[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ef2068aa-0732-820e-fbb4-299021ca54df@nvidia.com>
Date: Wed, 16 Aug 2023 03:24:14 +0300
From: Max Gurtovoy <mgurtovoy@...dia.com>
To: Aurelien Aptel <aaptel@...dia.com>, linux-nvme@...ts.infradead.org,
netdev@...r.kernel.org, sagi@...mberg.me, hch@....de, kbusch@...nel.org,
axboe@...com, chaitanyak@...dia.com, davem@...emloft.net, kuba@...nel.org
Cc: Ben Ben-Ishay <benishay@...dia.com>, aurelien.aptel@...il.com,
smalin@...dia.com, malin1024@...il.com, ogerlitz@...dia.com,
yorayz@...dia.com, borisp@...dia.com, galshalom@...dia.com
Subject: Re: [PATCH v12 05/26] iov_iter: skip copy if src == dst for direct
data placement
On 12/07/2023 19:14, Aurelien Aptel wrote:
> From: Ben Ben-Ishay <benishay@...dia.com>
>
> When using direct data placement (DDP) the NIC could write the payload
> directly into the destination buffer and constructs SKBs such that
> they point to this data. To skip copies when SKB data already resides
> in the destination buffer we check if (src == dst), and skip the copy
> when it's true.
>
> Signed-off-by: Ben Ben-Ishay <benishay@...dia.com>
> Signed-off-by: Boris Pismenny <borisp@...dia.com>
> Signed-off-by: Or Gerlitz <ogerlitz@...dia.com>
> Signed-off-by: Yoray Zack <yorayz@...dia.com>
> Signed-off-by: Shai Malin <smalin@...dia.com>
> Signed-off-by: Aurelien Aptel <aaptel@...dia.com>
> Reviewed-by: Chaitanya Kulkarni <kch@...dia.com>
> ---
> lib/iov_iter.c | 8 +++++++-
> 1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/lib/iov_iter.c b/lib/iov_iter.c
> index b667b1e2f688..1c9b10e1e1c8 100644
> --- a/lib/iov_iter.c
> +++ b/lib/iov_iter.c
> @@ -313,9 +313,15 @@ size_t _copy_to_iter(const void *addr, size_t bytes, struct iov_iter *i)
> return 0;
> if (user_backed_iter(i))
> might_fault();
> + /*
> + * When using direct data placement (DDP) the hardware writes
> + * data directly to the destination buffer, and constructs
> + * IOVs such that they point to this data.
> + * Thus, when the src == dst we skip the memcpy.
> + */
> iterate_and_advance(i, bytes, base, len, off,
> copyout(base, addr + off, len),
> - memcpy(base, addr + off, len)
> + (base != addr + off) && memcpy(base, addr + off, len)
> )
>
> return bytes;
Looks good,
Reviewed-by: Max Gurtovoy <mgurtovoy@...dia.com>
Powered by blists - more mailing lists