=== modified file 'drivers/net/macvtap.c' --- old/drivers/net/macvtap.c 2012-12-17 19:41:04 +0000 +++ new/drivers/net/macvtap.c 2012-12-17 23:03:50 +0000 @@ -536,7 +536,7 @@ static int zerocopy_sg_from_iovec(struct num_pages = get_user_pages_fast(base, size, 0, &page[i]); if (num_pages != size) { for (i = 0; i < num_pages; i++) - put_page(page[i]); + net_put_page(page[i]); return -EFAULT; } truesize = size * PAGE_SIZE; === modified file 'drivers/net/tun.c' --- old/drivers/net/tun.c 2012-12-17 19:41:04 +0000 +++ new/drivers/net/tun.c 2012-12-17 23:03:50 +0000 @@ -660,7 +660,7 @@ static int zerocopy_sg_from_iovec(struct num_pages = get_user_pages_fast(base, size, 0, &page[i]); if (num_pages != size) { for (i = 0; i < num_pages; i++) - put_page(page[i]); + net_put_page(page[i]); return -EFAULT; } truesize = size * PAGE_SIZE; === modified file 'drivers/net/vmxnet3/vmxnet3_drv.c' --- old/drivers/net/vmxnet3/vmxnet3_drv.c 2012-12-17 19:41:04 +0000 +++ new/drivers/net/vmxnet3/vmxnet3_drv.c 2012-12-17 23:03:50 +0000 @@ -1370,7 +1370,7 @@ vmxnet3_rq_cleanup(struct vmxnet3_rx_que rq->buf_info[ring_idx][i].page) { pci_unmap_page(adapter->pdev, rxd->addr, rxd->len, PCI_DMA_FROMDEVICE); - put_page(rq->buf_info[ring_idx][i].page); + net_put_page(rq->buf_info[ring_idx][i].page); rq->buf_info[ring_idx][i].page = NULL; } } === modified file 'drivers/net/xen-netback/netback.c' --- old/drivers/net/xen-netback/netback.c 2012-12-17 19:41:04 +0000 +++ new/drivers/net/xen-netback/netback.c 2012-12-17 23:03:50 +0000 @@ -1081,7 +1081,7 @@ static void xen_netbk_fill_frags(struct skb->truesize += txp->size; /* Take an extra reference to offset xen_netbk_idx_release */ - get_page(netbk->mmap_pages[pending_idx]); + net_get_page(netbk->mmap_pages[pending_idx]); xen_netbk_idx_release(netbk, pending_idx); } } @@ -1522,7 +1522,7 @@ static void xen_netbk_idx_release(struct xenvif_put(vif); netbk->mmap_pages[pending_idx]->mapping = 0; - put_page(netbk->mmap_pages[pending_idx]); + net_put_page(netbk->mmap_pages[pending_idx]); netbk->mmap_pages[pending_idx] = NULL; } === modified file 'include/linux/mm_types.h' --- old/include/linux/mm_types.h 2012-12-17 19:41:04 +0000 +++ new/include/linux/mm_types.h 2012-12-17 23:03:50 +0000 @@ -175,6 +175,17 @@ struct page { */ void *shadow; #endif + +#if defined(CONFIG_TCP_ZERO_COPY_TRANSFER_COMPLETION_NOTIFICATION) + /* + * Used to implement support for notification on zero-copy TCP transfer + * completion. It might look as not good to have this field here and + * it's better to have it in struct sk_buff, but it would make the code + * much more complicated and fragile, since all skb then would have to + * contain only pages with the same value in this field. + */ + void *net_priv; +#endif } /* * The struct page can be forced to be double word aligned so that atomic ops === modified file 'include/linux/net.h' --- old/include/linux/net.h 2012-12-17 19:41:04 +0000 +++ new/include/linux/net.h 2012-12-17 23:03:50 +0000 @@ -19,6 +19,7 @@ #define _LINUX_NET_H #include +#include #include #include #include /* For O_CLOEXEC and O_NONBLOCK */ @@ -270,6 +271,45 @@ extern int kernel_sock_ioctl(struct sock extern int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how); +#if defined(CONFIG_TCP_ZERO_COPY_TRANSFER_COMPLETION_NOTIFICATION) +/* Support for notification on zero-copy TCP transfer completion */ +typedef void (*net_get_page_callback_t)(struct page *page); +typedef void (*net_put_page_callback_t)(struct page *page); + +extern net_get_page_callback_t net_get_page_callback; +extern net_put_page_callback_t net_put_page_callback; + +extern int net_set_get_put_page_callbacks( + net_get_page_callback_t get_callback, + net_put_page_callback_t put_callback); + +/* + * See comment for net_set_get_put_page_callbacks() why those functions + * don't need any protection. + */ +static inline void net_get_page(struct