

Interested in commercial support? Inquire at support@linaro.org

Changelog 

View the full changelog

2013-01-10 Matthew Gretton-Dann <email address hidden>

 Merge from FSF GCC 4.7.3 (svn branches/gcc-4_7-branch 194772).

2013-01-10 Matthew Gretton-Dann <email address hidden>

 Merge from FSF arm/aarch64-4.7-branch r194220..r194808.

 Backport arm-aarch64-4.7 r194220:
 gcc/

 2012-12-05 Yufeng Zhang <email address hidden>

 * config/aarch64/aarch64.c (aarch64_mangle_type): New function.
 (TARGET_MANGLE_TYPE): Define.

 gcc/testsuite/

 2012-12-05 Yufeng Zhang <email address hidden>

 * g++.dg/abi/arm_va_list.C: Also test on aarch64*-*-*.

 Backport arm-aarch64-4.7 r194222:
 [AARCH64-4.7] Backport vectorize standard math patterns.

 gcc/

 * config/aarch64/aarch64-builtins.c
 (aarch64_builtin_vectorized_function): New.
 * config/aarch64/aarch64-protos.h
 (aarch64_builtin_vectorized_function): Declare.
 * config/aarch64/aarch64-simd-builtins.def (frintz, frintp): Add.
 (frintm, frinti, frintx, frinta, fcvtzs, fcvtzu): Likewise.
 (fcvtas, fcvtau, fcvtps, fcvtpu, fcvtms, fcvtmu): Likewise.
 * config/aarch64/aarch64-simd.md
 (aarch64_frint_<frint_suffix><mode>): New.
 (<frint_pattern><mode>2): Likewise.
 (aarch64_fcvt<frint_suffix><su><mode>): Likewise.
 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Likewise.
 * config/aarch64/aarch64.c (TARGET_VECTORIZE_BUILTINS): Define.
 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
 * config/aarch64/aarch64.md
 (btrunc<mode>2, ceil<mode>2, floor<mode>2)
 (round<mode>2, rint<mode>2, nearbyint<mode>2): Consolidate as...
 (<frint_pattern><mode>2): ...this.
 (lceil<su_optab><mode><mode>2, lfloor<su_optab><mode><mode>2)
 (lround<su_optab><mode><mode>2)
 (lrint<su_optab><mode><mode>2): Consolidate as...
 (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): ... this.
 * config/aarch64/iterators.md (fcvt_target): New.
 (FCVT_TARGET): Likewise.
 (FRINT): Likewise.
 (FCVT): Likewise.
 (frint_pattern): Likewise.
 (frint_suffix): Likewise.
 (fcvt_pattern): Likewise.

 gcc/testsuite/

 * gcc.dg/vect/vect-rounding-btrunc.c: New test.
 * gcc.dg/vect/vect-rounding-btruncf.c: Likewise.
 * gcc.dg/vect/vect-rounding-ceil.c: Likewise.
 * gcc.dg/vect/vect-rounding-ceilf.c: Likewise.
 * gcc.dg/vect/vect-rounding-floor.c: Likewise.
 * gcc.dg/vect/vect-rounding-floorf.c: Likewise.
 * gcc.dg/vect/vect-rounding-lceil.c: Likewise.
 * gcc.dg/vect/vect-rounding-lfloor.c: Likewise.
 * gcc.dg/vect/vect-rounding-nearbyint.c: Likewise.
 * gcc.dg/vect/vect-rounding-nearbyintf.c: Likewise.
 * gcc.dg/vect/vect-rounding-round.c: Likewise.
 * gcc.dg/vect/vect-rounding-roundf.c: Likewise.
 * target-supports.exp
 (check_effective_target_vect_call_btrunc): New.
 (check_effective_target_vect_call_btruncf): Likewise.
 (check_effective_target_vect_call_ceil): Likewise.
 (check_effective_target_vect_call_ceilf): Likewise.
 (check_effective_target_vect_call_floor): Likewise.
 (check_effective_target_vect_call_floorf): Likewise.
 (check_effective_target_vect_call_lceil): Likewise.
 (check_effective_target_vect_call_lfloor): Likewise.
 (check_effective_target_vect_call_nearbyint): Likewise.
 (check_effective_target_vect_call_nearbyintf): Likewise.
 (check_effective_target_vect_call_round): Likewise.
 (check_effective_target_vect_call_roundf): Likewise.

