From: Will Deacon <will@kernel.org>
To: linux-kernel@vger.kernel.org
Cc: Will Deacon <will@kernel.org>, Joel Fernandes <joelaf@google.com>,
Sami Tolvanen <samitolvanen@google.com>,
Nick Desaulniers <ndesaulniers@google.com>,
Kees Cook <keescook@chromium.org>, Marco Elver <elver@google.com>,
"Paul E. McKenney" <paulmck@kernel.org>,
Matt Turner <mattst88@gmail.com>,
Ivan Kokshaysky <ink@jurassic.park.msu.ru>,
Richard Henderson <rth@twiddle.net>,
Peter Zijlstra <peterz@infradead.org>,
Alan Stern <stern@rowland.harvard.edu>,
"Michael S. Tsirkin" <mst@redhat.com>,
Jason Wang <jasowang@redhat.com>, Arnd Bergmann <arnd@arndb.de>,
Boqun Feng <boqun.feng@gmail.com>,
Catalin Marinas <catalin.marinas@arm.com>,
Mark Rutland <mark.rutland@arm.com>,
linux-arm-kernel@lists.infradead.org,
linux-alpha@vger.kernel.org,
virtualization@lists.linux-foundation.org,
kernel-team@android.com
Subject: [PATCH v3 07/19] vhost: Remove redundant use of read_barrier_depends() barrier
Date: Fri, 10 Jul 2020 17:51:51 +0100 [thread overview]
Message-ID: <20200710165203.31284-8-will@kernel.org> (raw)
In-Reply-To: <20200710165203.31284-1-will@kernel.org>
Since commit 76ebbe78f739 ("locking/barriers: Add implicit
smp_read_barrier_depends() to READ_ONCE()"), there is no need to use
smp_read_barrier_depends() outside of the Alpha architecture code.
Unfortunately, there is precisely _one_ user in the vhost code, and
there isn't an obvious READ_ONCE() access making the barrier
redundant. However, on closer inspection (thanks, Jason), it appears
that vring synchronisation between the producer and consumer occurs via
the 'avail_idx' field, which is followed up by an rmb() in
vhost_get_vq_desc(), making the read_barrier_depends() redundant on
Alpha.
Jason says:
| I'm also confused about the barrier here, basically in driver side
| we did:
|
| 1) allocate pages
| 2) store pages in indirect->addr
| 3) smp_wmb()
| 4) increase the avail idx (somehow a tail pointer of vring)
|
| in vhost we did:
|
| 1) read avail idx
| 2) smp_rmb()
| 3) read indirect->addr
| 4) read from indirect->addr
|
| It looks to me even the data dependency barrier is not necessary
| since we have rmb() which is sufficient for us to the correct
| indirect->addr and driver are not expected to do any writing to
| indirect->addr after avail idx is increased
Remove the redundant barrier invocation.