page *page) +{ + if (page->net_priv != 0) + net_get_page_callback(page); + get_page(page); +} +static inline void net_put_page(struct page *page) +{ + if (page->net_priv != 0) + net_put_page_callback(page); + put_page(page); +} +#else +static inline void net_get_page(struct page *page) +{ + get_page(page); +} +static inline void net_put_page(struct page *page) +{ + put_page(page); +} +#endif /* CONFIG_TCP_ZERO_COPY_TRANSFER_COMPLETION_NOTIFICATION */ + #define MODULE_ALIAS_NETPROTO(proto) \ MODULE_ALIAS("net-pf-" __stringify(proto)) === modified file 'include/linux/skbuff.h' --- old/include/linux/skbuff.h 2012-12-17 19:41:04 +0000 +++ new/include/linux/skbuff.h 2012-12-17 23:03:50 +0000 @@ -1848,7 +1848,7 @@ static inline struct page *skb_frag_page */ static inline void __skb_frag_ref(skb_frag_t *frag) { - get_page(skb_frag_page(frag)); + net_get_page(skb_frag_page(frag)); } /** @@ -1871,7 +1871,7 @@ static inline void skb_frag_ref(struct s */ static inline void __skb_frag_unref(skb_frag_t *frag) { - put_page(skb_frag_page(frag)); + net_put_page(skb_frag_page(frag)); } /** === modified file 'net/Kconfig' --- old/net/Kconfig 2012-12-17 19:41:04 +0000 +++ new/net/Kconfig 2012-12-17 23:03:50 +0000 @@ -74,6 +74,18 @@ config INET Short answer: say Y. +config TCP_ZERO_COPY_TRANSFER_COMPLETION_NOTIFICATION + bool "TCP/IP zero-copy transfer completion notification" + depends on INET + default SCST_ISCSI + ---help--- + Adds support for sending a notification upon completion of a + zero-copy TCP/IP transfer. This can speed up certain TCP/IP + software. Currently this is only used by the iSCSI target driver + iSCSI-SCST. + + If unsure, say N. + if INET source "net/ipv4/Kconfig" source "net/ipv6/Kconfig" === modified file 'net/core/skbuff.c' --- old/net/core/skbuff.c 2012-12-17 19:41:04 +0000 +++ new/net/core/skbuff.c 2012-12-17 23:03:50 +0000 @@ -77,13 +77,13 @@ static struct kmem_cache *skbuff_fclone_ static void sock_pipe_buf_release(struct pipe_inode_info *pipe, struct pipe_buffer *buf) { - put_page(buf->page); + net_put_page(buf->page); } static void sock_pipe_buf_get(struct pipe_inode_info *pipe, struct pipe_buffer *buf) { - get_page(buf->page); + net_get_page(buf->page); } static int sock_pipe_buf_steal(struct pipe_inode_info *pipe, @@ -441,7 +441,7 @@ struct sk_buff *__netdev_alloc_skb(struc if (likely(data)) { skb = build_skb(data, fragsz); if (unlikely(!skb)) - put_page(virt_to_head_page(data)); + net_put_page(virt_to_head_page(data)); } } else { skb = __alloc_skb(length + NET_SKB_PAD, gfp_mask, @@ -494,7 +494,7 @@ static void skb_clone_fraglist(struct sk static void skb_free_head(struct sk_buff *skb) { if (skb->head_frag) - put_page(virt_to_head_page(skb->head)); + net_put_page(virt_to_head_page(skb->head)); else kfree(skb->head); } @@ -780,7 +780,7 @@ int skb_copy_ubufs(struct sk_buff *skb, if (!page) { while (head) { struct page *next = (struct page *)head->private; - put_page(head); + net_put_page(head); head = next; } return -ENOMEM; @@ -1615,7 +1615,7 @@ EXPORT_SYMBOL(skb_copy_bits); */ static void sock_spd_release(struct splice_pipe_desc *spd, unsigned int i) { - put_page(spd->pages[i]); + net_put_page(spd->pages[i]); } static struct page *linear_to_page(struct page *page, unsigned int *len, @@ -1668,7 +1668,7 @@ static bool spd_fill_page(struct splice_ spd->partial[spd->nr_pages - 1].len += *len; return false; } - get_page(page); + net_get_page(page); spd->pages[spd->nr_pages] = page; spd->partial[spd->nr_pages].len = *len; spd->partial[spd->nr_pages].offset = offset; === modified file 'net/core/sock.c' --- old/net/core/sock.c 2012-12-17 19:41:04 +0000 +++ new/net/core/sock.c 2012-12-17 23:03:50 +0000 @@ -1739,7 +1739,7 @@ bool sk_page_frag_refill(struct sock *sk } if (pfrag->offset < pfrag->size) return true; - put_page(pfrag->page); + net_put_page(pfrag->page); } /* We restrict high order allocations