All of lore.kernel.org
 help / color / mirror / Atom feed
From: Christoph Hellwig <hch@infradead.org>
To: Mikulas Patocka <mpatocka@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>,
	Zdenek Kabelac <zkabelac@redhat.com>,
	linux-block@vger.kernel.org, dm-devel@redhat.com
Subject: Re: [PATCH 2/4] brd: extend the rcu regions to cover read and write
Date: Tue, 20 Sep 2022 00:38:08 -0700	[thread overview]
Message-ID: <Yylt4A7B6dsn7+bu@infradead.org> (raw)
In-Reply-To: <alpine.LRH.2.02.2209160459250.543@file01.intranet.prod.int.rdu2.redhat.com>

>   * Look up and return a brd's page for a given sector.
> + * This must be called with the rcu lock held.

Please ad a rcu_read_lock_held() check then.

> -	rcu_read_lock();
>  	idx = sector >> PAGE_SECTORS_SHIFT; /* sector to page index */
>  	page = radix_tree_lookup(&brd->brd_pages, idx);
> -	rcu_read_unlock();
> -
> -	BUG_ON(page && page->index != idx);
>  
>  	return page;

No need for the page variable now.  In fact there is no real need
for this helper now, as all the callers really should operate on
the sector on the index anyway.

>  }
> @@ -88,7 +74,9 @@ static bool brd_insert_page(struct brd_d
>  	struct page *page;
>  	gfp_t gfp_flags;
>  
> +	rcu_read_lock();
>  	page = brd_lookup_page(brd, sector);
> +	rcu_read_unlock();
>  	if (page)
>  		return true;

So this looks odd, as we drop the rcu lock without doing anything,
but it actually turns out to be correct as brd_do_bvec does yet
another lookup of it.  So we get an initial look, and optional
insert and then another lookup.  Not very efficient and it might be
worth to fix brd_do_bvec up to avoid these extra lookups given
that you touch it anyway (as would be an radix tree to xarray
conversion).


WARNING: multiple messages have this Message-ID (diff)
From: Christoph Hellwig <hch@infradead.org>
To: Mikulas Patocka <mpatocka@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>,
	linux-block@vger.kernel.org, dm-devel@redhat.com,
	Zdenek Kabelac <zkabelac@redhat.com>
Subject: Re: [dm-devel] [PATCH 2/4] brd: extend the rcu regions to cover read and write
Date: Tue, 20 Sep 2022 00:38:08 -0700	[thread overview]
Message-ID: <Yylt4A7B6dsn7+bu@infradead.org> (raw)
In-Reply-To: <alpine.LRH.2.02.2209160459250.543@file01.intranet.prod.int.rdu2.redhat.com>

>   * Look up and return a brd's page for a given sector.
> + * This must be called with the rcu lock held.

Please ad a rcu_read_lock_held() check then.

> -	rcu_read_lock();
>  	idx = sector >> PAGE_SECTORS_SHIFT; /* sector to page index */
>  	page = radix_tree_lookup(&brd->brd_pages, idx);
> -	rcu_read_unlock();
> -
> -	BUG_ON(page && page->index != idx);
>  
>  	return page;

No need for the page variable now.  In fact there is no real need
for this helper now, as all the callers really should operate on
the sector on the index anyway.

>  }
> @@ -88,7 +74,9 @@ static bool brd_insert_page(struct brd_d
>  	struct page *page;
>  	gfp_t gfp_flags;
>  
> +	rcu_read_lock();
>  	page = brd_lookup_page(brd, sector);
> +	rcu_read_unlock();
>  	if (page)
>  		return true;

So this looks odd, as we drop the rcu lock without doing anything,
but it actually turns out to be correct as brd_do_bvec does yet
another lookup of it.  So we get an initial look, and optional
insert and then another lookup.  Not very efficient and it might be
worth to fix brd_do_bvec up to avoid these extra lookups given
that you touch it anyway (as would be an radix tree to xarray
conversion).

--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel


  reply	other threads:[~2022-09-20  7:38 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-09-16  8:58 [PATCH 0/4] brd: implement discard Mikulas Patocka
2022-09-16  8:58 ` [dm-devel] " Mikulas Patocka
2022-09-16  8:59 ` [PATCH 1/4] brd: make brd_insert_page return bool Mikulas Patocka
2022-09-16  8:59   ` [dm-devel] " Mikulas Patocka
2022-09-20  7:28   ` Christoph Hellwig
2022-09-20  7:28     ` [dm-devel] " Christoph Hellwig
2022-09-16  8:59 ` [PATCH 2/4] brd: extend the rcu regions to cover read and write Mikulas Patocka
2022-09-16  8:59   ` [dm-devel] " Mikulas Patocka
2022-09-20  7:38   ` Christoph Hellwig [this message]
2022-09-20  7:38     ` Christoph Hellwig
2022-09-16  9:00 ` [dm-devel] [PATCH 3/4] brd: enable discard Mikulas Patocka
2022-09-16  9:00   ` Mikulas Patocka
2022-09-20  7:39   ` [dm-devel] " Christoph Hellwig
2022-09-20  7:39     ` Christoph Hellwig
2022-09-20 17:47     ` [dm-devel] " Mikulas Patocka
2022-09-20 17:47       ` Mikulas Patocka
2022-09-16  9:00 ` [dm-devel] [PATCH 4/4] brd: implement secure erase and write zeroes Mikulas Patocka
2022-09-16  9:00   ` Mikulas Patocka
2022-09-20  7:29   ` Christoph Hellwig
2022-09-20  7:29     ` [dm-devel] " Christoph Hellwig
2022-09-20 17:46     ` Mikulas Patocka
2022-09-20 17:46       ` Mikulas Patocka

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=Yylt4A7B6dsn7+bu@infradead.org \
    --to=hch@infradead.org \
    --cc=axboe@kernel.dk \
    --cc=dm-devel@redhat.com \
    --cc=linux-block@vger.kernel.org \
    --cc=mpatocka@redhat.com \
    --cc=zkabelac@redhat.com \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.