Re: [PATCH v4] mm/page_owner.c: add llseek for page_owner

[Kassey Li <quic_yingangl@quicinc.com>]

On 8/8/2022 11:33 PM, Vlastimil Babka (SUSE) wrote:
> On 8/8/22 09:36, Kassey Li wrote:
>> There is usage to dump a given cma region page_owner
>> instead of all page's.
>>
>> This change allows to specify a ppos as start_pfn
>> by fseek.
>>
>> Any invalid ppos will be skipped, so it did not
>> broken the origin dump feature.
>>
>> Suggested-by: Vlastimil Babka (SUSE) <vbabka@kernel.org>
> 
> Thanks, but please change it to Vlastimil Babka <vbabka@suse.cz>
> It was some mistake between me and thunderbird that it was sent from @k.o.
   Ok, will change it.
> 
>> Signed-off-by: Kassey Li <quic_yingangl@quicinc.com>
>> ---
>>   mm/page_owner.c | 20 ++++++++++++++++++--
>>   1 file changed, 18 insertions(+), 2 deletions(-)
>>
>> diff --git a/mm/page_owner.c b/mm/page_owner.c
>> index e4c6f3f1695b..231b1877af99 100644
>> --- a/mm/page_owner.c
>> +++ b/mm/page_owner.c
>> @@ -497,8 +497,8 @@ read_page_owner(struct file *file, char __user *buf, size_t count, loff_t *ppos)
>>   		return -EINVAL;
>>   
>>   	page = NULL;
>> -	pfn = min_low_pfn + *ppos;
>>   
>> +	pfn = *ppos;
> 
> Maybe we could still optimize the typical case a bit such that when it's 0,
> we skip to min_low_pfn immediately.
> 
>>   	/* Find a valid PFN or the start of a MAX_ORDER_NR_PAGES area */
>>   	while (!pfn_valid(pfn) && (pfn & (MAX_ORDER_NR_PAGES - 1)) != 0)
>>   		pfn++;
>> @@ -561,7 +561,7 @@ read_page_owner(struct file *file, char __user *buf, size_t count, loff_t *ppos)
>>   			continue;
>>   
>>   		/* Record the next PFN to read in the file offset */
>> -		*ppos = (pfn - min_low_pfn) + 1;
>> +		*ppos = pfn + 1;
>>   
>>   		return print_page_owner(buf, count, pfn, page,
>>   				page_owner, handle);
>> @@ -570,6 +570,21 @@ read_page_owner(struct file *file, char __user *buf, size_t count, loff_t *ppos)
>>   	return 0;
>>   }
>>   
>> +static loff_t llseek_page_owner(struct file *file, loff_t offset, int whence)
>> +{
>> +	loff_t retval = 0;
>> +	switch (whence) {
>> +		case SEEK_CUR:
> 
> It's not correct for SEEK_CUR to behave like this, no?
> Can we perhaps reuse the existing mem_lseek() ?  
   agree.
> 
> Also we should update Documentation/mm/page_owner.rst to document the lseek
> support and that it works with a pfn.
   will do this in v5 patch set.
> 
> Thanks!
> 
>> +		case SEEK_SET:
>> +			file->f_pos = offset;
>> +			break;
>> +		default:
>> +			retval = -ENXIO;
>> +	}
>> +
>> +	return retval;
>> +}
>> +
>>   static void init_pages_in_zone(pg_data_t *pgdat, struct zone *zone)
>>   {
>>   	unsigned long pfn = zone->zone_start_pfn;
>> @@ -660,6 +675,7 @@ static void init_early_allocated_pages(void)
>>   
>>   static const struct file_operations proc_page_owner_operations = {
>>   	.read		= read_page_owner,
>> +	.llseek 	= llseek_page_owner,
>>   };
>>   
>>   static int __init pageowner_init(void)
>