On Mon, May 21, 2012 at 6:10 PM, Jim Meyering <j...@meyering.net> wrote:
> Blue Swirl wrote:
>> On Mon, May 21, 2012 at 10:03 AM, Jim Meyering <j...@meyering.net> wrote:
>>> From: Jim Meyering <meyer...@redhat.com>
>>>
>>> I noticed this commit,
>>>
>>> virtio-pci: add missing 'static'
>>>
>>> which made this change:
>>>
>>> > -const MemoryRegionPortio virtio_portio[] = {
>>> > +static const MemoryRegionPortio virtio_portio[] = {
>>>
>>> and wondered if there were other variables like that.
>>> The following command shows that there are:
>>> [note that there are probably more: this finds only those
>>> for which the variable name appears in only one source file. ]
>>
>> Also, only for files at the top level.
>
> Thanks. There were so many .o files, I assumed that all were at the top.
> Searching all .o files, I found many more:
>
> $ for i in $(nm -e $(find . -name '*.o')|sed -n 's/.* [BCDGRS] //p'|sort
> -u);\
> do test $(git grep -lw $i|wc -l) = 1 && echo $i;done
How about just checking the executables (*-softmmu/qemu-system-xyz and
linux-user/qemu-xyz)?
> BlockDeviceIoStatus_lookup
> DEFAULT_ATR
> SpiceQueryMouseMode_lookup
> __jit_debug_descriptor
> alpha_num_opcodes
> alpha_num_operands
> alpha_opcodes
> alpha_operands
> arg_table
> backend_enum_table
> bdrv_sheepdog
> bfd_mips_num_builtin_opcodes
> bfd_mips_num_opcodes
> bonito_state
> check_cpuid
> cris_cond15s
> cris_cores
> device_configs
> display_remote
> enforce_cpuid
> floatformat_arm_ext_big
> floatformat_arm_ext_littlebyte_bigword
> floatformat_i387_ext
> floatformat_i960_ext
> floatformat_ia64_quad_big
> floatformat_ia64_quad_little
> floatformat_ia64_spill_big
> floatformat_ia64_spill_little
> floatformat_ieee_double_big
> floatformat_ieee_double_little
> floatformat_ieee_double_littlebyte_bigword
> floatformat_ieee_single_big
> floatformat_ieee_single_little
> floatformat_m68881_ext
> floatformat_m88110_ext
> floatformat_m88110_harris_ext
> fsl_register_prefix
> fw_boot_order
> last_mapping_addr
> last_mapping_sym
> last_type
> leon3_generic_machine
> m68k_numaliases
> m68k_numopcodes
> m68k_opcode_aliases
> m68k_opcodes
> memory_region_transaction_depth
> mips_builtin_opcodes
> mips_fulong2e_machine
> mips_opcodes
> no_reboot
> num_powerpc_operands
> para_features
> powerpc_macros
> powerpc_num_macros
> powerpc_num_opcodes
> powerpc_operands
> pvr_register_prefix
> qemu_boot_opts
> qemu_global_mutex
> qemu_option_rom_opts
> register_prefix
> rtas_next
> s390_virtio_bus_info
> sh_table
> special_register_prefix
> test_image
> timers_state
> v9fs_synth_root
> virtcon_hds
> vmstate_bmdma_status
> vmstate_info_scsi_requests
> vmstate_rxtx_stats
> xen_platform_ioport
> xen_xcg
>
>> What about functions? ;-)
>
> I planned to check them separately, as I do in gnulib's
> sc_tight_scope syntax-check rule:
>
> http://git.sv.gnu.org/cgit/gnulib.git/tree/top/maint.mk#n1443
Interesting. We could do similar source level checks for patches in
checkpatch.pl (which should be a commit hook if it never generated any
false positives) and add the object file tests for example to the
'make check' test setup.
