* [PATCH mptcp-next] mptcp: drop tx skb cache
@ 2021-05-21 10:30 Paolo Abeni
2021-05-21 16:19 ` Paolo Abeni
0 siblings, 1 reply; 2+ messages in thread
From: Paolo Abeni @ 2021-05-21 10:30 UTC (permalink / raw)
To: mptcp
The mentioned cache was introduced to reduce the number of skb
allocation in atomic context, but the required complexity is
excessive.
This change remove the mentioned cache.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
---
formally submitting this one. I keept a VM running self-tests for ~24h
with no issues
---
net/mptcp/protocol.c | 89 ++------------------------------------------
1 file changed, 4 insertions(+), 85 deletions(-)
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index 446acfb85493..1114a914d845 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -903,22 +903,14 @@ static bool mptcp_frag_can_collapse_to(const struct mptcp_sock *msk,
df->data_seq + df->data_len == msk->write_seq;
}
-static int mptcp_wmem_with_overhead(struct sock *sk, int size)
+static int mptcp_wmem_with_overhead(int size)
{
- struct mptcp_sock *msk = mptcp_sk(sk);
- int ret, skbs;
-
- ret = size + ((sizeof(struct mptcp_data_frag) * size) >> PAGE_SHIFT);
- skbs = (msk->tx_pending_data + size) / msk->size_goal_cache;
- if (skbs < msk->skb_tx_cache.qlen)
- return ret;
-
- return ret + (skbs - msk->skb_tx_cache.qlen) * SKB_TRUESIZE(MAX_TCP_HEADER);
+ return size + ((sizeof(struct mptcp_data_frag) * size) >> PAGE_SHIFT);
}
static void __mptcp_wmem_reserve(struct sock *sk, int size)
{
- int amount = mptcp_wmem_with_overhead(sk, size);
+ int amount = mptcp_wmem_with_overhead(size);
struct mptcp_sock *msk = mptcp_sk(sk);
WARN_ON_ONCE(msk->wmem_reserved);
@@ -1213,49 +1205,8 @@ static struct sk_buff *__mptcp_do_alloc_tx_skb(struct sock *sk, gfp_t gfp)
return NULL;
}
-static bool mptcp_tx_cache_refill(struct sock *sk, int size,
- struct sk_buff_head *skbs, int *total_ts)
-{
- struct mptcp_sock *msk = mptcp_sk(sk);
- struct sk_buff *skb;
- int space_needed;
-
- if (unlikely(tcp_under_memory_pressure(sk))) {
- mptcp_mem_reclaim_partial(sk);
-
- /* under pressure pre-allocate at most a single skb */
- if (msk->skb_tx_cache.qlen)
- return true;
- space_needed = msk->size_goal_cache;
- } else {
- space_needed = msk->tx_pending_data + size -
- msk->skb_tx_cache.qlen * msk->size_goal_cache;
- }
-
- while (space_needed > 0) {
- skb = __mptcp_do_alloc_tx_skb(sk, sk->sk_allocation);
- if (unlikely(!skb)) {
- /* under memory pressure, try to pass the caller a
- * single skb to allow forward progress
- */
- while (skbs->qlen > 1) {
- skb = __skb_dequeue_tail(skbs);
- *total_ts -= skb->truesize;
- __kfree_skb(skb);
- }
- return skbs->qlen > 0;
- }
-
- *total_ts += skb->truesize;
- __skb_queue_tail(skbs, skb);
- space_needed -= msk->size_goal_cache;
- }
- return true;
-}
-
static bool __mptcp_alloc_tx_skb(struct sock *sk, struct sock *ssk, gfp_t gfp)
{
- struct mptcp_sock *msk = mptcp_sk(sk);
struct sk_buff *skb;
if (ssk->sk_tx_skb_cache) {
@@ -1266,22 +1217,6 @@ static bool __mptcp_alloc_tx_skb(struct sock *sk, struct sock *ssk, gfp_t gfp)
return true;
}
- skb = skb_peek(&msk->skb_tx_cache);
- if (skb) {
- if (likely(sk_wmem_schedule(ssk, skb->truesize))) {
- skb = __skb_dequeue(&msk->skb_tx_cache);