 Backport arm-aarch64-4.7 r194246:
 gcc/

 2012-12-05 Yufeng Zhang <email address hidden>

 * config/aarch64/aarch64.c (aarch64_simd_mangle_map_entry): New
 typedef.
 (aarch64_simd_mangle_map): New table.
 (aarch64_mangle_type): Locate and return the mangled name for
 a given AdvSIMD vector type.

 gcc/testsuite/

 2012-12-05 Yufeng Zhang <email address hidden>

 * g++.dg/abi/mangle-neon-aarch64.C: New test.

 Backport arm-aarch64-4.7 r194259:
 [AARCH64-4.7] Backport fix to slp-perm-8.c

 gcc/testsuite/

 Backport from mainline.
 2012-05-31 Greta Yorsh <email address hidden>

 * lib/target-supports.exp (check_effective_target_vect_char_mult): Add
 arm32 to targets.
 * gcc.dg/vect/slp-perm-8.c (main): Prevent vectorization
 of the initialization loop.
 (dg-final): Adjust the expected number of vectorized loops depending
 on vect_char_mult target selector.

 Backport arm-aarch64-4.7 r194260:
 [AARCH64] Implement Vector Permute Support.

 gcc/
 Backport from mainline.
 2012-12-05 James Greenhalgh <email address hidden>

 * config/aarch64/aarch64-protos.h
 (aarch64_split_combinev16qi): New.
 (aarch64_expand_vec_perm): Likewise.
 (aarch64_expand_vec_perm_const): Likewise.
 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): New.
 (vec_perm<mode>): Likewise.
 (aarch64_tbl1<mode>): Likewise.
 (aarch64_tbl2v16qi): Likewise.
 (aarch64_combinev16qi): New.
 * config/aarch64/aarch64.c
 (aarch64_vectorize_vec_perm_const_ok): New.
 (aarch64_split_combinev16qi): Likewise.
 (MAX_VECT_LEN): Define.
 (expand_vec_perm_d): New.
 (aarch64_expand_vec_perm_1): Likewise.
 (aarch64_expand_vec_perm): Likewise.
 (aarch64_evpc_tbl): Likewise.
 (aarch64_expand_vec_perm_const_1): Likewise.
 (aarch64_expand_vec_perm_const): Likewise.
 (aarch64_vectorize_vec_perm_const_ok): Likewise.
 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Likewise.
 * config/aarch64/iterators.md
 (unspec): Add UNSPEC_TBL, UNSPEC_CONCAT.
 (V_cmp_result): Add mapping for V2DF.

 gcc/testsuite/

 Backport from mainline.
 2012-12-05 James Greenhalgh <email address hidden>

 * lib/target-supports.exp
 (check_effective_target_vect_perm): Allow aarch64*-*-*.
 (check_effective_target_vect_perm_byte): Likewise.
 (check_effective_target_vect_perm_short): Likewise.
 (check_effective_target_vect_char_mult): Likewise.
 (check_effective_target_vect_extract_even_odd): Likewise.
 (check_effective_target_vect_interleave): Likewise.

 Backport arm-aarch64-4.7 r194261:
 [AARCH64-4.7] Add zip{1, 2}, uzp{1, 2}, trn{1, 2} support for vector permute.

 gcc/
 Backport from mainline.
 2012-12-05 James Greenhalgh <email address hidden>

 * config/aarch64/aarch64-simd-builtins.def: Add new builtins.
 * config/aarch64/aarch64-simd.md (simd_type): Add uzp.
 (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): New.
 * config/aarch64/aarch64.c (aarch64_evpc_trn): New.
 (aarch64_evpc_uzp): Likewise.
 (aarch64_evpc_zip): Likewise.
 (aarch64_expand_vec_perm_const_1): Check for trn, zip, uzp patterns.
 * config/aarch64/iterators.md (unspec): Add neccessary unspecs.
 (PERMUTE): New.
 (perm_insn): Likewise.
 (perm_hilo): Likewise.

 Backport arm-aarch64-4.7 r194553:
 [AARCH64] Backport support for TARGET_VECTORIZE_AUTOVECTORIZE_VECTOR_SIZES.

 gcc/

 * config/aarch64/aarch64.c
 (aarch64_autovectorize_vector_sizes): New.
 (TARGET_VECTORIZE_AUTOVECTORIZE_VECTOR_SIZES): Define.

 gcc/testsuite/

 * lib/target-supports.exp
 (check_effective_target_vect_multiple_sizes): Enable for AArch64.