Suggested-by: Jason Wang <jasowang@redhat.com>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Will Deacon <will@kernel.org>
---
drivers/vhost/vhost.c | 5 -----
1 file changed, 5 deletions(-)
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index d7b8df3edffc..74d135ee7e26 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -2092,11 +2092,6 @@ static int get_indirect(struct vhost_virtqueue *vq,
return ret;
}
iov_iter_init(&from, READ, vq->indirect, ret, len);
-
- /* We will use the result as an address to read from, so most
- * architectures only need a compiler barrier here. */
- read_barrier_depends();
-
count = len / sizeof desc;
/* Buffers are chained via a 16 bit next field, so
* we can have at most 2^16 of these. */
--
2.27.0.383.g050319c2ae-goog
WARNING: multiple messages have this Message-ID (diff)
From: Will Deacon <will@kernel.org>
To: linux-kernel@vger.kernel.org
Cc: Joel Fernandes <joelaf@google.com>,
Mark Rutland <mark.rutland@arm.com>,
"Michael S. Tsirkin" <mst@redhat.com>,
Peter Zijlstra <peterz@infradead.org>,
Catalin Marinas <catalin.marinas@arm.com>,
virtualization@lists.linux-foundation.org,
Will Deacon <will@kernel.org>, Arnd Bergmann <arnd@arndb.de>,
Alan Stern <stern@rowland.harvard.edu>,
Sami Tolvanen <samitolvanen@google.com>,
Matt Turner <mattst88@gmail.com>,
kernel-team@android.com, Marco Elver <elver@google.com>,
Kees Cook <keescook@chromium.org>,
"Paul E. McKenney" <paulmck@kernel.org>,
Boqun Feng <boqun.feng@gmail.com>,
Ivan Kokshaysky <ink@jurassic.park.msu.ru>,
linux-arm-kernel@lists.infradead.org,
Richard Henderson <rth@twiddle.net>,
Nick Desaulniers <ndesaulniers@google.com>,
linux-alpha@vger.kernel.org
Subject: [PATCH v3 07/19] vhost: Remove redundant use of read_barrier_depends() barrier
Date: Fri, 10 Jul 2020 17:51:51 +0100 [thread overview]
Message-ID: <20200710165203.31284-8-will@kernel.org> (raw)
In-Reply-To: <20200710165203.31284-1-will@kernel.org>
Since commit 76ebbe78f739 ("locking/barriers: Add implicit
smp_read_barrier_depends() to READ_ONCE()"), there is no need to use
smp_read_barrier_depends() outside of the Alpha architecture code.
Unfortunately, there is precisely _one_ user in the vhost code, and
there isn't an obvious READ_ONCE() access making the barrier
redundant. However, on closer inspection (thanks, Jason), it appears
that vring synchronisation between the producer and consumer occurs via
the 'avail_idx' field, which is followed up by an rmb() in
vhost_get_vq_desc(), making the read_barrier_depends() redundant on
Alpha.
Jason says:
| I'm also confused about the barrier here, basically in driver side
| we did:
|
| 1) allocate pages
| 2) store pages in indirect->addr
| 3) smp_wmb()
| 4) increase the avail idx (somehow a tail pointer of vring)
|
| in vhost we did:
|
| 1) read avail idx
| 2) smp_rmb()
| 3) read indirect->addr
| 4) read from indirect->addr
|
| It looks to me even the data dependency barrier is not necessary
| since we have rmb() which is sufficient for us to the correct
| indirect->addr and driver are not expected to do any writing to
| indirect->addr after avail idx is increased
Remove the redundant barrier invocation.
Suggested-by: Jason Wang <jasowang@redhat.com>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Will Deacon <will@kernel.org>
---
drivers/vhost/vhost.c | 5 -----
1 file changed, 5 deletions(-)
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index d7b8df3edffc..74d135ee7e26 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -2092,11 +2092,6 @@ static int get_indirect(struct vhost_virtqueue *vq,
return ret;
}
iov_iter_init(&from, READ, vq->indirect, ret, len);
-
- /* We will use the result as an address to read from, so most
- * architectures only need a compiler barrier here. */
- read_barrier_depends();
-
count = len / sizeof desc;
/* Buffers are chained via a 16 bit next field, so
* we can have at most 2^16 of these. */
--
2.27.0.383.g050319c2ae-goog
WARNING: multiple messages have this Message-ID (diff)
From: Will Deacon <will@kernel.org>
To: linux-kernel@vger.kernel.org
Cc: Joel Fernandes <joelaf@google.com>,
Mark Rutland <mark.rutland@arm.com>,
"Michael S. Tsirkin" <mst@redhat.com>,
Peter Zijlstra <peterz@infradead.org>,
Catalin Marinas <catalin.marinas@arm.com>,
Jason Wang <jasowang@redhat.com>,
virtualization@lists.linux-foundation.org,
Will Deacon <will@kernel.org>, Arnd Bergmann <arnd@arndb.de>,
Alan Stern <stern@rowland.harvard.edu>,
Sami Tolvanen <samitolvanen@google.com>,
Matt Turner <mattst88@gmail.com>,
kernel-team@android.com, Marco Elver <elver@google.com>,
Kees Cook <keescook@chromium.org>,
"Paul E. McKenney" <paulmck@kernel.org>,
Boqun Feng <boqun.feng@gmail.com>,
Ivan Kokshaysky <ink@jurassic.park.msu.ru>,
linux-arm-kernel@lists.infradead.org,
Richard Henderson <rth@twiddle.net>,
Nick Desaulniers <ndesaulniers@google.com>,
linux-alpha@vger.kernel.org
Subject: [PATCH v3 07/19] vhost: Remove redundant use of read_barrier_depends() barrier
Date: Fri, 10 Jul 2020 17:51:51 +0100 [thread overview]
Message-ID: <20200710165203.31284-8-will@kernel.org> (raw)
In-Reply-To: <20200710165203.31284-1-will@kernel.org>
Since commit 76ebbe78f739 ("locking/barriers: Add implicit
smp_read_barrier_depends() to READ_ONCE()"), there is no need to use
smp_read_barrier_depends() outside of the Alpha architecture code.
