MySQL Bugs: #104594: no need to maintain young and old area of LRU list

Bug #104594	no need to maintain young and old area of LRU list
Submitted:	11 Aug 2021 10:54	Modified:	12 Aug 2021 12:17
Reporter:	alex xing (OCA)	Email Updates:
Status:	Verified	Impact on me:	None
Category:	MySQL Server: InnoDB storage engine	Severity:	S5 (Performance)
Version:	8.0	OS:	Any
Assigned to:		CPU Architecture:	Any

Description:
When BUF_FLUSH_SINGLE_PAGE or BUF_FLUSH_SINGLE_PAGE is used, this page will be removed from the LRU.  If the page is accessed by the user thread during flush, there is no need to move the page to the Young area, which will cause the expensive cost of  LRU_list_mutex.

How to repeat:
just read the code 

Suggested fix:
in Buf_fetch<T>::single_page, perhaps we  can determine the buF_IO_fix and flush_type of the page before calling buf_page_make_YOUNG_if_needed.

a simple patch to describe the optimization (*) I confirm the code being submitted is offered under the terms of the OCA, and that I am authorized to contribute it.

Contribution: commit.patch (text/plain), 720 bytes.

the above is just a simple patch, Extra attention may be required for compressed pages

Hi Mr. xing,

Thank you very much for your performance improvement report.

We do have couple of questions regarding your report.

First of all, have you measured what impact on performance your patch has with multiple concurrent reads ???

Second, we took a look at your patch and it seems that it is not complete ....

Can you complete your request by specifying exactly what else should be done to make a patch complete.

Thanks in advance.

new patch

Attachment: commit_new .patch (text/plain), 696 bytes.

new patch (*) I confirm the code being submitted is offered under the terms of the OCA, and that I am authorized to contribute it.

Contribution: commit_new .patch (text/plain), 696 bytes.

Hi Mr. xing,

Thank you for your new patch.

We found that it is quite useful and we are thanking you for it.

Verified as reported.

new patch (*) I confirm the code being submitted is offered under the terms of the OCA, and that I am authorized to contribute it.

Contribution: optimize.patch (text/plain), 712 bytes.

Thank you, Mr. xing.

Your patch will be treated as it should be, because it is coming from an OCA contributor.