linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Allan McAleavy <allan.mcaleavy@gmail.com>
To: linux-mm@kvack.org
Subject: Re: Page Cache Monitoring ( Hit/Miss )
Date: Fri, 11 Dec 2015 15:22:31 +0000	[thread overview]
Message-ID: <CAF6XsOesbGN=rH0g_2JXMeyovbVPyKYvPQ-jgoVbh8cZkhyFiA@mail.gmail.com> (raw)
In-Reply-To: <CAF6XsOeYWvuNm=uuMCM4YD4a2dCoBe6TvimygPKRe4PMiHwQmw@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 2077 bytes --]

We add a page to the page cache using add_to_page_cache_lru based on the
following assumptions

READ MISS - So add an entry
no_cached_page - so we need to create one from
do_generic_file_read - during a file read operation
page_cache_read - calls add_to_page_cache_lru

CREATE A PAGE
struct page *__page_cache_alloc(gfp_t gfp)

page has flags which identify if dirty or free

Writes
add_to_page_cache
writepage / writepages
set_page_dirty_buffers
do_generic_mapping_read  - ASYNC read of pages in readahed?
__block_write_full_page

READ
find_get_pages


mark_page_accessed() for measuring cache accesses
mark_buffer_dirty() for measuring cache writes
add_to_page_cache_lru() for measuring page additions
account_page_dirtied() for measuring page dirties


 (mark_page_accessed - mark_buffer_dirty) & misses = (add_to_page_cache_lru
- account_page_dirtied),
 from this I then work out the hit ratio etc. Is there any other key
functions I should be tracing?

add_to_page_cache_lru
lru_cache_add
swap.c
filemap.c
vmscan.c

Functions used
lru_cache_add_active_or_unevictable
add_to_page_cache_lru
putback_lru_page

So best use lru_cache_add for additions.

account_page_dirtied() for measuring page dirties
set_page_dirty - calls above
__set_page_dirty_nobuffers - calls above also

mark_buffer_dirty() for measuring cache writes - this calls
__set_page_dirty ( are we getting twice the calls here? )
mark_page_accessed - calls SetPageActive

On Thu, Dec 10, 2015 at 9:42 AM, Allan McAleavy <allan.mcaleavy@gmail.com>
wrote:

> Hi Folks,
>
> I am working on a rewrite of Brendan Greggs original cachestat (ftrace)
> script into bcc. What I was looking for was a steer in the right direction
> for what functions to trace. At present I trace the following.
>
> add_to_page_cache_lru
> account_page_dirtied
> mark_page_accessed
> mark_buffer_dirty
>
> Where total = (mark_page_accessed - mark_buffer_dirty) & misses =
> (add_to_page_cache_lru - account_page_dirtied), from this I then work out
> the hit ratio etc. Is there any other key functions I should be tracing?
>
> Thanks
>

[-- Attachment #2: Type: text/html, Size: 3778 bytes --]

      reply	other threads:[~2015-12-11 15:22 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-12-10  9:42 Allan McAleavy
2015-12-11 15:22 ` Allan McAleavy [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAF6XsOesbGN=rH0g_2JXMeyovbVPyKYvPQ-jgoVbh8cZkhyFiA@mail.gmail.com' \
    --to=allan.mcaleavy@gmail.com \
    --cc=linux-mm@kvack.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox