- Kernel - mailweb.openeuler.org

[PATCH OLK-6.6 0/1] arm64: Add non nmi ipi backtrace support
by Liao Chen 22 Dec '23

22 Dec '23

*** BLURB HERE *** Li Zhengyu (1): arm64: Add non nmi ipi backtrace support arch/arm64/kernel/ipi_nmi.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-) -- 2.34.1

2 2

[PATCH OLK-6.6] tcp_comp: implement tcp compression
by Lu Wei 22 Dec '23

22 Dec '23

hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I4PNEK CVE: NA ------------------------------------------------- Signed-off-by: Lu Wei <luwei32(a)huawei.com> --- arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + include/linux/tcp.h | 6 +- include/net/inet_sock.h | 3 +- include/net/sock.h | 1 + include/net/tcp.h | 40 ++ net/ipv4/Kconfig | 10 + net/ipv4/Makefile | 1 + net/ipv4/syncookies.c | 2 + net/ipv4/sysctl_net_ipv4.c | 42 ++ net/ipv4/tcp.c | 5 + net/ipv4/tcp_comp.c | 912 +++++++++++++++++++++++++ net/ipv4/tcp_input.c | 26 + net/ipv4/tcp_ipv4.c | 2 + net/ipv4/tcp_minisocks.c | 3 + net/ipv4/tcp_output.c | 52 ++ net/ipv6/syncookies.c | 2 + 17 files changed, 1107 insertions(+), 2 deletions(-) create mode 100644 net/ipv4/tcp_comp.c diff --git a/arch/arm64/configs/openeuler_defconfig b/arch/arm64/configs/openeuler_defconfig index 33ba39711884..abc33d2c29a6 100644 --- a/arch/arm64/configs/openeuler_defconfig +++ b/arch/arm64/configs/openeuler_defconfig @@ -1222,6 +1222,7 @@ CONFIG_DEFAULT_CUBIC=y # CONFIG_DEFAULT_RENO is not set CONFIG_DEFAULT_TCP_CONG="cubic" CONFIG_TCP_MD5SIG=y +CONFIG_TCP_COMP=y CONFIG_IPV6=y CONFIG_IPV6_ROUTER_PREF=y CONFIG_IPV6_ROUTE_INFO=y diff --git a/arch/x86/configs/openeuler_defconfig b/arch/x86/configs/openeuler_defconfig index 44040b835333..177a6fdcce58 100644 --- a/arch/x86/configs/openeuler_defconfig +++ b/arch/x86/configs/openeuler_defconfig @@ -1244,6 +1244,7 @@ CONFIG_DEFAULT_CUBIC=y # CONFIG_DEFAULT_RENO is not set CONFIG_DEFAULT_TCP_CONG="cubic" CONFIG_TCP_MD5SIG=y +CONFIG_TCP_COMP=y CONFIG_IPV6=y CONFIG_IPV6_ROUTER_PREF=y CONFIG_IPV6_ROUTE_INFO=y diff --git a/include/linux/tcp.h b/include/linux/tcp.h index 3c5efeeb024f..37ae82ab7795 100644 --- a/include/linux/tcp.h +++ b/include/linux/tcp.h @@ -123,7 +123,8 @@ struct tcp_options_received { snd_wscale : 4, /* Window scaling received from sender */ rcv_wscale : 4; /* Window scaling to send to receiver */ u8 saw_unknown:1, /* Received unknown option */ - unused:7; + comp_ok:1, /* COMP seen on SYN packet */ + unused:6; u8 num_sacks; /* Number of SACK blocks */ u16 user_mss; /* mss requested by user in ioctl */ u16 mss_clamp; /* Maximal mss, negotiated at connection setup */ @@ -136,6 +137,9 @@ static inline void tcp_clear_options(struct tcp_options_received *rx_opt) #if IS_ENABLED(CONFIG_SMC) rx_opt->smc_ok = 0; #endif +#if IS_ENABLED(CONFIG_TCP_COMP) + rx_opt->comp_ok = 0; +#endif } /* This is the max number of SACKS that we'll generate and process. It's safe diff --git a/include/net/inet_sock.h b/include/net/inet_sock.h index 2de0e4d4a027..bb09ac95cc9d 100644 --- a/include/net/inet_sock.h +++ b/include/net/inet_sock.h @@ -87,7 +87,8 @@ struct inet_request_sock { ecn_ok : 1, acked : 1, no_srccheck: 1, - smc_ok : 1; + smc_ok : 1, + comp_ok : 1; u32 ir_mark; union { struct ip_options_rcu __rcu *ireq_opt; diff --git a/include/net/sock.h b/include/net/sock.h index 7753354d59c0..c86845136ec5 100644 --- a/include/net/sock.h +++ b/include/net/sock.h @@ -959,6 +959,7 @@ enum sock_flags { SOCK_XDP, /* XDP is attached */ SOCK_TSTAMP_NEW, /* Indicates 64 bit timestamps always */ SOCK_RCVMARK, /* Receive SO_MARK ancillary data with packet */ + SOCK_COMP, }; #define SK_FLAGS_TIMESTAMP ((1UL << SOCK_TIMESTAMP) | (1UL << SOCK_TIMESTAMPING_RX_SOFTWARE)) diff --git a/include/net/tcp.h b/include/net/tcp.h index e9d387fffe22..cb33a2c46b2f 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -195,6 +195,7 @@ void tcp_time_wait(struct sock *sk, int state, int timeo); */ #define TCPOPT_FASTOPEN_MAGIC 0xF989 #define TCPOPT_SMC_MAGIC 0xE2D4C3D9 +#define TCPOPT_COMP_MAGIC 0x7954 /* * TCP option lengths @@ -208,6 +209,7 @@ void tcp_time_wait(struct sock *sk, int state, int timeo); #define TCPOLEN_FASTOPEN_BASE 2 #define TCPOLEN_EXP_FASTOPEN_BASE 4 #define TCPOLEN_EXP_SMC_BASE 6 +#define TCPOLEN_EXP_COMP_BASE 4 /* But this is what stacks really send out. */ #define TCPOLEN_TSTAMP_ALIGNED 12 @@ -2531,4 +2533,42 @@ static inline u64 tcp_transmit_time(const struct sock *sk) return 0; } +#if IS_ENABLED(CONFIG_TCP_COMP) +extern struct static_key_false tcp_have_comp; + +extern unsigned long *sysctl_tcp_compression_ports; +extern int sysctl_tcp_compression_local; + +bool tcp_syn_comp_enabled(struct sock *sk); +bool tcp_synack_comp_enabled(struct sock *sk, + const struct inet_request_sock *ireq); +void tcp_init_compression(struct sock *sk); +void tcp_cleanup_compression(struct sock *sk); +int tcp_comp_init(void); +#else +static inline bool tcp_syn_comp_enabled(struct tcp_sock *tp) +{ + return false; +} + +static inline bool tcp_synack_comp_enabled(struct sock *sk, + const struct inet_request_sock *ireq) +{ + return false; +} + +static inline void tcp_init_compression(struct sock *sk) +{ +} + +static inline void tcp_cleanup_compression(struct sock *sk) +{ +} + +static inline int tcp_comp_init(void) +{ + return 0; +} +#endif + #endif /* _TCP_H */ diff --git a/net/ipv4/Kconfig b/net/ipv4/Kconfig index 2dfb12230f08..a4405ea38338 100644 --- a/net/ipv4/Kconfig +++ b/net/ipv4/Kconfig @@ -751,3 +751,13 @@ config TCP_MD5SIG on the Internet. If unsure, say N. + +config TCP_COMP + bool "TCP: Transport Layer Compression support" + depends on CRYPTO_ZSTD=y + select STREAM_PARSER + help + Enable kernel payload compression support for TCP protocol. This allows + payload compression handling of the TCP protocol to be done in-kernel. + + If unsure, say Y. diff --git a/net/ipv4/Makefile b/net/ipv4/Makefile index b18ba8ef93ad..aa458d9f534a 100644 --- a/net/ipv4/Makefile +++ b/net/ipv4/Makefile @@ -65,6 +65,7 @@ obj-$(CONFIG_TCP_CONG_ILLINOIS) += tcp_illinois.o obj-$(CONFIG_NET_SOCK_MSG) += tcp_bpf.o obj-$(CONFIG_BPF_SYSCALL) += udp_bpf.o obj-$(CONFIG_NETLABEL) += cipso_ipv4.o +obj-$(CONFIG_TCP_COMP) += tcp_comp.o obj-$(CONFIG_XFRM) += xfrm4_policy.o xfrm4_state.o xfrm4_input.o \ xfrm4_output.o xfrm4_protocol.o diff --git a/net/ipv4/syncookies.c b/net/ipv4/syncookies.c index 3b4dafefb4b0..42c577bfc26b 100644 --- a/net/ipv4/syncookies.c +++ b/net/ipv4/syncookies.c @@ -390,6 +390,8 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb) if (IS_ENABLED(CONFIG_SMC)) ireq->smc_ok = 0; + if (IS_ENABLED(CONFIG_TCP_COMP)) + ireq->comp_ok = 0; ireq->ir_iif = inet_request_bound_dev_if(sk, skb); diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index b17eb28a9690..f212133b05e9 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -469,6 +469,30 @@ static int proc_fib_multipath_hash_fields(struct ctl_table *table, int write, } #endif +#if IS_ENABLED(CONFIG_TCP_COMP) +static int proc_tcp_compression_ports(struct ctl_table *table, int write, + void __user *buffer, size_t *lenp, + loff_t *ppos) +{ + unsigned long *bitmap = *(unsigned long **)table->data; + unsigned long bitmap_len = table->maxlen; + int ret; + + ret = proc_do_large_bitmap(table, write, buffer, lenp, ppos); + if (write && ret == 0) { + if (bitmap_empty(bitmap, bitmap_len)) { + if (static_key_enabled(&tcp_have_comp)) + static_branch_disable(&tcp_have_comp); + } else { + if (!static_key_enabled(&tcp_have_comp)) + static_branch_enable(&tcp_have_comp); + } + } + + return ret; +} +#endif + static struct ctl_table ipv4_table[] = { { .procname = "tcp_max_orphans", @@ -587,6 +611,24 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_dointvec, }, +#if IS_ENABLED(CONFIG_TCP_COMP) + { + .procname = "tcp_compression_ports", + .data = &sysctl_tcp_compression_ports, + .maxlen = 65536, + .mode = 0644, + .proc_handler = proc_tcp_compression_ports, + }, + { + .procname = "tcp_compression_local", + .data = &sysctl_tcp_compression_local, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = SYSCTL_ZERO, + .extra2 = SYSCTL_ONE, + }, +#endif { } }; diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index 3d3a24f79573..2703be8a7316 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -302,6 +302,10 @@ DEFINE_STATIC_KEY_FALSE(tcp_have_smc); EXPORT_SYMBOL(tcp_have_smc); #endif +#if IS_ENABLED(CONFIG_TCP_COMP) +DEFINE_STATIC_KEY_FALSE(tcp_have_comp); +#endif + /* * Current number of TCP sockets. */ @@ -4707,5 +4711,6 @@ void __init tcp_init(void) tcp_metrics_init(); BUG_ON(tcp_register_congestion_control(&tcp_reno) != 0); tcp_tasklet_init(); + tcp_comp_init(); mptcp_init(); } diff --git a/net/ipv4/tcp_comp.c b/net/ipv4/tcp_comp.c new file mode 100644 index 000000000000..53157a413b58 --- /dev/null +++ b/net/ipv4/tcp_comp.c @@ -0,0 +1,912 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * TCP compression support + * + * Copyright(c) 2021 Huawei Technologies Co., Ltd + */ + +#include <linux/skmsg.h> +#include <linux/zstd.h> + +#define TCP_COMP_MAX_PADDING 64 +#define TCP_COMP_DATA_SIZE 65536 +#define TCP_COMP_SCRATCH_SIZE (TCP_COMP_DATA_SIZE - 1) +#define TCP_COMP_MAX_CSIZE (TCP_COMP_SCRATCH_SIZE + TCP_COMP_MAX_PADDING) +#define TCP_COMP_ALLOC_ORDER get_order(TCP_COMP_DATA_SIZE) +#define TCP_COMP_MAX_WINDOWLOG 17 +#define TCP_COMP_MAX_INPUT (1 << TCP_COMP_MAX_WINDOWLOG) + +#define TCP_COMP_SEND_PENDING 1 +#define ZSTD_COMP_DEFAULT_LEVEL 1 + +static unsigned long tcp_compression_ports[65536 / 8]; + +unsigned long *sysctl_tcp_compression_ports = tcp_compression_ports; +int sysctl_tcp_compression_local __read_mostly; + +static struct proto tcp_prot_override; + +struct tcp_comp_context_tx { + ZSTD_CStream *cstream; + void *cworkspace; + void *plaintext_data; + void *compressed_data; + struct sk_msg msg; + bool in_tcp_sendpages; +}; + +struct tcp_comp_context_rx { + ZSTD_DStream *dstream; + void *dworkspace; + void *plaintext_data; + + struct strparser strp; + void (*saved_data_ready)(struct sock *sk); + struct sk_buff *pkt; + struct sk_buff *dpkt; +}; + +struct tcp_comp_context { + struct rcu_head rcu; + + struct proto *sk_proto; + void (*sk_write_space)(struct sock *sk); + + struct tcp_comp_context_tx tx; + struct tcp_comp_context_rx rx; + + unsigned long flags; +}; + +static bool tcp_comp_is_write_pending(struct tcp_comp_context *ctx) +{ + return test_bit(TCP_COMP_SEND_PENDING, &ctx->flags); +} + +static void tcp_comp_err_abort(struct sock *sk, int err) +{ + sk->sk_err = err; + sk->sk_error_report(sk); +} + +static bool tcp_comp_enabled(__be32 saddr, __be32 daddr, int port) +{ + if (!sysctl_tcp_compression_local && + (saddr == daddr || ipv4_is_loopback(daddr))) + return false; + + return test_bit(port, sysctl_tcp_compression_ports); +} + +bool tcp_syn_comp_enabled(struct sock *sk) +{ + struct inet_sock *inet = inet_sk(sk); + + return tcp_comp_enabled(inet->inet_saddr, inet->inet_daddr, + ntohs(inet->inet_dport)); +} + +bool tcp_synack_comp_enabled(struct sock *sk, + const struct inet_request_sock *ireq) +{ + struct inet_sock *inet = inet_sk(sk); + + if (!ireq->comp_ok) + return false; + + return tcp_comp_enabled(ireq->ir_loc_addr, ireq->ir_rmt_addr, + ntohs(inet->inet_sport)); +} + +static struct tcp_comp_context *comp_get_ctx(const struct sock *sk) +{ + struct inet_connection_sock *icsk = inet_csk(sk); + + return (__force void *)icsk->icsk_ulp_data; +} + +static int tcp_comp_tx_context_init(struct tcp_comp_context *ctx) +{ + ZSTD_parameters params; + int csize; + + params = ZSTD_getParams(ZSTD_COMP_DEFAULT_LEVEL, PAGE_SIZE, 0); + csize = zstd_cstream_workspace_bound(&params.cParams); + if (csize <= 0) + return -EINVAL; + + ctx->tx.cworkspace = kmalloc(csize, GFP_KERNEL); + if (!ctx->tx.cworkspace) + return -ENOMEM; + + ctx->tx.cstream = zstd_init_cstream(&params, 0, ctx->tx.cworkspace, + csize); + if (!ctx->tx.cstream) + goto err_cstream; + + ctx->tx.plaintext_data = kvmalloc(TCP_COMP_SCRATCH_SIZE, GFP_KERNEL); + if (!ctx->tx.plaintext_data) + goto err_cstream; + + ctx->tx.compressed_data = kvmalloc(TCP_COMP_MAX_CSIZE, GFP_KERNEL); + if (!ctx->tx.compressed_data) + goto err_compressed; + + return 0; + +err_compressed: + kvfree(ctx->tx.plaintext_data); + ctx->tx.plaintext_data = NULL; +err_cstream: + kfree(ctx->tx.cworkspace); + ctx->tx.cworkspace = NULL; + + return -ENOMEM; +} + +static void *tcp_comp_get_tx_stream(struct sock *sk) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + + if (!ctx->tx.plaintext_data) + tcp_comp_tx_context_init(ctx); + + return ctx->tx.plaintext_data; +} + +static int alloc_compressed_msg(struct sock *sk, int len) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + struct sk_msg *msg = &ctx->tx.msg; + + sk_msg_init(msg); + + return sk_msg_alloc(sk, msg, len, 0); +} + +static int memcopy_from_iter(struct sock *sk, struct iov_iter *from, int copy) +{ + void *dest; + int rc; + + dest = tcp_comp_get_tx_stream(sk); + if (!dest) + return -ENOSPC; + + if (sk->sk_route_caps & NETIF_F_NOCACHE_COPY) + rc = copy_from_iter_nocache(dest, copy, from); + else + rc = copy_from_iter(dest, copy, from); + + if (rc != copy) + rc = -EFAULT; + + return rc; +} + +static int memcopy_to_msg(struct sock *sk, int bytes) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + struct sk_msg *msg = &ctx->tx.msg; + int i = msg->sg.curr; + struct scatterlist *sge; + u32 copy, buf_size; + void *from, *to; + + from = ctx->tx.compressed_data; + do { + sge = sk_msg_elem(msg, i); + /* This is possible if a trim operation shrunk the buffer */ + if (msg->sg.copybreak >= sge->length) { + msg->sg.copybreak = 0; + sk_msg_iter_var_next(i); + if (i == msg->sg.end) + break; + sge = sk_msg_elem(msg, i); + } + buf_size = sge->length - msg->sg.copybreak; + copy = (buf_size > bytes) ? bytes : buf_size; + to = sg_virt(sge) + msg->sg.copybreak; + msg->sg.copybreak += copy; + memcpy(to, from, copy); + bytes -= copy; + from += copy; + if (!bytes) + break; + msg->sg.copybreak = 0; + sk_msg_iter_var_next(i); + } while (i != msg->sg.end); + + msg->sg.curr = i; + return bytes; +} + +static int tcp_comp_compress_to_msg(struct sock *sk, int bytes) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + ZSTD_outBuffer outbuf; + ZSTD_inBuffer inbuf; + size_t ret; + + inbuf.src = ctx->tx.plaintext_data; + outbuf.dst = ctx->tx.compressed_data; + inbuf.size = bytes; + outbuf.size = TCP_COMP_MAX_CSIZE; + inbuf.pos = 0; + outbuf.pos = 0; + + ret = ZSTD_compressStream(ctx->tx.cstream, &outbuf, &inbuf); + if (ZSTD_isError(ret)) + return -EIO; + + ret = ZSTD_flushStream(ctx->tx.cstream, &outbuf); + if (ZSTD_isError(ret)) + return -EIO; + + if (inbuf.pos != inbuf.size) + return -EIO; + + if (memcopy_to_msg(sk, outbuf.pos)) + return -EIO; + + sk_msg_trim(sk, &ctx->tx.msg, outbuf.pos); + + return 0; +} + +static int tcp_comp_push_msg(struct sock *sk, struct sk_msg *msg, int flags) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + struct msghdr mh; + struct bio_vec bvec; + struct scatterlist *sg; + int ret, offset; + struct page *p; + size_t size; + + ctx->tx.in_tcp_sendpages = true; + while (1) { + sg = sk_msg_elem(msg, msg->sg.start); + offset = sg->offset; + size = sg->length; + p = sg_page(sg); +retry: + memset(&mh, 0, sizeof(struct msghdr)); + memset(&bvec, 0, sizeof(struct bio_vec)); + + mh.msg_flags = flags | MSG_SPLICE_PAGES; + bvec_set_page(&bvec, p, size, offset); + iov_iter_bvec(&mh.msg_iter, ITER_SOURCE, &bvec, 1, size); + + ret = tcp_sendmsg_locked(sk, &mh, size); + if (ret != size) { + if (ret > 0) { + sk_mem_uncharge(sk, ret); + sg->offset += ret; + sg->length -= ret; + size -= ret; + offset += ret; + goto retry; + } + ctx->tx.in_tcp_sendpages = false; + return ret; + } + + sk_mem_uncharge(sk, ret); + msg->sg.size -= size; + put_page(p); + sk_msg_iter_next(msg, start); + if (msg->sg.start == msg->sg.end) + break; + } + + clear_bit(TCP_COMP_SEND_PENDING, &ctx->flags); + ctx->tx.in_tcp_sendpages = false; + + return 0; +} + +static int tcp_comp_push(struct sock *sk, int bytes, int flags) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + int ret; + + ret = tcp_comp_compress_to_msg(sk, bytes); + if (ret < 0) { + pr_debug("%s: failed to compress sg\n", __func__); + return ret; + } + + set_bit(TCP_COMP_SEND_PENDING, &ctx->flags); + + ret = tcp_comp_push_msg(sk, &ctx->tx.msg, flags); + if (ret) { + pr_debug("%s: failed to tcp_comp_push_sg\n", __func__); + return ret; + } + + return 0; +} + +static int wait_on_pending_writer(struct sock *sk, long *timeo) +{ + DEFINE_WAIT_FUNC(wait, woken_wake_function); + int ret = 0; + + add_wait_queue(sk_sleep(sk), &wait); + while (1) { + if (!*timeo) { + ret = -EAGAIN; + break; + } + + if (signal_pending(current)) { + ret = sock_intr_errno(*timeo); + break; + } + + if (sk_wait_event(sk, timeo, !sk->sk_write_pending, &wait)) + break; + } + remove_wait_queue(sk_sleep(sk), &wait); + + return ret; +} + +static int tcp_comp_push_pending_msg(struct sock *sk, int flags) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + struct sk_msg *msg = &ctx->tx.msg; + + if (msg->sg.start == msg->sg.end) + return 0; + + return tcp_comp_push_msg(sk, msg, flags); +} + +static int tcp_comp_complete_pending_work(struct sock *sk, int flags, + long *timeo) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + int ret = 0; + + if (unlikely(sk->sk_write_pending)) + ret = wait_on_pending_writer(sk, timeo); + + if (!ret && tcp_comp_is_write_pending(ctx)) + ret = tcp_comp_push_pending_msg(sk, flags); + + return ret; +} + +static int tcp_comp_sendmsg(struct sock *sk, struct msghdr *msg, size_t size) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + int copied = 0, err = 0; + size_t try_to_copy; + int required_size; + long timeo; + + lock_sock(sk); + + timeo = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT); + + err = tcp_comp_complete_pending_work(sk, msg->msg_flags, &timeo); + if (err) + goto out_err; + + while (msg_data_left(msg)) { + if (sk->sk_err) { + err = -sk->sk_err; + goto out_err; + } + + try_to_copy = msg_data_left(msg); + if (try_to_copy > TCP_COMP_SCRATCH_SIZE) + try_to_copy = TCP_COMP_SCRATCH_SIZE; + required_size = try_to_copy + TCP_COMP_MAX_PADDING; + + if (!sk_stream_memory_free(sk)) + goto wait_for_sndbuf; + +alloc_compressed: + err = alloc_compressed_msg(sk, required_size); + if (err) { + if (err != -ENOSPC) + goto wait_for_memory; + goto out_err; + } + + err = memcopy_from_iter(sk, &msg->msg_iter, try_to_copy); + if (err < 0) + goto out_err; + + copied += try_to_copy; + + err = tcp_comp_push(sk, try_to_copy, msg->msg_flags); + if (err < 0) { + if (err == -ENOMEM) + goto wait_for_memory; + if (err != -EAGAIN) + tcp_comp_err_abort(sk, EBADMSG); + goto out_err; + } + + continue; +wait_for_sndbuf: + set_bit(SOCK_NOSPACE, &sk->sk_socket->flags); +wait_for_memory: + err = sk_stream_wait_memory(sk, &timeo); + if (err) + goto out_err; + if (ctx->tx.msg.sg.size < required_size) + goto alloc_compressed; + } + +out_err: + err = sk_stream_error(sk, msg->msg_flags, err); + + release_sock(sk); + + return copied ? copied : err; +} + +static struct sk_buff *comp_wait_data(struct sock *sk, int flags, + long timeo, int *err) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + struct sk_buff *skb; + DEFINE_WAIT_FUNC(wait, woken_wake_function); + + while (!(skb = ctx->rx.pkt)) { + if (sk->sk_err) { + *err = sock_error(sk); + return NULL; + } + + if (!skb_queue_empty(&sk->sk_receive_queue)) { + __strp_unpause(&ctx->rx.strp); + if (ctx->rx.pkt) + return ctx->rx.pkt; + } + + if (sk->sk_shutdown & RCV_SHUTDOWN) + return NULL; + + if (sock_flag(sk, SOCK_DONE)) + return NULL; + + if ((flags & MSG_DONTWAIT) || !timeo) { + *err = -EAGAIN; + return NULL; + } + + add_wait_queue(sk_sleep(sk), &wait); + sk_set_bit(SOCKWQ_ASYNC_WAITDATA, sk); + sk_wait_event(sk, &timeo, ctx->rx.pkt != skb, &wait); + sk_clear_bit(SOCKWQ_ASYNC_WAITDATA, sk); + remove_wait_queue(sk_sleep(sk), &wait); + + /* Handle signals */ + if (signal_pending(current)) { + *err = sock_intr_errno(timeo); + return NULL; + } + } + + return skb; +} + +static bool comp_advance_skb(struct sock *sk, struct sk_buff *skb, + unsigned int len) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + struct strp_msg *rxm = strp_msg(skb); + + if (len < rxm->full_len) { + rxm->offset += len; + rxm->full_len -= len; + return false; + } + + /* Finished with message */ + ctx->rx.pkt = NULL; + kfree_skb(skb); + __strp_unpause(&ctx->rx.strp); + + return true; +} + +static bool comp_advance_dskb(struct sock *sk, struct sk_buff *skb, + unsigned int len) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + struct strp_msg *rxm = strp_msg(skb); + + if (len < rxm->full_len) { + rxm->offset += len; + rxm->full_len -= len; + return false; + } + + /* Finished with message */ + ctx->rx.dpkt = NULL; + kfree_skb(skb); + return true; +} + +static int tcp_comp_rx_context_init(struct tcp_comp_context *ctx) +{ + int dsize; + + dsize = zstd_dstream_workspace_bound(TCP_COMP_MAX_INPUT); + if (dsize <= 0) + return -EINVAL; + + ctx->rx.dworkspace = kmalloc(dsize, GFP_KERNEL); + if (!ctx->rx.dworkspace) + return -ENOMEM; + + ctx->rx.dstream = zstd_init_dstream(TCP_COMP_MAX_INPUT, + ctx->rx.dworkspace, dsize); + if (!ctx->rx.dstream) + goto err_dstream; + + ctx->rx.plaintext_data = kvmalloc(TCP_COMP_MAX_CSIZE * 32, GFP_KERNEL); + if (!ctx->rx.plaintext_data) + goto err_dstream; + + return 0; + +err_dstream: + kfree(ctx->rx.dworkspace); + ctx->rx.dworkspace = NULL; + + return -ENOMEM; +} + +static void *tcp_comp_get_rx_stream(struct sock *sk) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + + if (!ctx->rx.plaintext_data) + tcp_comp_rx_context_init(ctx); + + return ctx->rx.plaintext_data; +} + +static int tcp_comp_decompress(struct sock *sk, struct sk_buff *skb, int flags) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + struct strp_msg *rxm = strp_msg(skb); + size_t ret, compressed_len = 0; + int nr_frags_over = 0; + ZSTD_outBuffer outbuf; + ZSTD_inBuffer inbuf; + struct sk_buff *nskb; + int len, plen; + void *to; + + to = tcp_comp_get_rx_stream(sk); + if (!to) + return -ENOSPC; + + if (skb_linearize_cow(skb)) + return -ENOMEM; + + nskb = skb_copy(skb, GFP_KERNEL); + if (!nskb) + return -ENOMEM; + + while (compressed_len < (skb->len - rxm->offset)) { + if (skb_shinfo(nskb)->nr_frags >= MAX_SKB_FRAGS) + break; + + len = 0; + plen = skb->len - rxm->offset - compressed_len; + if (plen > TCP_COMP_MAX_CSIZE) + plen = TCP_COMP_MAX_CSIZE; + + inbuf.src = (char *)skb->data + rxm->offset + compressed_len; + inbuf.pos = 0; + inbuf.size = plen; + + outbuf.dst = ctx->rx.plaintext_data; + outbuf.pos = 0; + outbuf.size = MAX_SKB_FRAGS * TCP_COMP_DATA_SIZE; + outbuf.size -= skb_shinfo(nskb)->nr_frags * TCP_COMP_DATA_SIZE; + + ret = ZSTD_decompressStream(ctx->rx.dstream, &outbuf, &inbuf); + if (ZSTD_isError(ret)) { + kfree_skb(nskb); + return -EIO; + } + + if (!compressed_len) { + len = outbuf.pos - skb->len; + if (len > skb_tailroom(nskb)) + len = skb_tailroom(nskb); + + __skb_put(nskb, len); + + len += skb->len; + skb_copy_to_linear_data(nskb, to, len); + } + + while ((to += len, outbuf.pos -= len) > 0) { + struct page *pages; + skb_frag_t *frag; + + if (skb_shinfo(nskb)->nr_frags >= MAX_SKB_FRAGS) { + nr_frags_over = 1; + break; + } + + frag = skb_shinfo(nskb)->frags + + skb_shinfo(nskb)->nr_frags; + pages = alloc_pages(__GFP_NOWARN | GFP_KERNEL | __GFP_COMP, + TCP_COMP_ALLOC_ORDER); + if (!pages) { + kfree_skb(nskb); + return -ENOMEM; + } + + frag->bv_page = pages; + len = PAGE_SIZE << TCP_COMP_ALLOC_ORDER; + if (outbuf.pos < len) + len = outbuf.pos; + + frag->bv_offset = 0; + skb_frag_size_set(frag, len); + memcpy(skb_frag_address(frag), to, len); + + nskb->truesize += len; + nskb->data_len += len; + nskb->len += len; + skb_shinfo(nskb)->nr_frags++; + } + + if (nr_frags_over) + break; + + compressed_len += inbuf.pos; + } + + ctx->rx.dpkt = nskb; + rxm = strp_msg(nskb); + rxm->full_len = nskb->len; + rxm->offset = 0; + comp_advance_skb(sk, skb, compressed_len); + + return 0; +} + +static int tcp_comp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, + int flags, int *addr_len) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + struct strp_msg *rxm; + struct sk_buff *skb; + ssize_t copied = 0; + int target, err = 0; + long timeo; + + if (unlikely(flags & MSG_ERRQUEUE)) + return sock_recv_errqueue(sk, msg, len, SOL_IP, IP_RECVERR); + + lock_sock(sk); + + target = sock_rcvlowat(sk, flags & MSG_WAITALL, len); + timeo = sock_rcvtimeo(sk, flags & MSG_WAITALL); + + do { + int chunk = 0; + + if (!ctx->rx.dpkt) { + skb = comp_wait_data(sk, flags, timeo, &err); + if (!skb) + goto recv_end; + + err = tcp_comp_decompress(sk, skb, flags); + if (err < 0) { + goto recv_end; + } + } + + skb = ctx->rx.dpkt; + rxm = strp_msg(skb); + chunk = min_t(unsigned int, rxm->full_len, len); + err = skb_copy_datagram_msg(skb, rxm->offset, msg, + chunk); + if (err < 0) + goto recv_end; + + copied += chunk; + len -= chunk; + if (likely(!(flags & MSG_PEEK))) + comp_advance_dskb(sk, skb, chunk); + else + break; + + if (copied >= target && !ctx->rx.dpkt) + break; + } while (len > 0); + +recv_end: + release_sock(sk); + return copied ? : err; +} + +bool comp_stream_read(struct sock *sk) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + + if (!ctx) + return false; + + if (ctx->rx.pkt || ctx->rx.dpkt) + return true; + + return false; +} + +static void comp_data_ready(struct sock *sk) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + + strp_data_ready(&ctx->rx.strp); +} + +static void comp_queue(struct strparser *strp, struct sk_buff *skb) +{ + struct tcp_comp_context *ctx = comp_get_ctx(strp->sk); + + ctx->rx.pkt = skb; + strp_pause(strp); + ctx->rx.saved_data_ready(strp->sk); +} + +static int comp_read_size(struct strparser *strp, struct sk_buff *skb) +{ + struct strp_msg *rxm = strp_msg(skb); + + if (rxm->offset > skb->len) + return 0; + + return skb->len - rxm->offset; +} + +void comp_setup_strp(struct sock *sk, struct tcp_comp_context *ctx) +{ + struct strp_callbacks cb; + + memset(&cb, 0, sizeof(cb)); + cb.rcv_msg = comp_queue; + cb.parse_msg = comp_read_size; + strp_init(&ctx->rx.strp, sk, &cb); + + write_lock_bh(&sk->sk_callback_lock); + ctx->rx.saved_data_ready = sk->sk_data_ready; + sk->sk_data_ready = comp_data_ready; + write_unlock_bh(&sk->sk_callback_lock); + + strp_check_rcv(&ctx->rx.strp); +} + +static void tcp_comp_write_space(struct sock *sk) +{ + struct tcp_comp_context *ctx = comp_get_ctx(sk); + + if (ctx->tx.in_tcp_sendpages) { + ctx->sk_write_space(sk); + return; + } + + if (!sk->sk_write_pending && tcp_comp_is_write_pending(ctx)) { + gfp_t sk_allocation = sk->sk_allocation; + int rc; + + sk->sk_allocation = GFP_ATOMIC; + rc = tcp_comp_push_pending_msg(sk, MSG_DONTWAIT | MSG_NOSIGNAL); + sk->sk_allocation = sk_allocation; + + if (rc < 0) + return; + } + + ctx->sk_write_space(sk); +} + +void tcp_init_compression(struct sock *sk) +{ + struct inet_connection_sock *icsk = inet_csk(sk); + struct tcp_comp_context *ctx = NULL; + struct sk_msg *msg = NULL; + struct tcp_sock *tp = tcp_sk(sk); + + if (!tp->rx_opt.comp_ok) + return; + + ctx = kzalloc(sizeof(*ctx), GFP_ATOMIC); + if (!ctx) + return; + + msg = &ctx->tx.msg; + sk_msg_init(msg); + + ctx->sk_write_space = sk->sk_write_space; + ctx->sk_proto = sk->sk_prot; + WRITE_ONCE(sk->sk_prot, &tcp_prot_override); + sk->sk_write_space = tcp_comp_write_space; + + rcu_assign_pointer(icsk->icsk_ulp_data, ctx); + + sock_set_flag(sk, SOCK_COMP); + comp_setup_strp(sk, ctx); +} + +static void tcp_comp_context_tx_free(struct tcp_comp_context *ctx) +{ + kfree(ctx->tx.cworkspace); + ctx->tx.cworkspace = NULL; + + kvfree(ctx->tx.plaintext_data); + ctx->tx.plaintext_data = NULL; + + kvfree(ctx->tx.compressed_data); + ctx->tx.compressed_data = NULL; +} + +static void tcp_comp_context_rx_free(struct tcp_comp_context *ctx) +{ + kfree(ctx->rx.dworkspace); + ctx->rx.dworkspace = NULL; + + kvfree(ctx->rx.plaintext_data); + ctx->rx.plaintext_data = NULL; +} + +static void tcp_comp_context_free(struct rcu_head *head) +{ + struct tcp_comp_context *ctx; + + ctx = container_of(head, struct tcp_comp_context, rcu); + + tcp_comp_context_tx_free(ctx); + tcp_comp_context_rx_free(ctx); + strp_done(&ctx->rx.strp); + kfree(ctx); +} + +void tcp_cleanup_compression(struct sock *sk) +{ + struct inet_connection_sock *icsk = inet_csk(sk); + struct tcp_comp_context *ctx = comp_get_ctx(sk); + + if (!ctx || !sock_flag(sk, SOCK_COMP)) + return; + + if (ctx->rx.pkt) { + kfree_skb(ctx->rx.pkt); + ctx->rx.pkt = NULL; + } + + if (ctx->rx.dpkt) { + kfree_skb(ctx->rx.dpkt); + ctx->rx.dpkt = NULL; + } + strp_stop(&ctx->rx.strp); + + rcu_assign_pointer(icsk->icsk_ulp_data, NULL); + call_rcu(&ctx->rcu, tcp_comp_context_free); +} + +int tcp_comp_init(void) +{ + tcp_prot_override = tcp_prot; + tcp_prot_override.sendmsg = tcp_comp_sendmsg; + tcp_prot_override.recvmsg = tcp_comp_recvmsg; + tcp_prot_override.sock_is_readable = comp_stream_read; + + return 0; +} diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index fd5c13c1fbc8..a5adf4822663 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -4012,6 +4012,24 @@ static bool smc_parse_options(const struct tcphdr *th, return false; } +static bool tcp_parse_comp_option(const struct tcphdr *th, + struct tcp_options_received *opt_rx, + const unsigned char *ptr, + int opsize) +{ +#if IS_ENABLED(CONFIG_TCP_COMP) + if (static_branch_unlikely(&tcp_have_comp)) { + if (th->syn && !(opsize & 1) && + opsize >= TCPOLEN_EXP_COMP_BASE && + get_unaligned_be16(ptr) == TCPOPT_COMP_MAGIC) { + opt_rx->comp_ok = 1; + return true; + } + } +#endif + return false; +} + /* Try to parse the MSS option from the TCP header. Return 0 on failure, clamped * value on success. */ @@ -4171,6 +4189,10 @@ void tcp_parse_options(const struct net *net, if (smc_parse_options(th, opt_rx, ptr, opsize)) break; + if (tcp_parse_comp_option(th, opt_rx, ptr, + opsize)) + break; + opt_rx->saw_unknown = 1; break; @@ -6097,6 +6119,7 @@ void tcp_init_transfer(struct sock *sk, int bpf_op, struct sk_buff *skb) /* Initialize congestion control unless BPF initialized it already: */ if (!icsk->icsk_ca_initialized) tcp_init_congestion_control(sk); + tcp_init_compression(sk); tcp_init_buffer_space(sk); } @@ -6835,6 +6858,9 @@ static void tcp_openreq_init(struct request_sock *req, ireq->smc_ok = rx_opt->smc_ok && !(tcp_sk(sk)->smc_hs_congested && tcp_sk(sk)->smc_hs_congested(sk)); #endif +#if IS_ENABLED(CONFIG_TCP_COMP) + ireq->comp_ok = rx_opt->comp_ok; +#endif } struct request_sock *inet_reqsk_alloc(const struct request_sock_ops *ops, diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 4167e8a48b60..9cbf23c1c039 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2304,6 +2304,8 @@ void tcp_v4_destroy_sock(struct sock *sk) tcp_cleanup_congestion_control(sk); + tcp_cleanup_compression(sk); + tcp_cleanup_ulp(sk); /* Cleanup up the write buffer. */ diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index b98d476f1594..d67a84c114ce 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -543,6 +543,9 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, newtp->rcv_ssthresh = req->rsk_rcv_wnd; newtp->rcv_wnd = req->rsk_rcv_wnd; newtp->rx_opt.wscale_ok = ireq->wscale_ok; +#if IS_ENABLED(CONFIG_TCP_COMP) + newtp->rx_opt.comp_ok = ireq->comp_ok; +#endif if (newtp->rx_opt.wscale_ok) { newtp->rx_opt.snd_wscale = ireq->snd_wscale; newtp->rx_opt.rcv_wscale = ireq->rcv_wscale; diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index 1917c62ad3bf..eee34f6c7643 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -422,6 +422,7 @@ static inline bool tcp_urg_mode(const struct tcp_sock *tp) #define OPTION_FAST_OPEN_COOKIE BIT(8) #define OPTION_SMC BIT(9) #define OPTION_MPTCP BIT(10) +#define OPTION_COMP BIT(11) static void smc_options_write(__be32 *ptr, u16 *options) { @@ -438,6 +439,19 @@ static void smc_options_write(__be32 *ptr, u16 *options) #endif } +static void comp_options_write(__be32 *ptr, u16 *options) +{ +#if IS_ENABLED(CONFIG_TCP_COMP) + if (static_branch_unlikely(&tcp_have_comp)) { + if (unlikely(OPTION_COMP & *options)) { + *ptr++ = htonl((TCPOPT_EXP << 24) | + (TCPOLEN_EXP_COMP_BASE << 16) | + (TCPOPT_COMP_MAGIC)); + } + } +#endif +} + struct tcp_out_options { u16 options; /* bit field of OPTION_* */ u16 mss; /* 0 to disable */ @@ -711,6 +725,8 @@ static void tcp_options_write(struct tcphdr *th, struct tcp_sock *tp, smc_options_write(ptr, &options); mptcp_options_write(th, ptr, tp, opts); + + comp_options_write(ptr, &options); } static void smc_set_option(const struct tcp_sock *tp, @@ -729,6 +745,39 @@ static void smc_set_option(const struct tcp_sock *tp, #endif } +static void comp_set_option(struct sock *sk, + struct tcp_out_options *opts, + unsigned int *remaining) +{ +#if IS_ENABLED(CONFIG_TCP_COMP) + if (static_branch_unlikely(&tcp_have_comp)) { + if (tcp_syn_comp_enabled(sk)) { + if (*remaining >= TCPOLEN_EXP_COMP_BASE) { + opts->options |= OPTION_COMP; + *remaining -= TCPOLEN_EXP_COMP_BASE; + } + } + } +#endif +} + +static void comp_set_option_cond(struct sock *sk, + const struct inet_request_sock *ireq, + struct tcp_out_options *opts, + unsigned int *remaining) +{ +#if IS_ENABLED(CONFIG_TCP_COMP) + if (static_branch_unlikely(&tcp_have_comp)) { + if (tcp_synack_comp_enabled(sk, ireq)) { + if (*remaining >= TCPOLEN_EXP_COMP_BASE) { + opts->options |= OPTION_COMP; + *remaining -= TCPOLEN_EXP_COMP_BASE; + } + } + } +#endif +} + static void smc_set_option_cond(const struct tcp_sock *tp, const struct inet_request_sock *ireq, struct tcp_out_options *opts, @@ -830,6 +879,7 @@ static unsigned int tcp_syn_options(struct sock *sk, struct sk_buff *skb, } smc_set_option(tp, opts, &remaining); + comp_set_option(sk, opts, &remaining); if (sk_is_mptcp(sk)) { unsigned int size; @@ -910,6 +960,8 @@ static unsigned int tcp_synack_options(const struct sock *sk, smc_set_option_cond(tcp_sk(sk), ireq, opts, &remaining); + comp_set_option_cond((struct sock *)sk, ireq, opts, &remaining); + bpf_skops_hdr_opt_len((struct sock *)sk, skb, req, syn_skb, synack_type, opts, &remaining); diff --git a/net/ipv6/syncookies.c b/net/ipv6/syncookies.c index 8698b49dfc8d..d17389f1afe0 100644 --- a/net/ipv6/syncookies.c +++ b/net/ipv6/syncookies.c @@ -217,6 +217,8 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb) treq->txhash = net_tx_rndhash(); if (IS_ENABLED(CONFIG_SMC)) ireq->smc_ok = 0; + if (IS_ENABLED(CONFIG_TCP_COMP)) + ireq->comp_ok = 0; /* * We need to lookup the dst_entry to get the correct window size. -- 2.34.1

