[PATCH v5 18/20] net-lwip: add TFTP_BLOCKSIZE

Ilias Apalodimas ilias.apalodimas at linaro.org
Mon Jul 29 14:29:03 CEST 2024


Hi Jerome,

On Thu, 25 Jul 2024 at 15:59, Jerome Forissier
<jerome.forissier at linaro.org> wrote:
>
> Add support for setting the TFTP block size. The default value (1468)
> is fine for Ethernet and allows a better throughput than the TFTP
> default (512), if the server supports the blksize option of course.
>
> I tested this change with qemu_arm64_lwip_defconfig. The throughput is
> now 875 KiB/s vs. 313 KiB/s before. That is still a low number, but I
> think we can't expect more without implementing the windowsize option.
>
> Signed-off-by: Jerome Forissier <jerome.forissier at linaro.org>
> ---
>  net/Kconfig     | 20 ++++++++++----------
>  net/lwip/tftp.c |  2 ++
>  2 files changed, 12 insertions(+), 10 deletions(-)
>
> diff --git a/net/Kconfig b/net/Kconfig
> index 952690d677d..7e5406bb923 100644
> --- a/net/Kconfig
> +++ b/net/Kconfig
> @@ -69,16 +69,6 @@ config SYS_FAULT_ECHO_LINK_DOWN
>           this option is active, then CONFIG_SYS_FAULT_MII_ADDR also needs to
>           be configured.
>
> -config TFTP_BLOCKSIZE
> -       int "TFTP block size"
> -       default 1468
> -       help
> -         Default TFTP block size.
> -         The MTU is typically 1500 for ethernet, so a TFTP block of
> -         1468 (MTU minus eth.hdrs) provides a good throughput with
> -         almost-MTU block sizes.
> -         You can also activate CONFIG_IP_DEFRAG to set a larger block.
> -

Why do you have to move the config option?

Cheers
/Ilias
>  config TFTP_PORT
>         bool "Set TFTP UDP source/destination ports via the environment"
>         help
> @@ -263,4 +253,14 @@ config SYS_RX_ETH_BUFFER
>           since all buffers can be full shortly after enabling the interface on
>           high Ethernet traffic.
>
> +config TFTP_BLOCKSIZE
> +       int "TFTP block size"
> +       default 1468
> +       help
> +         Default TFTP block size.
> +         The MTU is typically 1500 for ethernet, so a TFTP block of
> +         1468 (MTU minus eth.hdrs) provides a good throughput with
> +         almost-MTU block sizes.
> +         You can also activate CONFIG_IP_DEFRAG to set a larger block.
> +
>  endif   # if NET || NET_LWIP
> diff --git a/net/lwip/tftp.c b/net/lwip/tftp.c
> index 78de0bd0dba..1fe948c84ec 100644
> --- a/net/lwip/tftp.c
> +++ b/net/lwip/tftp.c
> @@ -136,6 +136,8 @@ static int tftp_loop(struct udevice *udev, ulong addr, char *fname,
>         if (!(err == ERR_OK || err == ERR_USE))
>                 log_err("tftp_init_client err: %d\n", err);
>
> +       tftp_client_set_blksize(CONFIG_TFTP_BLOCKSIZE);
> +
>         ctx.start_time = get_timer(0);
>         err = tftp_get(&ctx, &srvip, srvport, fname, TFTP_MODE_OCTET);
>         /* might return different errors, like routing problems */
> --
> 2.40.1
>


More information about the U-Boot mailing list