From: Bjorn Helgaas <bhelgaas@google.com>
To: Allen Hubbe <Allen.Hubbe@emc.com>
Cc: linux-ntb@googlegroups.com, linux-kernel@vger.kernel.org,
linux-pci@vger.kernel.org, Jon Mason <jdmason@kudzu.us>,
Dave Jiang <dave.jiang@intel.com>
Subject: Re: [PATCH v3 15/18] NTB: Improve performance with write combining
Date: Tue, 9 Jun 2015 11:46:32 -0500 [thread overview]
Message-ID: <20150609164632.GG23119@google.com> (raw)
In-Reply-To: <bc5eb3dd58e308b06a3a7464627704210c175394.1433838377.git.Allen.Hubbe@emc.com>
On Tue, Jun 09, 2015 at 05:44:42AM -0400, Allen Hubbe wrote:
> From: Dave Jiang <dave.jiang@intel.com>
>
> Changing the memory window BAR mappings to write combining significantly
> boosts the performance. We will also use memcpy that utilizies
s/utilizies/uses/
> non-temporal store which showed performance improement when doing
s/improement/improvement/
> non-cached memcpys.
>
> Signed-off-by: Dave Jiang <dave.jiang@intel.com>
> ---
> drivers/ntb/ntb_transport.c | 9 +++++++--
> 1 file changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/ntb/ntb_transport.c b/drivers/ntb/ntb_transport.c
> index dc14ec81c43e..528ea0af6d54 100644
> --- a/drivers/ntb/ntb_transport.c
> +++ b/drivers/ntb/ntb_transport.c
> @@ -58,6 +58,7 @@
> #include <linux/pci.h>
> #include <linux/slab.h>
> #include <linux/types.h>
> +#include <linux/uaccess.h>
> #include "linux/ntb.h"
> #include "linux/ntb_transport.h"
>
> @@ -993,7 +994,7 @@ static int ntb_transport_probe(struct ntb_client *self, struct ntb_dev *ndev)
> if (rc)
> goto err1;
>
> - mw->vbase = ioremap(mw->phys_addr, mw->phys_size);
> + mw->vbase = ioremap_wc(mw->phys_addr, mw->phys_size);
> if (!mw->vbase) {
> rc = -ENOMEM;
> goto err1;
> @@ -1375,7 +1376,11 @@ static void ntb_tx_copy_callback(void *data)
>
> static void ntb_memcpy_tx(struct ntb_queue_entry *entry, void __iomem *offset)
> {
> - memcpy_toio(offset, entry->buf, entry->len);
> + /*
> + * Using non-temporal mov to improve performance on non-cached
> + * writes. Even though we aren't actually copying from user space.
> + */
> + __copy_from_user_inatomic_nocache(offset, entry->buf, entry->len);
__copy_from_user_inatomic_nocache() only exists on x86, but this file looks
like it's intended to be arch-agnostic?
>
> /* Ensure that the data is fully copied out before setting the flags */
> wmb();
> --
> 2.4.0.rc0.43.gcf8a8c6
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-pci" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
next prev parent reply other threads:[~2015-06-09 16:46 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-06-09 9:44 [PATCH v3 00/18] NTB: Add NTB hardware abstraction layer Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 01/18] NTB: Move files in preparation for NTB abstraction Allen Hubbe
2015-06-09 16:16 ` Bjorn Helgaas
2015-06-09 16:42 ` Hubbe, Allen
2015-06-09 9:44 ` [PATCH v3 02/18] NTB: Add NTB hardware abstraction layer Allen Hubbe
2015-06-09 16:23 ` Bjorn Helgaas
2015-06-09 16:43 ` Hubbe, Allen
2015-06-09 16:53 ` Bjorn Helgaas
2015-06-09 17:32 ` Hubbe, Allen
2015-06-09 9:44 ` [PATCH v3 03/18] NTB: Read peer info from local SPAD in transport Allen Hubbe
2015-06-09 16:24 ` Bjorn Helgaas
2015-06-09 9:44 ` [PATCH v3 04/18] NTB: Enable link for Intel root port mode in probe Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 05/18] NTB: Check the device ID to set errata flags Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 06/18] NTB: Differentiate transport link down messages Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 07/18] NTB: Do not advance transport RX on link down Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 08/18] NTB: Reset transport QP link stats on down Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 09/18] NTB: Add parameters for Intel SNB B2B addresses Allen Hubbe
2015-06-09 16:42 ` Bjorn Helgaas
2015-06-09 16:59 ` Hubbe, Allen
2015-06-09 9:44 ` [PATCH v3 10/18] NTB: Add ping pong test client Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 11/18] NTB: Add tool " Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 12/18] NTB: Rate limit ntb_qp_link_work Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 13/18] NTB: Use NUMA memory and DMA chan in transport Allen Hubbe
2015-06-09 16:43 ` Bjorn Helgaas
2015-06-09 9:44 ` [PATCH v3 14/18] NTB: Use NUMA memory in Intel driver Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 15/18] NTB: Improve performance with write combining Allen Hubbe
2015-06-09 16:46 ` Bjorn Helgaas [this message]
2015-06-09 9:44 ` [PATCH v3 16/18] NTB: Default to CPU memcpy for performance Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 17/18] NTB: Rename intel hw to proper platform names Allen Hubbe
2015-06-09 9:44 ` [PATCH v3 18/18] NTB: Increase transport MTU to 64k from 16k Allen Hubbe
2015-06-09 16:04 ` Bjorn Helgaas
2015-06-09 16:45 ` Jiang, Dave
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=20150609164632.GG23119@google.com \
--to=bhelgaas@google.com \
--cc=Allen.Hubbe@emc.com \
--cc=dave.jiang@intel.com \
--cc=jdmason@kudzu.us \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-ntb@googlegroups.com \
--cc=linux-pci@vger.kernel.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;
as well as URLs for NNTP newsgroup(s).