Allow the user to specify functions for allocating memory for
splay tree roots and nodes.
* splay-tree.h (splay_tree_allocate_fn, splay_tree_deallocate_fn):
New types.
(splay_tree): New fields: `allocate', `deallocate', and
`allocate_data'.
(splay_tree_new_with_allocator): New function declaration.
libiberty:
* splay-tree.c (splay_tree_xmalloc_allocate,
splay_tree_xmalloc_deallocate): New functions.
(splay_tree_new): Call splay_tree_new_with_allocator, passing the
above functions and a dummy data pointer.
(splay_tree_new_with_allocator): New function.
(splay_tree_delete_helper, splay_tree_delete, splay_tree_insert,
splay_tree_remove): Use the splay tree's allocation and
deallocation functions.
* ppc-linux-nat.c (PTRACE_GETVRREGS, PTRACE_SETVRREGS): Define.
(have_ptrace_getvrregs): Define for run time checks.
(gdb_vrregset_t): New type for Altivec register handling.
(fetch_register, store_register): Fetch/store altivec register
when needed.
(fetch_altivec_register, store_altivec_register): New functions.
(supply_vrregset, fill_vrregset): New functions.
(fetch_altivec_registers, store_altivec_registers): New functions.
(fetch_ppc_registers, store_ppc_registers): Fetch/store altivec
registers as well.
to the x86-64 backend. Get a closer match to elf32-i386.
(struct elf64_x86_64_dyn_relocs): Rename from
elf64_x86_64_pcrel_relocs_copied, add additional fields. Change
all users.
(struct elf64_x86_64_link_hash_table): Add short cuts to some
sections.
(link_hash_newfunc): Rename from elf64_x86_64_link_hash_newfunc,
remove casts, initialize new hash members.
(create_got_section): New.
(elf64_x86_64_create_dynamic_sections): New.
(elf64_x86_64_copy_indirect_symbol): New.
(elf64_x86_64_check_relocs): Don't allocate space for dynamic
relocs, .got or .relgot here but do it in allocate_dynrelocs.
Reference count possible .plt and .got entries. Don't test input
section SEC_READONLY here to try to avoid copy relocs, and keep
dyn_relocs regardless of ELF_LINK_NON_GOT_REF. Don't set
DF_TEXTREL here. Delay setting of variables until needed. Cache
pointer to "sreloc" section in elf_section_data. Tweak condition
under which .got created. Report files with bad relocation
section names.
(elf64_x86_64_gc_sweep_hook): Sweep dyn_relocs and local_dynrel.
Reference count possible .plt entries. Don't deallocate .got and
.relgot space here.
(elf64_x86_64_adjust_dynamic_symbol): Handle nocopyreloc. Don't
do copy reloc processing for weakdefs. Remove redundant casts and
aborts. Delay setting of vars until needed. Move creation of
dynamic symbols and allocation of .plt and .rela.plt to
allocate_dynrelocs. Replace BFD_ASSERT with abort.
(WILL_CALL_FINISH_DYNAMIC_SYMBOL): New.
(allocate_dynrelocs): New.
(readonly_dynrelocs): New.
(elf64_x86_64_size_dynamic_sections): Call readonly_dynrelocs.
Allocate space for dyn relocs. Replace BFD_ASSERT with abort.
Zero out the dynamic allocated content space.
(elf64_x86_64_discard_copies): Removed.
(elf64_x86_64_relocate_section): Make use of dynamic section
short-cuts. Localise vars, and delay setting. Better error
reporting, replace BFD_ASSERT with abort. Check
ELF_LINK_HASH_DEF_DYNAMIC to see if a symbol is not defined in the
regular object file and tread the weak definition as the normal
one. Don't discard relocs for undefweak or undefined symbols and
check !DEF_REGULAR as well as DEF_DYNAMIC in test for avoided copy
relocs.
(elf64_x86_64_finish_dynamic_symbol): Don't copy relocs for
symbols that have been forced local. Use same test to decide if
we can use a relative reloc for got as relocate_section. Expand
SHN_UNDEF comment. Move expressions out of function calls.
Replace BFD_ASSERT with abort.
(bfd_elf64_bfd_final_link): Removed.
(elf_backend_copy_indirect_symbol): Define.
* armnbsd-nat.c (supply_gregset): New function. Common code to
supply the integer register set.
(supply_fparegset): New function. Similar for FPA registers.
(fetch_regs, fetch_fp_regs): Use them.
(fetch_core_registers): Likewise.
(fetch_elfcore_registers): New function.
(arm_netbsd_elfcore_fns): New core-file type specification.
(_initialize_arm_netbsd_nat): Register it.
(FETCH_INFERIOR_REGISTERS): Just error if this isn't defined.
(fetch_regs, fetch_fp_regs, store_regs, store_fp_regs): Add explicit
'void' to declaration, to shut up ARI.
(fetch_core_registers): Make static. Rewrite using supply_register.
(arm_netbsd_core_fns): New core-file type specification.
(_initialize_arm_netbsd_nat): New function.
signed, unsigned and straight char.
(padding_char, padding_short, padding_int, padding_long,
padding_long_long, padding_float, padding_double,
padding_long_double): New global variables.
(fill, fill_structs): New functions.
* gdb.base/sizeof.exp: Check for signed and unsigned char. Check
for correctly sized writes. Update copyright.
(get_valueof): New procedure.
(get_sizeof): Call get_valueof.
(check_valueof): New procedure.
(check_padding): New procedure.
* Makefile.am: Add new files earmelf_fbsd, eelf32ppc_fbsd,
eelf_i386_fbsd, eelf64_ia64_fbsd, eelf_x86_64_fbsd, eelf64_sparc_fbsd,
and eelf64alpha_fbsd.
(left this out of last patch)
(struct gdbarch_tdep): Add fp_model.
* arm-tdep.c (arm_gdbarch_init): Set fp_model in tdep. Defer setting
up floating-point conversions until we know the floating-point model
in use by the inferior. Don't complain about being unable to
determine the ABI of the inferior when we don't have one.
(arm_extract_return_value): Support different floating-point models.
(arm_store_return_value): Likewise.
* armnbsd-tdep.c (arm_netbsd_aout_init_abi): Set fp_model in tdep to
ARM_FLOAT_SOFT.
(arm_netbsd_elf_init_abi): Set fp_model to ARM_FLOAT_SOFT_VFP.
* config/tc-s390.c (md_parse_option): Add switches -m31 and -m64.
Make bit size independent of architecture switch.
(md_begin): Add warning for -m64 with -Aesa.
(s390_md_end): Use renamed architecture defines.