Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit aa133076 authored by Wu Fengguang's avatar Wu Fengguang Committed by David S. Miller
Browse files

tcp: replace hard coded GFP_KERNEL with sk_allocation 



This fixed a lockdep warning which appeared when doing stress
memory tests over NFS:

	inconsistent {RECLAIM_FS-ON-W} -> {IN-RECLAIM_FS-W} usage.

	page reclaim => nfs_writepage => tcp_sendmsg => lock sk_lock

	mount_root => nfs_root_data => tcp_close => lock sk_lock =>
			tcp_send_fin => alloc_skb_fclone => page reclaim

David raised a concern that if the allocation fails in tcp_send_fin(), and it's
GFP_ATOMIC, we are going to yield() (which sleeps) and loop endlessly waiting
for the allocation to succeed.

But fact is, the original GFP_KERNEL also sleeps. GFP_ATOMIC+yield() looks
weird, but it is no worse the implicit sleep inside GFP_KERNEL. Both could
loop endlessly under memory pressure.

CC: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
CC: David S. Miller <davem@davemloft.net>
CC: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: default avatarWu Fengguang <fengguang.wu@intel.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent 05c6a8d7

include/net/tcp.h

+1 −1
Original line number Diff line number Diff line
@@ -1186,7 +1186,7 @@ extern int tcp_v4_md5_do_del(struct sock *sk, 
#define tcp_twsk_md5_key(twsk)	NULL

#endif



extern struct tcp_md5sig_pool	**tcp_alloc_md5sig_pool(void);

extern struct tcp_md5sig_pool	**tcp_alloc_md5sig_pool(struct sock *);

extern void			tcp_free_md5sig_pool(void);



extern struct tcp_md5sig_pool	*__tcp_get_md5sig_pool(int cpu);

net/ipv4/tcp.c

+5 −5
Original line number Diff line number Diff line
@@ -1839,7 +1839,7 @@ void tcp_close(struct sock *sk, long timeout) 
		/* Unread data was tossed, zap the connection. */

		NET_INC_STATS_USER(sock_net(sk), LINUX_MIB_TCPABORTONCLOSE);

		tcp_set_state(sk, TCP_CLOSE);

		tcp_send_active_reset(sk, GFP_KERNEL);

		tcp_send_active_reset(sk, sk->sk_allocation);

	} else if (sock_flag(sk, SOCK_LINGER) && !sk->sk_lingertime) {

		/* Check zero linger _after_ checking for unread data. */

		sk->sk_prot->disconnect(sk, 0);
@@ -2658,7 +2658,7 @@ void tcp_free_md5sig_pool(void) 


EXPORT_SYMBOL(tcp_free_md5sig_pool);



static struct tcp_md5sig_pool **__tcp_alloc_md5sig_pool(void)

static struct tcp_md5sig_pool **__tcp_alloc_md5sig_pool(struct sock *sk)

{

	int cpu;

	struct tcp_md5sig_pool **pool;
@@ -2671,7 +2671,7 @@ static struct tcp_md5sig_pool **__tcp_alloc_md5sig_pool(void) 
		struct tcp_md5sig_pool *p;

		struct crypto_hash *hash;



		p = kzalloc(sizeof(*p), GFP_KERNEL);

		p = kzalloc(sizeof(*p), sk->sk_allocation);

		if (!p)

			goto out_free;

		*per_cpu_ptr(pool, cpu) = p;
@@ -2688,7 +2688,7 @@ static struct tcp_md5sig_pool **__tcp_alloc_md5sig_pool(void) 
	return NULL;

}



struct tcp_md5sig_pool **tcp_alloc_md5sig_pool(void)

struct tcp_md5sig_pool **tcp_alloc_md5sig_pool(struct sock *sk)

{

	struct tcp_md5sig_pool **pool;

	int alloc = 0;
@@ -2709,7 +2709,7 @@ struct tcp_md5sig_pool **tcp_alloc_md5sig_pool(void) 


	if (alloc) {

		/* we cannot hold spinlock here because this may sleep. */

		struct tcp_md5sig_pool **p = __tcp_alloc_md5sig_pool();

		struct tcp_md5sig_pool **p = __tcp_alloc_md5sig_pool(sk);

		spin_lock_bh(&tcp_md5sig_pool_lock);

		if (!p) {

			tcp_md5sig_users--;

net/ipv4/tcp_ipv4.c

+4 −3
Original line number Diff line number Diff line
@@ -886,7 +886,7 @@ int tcp_v4_md5_do_add(struct sock *sk, __be32 addr, 
			}

			sk->sk_route_caps &= ~NETIF_F_GSO_MASK;

		}

		if (tcp_alloc_md5sig_pool() == NULL) {

		if (tcp_alloc_md5sig_pool(sk) == NULL) {

			kfree(newkey);

			return -ENOMEM;

		}
@@ -1007,8 +1007,9 @@ static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval, 


	if (!tcp_sk(sk)->md5sig_info) {

		struct tcp_sock *tp = tcp_sk(sk);

		struct tcp_md5sig_info *p = kzalloc(sizeof(*p), GFP_KERNEL);

		struct tcp_md5sig_info *p;



		p = kzalloc(sizeof(*p), sk->sk_allocation);

		if (!p)

			return -EINVAL;
@@ -1016,7 +1017,7 @@ static int tcp_v4_parse_md5_keys(struct sock *sk, char __user *optval, 
		sk->sk_route_caps &= ~NETIF_F_GSO_MASK;

	}



	newkey = kmemdup(cmd.tcpm_key, cmd.tcpm_keylen, GFP_KERNEL);

	newkey = kmemdup(cmd.tcpm_key, cmd.tcpm_keylen, sk->sk_allocation);

	if (!newkey)

		return -ENOMEM;

	return tcp_v4_md5_do_add(sk, sin->sin_addr.s_addr,

net/ipv4/tcp_minisocks.c

+1 −1
Original line number Diff line number Diff line
@@ -322,7 +322,7 @@ void tcp_time_wait(struct sock *sk, int state, int timeo) 
			if (key != NULL) {

				memcpy(&tcptw->tw_md5_key, key->key, key->keylen);

				tcptw->tw_md5_keylen = key->keylen;

				if (tcp_alloc_md5sig_pool() == NULL)

				if (tcp_alloc_md5sig_pool(sk) == NULL)

					BUG();

			}

		} while (0);

net/ipv4/tcp_output.c

+3 −2
Original line number Diff line number Diff line
@@ -2135,7 +2135,8 @@ void tcp_send_fin(struct sock *sk) 
	} else {

		/* Socket is locked, keep trying until memory is available. */

		for (;;) {

			skb = alloc_skb_fclone(MAX_TCP_HEADER, GFP_KERNEL);

			skb = alloc_skb_fclone(MAX_TCP_HEADER,

					       sk->sk_allocation);

			if (skb)

				break;

			yield();
@@ -2388,7 +2389,7 @@ int tcp_connect(struct sock *sk) 
	sk->sk_wmem_queued += buff->truesize;

	sk_mem_charge(sk, buff->truesize);

	tp->packets_out += tcp_skb_pcount(buff);

	tcp_transmit_skb(sk, buff, 1, GFP_KERNEL);

	tcp_transmit_skb(sk, buff, 1, sk->sk_allocation);



	/* We change tp->snd_nxt after the tcp_transmit_skb() call

	 * in order to make this packet get counted in tcpOutSegs.