to users that can afford to wait */ @@ -2439,7 +2439,7 @@ void sk_common_release(struct sock *sk) sk_refcnt_debug_release(sk); if (sk->sk_frag.page) { - put_page(sk->sk_frag.page); + net_put_page(sk->sk_frag.page); sk->sk_frag.page = NULL; } === modified file 'net/ipv4/Makefile' --- old/net/ipv4/Makefile 2012-12-17 19:41:04 +0000 +++ new/net/ipv4/Makefile 2012-12-17 23:03:50 +0000 @@ -51,6 +51,7 @@ obj-$(CONFIG_TCP_CONG_YEAH) += tcp_yeah. obj-$(CONFIG_TCP_CONG_ILLINOIS) += tcp_illinois.o obj-$(CONFIG_MEMCG_KMEM) += tcp_memcontrol.o obj-$(CONFIG_NETLABEL) += cipso_ipv4.o +obj-$(CONFIG_TCP_ZERO_COPY_TRANSFER_COMPLETION_NOTIFICATION) += tcp_zero_copy.o obj-$(CONFIG_XFRM) += xfrm4_policy.o xfrm4_state.o xfrm4_input.o \ xfrm4_output.o === modified file 'net/ipv4/ip_output.c' --- old/net/ipv4/ip_output.c 2012-12-17 19:41:04 +0000 +++ new/net/ipv4/ip_output.c 2012-12-17 23:03:50 +0000 @@ -1002,7 +1002,7 @@ alloc_new_skb: __skb_fill_page_desc(skb, i, pfrag->page, pfrag->offset, 0); skb_shinfo(skb)->nr_frags = ++i; - get_page(pfrag->page); + net_get_page(pfrag->page); } copy = min_t(int, copy, pfrag->size - pfrag->offset); if (getfrag(from, @@ -1223,7 +1223,7 @@ ssize_t ip_append_page(struct sock *sk, if (skb_can_coalesce(skb, i, page, offset)) { skb_frag_size_add(&skb_shinfo(skb)->frags[i-1], len); } else if (i < MAX_SKB_FRAGS) { - get_page(page); + net_get_page(page); skb_fill_page_desc(skb, i, page, offset, len); } else { err = -EMSGSIZE; === modified file 'net/ipv4/tcp.c' --- old/net/ipv4/tcp.c 2012-12-17 19:41:04 +0000 +++ new/net/ipv4/tcp.c 2012-12-17 23:03:50 +0000 @@ -891,7 +891,7 @@ new_segment: if (can_coalesce) { skb_frag_size_add(&skb_shinfo(skb)->frags[i - 1], copy); } else { - get_page(page); + net_get_page(page); skb_fill_page_desc(skb, i, page, offset, copy); } @@ -1179,7 +1179,7 @@ new_segment: } else { skb_fill_page_desc(skb, i, pfrag->page, pfrag->offset, copy); - get_page(pfrag->page); + net_get_page(pfrag->page); } pfrag->offset += copy; } === added file 'net/ipv4/tcp_zero_copy.c' --- old/net/ipv4/tcp_zero_copy.c 1970-01-01 00:00:00 +0000 +++ new/net/ipv4/tcp_zero_copy.c 2012-12-17 23:03:50 +0000 @@ -0,0 +1,50 @@ +/* + * Support routines for TCP zero copy transmit + * + * Created by Vladislav Bolkhovitin + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * version 2 as published by the Free Software Foundation. + */ + +#include +#include + +net_get_page_callback_t net_get_page_callback __read_mostly; +EXPORT_SYMBOL_GPL(net_get_page_callback); + +net_put_page_callback_t net_put_page_callback __read_mostly; +EXPORT_SYMBOL_GPL(net_put_page_callback); + +/* + * Caller of this function must ensure that at the moment when it's called + * there are no pages in the system with net_priv field set to non-zero + * value. Hence, this function, as well as net_get_page() and net_put_page(), + * don't need any protection. + */ +int net_set_get_put_page_callbacks( + net_get_page_callback_t get_callback, + net_put_page_callback_t put_callback) +{ + int res = 0; + + if ((net_get_page_callback != NULL) && (get_callback != NULL) && + (net_get_page_callback != get_callback)) { + res = -EBUSY; + goto out; + } + + if ((net_put_page_callback != NULL) && (put_callback != NULL) && + (net_put_page_callback != put_callback)) { + res = -EBUSY; + goto out; + } + + net_get_page_callback = get_callback; + net_put_page_callback = put_callback; + +out: + return res; +} +EXPORT_SYMBOL_GPL(net_set_get_put_page_callbacks); === modified file 'net/ipv6/ip6_output.c' --- old/net/ipv6/ip6_output.c 2012-12-17 19:41:04 +0000 +++ new/net/ipv6/ip6_output.c 2012-12-17 23:03:50 +0000 @@ -1517,7 +1517,7 @@ alloc_new_skb: __skb_fill_page_desc(skb, i, pfrag->page, pfrag->offset, 0); skb_shinfo(skb)->nr_frags = ++i; - get_page(pfrag->page); + net_get_page(pfrag->page); } copy = min_t(int, copy, pfrag->size - pfrag->offset); if (getfrag(from,