2 1

[PATCH OLK-6.6 00/10] ACPI & PCI bugfix from 22.03 SP3
by Xiongfeng Wang 22 Dec '23

22 Dec '23

Xiongfeng Wang (10): pci: do not save 'PCI_BRIDGE_CTL_BUS_RESET' PCI: check BIR before mapping MSI-X Table PCI: Fail MSI-X mapping if MSI-X Table offset is out of range of BAR space sysrq: avoid concurrently info printing by 'sysrq-trigger' PCI: Add MCFG quirks for some Hisilicon Chip host controllers PCI: add a member in 'struct pci_bus' to record the original 'pci_ops' PCI/AER: increments pci bus reference count in aer-inject process ntp: Avoid undefined behaviour in second_overflow() hinic: ethtool: Allow userspace to set more aggregation params PCI/sysfs: Take reference on device to be removed drivers/acpi/pci_mcfg.c | 4 ++++ drivers/net/ethernet/huawei/hinic/hinic_ethtool.c | 10 +++++----- drivers/pci/bus.c | 2 ++ drivers/pci/msi/msi.c | 11 +++++++++++ drivers/pci/pci-sysfs.c | 9 +++++++-- drivers/pci/pci.c | 3 +++ drivers/pci/pcie/aer_inject.c | 9 +++++++++ drivers/pci/probe.c | 12 +++++++++--- drivers/tty/sysrq.c | 6 ++++++ include/linux/pci.h | 1 + kernel/time/ntp.c | 2 ++ 11 files changed, 59 insertions(+), 10 deletions(-) -- 2.20.1

