On 08/19/2015 05:24 PM, Eric Dumazet wrote:
> On Wed, 2015-08-19 at 15:59 +0300, Emmanuel Grumbach wrote:
>> This allows to release the backpressure on the socket only
>> when the last segment is released.
>> Now the truesize looks like this:
>> if the truesize of the original skb is 65420, all the
>> segments will have a truesize of 704 (skb itself) and the
>> last one will have 65420.
>>
>> Change-Id: I3c894cf2afc0aedfe7b2a5b992ba41653ff79c0e
>> Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@xxxxxxxxx>
>> ---
>> drivers/net/wireless/iwlwifi/mvm/tx.c | 17 ++++++++++++++++-
>> 1 file changed, 16 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/net/wireless/iwlwifi/mvm/tx.c b/drivers/net/wireless/iwlwifi/mvm/tx.c
>> index 5046833..046e50d 100644
>> --- a/drivers/net/wireless/iwlwifi/mvm/tx.c
>> +++ b/drivers/net/wireless/iwlwifi/mvm/tx.c
>> @@ -764,7 +764,7 @@ static int iwl_mvm_tx_tso(struct iwl_mvm *mvm, struct sk_buff *skb_gso,
>> bool ipv6 = skb_shinfo(skb_gso)->gso_type & SKB_GSO_TCPV6;
>> struct iwl_lso_splitter s = {};
>> struct page *hdr_page;
>> - unsigned int mpdu_sz;
>> + unsigned int mpdu_sz, sum_truesize = 0;
>> u8 *hdr_page_pos, *qc, tid;
>> int i, ret;
>>
>> @@ -898,6 +898,7 @@ static int iwl_mvm_tx_tso(struct iwl_mvm *mvm, struct sk_buff *skb_gso,
>> mpdu_sz, tcp_hdrlen(skb_gso));
>>
>> __skb_queue_tail(mpdus_skb, skb_gso);
>> + sum_truesize += skb_gso->truesize;
>>
>> /* mss bytes have been consumed from the data */
>> s.gso_payload_pos = s.mss;
>> @@ -1034,6 +1035,20 @@ static int iwl_mvm_tx_tso(struct iwl_mvm *mvm, struct sk_buff *skb_gso,
>> }
>>
>> __skb_queue_tail(mpdus_skb, skb);
>> + sum_truesize += skb->truesize;
>> + }
>> +
>> + /* Release the backpressure on the socket only when
>> + * the last segment is released.
>> + */
>> + if (skb_gso->destructor == sock_wfree) {
>> + struct sk_buff *tail = mpdus_skb->prev;
>> +
>> + swap(tail->truesize, skb_gso->truesize);
>> + swap(tail->destructor, skb_gso->destructor);
>> + swap(tail->sk, skb_gso->sk);
>> + atomic_add(sum_truesize - skb_gso->truesize,
>> + &skb_gso->sk->sk_wmem_alloc);
>> }
>>
>> ret = 0;
>
> Using existing net/core/tso.c helpers would avoid using this.
Hm.. how would net/core/tso.c avoid this?
I can't see anything related to truesize there.
Note that this work since it is guaranteed that we release the skbs in
order.
>
> (BTW TCP packets do not have sock_wfree as destructor but tcp_wfree(),
> yet we want backpressure mostly for TCP stack (TCP Small Queues))
>
>
I am not sure I follow here.
You want me to test:
if (skb_gso->destructor == tcp_wfree) ?
I checked that code using iperf and saw that I don't get into this if,
but I (probably wrongly) assumed that other applications would set a
flag on the socket (forgive my ignorance) that would make this if be taken.
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
