Re: [RFC] arm: add __initbss section attribute

From: yalin wang
Date: Tue Oct 13 2015 - 05:51:50 EST



> On Oct 13, 2015, at 17:40, Arnd Bergmann <arnd@xxxxxxxx> wrote:
>
> On Tuesday 13 October 2015 15:33:10 yalin wang wrote:
>>
>>> On Oct 13, 2015, at 04:04, Sam Ravnborg <sam@xxxxxxxxxxxx> wrote:
>>>
>>>> --- a/include/asm-generic/vmlinux.lds.h
>>>> +++ b/include/asm-generic/vmlinux.lds.h
>>>>
>>>> -#define BSS_SECTION(sbss_align, bss_align, stop_align) \
>>>> +#define BSS_SECTION(sbss_align, bss_align, initbss_align, stop_align) \
>>>
>>> A few comments:
>>>
>>> 1) - please align the backslash at the end of the
>>> line with the backslash above it.
>>> 2) - you need to fix all the remaining users of BSS_SECTION.
>>> 3) - do we really need the flexibility to specify an alignment (stop_align)?
>>> If not - drop the extra argument.
>>>
>>> Sam
>> i change lots of __initdata to __initbss to test it on ARM arch,
>>
>
> Ok, I found my mistake in the script I used to calculate the savings,
> here is the correct output showing all uninitialized variables in
> multi_v7_defconfig:
>
> 4 done.44688
> 1024 boot_command_line
> 1024 tmp_cmdline.44689
> 4 late_time_init
> 4 root_mount_data
> 4 root_fs_names
> 4 rd_doload
> 4 root_delay
> 64 saved_root_name
> 4 root_device_name
> 4 message
> 4 byte_count
> 4 victim
> 4 collected
> 8 this_header
> 4 state
> 4 collect
> 4 remains
> 4 next_state
> 8 header_buf
> 8 next_header
> 4 do_retain_initrd
> 4 name_len
> 4 body_len
> 4 gid
> 4 uid
> 4 mtime
> 4 wfd
> 4 vcollected
> 4 ino
> 4 mode
> 4 nlink
> 4 major
> 4 minor
> 4 rdev
> 4 symlink_buf
> 4 name_buf
> 64 msg_buf.29770
> 128 head
> 4 machine_desc
> 4 usermem.34390
> 4 __atags_pointer
> 1024 cmd_line
> 1024 default_command_line
> 1536 atags_copy
> 4 dma_mmu_remap_num
> 64 dma_mmu_remap
> 4 phys_initrd_start
> 4024 phys_initrd_size
> 4096 bm_pte
> 4 ecc_mask
> 4 initial_pmd_value
> 4 arm_lowmem_limit
> 24 s5p_mfc_mem
> 28 tx_pad_name
> 4 use_gptimer_clksrc
> 4 omap_table_init
> 4 mpurate
> 4 am35xx_aes_hwmod_ocp_ifs
> 4 am35xx_sham_hwmod_ocp_ifs
> 4 dra72x_hwmod_ocp_ifs
> 4 rx51_vibra_data
> 4 num_special_pds
> 128 special_pds
> 4 main_extable_sort_needed
> 4 new_log_buf_len
> 24 opts.36513
> 512 smap.22469
> 512 dmap.22470
> 64 group_map.22521
> 64 group_cnt.22522
> 4 pcpu_chosen_fc
> 4 vmlist
> 4 vm_init_off.26867
> 4 reset_managed_pages_done
> 116 boot_kmem_cache_node.32827
> 116 boot_kmem_cache.32826
> 4 dhash_entries
> 4 ihash_entries
> 4 mhash_entries
> 4 mphash_entries
> 256 nfs_root_parms
> 1028 nfs_export_path
> 1028 nfs_root_device
> 4 gic_cnt
> 4 threshold_index
> 4096 ata_force_param_buf
> 4 mtd_devs
> 2432 mtd_dev_param
> 8 m68k_probes
> 8 isa_probes
> 4 arch_timers_present
> 4 dt_root_size_cells
> 4 dt_root_addr_cells
> 4 imx_keep_uart_clocks
> 4 imx_uart_clocks
> 4 mt8173_top_clk_data
> 4 mt8173_pll_clk_data
> 4 cpg_mode
> 4 cpg_mode_rates
> 4 cpg_mode_divs
> 4 cpg_mode
> 4 cpg_mode
> 4 thash_entries
> 4 uhash_entries
> 4 ic_got_reply
> 4 ic_first_dev
> 4 ic_dev
> 16 user_dev_name
> 4 ic_dhcp_msgtype
> 4 ic_host_name_set
> 4 ic_dev_mtu
> 4 ic_set_manually
> 4 ic_enable
> 256 vendor_class_identifier
> 4 ic_proto_have_if
> 4 dma_reserve
> 4 nr_kernel_pages
> 4 nr_all_pages
> 196 __clk_of_table_sentinel
> 200 __rmem_of_table_sentinel
> 196 __clksrc_of_table_sentinel
> 200 __iommu_of_table_sentinel
> 8 __cpu_method_of_table_sentinel
> 8 __cpuidle_method_of_table_sentinel
> 196 irqchip_of_match_end
> 32 __earlycon_table_sentinel
> 200 __earlycon_of_table_sentinel
> 6 __irf_end
>
>
> 26398 total
i am curious about your scripts ,
could you show me ?
Thank you !




--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/