 Backport arm-aarch64-4.7 r194673:
 Make zero_extends explicit for common AArch64 SI mode patterns

 Backport arm-aarch64-4.7 r194808:
 [AArch64] Backport: Fix some warnings about unused variables.

 gcc/
 * config/aarch64/aarch64.c (aarch64_simd_attr_length_move):
 Remove unused variables.
 (aarch64_split_compare_and_swap): Likewise.

2012-12-20 Brice Dobry <email address hidden>

 Backport from mainline r191181

 2012-09-11 Tobias Burnus <email address hidden>

 * doc/sourcebuild.texi (arm_neon_v2_ok): Fix @anchor.

2012-12-20 Brice Dobry <email address hidden>

 Blueprints: backport-the-fma-intrinsic, fused-multiply-add-support

 Backport from mainline r192560

 2012-10-18 Matthew Gretton-Dann <email address hidden>
             Ramana Radhakrishnan <email address hidden>

 * config/arm/arm.c (neon_builtin_data): Add vfma and vfms
 builtins.
 * config/arm/neon-docgen.ml (intrinsic_groups): Add
 fused-multiply-* groups.
 * config/neon-gen.ml (print_feature_test_start): New function.
 (print_feature_test_end): Likewise.
 (print_variant): Print feature test macros.
 * config/arm/neon-testgen.ml (emit_prologue): Allow different
 tests to require different effective targets.
 (effective_target): New function.
 (test_intrinsic): Specify correct effective targets.
 * gcc/config/arm/neon.md (fma<VCVTF:mode>4_intrinsic): New pattern.
 (fmsub<VCVTF:mode>4_intrinsic): Likewise.
 (neon_vfma<VCVFT:mode>): New expand.
 (neon_vfms<VCVFT:mode>): Likewise.
 * config/neon.ml (opcode): Add Vfma and Vfms.
 (features): Add Requires_feature.
 (ops): Add VFMA and VFMS intrinsics.
 * config/arm/arm_neon.h: Regenerate.
 * doc/arm-neon-intrinsics.texi: Likewise.

2012-12-20 Brice Dobry <email address hidden>

 Blueprints: backport-the-fma-intrinsic, fused-multiply-add-support

 Backport from mainline r191180

 2012-09-11 Ramana Radhakrishnan <email address hidden>
 Matthew Gretton-Dann <email address hidden>

 * config/arm/neon.md (fma<VCVTF:mode>4): New pattern.
 (*fmsub<VCVTF:mode>4): Likewise.
 * doc/sourcebuild.texi (arm_neon_v2_ok, arm_neon_v2_hw): Document it.

2012-12-20 Brice Dobry <email address hidden>

 Blueprints: backport-the-fma-intrinsic, fused-multiply-add-support

 Backport from mainline r189283

 2012-07-05 Matthew Gretton-Dann <email address hidden>

 * config/arm/iterators.md (SDF): New mode iterator.
 (V_if_elem): Add support for SF and DF modes.
 (V_reg): Likewise.
 (F_constraint): New mode iterator attribute.
 (F_fma_type): Likewise.
 config/arm/vfp.md (fma<SDF:mode>4): New pattern.
 (*fmsub<SDF:mode>4): Likewise.
 (*fmnsub<SDF:mode>4): Likewise.
 (*fmnadd<SDF:mode>4): Likewise.

2012-12-20 Brice Dobry <email address hidden>

 Blueprints: backport-the-fma-intrinsic, fused-multiply-add-support

 Partial backport from mainline r188946

 2012-06-25 Matthew Gretton-Dann <email address hidden>
             James Greenhalgh <email address hidden>
 * config/arm/arm.h (TARGET_CPU_CPP_BUILTINS): Add new built-ins.
 (TARGET_FMA): New macro.