Unfortunately, there is precisely _one_ user in the vhost code, and
there isn't an obvious READ_ONCE() access making the barrier
redundant. However, on closer inspection (thanks, Jason), it appears
that vring synchronisation between the producer and consumer occurs via
the 'avail_idx' field, which is followed up by an rmb() in
vhost_get_vq_desc(), making the read_barrier_depends() redundant on
Alpha.
Jason says:
| I'm also confused about the barrier here, basically in driver side
| we did:
|
| 1) allocate pages
| 2) store pages in indirect->addr
| 3) smp_wmb()
| 4) increase the avail idx (somehow a tail pointer of vring)
|
| in vhost we did:
|
| 1) read avail idx
| 2) smp_rmb()
| 3) read indirect->addr
| 4) read from indirect->addr
|
| It looks to me even the data dependency barrier is not necessary
| since we have rmb() which is sufficient for us to the correct
| indirect->addr and driver are not expected to do any writing to
| indirect->addr after avail idx is increased
Remove the redundant barrier invocation.
Suggested-by: Jason Wang <jasowang@redhat.com>
Acked-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Will Deacon <will@kernel.org>
---
drivers/vhost/vhost.c | 5 -----
1 file changed, 5 deletions(-)
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index d7b8df3edffc..74d135ee7e26 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -2092,11 +2092,6 @@ static int get_indirect(struct vhost_virtqueue *vq,
return ret;
}
iov_iter_init(&from, READ, vq->indirect, ret, len);
-
- /* We will use the result as an address to read from, so most
- * architectures only need a compiler barrier here. */
- read_barrier_depends();
-
count = len / sizeof desc;
/* Buffers are chained via a 16 bit next field, so
* we can have at most 2^16 of these. */
--
2.27.0.383.g050319c2ae-goog
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
next prev parent reply other threads:[~2020-07-10 16:52 UTC|newest]
Thread overview: 81+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-10 16:51 [PATCH 00/18] Allow architectures to override __READ_ONCE() Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 01/19] tools: bpf: Use local copy of headers including uapi/linux/filter.h Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 02/19] compiler.h: Split {READ,WRITE}_ONCE definitions out into rwonce.h Will Deacon
2020-07-10 16:51 ` [PATCH v3 02/19] compiler.h: Split {READ, WRITE}_ONCE " Will Deacon
2020-07-10 16:51 ` [PATCH v3 02/19] compiler.h: Split {READ,WRITE}_ONCE " Will Deacon
2020-07-13 12:23 ` boqun.feng
2020-07-13 12:23 ` [PATCH v3 02/19] compiler.h: Split {READ, WRITE}_ONCE " boqun.feng
2020-07-13 12:23 ` [PATCH v3 02/19] compiler.h: Split {READ,WRITE}_ONCE " boqun.feng
2020-07-20 15:55 ` Will Deacon
2020-07-20 15:55 ` [PATCH v3 02/19] compiler.h: Split {READ, WRITE}_ONCE " Will Deacon
2020-07-20 15:55 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 03/19] asm/rwonce: Allow __READ_ONCE to be overridden by the architecture Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 04/19] alpha: Override READ_ONCE() with barriered implementation Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 05/19] asm/rwonce: Remove smp_read_barrier_depends() invocation Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 06/19] asm/rwonce: Don't pull <asm/barrier.h> into 'asm-generic/rwonce.h' Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 17:06 ` Nick Desaulniers