- if (WARN_ON_ONCE(!skb))
- return false;
-
- mptcp_wmem_uncharge(sk, skb->truesize);
- ssk->sk_tx_skb_cache = skb;
- return true;
- }
-
- /* over memory limit, no point to try to allocate a new skb */
- return false;
- }
-
skb = __mptcp_do_alloc_tx_skb(sk, gfp);
if (!skb)
return false;
@@ -1297,7 +1232,6 @@ static bool __mptcp_alloc_tx_skb(struct sock *sk, struct sock *ssk, gfp_t gfp)
static bool mptcp_must_reclaim_memory(struct sock *sk, struct sock *ssk)
{
return !ssk->sk_tx_skb_cache &&
- !skb_peek(&mptcp_sk(sk)->skb_tx_cache) &&
tcp_under_memory_pressure(sk);
}
@@ -1340,7 +1274,6 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
/* compute send limit */
info->mss_now = tcp_send_mss(ssk, &info->size_goal, info->flags);
avail_size = info->size_goal;
- msk->size_goal_cache = info->size_goal;
skb = tcp_write_queue_tail(ssk);
if (skb) {
/* Limit the write to the size available in the
@@ -1689,7 +1622,6 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
while (msg_data_left(msg)) {
int total_ts, frag_truesize = 0;
struct mptcp_data_frag *dfrag;
- struct sk_buff_head skbs;
bool dfrag_collapsed;
size_t psize, offset;
@@ -1722,16 +1654,10 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
psize = pfrag->size - offset;
psize = min_t(size_t, psize, msg_data_left(msg));
total_ts = psize + frag_truesize;
- __skb_queue_head_init(&skbs);
- if (!mptcp_tx_cache_refill(sk, psize, &skbs, &total_ts))
- goto wait_for_memory;
- if (!mptcp_wmem_alloc(sk, total_ts)) {
- __skb_queue_purge(&skbs);
+ if (!mptcp_wmem_alloc(sk, total_ts))
goto wait_for_memory;
- }
- skb_queue_splice_tail(&skbs, &msk->skb_tx_cache);
if (copy_page_from_iter(dfrag->page, offset, psize,
&msg->msg_iter) != psize) {
mptcp_wmem_uncharge(sk, psize + frag_truesize);
@@ -2460,13 +2386,11 @@ static int __mptcp_init_sock(struct sock *sk)
INIT_LIST_HEAD(&msk->rtx_queue);
INIT_WORK(&msk->work, mptcp_worker);
__skb_queue_head_init(&msk->receive_queue);
- __skb_queue_head_init(&msk->skb_tx_cache);
msk->out_of_order_queue = RB_ROOT;
msk->first_pending = NULL;
msk->wmem_reserved = 0;
msk->rmem_released = 0;
msk->tx_pending_data = 0;
- msk->size_goal_cache = TCP_BASE_MSS;
msk->ack_hint = NULL;
msk->first = NULL;
@@ -2527,15 +2451,10 @@ static void __mptcp_clear_xmit(struct sock *sk)
{
struct mptcp_sock *msk = mptcp_sk(sk);
struct mptcp_data_frag *dtmp, *dfrag;
- struct sk_buff *skb;
WRITE_ONCE(msk->first_pending, NULL);
list_for_each_entry_safe(dfrag, dtmp, &msk->rtx_queue, list)
dfrag_clear(sk, dfrag);
- while ((skb = __skb_dequeue(&msk->skb_tx_cache)) != NULL) {
- sk->sk_forward_alloc += skb->truesize;
- kfree_skb(skb);
- }
}
static void mptcp_cancel_work(struct sock *sk)
--
2.26.3
^ permalink raw reply related [flat|nested] 2+ messages in thread
* Re: [PATCH mptcp-next] mptcp: drop tx skb cache
2021-05-21 10:30 [PATCH mptcp-next] mptcp: drop tx skb cache Paolo Abeni
@ 2021-05-21 16:19 ` Paolo Abeni
0 siblings, 0 replies; 2+ messages in thread
From: Paolo Abeni @ 2021-05-21 16:19 UTC (permalink / raw)
To: mptcp
On Fri, 2021-05-21 at 12:30 +0200, Paolo Abeni wrote:
> The mentioned cache was introduced to reduce the number of skb
> allocation in atomic context, but the required complexity is
> excessive.