2012-12-20 Ulrich Weigand <email address hidden>

 Backport from mainline

 gcc/
 2012-12-17 Andrew Stubbs <email address hidden>
      Ulrich Weigand <email address hidden>

 * config/arm/arm.md (zero_extend<mode>di2): Add extra alternatives
 for NEON registers.
 Add alternative for one-instruction extend-in-place.
 (extend<mode>di2): Likewise.
 Add constraints for Thumb-mode memory loads.
 Prevent extend splitters doing NEON alternatives.
 * config/arm/iterators.md (qhs_extenddi_cstr, qhs_zextenddi_cstr):
 Adjust constraints to add new alternatives.
 * config/arm/neon.md: Add splitters for zero- and sign-extend.

 gcc/testsuite/
 2012-12-17 Andrew Stubbs <email address hidden>
      Ulrich Weigand <email address hidden>

 * gcc.target/arm/neon-extend-1.c: New file.
 * gcc.target/arm/neon-extend-2.c: New file.

 gcc/testsuite/
 2012-10-01 Ulrich Weigand <email address hidden>

 * gcc.dg/lower-subreg-1.c: Disable on arm*-*-* targets.

 gcc/
 2012-09-27 Ulrich Weigand <email address hidden>

 * lower-subreg.c (enum classify_move_insn): Rename
 SIMPLE_PSEUDO_REG_MOVE to DECOMPOSABLE_SIMPLE_MOVE.
 (find_decomposable_subregs): Update.
 (decompose_multiword_subregs): Add DECOMPOSE_COPIES parameter.
 Only mark pseudo-to-pseudo copies as DECOMPOSABLE_SIMPLE_MOVE
 if that parameter is true.
 (rest_of_handle_lower_subreg): Call decompose_multiword_subregs
 with DECOMPOSE_COPIES false.
 (rest_of_handle_lower_subreg2): Call decompose_multiword_subregs
 with DECOMPOSE_COPIES true.

 gcc/testsuite/
 2012-09-27 Ulrich Weigand <email address hidden>

 * gcc.dg/lower-subreg-1.c: Disable on arm-*-* targets.

2012-12-19 Christophe Lyon <email address hidden>

 gcc/testsuite/
 * gcc.target/arm/builtin-bswap16-1.c: Replace armv6 by armv7a to
 avoid failure when testing on hard-float+thumb target.
 * gcc.target/arm/builtin-bswap-1.c: Likewise.

 Backport from mainline r191760:

 2012-09-21 Christophe Lyon <email address hidden>

 gcc/
 * tree-ssa-math-opts.c (bswap_stats): Add found_16bit field.
 (execute_optimize_bswap): Add support for builtin_bswap16.

 gcc/testsuite/
 * gcc.target/arm/builtin-bswap16-1.c: New testcase.

 Backport from mainline r188526:

 2012-06-13 Alexandre Oliva <email address hidden>

 gcc/
 * common.opt (ftree-coalesce-inlined-vars): New.
 (ftree-coalesce-vars): New.
 * doc/invoke.texi: Document them.
 * tree-ssa-copyrename.c (copy_rename_partition_coalesce):
 Implement them.

 gcc/testsuite/
 * g++.dg/tree-ssa/ivopts-2.C: Adjust for coalescing.
 * gcc.dg/tree-ssa/forwprop-11.c: Likewise.
 * gcc.dg/tree-ssa/ssa-fre-1.c: Likewise.

 Backport from mainline r191243:

 2012-09-13 Christophe Lyon <email address hidden>
     Richard Earnshaw <email address hidden>

 gcc/
 * config/arm/arm.md (arm_rev): Factorize thumb1, thumb2 and arm
 variants for rev instruction..
 (thumb1_rev): Delete pattern.
 (arm_revsh): New pattern to support builtin_bswap16.
 (arm_rev16, bswaphi2): Likewise.

 gcc/testsuite/
 * gcc.target/arm/builtin-bswap-1.c: New testcase.

 Backport from mainline r186308:

 PR target/52624
 gcc/
 * doc/extend.texi (Other Builtins): Document __builtin_bswap16.
 (PowerPC AltiVec/VSX Built-in Functions): Remove it.
 * doc/md.texi (Standard Names): Add bswap.
 * builtin-types.def (BT_UINT16): New primitive type.
 (BT_FN_UINT16_UINT16): New function type.
 * builtins.def (BUILT_IN_BSWAP16): New.
 * builtins.c (expand_builtin_bswap): Add TARGET_MODE argument.
 (expand_builtin) <BUILT_IN_BSWAP16>: New case. Pass TARGET_MODE to
 expand_builtin_bswap.
 (fold_builtin_bswap): Add BUILT_IN_BSWAP16 case.
 (fold_builtin_1): Likewise.
 (is_inexpensive_builtin): Likewise.
 * optabs.c (expand_unop): Deal with bswap in HImode specially. Add
 missing bits for bswap to libcall code.
 * tree.c (build_common_tree_nodes): Build uint16_type_node.
 * tree.h (enum tree_index): Add TI_UINT16_TYPE.
 (uint16_type_node): New define.
 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_BSWAP_HI): Delete.
 * config/rs6000/rs6000.c (rs6000_expand_builtin): Remove handling of
 above builtin.
 (rs6000_init_builtins): Likewise.
 * config/rs6000/rs6000.md (bswaphi2): Add TARGET_POWERPC predicate.

