[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <1497612396.10546.7.camel@perches.com>
Date: Fri, 16 Jun 2017 04:26:36 -0700
From: Joe Perches <joe@...ches.com>
To: Johannes Berg <johannes@...solutions.net>,
Bjørn Mork <bjorn@...k.no>
Cc: netdev@...r.kernel.org
Subject: Re: [RFC 1/3] networking: make skb_put & friends return void
pointers
On Fri, 2017-06-16 at 13:03 +0200, Johannes Berg wrote:
> On Fri, 2017-06-16 at 03:39 -0700, Joe Perches wrote:
> >
> > I would suggest avoiding using skb_put_u8 as it
> > would suggest there should be skb_put_u16 etc
>
> If there are enough users, I don't know why we wouldn't add it? Though
> it should be _unaligned_le16 and _unaligned_be16, if at all :)
Maybe.
Avoid api explosion can be good.
Maybe using skb_put_data() and optimizing away size
using __builtin_constant_p and a switch statement
could do more or less the same thing.
$ git grep -P "\*\s*\(.*\*\s*\)\s*skb_put" *
drivers/net/wireless/intel/ipw2x00/ipw2200.c: *(__le16*)skb_put(dst, sizeof(u16)) = cpu_to_le16(
drivers/net/wireless/intel/ipw2x00/ipw2200.c: *(__le16*)skb_put(dst, sizeof(u16)) =
drivers/net/wireless/intel/ipw2x00/ipw2200.c: *(__le16*)skb_put(dst, sizeof(u16)) =
drivers/net/wireless/intel/ipw2x00/ipw2200.c: *(__le16*)skb_put(dst, sizeof(u16)) =
drivers/s390/net/ctcm_fsms.c: *((__u16 *)skb_put(ch->trans_skb, 2)) = ch->collect_len + 2;
drivers/atm/fore200e.c: *((u32*)skb_put(skb, 4)) = cell_header;
net/l2tp/l2tp_ip.c: *((__be32 *) skb_put(skb, 4)) = 0;
net/atm/signaling.c: *(struct atmsvc_msg *)skb_put(skb, sizeof(struct atmsvc_msg)) = *msg;
Powered by blists - more mailing lists