>
> This change remove the mentioned cache.
>
> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
> ---
> formally submitting this one. I keept a VM running self-tests for ~24h
> with no issues
> ---
> net/mptcp/protocol.c | 89 ++------------------------------------------
> 1 file changed, 4 insertions(+), 85 deletions(-)
>
> diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
> index 446acfb85493..1114a914d845 100644
> --- a/net/mptcp/protocol.c
> +++ b/net/mptcp/protocol.c
> @@ -903,22 +903,14 @@ static bool mptcp_frag_can_collapse_to(const struct mptcp_sock *msk,
> df->data_seq + df->data_len == msk->write_seq;
> }
>
> -static int mptcp_wmem_with_overhead(struct sock *sk, int size)
> +static int mptcp_wmem_with_overhead(int size)
> {
> - struct mptcp_sock *msk = mptcp_sk(sk);
> - int ret, skbs;
> -
> - ret = size + ((sizeof(struct mptcp_data_frag) * size) >> PAGE_SHIFT);
> - skbs = (msk->tx_pending_data + size) / msk->size_goal_cache;
> - if (skbs < msk->skb_tx_cache.qlen)
> - return ret;
> -
> - return ret + (skbs - msk->skb_tx_cache.qlen) * SKB_TRUESIZE(MAX_TCP_HEADER);
> + return size + ((sizeof(struct mptcp_data_frag) * size) >> PAGE_SHIFT);
> }
>
> static void __mptcp_wmem_reserve(struct sock *sk, int size)
> {
> - int amount = mptcp_wmem_with_overhead(sk, size);
> + int amount = mptcp_wmem_with_overhead(size);
> struct mptcp_sock *msk = mptcp_sk(sk);
>
> WARN_ON_ONCE(msk->wmem_reserved);
> @@ -1213,49 +1205,8 @@ static struct sk_buff *__mptcp_do_alloc_tx_skb(struct sock *sk, gfp_t gfp)
> return NULL;
> }
>
> -static bool mptcp_tx_cache_refill(struct sock *sk, int size,
> - struct sk_buff_head *skbs, int *total_ts)
> -{
> - struct mptcp_sock *msk = mptcp_sk(sk);
> - struct sk_buff *skb;
> - int space_needed;
> -
> - if (unlikely(tcp_under_memory_pressure(sk))) {
> - mptcp_mem_reclaim_partial(sk);
> -
> - /* under pressure pre-allocate at most a single skb */
> - if (msk->skb_tx_cache.qlen)
> - return true;
> - space_needed = msk->size_goal_cache;
> - } else {
> - space_needed = msk->tx_pending_data + size -
> - msk->skb_tx_cache.qlen * msk->size_goal_cache;
> - }
> -
> - while (space_needed > 0) {
> - skb = __mptcp_do_alloc_tx_skb(sk, sk->sk_allocation);
> - if (unlikely(!skb)) {
> - /* under memory pressure, try to pass the caller a
> - * single skb to allow forward progress
> - */
> - while (skbs->qlen > 1) {
> - skb = __skb_dequeue_tail(skbs);
> - *total_ts -= skb->truesize;
> - __kfree_skb(skb);
> - }
> - return skbs->qlen > 0;
> - }
> -
> - *total_ts += skb->truesize;
> - __skb_queue_tail(skbs, skb);
> - space_needed -= msk->size_goal_cache;
> - }
> - return true;
> -}
> -
> static bool __mptcp_alloc_tx_skb(struct sock *sk, struct sock *ssk, gfp_t gfp)
> {
> - struct mptcp_sock *msk = mptcp_sk(sk);
> struct sk_buff *skb;