2 11

[PATCH OLK-6.6 0/1] cgroup: Return ERSCH when add Z process into task
by chenridong 22 Dec '23

22 Dec '23

Lu Jialin (1): cgroup: Return ERSCH when add Z process into task kernel/cgroup/cgroup.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) -- 2.34.1

2 2

[PATCH] memcg: support ksm merge any mode per cgroup
by Nanyong Sun 22 Dec '23

22 Dec '23

hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I8OIQR ---------------------------------------------------------------------- Add control file "memory.ksm" to enable ksm per cgroup. Echo to 1 will set all tasks currently in the cgroup to ksm merge any mode, which means ksm gets enabled for all vma's of a process. Meanwhile echo to 0 will disable ksm for them and unmerge the merged pages. Cat the file will show the above state and ksm related profits of this cgroup. Signed-off-by: Nanyong Sun <sunnanyong(a)huawei.com> --- .../admin-guide/cgroup-v1/memory.rst | 1 + mm/memcontrol.c | 110 +++++++++++++++++- 2 files changed, 109 insertions(+), 2 deletions(-) diff --git a/Documentation/admin-guide/cgroup-v1/memory.rst b/Documentation/admin-guide/cgroup-v1/memory.rst index ff456871bf4b..3fdb48435e8e 100644 --- a/Documentation/admin-guide/cgroup-v1/memory.rst +++ b/Documentation/admin-guide/cgroup-v1/memory.rst @@ -109,6 +109,7 @@ Brief summary of control files. memory.kmem.tcp.failcnt show the number of tcp buf memory usage hits limits memory.kmem.tcp.max_usage_in_bytes show max tcp buf memory usage recorded + memory.ksm set/show ksm merge any mode ==================================== ========================================== 1. History diff --git a/mm/memcontrol.c b/mm/memcontrol.c index 2d9a873e5522..30cafc2f22c1 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -73,6 +73,7 @@ #include <linux/uaccess.h> #include <trace/events/vmscan.h> +#include <linux/ksm.h> struct cgroup_subsys memory_cgrp_subsys __read_mostly; EXPORT_SYMBOL(memory_cgrp_subsys); @@ -242,10 +243,15 @@ enum res_type { iter != NULL; \ iter = mem_cgroup_iter(NULL, iter, NULL)) +static inline bool __task_is_dying(struct task_struct *task) +{ + return tsk_is_oom_victim(task) || fatal_signal_pending(task) || + (task->flags & PF_EXITING); +} + static inline bool task_is_dying(void) { - return tsk_is_oom_victim(current) || fatal_signal_pending(current) || - (current->flags & PF_EXITING); + return __task_is_dying(current); } /* Some nice accessors for the vmpressure. */ @@ -5100,6 +5106,98 @@ static int mem_cgroup_slab_show(struct seq_file *m, void *p) } #endif +#ifdef CONFIG_KSM +static int memcg_set_ksm_for_tasks(struct mem_cgroup *memcg, bool enable) +{ + struct task_struct *task; + struct mm_struct *mm; + struct css_task_iter it; + int ret = 0; + + css_task_iter_start(&memcg->css, CSS_TASK_ITER_PROCS, &it); + while (!ret && (task = css_task_iter_next(&it))) { + if (__task_is_dying(task)) + continue; + + mm = get_task_mm(task); + if (!mm) + continue; + + if (mmap_write_lock_killable(mm)) { + mmput(mm); + continue; + } + + if (enable) + ret = ksm_enable_merge_any(mm); + else + ret = ksm_disable_merge_any(mm); + + mmap_write_unlock(mm); + mmput(mm); + } + css_task_iter_end(&it); + + return ret; +} + +static int memory_ksm_show(struct seq_file *m, void *v) +{ + unsigned long ksm_merging_pages = 0; + unsigned long ksm_rmap_items = 0; + long ksm_process_profits = 0; + unsigned int tasks = 0; + struct task_struct *task; + struct mm_struct *mm; + struct css_task_iter it; + struct mem_cgroup *memcg = mem_cgroup_from_seq(m); + + css_task_iter_start(&memcg->css, CSS_TASK_ITER_PROCS, &it); + while ((task = css_task_iter_next(&it))) { + mm = get_task_mm(task); + if (!mm) + continue; + + if (test_bit(MMF_VM_MERGE_ANY, &mm->flags)) + tasks++; + + ksm_rmap_items += mm->ksm_rmap_items; + ksm_merging_pages += mm->ksm_merging_pages; + ksm_process_profits += ksm_process_profit(mm); + mmput(mm); + } + css_task_iter_end(&it); + + seq_printf(m, "merge any tasks: %u\n", tasks); + seq_printf(m, "ksm_rmap_items %lu\n", ksm_rmap_items); + seq_printf(m, "ksm_merging_pages %lu\n", ksm_merging_pages); + seq_printf(m, "ksm_process_profits %ld\n", ksm_process_profits); + return 0; +} + +static ssize_t memory_ksm_write(struct kernfs_open_file *of, char *buf, + size_t nbytes, loff_t off) +{ + bool enable; + int err; + struct mem_cgroup *memcg = mem_cgroup_from_css(of_css(of)); + + buf = strstrip(buf); + if (!buf) + return -EINVAL; + + err = kstrtobool(buf, &enable); + if (err) + return err; + + err = memcg_set_ksm_for_tasks(memcg, enable); + if (err) + return err; + + return nbytes; +} +#endif /* CONFIG_KSM */ + static int memory_stat_show(struct seq_file *m, void *v); #ifdef CONFIG_MEMCG_V1_RECLAIM @@ -5337,6 +5435,14 @@ static struct cftype mem_cgroup_legacy_files[] = { .name = "reclaim", .write = memory_reclaim, }, +#endif +#ifdef CONFIG_KSM + { + .name = "ksm", + .flags = CFTYPE_NOT_ON_ROOT, + .write = memory_ksm_write, + .seq_show = memory_ksm_show, + }, #endif { }, /* terminate */ }; -- 2.25.1

1 0

[PATCH OLK-6.6 v4 0/7] arm64: Add support to turn an IPI as NMI
by Liao Chen 22 Dec '23

22 Dec '23

With pseudo NMIs support available its possible to configure SGIs to be triggered as pseudo NMIs running in NMI context. And kernel features such as: - NMI backtrace can leverage IPI turned as NMI to get a backtrace of CPU stuck in hard lockup using magic SYSRQ. - kgdb relies on NMI support to round up CPUs which are stuck in hard lockup state with interrupts disabled. This patch-set adds framework to turn an IPI as NMI which can be triggered as a pseudo NMI which in turn invokes registered NMI handlers. After this patch-set we should be able to get a backtrace for a CPU stuck in HARDLOCKUP. Sumit Garg (7): arm64: Add framework to turn IPI as NMI irqchip/gic-v3: Enable support for SGIs to act as NMIs arm64: smp: Assign and setup an IPI as NMI nmi: backtrace: Allow runtime arch specific override arm64: ipi_nmi: Add support for NMI backtrace kgdb: Expose default CPUs roundup fallback mechanism arm64: kgdb: Roundup cpus using IPI as NMI arch/arm/include/asm/irq.h | 2 +- arch/arm/kernel/smp.c | 3 +- arch/arm64/include/asm/irq.h | 6 +++ arch/arm64/include/asm/nmi.h | 17 +++++++ arch/arm64/kernel/Makefile | 2 +- arch/arm64/kernel/ipi_nmi.c | 84 ++++++++++++++++++++++++++++++++ arch/arm64/kernel/kgdb.c | 18 +++++++ arch/arm64/kernel/smp.c | 8 +++ arch/mips/include/asm/irq.h | 2 +- arch/mips/kernel/process.c | 3 +- arch/powerpc/include/asm/irq.h | 2 +- arch/powerpc/kernel/stacktrace.c | 3 +- arch/sparc/include/asm/irq_64.h | 2 +- arch/sparc/kernel/process_64.c | 4 +- arch/x86/include/asm/irq.h | 2 +- arch/x86/kernel/apic/hw_nmi.c | 3 +- drivers/irqchip/irq-gic-v3.c | 29 ++++++++--- include/linux/kgdb.h | 12 +++++ include/linux/nmi.h | 12 ++--- kernel/debug/debug_core.c | 8 ++- 20 files changed, 194 insertions(+), 28 deletions(-) create mode 100644 arch/arm64/include/asm/nmi.h create mode 100644 arch/arm64/kernel/ipi_nmi.c -- 2.34.1