2020-07-10 17:06 ` Nick Desaulniers
2020-07-10 17:06 ` Nick Desaulniers
2020-07-10 17:15 ` Will Deacon
2020-07-10 17:15 ` Will Deacon
2020-07-10 17:15 ` Will Deacon
2020-07-10 16:51 ` Will Deacon [this message]
2020-07-10 16:51 ` [PATCH v3 07/19] vhost: Remove redundant use of read_barrier_depends() barrier Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-13 11:27 ` Michael S. Tsirkin
2020-07-13 11:27 ` Michael S. Tsirkin
2020-07-13 11:27 ` Michael S. Tsirkin
2020-07-10 16:51 ` [PATCH v3 08/19] alpha: Replace smp_read_barrier_depends() usage with smp_[r]mb() Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 09/19] locking/barriers: Remove definitions for [smp_]read_barrier_depends() Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 10/19] Documentation/barriers: Remove references to [smp_]read_barrier_depends() Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 11/19] Documentation/barriers/kokr: " Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 12/19] tools/memory-model: Remove smp_read_barrier_depends() from informal doc Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 13/19] include/linux: Remove smp_read_barrier_depends() from comments Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 14/19] checkpatch: Remove checks relating to [smp_]read_barrier_depends() Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` [PATCH v3 15/19] arm64: Reduce the number of header files pulled into vmlinux.lds.S Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:51 ` Will Deacon
2020-07-10 16:52 ` [PATCH v3 16/19] arm64: alternatives: Split up alternative.h Will Deacon
2020-07-10 16:52 ` Will Deacon
2020-07-10 16:52 ` Will Deacon
2020-07-10 16:52 ` [PATCH v3 17/19] arm64: cpufeatures: Add capability for LDAPR instruction Will Deacon
2020-07-10 16:52 ` Will Deacon
2020-07-10 16:52 ` Will Deacon
2020-07-10 16:52 ` [PATCH v3 18/19] arm64: alternatives: Remove READ_ONCE() usage during patch operation Will Deacon
2020-07-10 16:52 ` Will Deacon
2020-07-10 16:52 ` Will Deacon
2020-07-10 16:52 ` [PATCH v3 19/19] arm64: lto: Strengthen READ_ONCE() to acquire when CONFIG_LTO=y Will Deacon
2020-07-10 16:52 ` Will Deacon
2020-07-10 16:52 ` Will Deacon
2020-07-28 20:40 ` Pavel Machek
2020-07-28 20:40 ` Pavel Machek
2020-07-28 20:40 ` Pavel Machek
2020-07-13 10:34 ` [PATCH 00/18] Allow architectures to override __READ_ONCE() Peter Zijlstra
2020-07-13 10:34 ` Peter Zijlstra
2020-07-13 10:34 ` Peter Zijlstra
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=20200710165203.31284-8-will@kernel.org \
--to=will@kernel.org \
--cc=arnd@arndb.de \
--cc=boqun.feng@gmail.com \
--cc=catalin.marinas@arm.com \
--cc=elver@google.com \
--cc=ink@jurassic.park.msu.ru \
--cc=jasowang@redhat.com \
--cc=joelaf@google.com \
--cc=keescook@chromium.org \
--cc=kernel-team@android.com \
--cc=linux-alpha@vger.kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mark.rutland@arm.com \
--cc=mattst88@gmail.com \
--cc=mst@redhat.com \
--cc=ndesaulniers@google.com \
--cc=paulmck@kernel.org \
--cc=peterz@infradead.org \
--cc=rth@twiddle.net \
--cc=samitolvanen@google.com \
--cc=stern@rowland.harvard.edu \
--cc=virtualization@lists.linux-foundation.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 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.