From: David Sterba <dsterba@suse.cz>
To: Nikolay Borisov <nborisov@suse.com>
Cc: David Sterba <dsterba@suse.com>, linux-btrfs@vger.kernel.org
Subject: Re: [PATCH 2/5] btrfs: set blocking_writers directly, no increment or decrement
Date: Fri, 18 Oct 2019 19:31:03 +0200 [thread overview]
Message-ID: <20191018173103.GE3001@twin.jikos.cz> (raw)
In-Reply-To: <75bf572e-34ce-5f99-21b0-18359ce5581c@suse.com>
On Fri, Oct 18, 2019 at 03:08:43PM +0300, Nikolay Borisov wrote:
>
>
> On 17.10.19 г. 22:38 ч., David Sterba wrote:
> > The increment and decrement was inherited from previous version that
> > used atomics, switched in commit 06297d8cefca ("btrfs: switch
> > extent_buffer blocking_writers from atomic to int"). The only possible
> > values are 0 and 1 so we can set them directly.
> >
> > The generated assembly (gcc 9.x) did the direct value assignment in
> > btrfs_set_lock_blocking_write:
> >
> > 5d: test %eax,%eax
> > 5f: je 62 <btrfs_set_lock_blocking_write+0x22>
> > 61: retq
> >
> > - 62: lock incl 0x44(%rdi)
> > - 66: add $0x50,%rdi
> > - 6a: jmpq 6f <btrfs_set_lock_blocking_write+0x2f>
> >
> > + 62: movl $0x1,0x44(%rdi)
> > + 69: add $0x50,%rdi
> > + 6d: jmpq 72 <btrfs_set_lock_blocking_write+0x32>
> >
> > The part in btrfs_tree_unlock did a decrement because
> > BUG_ON(blockers > 1) is probably not a strong hint for the compiler, but
> > otherwise the output looks safe:
> >
> > - lock decl 0x44(%rdi)
> >
> > + sub $0x1,%eax
> > + mov %eax,0x44(%rdi)
>
> I find it strange that plain inc/decs had the lock prefix in the
> resulting assembly. The sub instruction can have the lock prefix but
> here the compiler chooses not to. While this doesn't mean it's buggy I'm
> just curious what's happening.
The asm snippet is from the patch that did atomics -> int, which was
atomic_dec(blocking_writers) -> blocking_writers--, yet the result was
plain store.
It's probably not too clear from the context that it referes to the
menined commit 06297d8cefca.
next prev parent reply other threads:[~2019-10-18 17:30 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-17 19:38 [PATCH 0/5] Extent buffer locking and documentation David Sterba
2019-10-17 19:38 ` [PATCH 1/5] btrfs: merge blocking_writers branches in btrfs_tree_read_lock David Sterba
2019-10-17 19:38 ` [PATCH 2/5] btrfs: set blocking_writers directly, no increment or decrement David Sterba
2019-10-18 12:08 ` Nikolay Borisov
2019-10-18 17:31 ` David Sterba [this message]
2019-10-17 19:38 ` [PATCH 3/5] btrfs: access eb::blocking_writers according to ACCESS_ONCE policies David Sterba
2019-10-23 8:42 ` Nikolay Borisov
2019-10-29 21:33 ` David Sterba
2019-10-17 19:39 ` [PATCH 4/5] btrfs: serialize blocking_writers updates David Sterba
2019-10-23 9:57 ` Nikolay Borisov
2019-10-29 17:51 ` David Sterba
2019-10-29 18:48 ` Nikolay Borisov
2019-10-29 21:15 ` David Sterba
2019-10-17 19:39 ` [PATCH 5/5] btrfs: document extent buffer locking David Sterba
2019-10-18 0:17 ` Qu Wenruo
2019-10-18 11:56 ` David Sterba
2019-10-22 9:53 ` Nikolay Borisov
2019-10-22 10:41 ` David Sterba
2019-10-23 11:16 ` Nikolay Borisov
2019-10-29 17:33 ` David Sterba
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=20191018173103.GE3001@twin.jikos.cz \
--to=dsterba@suse.cz \
--cc=dsterba@suse.com \
--cc=linux-btrfs@vger.kernel.org \
--cc=nborisov@suse.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).