>
> if (ssk->sk_tx_skb_cache) {
> @@ -1266,22 +1217,6 @@ static bool __mptcp_alloc_tx_skb(struct sock *sk, struct sock *ssk, gfp_t gfp)
> return true;
> }
>
> - skb = skb_peek(&msk->skb_tx_cache);
> - if (skb) {
> - if (likely(sk_wmem_schedule(ssk, skb->truesize))) {
> - skb = __skb_dequeue(&msk->skb_tx_cache);
> - if (WARN_ON_ONCE(!skb))
> - return false;
> -
> - mptcp_wmem_uncharge(sk, skb->truesize);
> - ssk->sk_tx_skb_cache = skb;
> - return true;
> - }
> -
> - /* over memory limit, no point to try to allocate a new skb */
> - return false;
> - }
> -
> skb = __mptcp_do_alloc_tx_skb(sk, gfp);
> if (!skb)
> return false;
> @@ -1297,7 +1232,6 @@ static bool __mptcp_alloc_tx_skb(struct sock *sk, struct sock *ssk, gfp_t gfp)
> static bool mptcp_must_reclaim_memory(struct sock *sk, struct sock *ssk)
> {
> return !ssk->sk_tx_skb_cache &&
> - !skb_peek(&mptcp_sk(sk)->skb_tx_cache) &&
> tcp_under_memory_pressure(sk);
> }
>
> @@ -1340,7 +1274,6 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk,
> /* compute send limit */
> info->mss_now = tcp_send_mss(ssk, &info->size_goal, info->flags);
> avail_size = info->size_goal;
> - msk->size_goal_cache = info->size_goal;
> skb = tcp_write_queue_tail(ssk);
> if (skb) {
> /* Limit the write to the size available in the
> @@ -1689,7 +1622,6 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
> while (msg_data_left(msg)) {
> int total_ts, frag_truesize = 0;
> struct mptcp_data_frag *dfrag;
> - struct sk_buff_head skbs;
> bool dfrag_collapsed;
> size_t psize, offset;
>
> @@ -1722,16 +1654,10 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
> psize = pfrag->size - offset;
> psize = min_t(size_t, psize, msg_data_left(msg));
> total_ts = psize + frag_truesize;
> - __skb_queue_head_init(&skbs);
> - if (!mptcp_tx_cache_refill(sk, psize, &skbs, &total_ts))
> - goto wait_for_memory;
>
> - if (!mptcp_wmem_alloc(sk, total_ts)) {
> - __skb_queue_purge(&skbs);
> + if (!mptcp_wmem_alloc(sk, total_ts))
> goto wait_for_memory;
> - }
>
> - skb_queue_splice_tail(&skbs, &msk->skb_tx_cache);
> if (copy_page_from_iter(dfrag->page, offset, psize,
> &msg->msg_iter) != psize) {
> mptcp_wmem_uncharge(sk, psize + frag_truesize);
> @@ -2460,13 +2386,11 @@ static int __mptcp_init_sock(struct sock *sk)
> INIT_LIST_HEAD(&msk->rtx_queue);
> INIT_WORK(&msk->work, mptcp_worker);
> __skb_queue_head_init(&msk->receive_queue);
> - __skb_queue_head_init(&msk->skb_tx_cache);
whoops, I forgot to remove the 'skb_tx_cache' field from the msk.
v2 is coming...
/P
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2021-05-21 16:19 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-05-21 10:30 [PATCH mptcp-next] mptcp: drop tx skb cache Paolo Abeni
2021-05-21 16:19 ` Paolo Abeni
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).