Re: [PATCH 2/6] writeback: support retrieving per group debug writeback stats of bdi

[Kemeng Shi]

on 3/26/2024 8:24 PM, Jan Kara wrote:
> On Wed 20-03-24 19:02:18, Kemeng Shi wrote:
>> Add /sys/kernel/debug/bdi/xxx/wb_stats to show per group writeback stats
>> of bdi.
>>
>> Signed-off-by: Kemeng Shi <shikemeng@xxxxxxxxxxxxxxx>
> 
> ...
> 
>> +unsigned long wb_calc_cg_thresh(struct bdi_writeback *wb)
>> +{
>> +	struct dirty_throttle_control gdtc = { GDTC_INIT_NO_WB };
>> +	struct dirty_throttle_control mdtc = { MDTC_INIT(wb, &gdtc) };
>> +	unsigned long filepages, headroom, writeback;
>> +
>> +	gdtc.avail = global_dirtyable_memory();
>> +	gdtc.dirty = global_node_page_state(NR_FILE_DIRTY) +
>> +		     global_node_page_state(NR_WRITEBACK);
>> +
>> +	mem_cgroup_wb_stats(wb, &filepages, &headroom,
>> +			    &mdtc.dirty, &writeback);
>> +	mdtc.dirty += writeback;
>> +	mdtc_calc_avail(&mdtc, filepages, headroom);
>> +	domain_dirty_limits(&mdtc);
>> +
>> +	return __wb_calc_thresh(&mdtc, mdtc.thresh);
>> +}
> 
> I kind of wish we didn't replicate this logic from balance_dirty_pages()
> and wb_over_bg_thresh() into yet another place. But the refactoring would
> be kind of difficult. So OK.
Thanks for review the code.
I have considered about this. It's difficult and will introduce a lot
change to non-debug code which make this series uneasy to review.
I will submit another patch for refactoring if I could find a way to
clean the code.

Kemeng
> 
> 								Honza
>