netdev - Re: [RFC v2 1/3] iwlwifi: mvm: add real TSO implementation

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <0BA3FCBA62E2DC44AF3030971E174FB32E825194@hasmsx107.ger.corp.intel.com>
Date:	Wed, 19 Aug 2015 17:05:23 +0000
From:	"Grumbach, Emmanuel" <emmanuel.grumbach@...el.com>
To:	Eric Dumazet <eric.dumazet@...il.com>
CC:	"linux-wireless@...r.kernel.org" <linux-wireless@...r.kernel.org>,
	"ido@...ery.com" <ido@...ery.com>,
	"netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: [RFC v2 1/3] iwlwifi: mvm: add real TSO implementation



On 08/19/2015 05:18 PM, Eric Dumazet wrote:
> On Wed, 2015-08-19 at 15:59 +0300, Emmanuel Grumbach wrote:
>> The segmentation is done completely in software. The
>> driver creates several MPDUs out of a single large send.
>> Each MPDU is a newly allocated SKB.
>> A page is allocated to create the headers that need to be
>> duplicated (SNAP / IP / TCP). The WiFi header is in the
>> header of the newly created SKBs.
>>
>> type=feature
>>
>> Change-Id: I238ffa79cacc5bbdacdfbf3e9673c8d4f02b462a
>> Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@...el.com>
>> ---
>>  drivers/net/wireless/iwlwifi/mvm/tx.c | 513 +++++++++++++++++++++++++++++++---
>>  1 file changed, 481 insertions(+), 32 deletions(-)
> 
> Ouch dynamic allocations while doing xmit are certainly not needed.
> Your driver should pre-allocated space for headers.

This is right as long as you don't need *several* headers in one single
skb. In the case of A-MSDU, I need to have several TCP / IP / SNAP
headers in the same skb. At least that's how my HW layer in the driver
is built. See the other thread.

> 
> Drivers willing to implement tso have to use net/core/tso.c provided
> helpers.
> 
> $ git grep -n tso_build_hdr
> drivers/net/ethernet/cavium/thunder/nicvf_queues.c:1030:                tso_build_hdr(skb, hdr, &tso, data_left, total_len == 0);
> drivers/net/ethernet/freescale/fec_main.c:729:          tso_build_hdr(skb, hdr, &tso, data_left, total_len == 0);
> drivers/net/ethernet/marvell/mv643xx_eth.c:842:         tso_build_hdr(skb, hdr, &tso, data_left, total_len == 0);
> drivers/net/ethernet/marvell/mvneta.c:1650:             tso_build_hdr(skb, hdr, &tso, data_left, total_len == 0);
> include/net/tso.h:15:void tso_build_hdr(struct sk_buff *skb, char *hdr, struct tso_t *tso,
> net/core/tso.c:14:void tso_build_hdr(struct sk_buff *skb, char *hdr, struct tso_t *tso,
> net/core/tso.c:37:EXPORT_SYMBOL(tso_build_hdr);
> 
> 

This looks promising indeed. I'll take a close look.
Thanks a bunch.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html