 gcc/c-family/
 * c-common.h (uint16_type_node): Rename into...
 (c_uint16_type_node): ...this.
 * c-common.c (c_common_nodes_and_builtins): Adjust for above renaming.
 * c-cppbuiltin.c (builtin_define_stdint_macros): Likewise.

 gcc/testsuite/
 * gcc.dg/builtin-bswap-1.c: Test __builtin_bswap16 & __builtin_bswap64.
 * gcc.dg/builtin-bswap-4.c: Test __builtin_bswap16.
 * gcc.dg/builtin-bswap-5.c: Likewise.
 * gcc.target/i386/builtin-bswap-4.c: New test.

2012-12-17 Ulrich Weigand <email address hidden>

 LP 1088898

 Backport from mainline

 gcc/
 2012-09-24 Richard Guenther <email address hidden>

 PR tree-optimization/54684
 * tree-ssa-ccp.c (optimize_unreachable): Properly update stmts.

 gcc/testsuite/
 2012-09-24 Richard Guenther <email address hidden>

 PR tree-optimization/54684
 * g++.dg/torture/pr54684.C: New testcase.

2012-12-14 Michael Hope <email address hidden>

 Backport from mainline r192569:

 gcc/
 2012-10-18 Matthew Gretton-Dann <email address hidden>
     Ramana Radhakrishnan <email address hidden>
     Sameera Deshpande <email address hidden>

 * config/arm/cortex-a15-neon.md: New file.
 * config/arm/cortex-a15.md (cortex_a15_call): Adjust reservation.
 (cortex_a15_load1): Likewise.
 (cortex_a15_load3): Likewise.
 (cortex_a15_store1): Likewise.
 (cortex_a15_store3): Likewise.
 (cortex-a15-neon.md): Include.

2012-12-14 Michael Hope <email address hidden>

 Backport from mainline r193724:

 gcc/
 2012-11-20 Kyrylo Tkachov <email address hidden>

 * config/arm/arm.md (*arm_abssi2): Define predicable attribute.
 (*arm_neg_abssi2): Likewise.
 * config/arm/thumb2.md (*thumb2_abssi2): Likewise.
 (*thumb2_neg_abssi2): Likewise.

 Backport from mainline r194398:

 gcc/
 2012-12-11 Kyrylo Tkachov <email address hidden>

 PR target/55642
 * config/arm/thumb2.md (*thumb2_abssi2):
 Set ce_count attribute to 2.
 (*thumb2_neg_abssi2): Likewise.

 gcc/testsuite/
 2012-12-11 Kyrylo Tkachov <email address hidden>

 PR target/55642
 * gcc.target/arm/pr55642.c: New testcase.

7 blueprints and 3 bugs targeted

Blueprint Priority Assignee Delivery
Initial aarch64 backport Initial aarch64 backport 4 High Matthew Gretton-Dann  11 Implemented
aarch64 baremetal testing aarch64 baremetal testing 4 High Matthew Gretton-Dann  11 Implemented
Add support for the rev16 and revsh instructions. Add support for the rev16 and revsh instructions. 3 Medium Christophe Lyon  11 Implemented
Backport the A15 NEON pipeline Backport the A15 NEON pipeline 3 Medium Zhenqiang Chen  11 Implemented
Backport the FMA intrinsic Backport the FMA intrinsic 3 Medium Brice Dobry  11 Implemented
Better extending core to NEON transfers Better extending core to NEON transfers 3 Medium Ulrich Weigand  11 Implemented
Fused multiply-add support Fused multiply-add support 3 Medium Brice Dobry  11 Implemented
Bug report Importance Assignee Status
1088898 #1088898 regression of x86 gcc bootstrap with Linaro sourcebase 3 High Michael Hope  10 Fix Released
1067766 #1067766 Backport support for arm-linux-gnueabihf to GCC Linaro 4 Medium Matthew Gretton-Dann  10 Fix Released
1084010 #1084010 __atomic_load doesn't match ACQUIRE memory model 1 Undecided Yvan Roux  10 Fix Released
This milestone contains Public information
Everyone can see this information.