Re: [PATCH] powerpc/pseries: explicitly reschedule during drmem_lmb list traversal

[Nathan Lynch <nathanl@linux.ibm.com>]

Nathan Lynch <nathanl@linux.ibm.com> writes:
> Michael Ellerman <mpe@ellerman.id.au> writes:
>> One thought, which I possibly should not put in writing, is that we
>> could use the alignment of the pointer as a poor man's substitute for a
>> counter, eg:
>>
>> +static inline struct drmem_lmb *drmem_lmb_next(struct drmem_lmb *lmb)
>> +{
>> +	if (lmb % PAGE_SIZE == 0)
>> +		cond_resched();
>> +
>> +	return ++lmb;
>> +}
>>
>> I think the lmbs are allocated in a block, so I think that will work.
>> Maybe PAGE_SIZE is not the right size to use, but you get the idea.
>>
>> Gross I know, but might be OK as short term solution?
>
> OK, looking into this.

To follow up:

I wasn't able to measure more than ~1% difference in DLPAR memory
performance with my original version of this, but that was on a
relatively small configuration - hundreds of elements in the array as
opposed to thousands. I took an educated guess at an appropriate
interval and posted v2:

https://lore.kernel.org/linuxppc-dev/20200812012005.1919255-1-nathanl@linux.ibm.com/