diff options
| author | Takashi Iwai <tiwai@suse.de> | 2020-03-30 14:43:00 +0300 |
|---|---|---|
| committer | Takashi Iwai <tiwai@suse.de> | 2020-03-30 14:43:00 +0300 |
| commit | 3c22baeab40b2f8e75907cfd7aa69147d5343d2c (patch) | |
| tree | e8300e52242ae3a9c7bfe6089ee7125324095e27 /drivers/net/wireguard/send.c | |
| parent | aa21c3d4b941739651e77747d2f7a20a6c1d87bc (diff) | |
| parent | 1c521d7e62262793789845989edca57dea24eb7d (diff) | |
| download | linux-3c22baeab40b2f8e75907cfd7aa69147d5343d2c.tar.xz | |
Merge tag 'asoc-v5.7' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-linus
ASoC: Updates for v5.7
This is a very big update for the core since Morimoto-san has been
rather busy continuing his refactorings to clean up a lot of the cruft
that we have accumilated over the years. We've also gained several new
drivers, including initial (but still not complete) parts of the Intel
SoundWire support.
- Lots of refactorings to modernize the code from Morimoto-san.
- Conversion of SND_SOC_ALL_CODECS to use imply from Geert Uytterhoeven.
- Continued refactoring and fixing of the Intel support.
- Soundwire and more advanced clocking support for Realtek RT5682.
- Support for amlogic GX, Meson 8, Meson 8B and T9015 DAC, Broadcom
DSL/PON, Ingenic JZ4760 and JZ4770, Realtek RL6231, and TI TAS2563 and
TLV320ADCX140.
Diffstat (limited to 'drivers/net/wireguard/send.c')
| -rw-r--r-- | drivers/net/wireguard/send.c | 16 |
1 files changed, 11 insertions, 5 deletions
diff --git a/drivers/net/wireguard/send.c b/drivers/net/wireguard/send.c index c13260563446..7348c10cbae3 100644 --- a/drivers/net/wireguard/send.c +++ b/drivers/net/wireguard/send.c @@ -143,16 +143,22 @@ static void keep_key_fresh(struct wg_peer *peer) static unsigned int calculate_skb_padding(struct sk_buff *skb) { + unsigned int padded_size, last_unit = skb->len; + + if (unlikely(!PACKET_CB(skb)->mtu)) + return ALIGN(last_unit, MESSAGE_PADDING_MULTIPLE) - last_unit; + /* We do this modulo business with the MTU, just in case the networking * layer gives us a packet that's bigger than the MTU. In that case, we * wouldn't want the final subtraction to overflow in the case of the - * padded_size being clamped. + * padded_size being clamped. Fortunately, that's very rarely the case, + * so we optimize for that not happening. */ - unsigned int last_unit = skb->len % PACKET_CB(skb)->mtu; - unsigned int padded_size = ALIGN(last_unit, MESSAGE_PADDING_MULTIPLE); + if (unlikely(last_unit > PACKET_CB(skb)->mtu)) + last_unit %= PACKET_CB(skb)->mtu; - if (padded_size > PACKET_CB(skb)->mtu) - padded_size = PACKET_CB(skb)->mtu; + padded_size = min(PACKET_CB(skb)->mtu, + ALIGN(last_unit, MESSAGE_PADDING_MULTIPLE)); return padded_size - last_unit; } |