2 8

[PATCH OLK-6.6 v4 0/7] arm64: Add support to turn an IPI as NMI
by Liao Chen 22 Dec '23

22 Dec '23

With pseudo NMIs support available its possible to configure SGIs to be triggered as pseudo NMIs running in NMI context. And kernel features such as: - NMI backtrace can leverage IPI turned as NMI to get a backtrace of CPU stuck in hard lockup using magic SYSRQ. - kgdb relies on NMI support to round up CPUs which are stuck in hard lockup state with interrupts disabled. This patch-set adds framework to turn an IPI as NMI which can be triggered as a pseudo NMI which in turn invokes registered NMI handlers. After this patch-set we should be able to get a backtrace for a CPU stuck in HARDLOCKUP. Sumit Garg (7): arm64: Add framework to turn IPI as NMI irqchip/gic-v3: Enable support for SGIs to act as NMIs arm64: smp: Assign and setup an IPI as NMI nmi: backtrace: Allow runtime arch specific override arm64: ipi_nmi: Add support for NMI backtrace kgdb: Expose default CPUs roundup fallback mechanism arm64: kgdb: Roundup cpus using IPI as NMI arch/arm/include/asm/irq.h | 2 +- arch/arm/kernel/smp.c | 3 +- arch/arm64/include/asm/irq.h | 6 +++ arch/arm64/include/asm/nmi.h | 17 +++++++ arch/arm64/kernel/Makefile | 2 +- arch/arm64/kernel/ipi_nmi.c | 84 ++++++++++++++++++++++++++++++++ arch/arm64/kernel/kgdb.c | 18 +++++++ arch/arm64/kernel/smp.c | 8 +++ arch/mips/include/asm/irq.h | 2 +- arch/mips/kernel/process.c | 3 +- arch/powerpc/include/asm/irq.h | 2 +- arch/powerpc/kernel/stacktrace.c | 3 +- arch/sparc/include/asm/irq_64.h | 2 +- arch/sparc/kernel/process_64.c | 4 +- arch/x86/include/asm/irq.h | 2 +- arch/x86/kernel/apic/hw_nmi.c | 3 +- drivers/irqchip/irq-gic-v3.c | 29 ++++++++--- include/linux/kgdb.h | 12 +++++ include/linux/nmi.h | 12 ++--- kernel/debug/debug_core.c | 8 ++- 20 files changed, 194 insertions(+), 28 deletions(-) create mode 100644 arch/arm64/include/asm/nmi.h create mode 100644 arch/arm64/kernel/ipi_nmi.c -- 2.34.1

1 1

[PATCH OLK-6.6 v4 1/7] arm64: Add framework to turn IPI as NMI
by Liao Chen 22 Dec '23

22 Dec '23

From: Sumit Garg <sumit.garg(a)linaro.org> maillist inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I7R4EN CVE: NA Reference: https://lore.kernel.org/all/1604317487-14543-1-git-send-email-sumit.garg@li… ------------------------------------------------- Introduce framework to turn an IPI as NMI using pseudo NMIs. The main motivation for this feature is to have an IPI that can be leveraged to invoke NMI functions on other CPUs. And current prospective users are NMI backtrace and KGDB CPUs round-up whose support is added via future patches. Signed-off-by: Sumit Garg <sumit.garg(a)linaro.org> Signed-off-by: Wei Li <liwei391(a)huawei.com> Reviewed-by: Xie XiuQi <xiexiuqi(a)huawei.com> Signed-off-by: Zheng Zengkai <zhengzengkai(a)huawei.com> Signed-off-by: Ruan Jinjie <ruanjinjie(a)huawei.com> Signed-off-by: Liao Chen <liaochen4(a)huawei.com> --- arch/arm64/include/asm/nmi.h | 17 ++++++++++ arch/arm64/kernel/ipi_nmi.c | 65 ++++++++++++++++++++++++++++++++++++ 2 files changed, 82 insertions(+) create mode 100644 arch/arm64/include/asm/nmi.h create mode 100644 arch/arm64/kernel/ipi_nmi.c diff --git a/arch/arm64/include/asm/nmi.h b/arch/arm64/include/asm/nmi.h new file mode 100644 index 000000000000..4cd14b6af88b --- /dev/null +++ b/arch/arm64/include/asm/nmi.h @@ -0,0 +1,17 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +#ifndef __ASM_NMI_H +#define __ASM_NMI_H + +#ifndef __ASSEMBLER__ + +#include <linux/cpumask.h> + +extern bool arm64_supports_nmi(void); +extern void arm64_send_nmi(cpumask_t *mask); + +void set_smp_dynamic_ipi(int ipi); +void dynamic_ipi_setup(int cpu); +void dynamic_ipi_teardown(int cpu); + +#endif /* !__ASSEMBLER__ */ +#endif diff --git a/arch/arm64/kernel/ipi_nmi.c b/arch/arm64/kernel/ipi_nmi.c new file mode 100644 index 000000000000..a945dcf8015f --- /dev/null +++ b/arch/arm64/kernel/ipi_nmi.c @@ -0,0 +1,65 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * NMI support for IPIs + * + * Copyright (C) 2020 Linaro Limited + * Author: Sumit Garg <sumit.garg(a)linaro.org> + */ + +#include <linux/interrupt.h> +#include <linux/irq.h> +#include <linux/smp.h> + +#include <asm/nmi.h> + +static struct irq_desc *ipi_nmi_desc __read_mostly; +static int ipi_nmi_id __read_mostly; + +bool arm64_supports_nmi(void) +{ + if (ipi_nmi_desc) + return true; + + return false; +} + +void arm64_send_nmi(cpumask_t *mask) +{ + if (WARN_ON_ONCE(!ipi_nmi_desc)) + return; + + __ipi_send_mask(ipi_nmi_desc, mask); +} + +static irqreturn_t ipi_nmi_handler(int irq, void *data) +{ + /* nop, NMI handlers for special features can be added here. */ + + return IRQ_NONE; +} + +void dynamic_ipi_setup(int cpu) +{ + if (!ipi_nmi_desc) + return; + + if (!prepare_percpu_nmi(ipi_nmi_id)) + enable_percpu_nmi(ipi_nmi_id, IRQ_TYPE_NONE); +} + +void dynamic_ipi_teardown(int cpu) +{ + if (!ipi_nmi_desc) + return; + + disable_percpu_nmi(ipi_nmi_id); + teardown_percpu_nmi(ipi_nmi_id); +} + +void __init set_smp_dynamic_ipi(int ipi) +{ + if (!request_percpu_nmi(ipi, ipi_nmi_handler, "IPI", &cpu_number)) { + ipi_nmi_desc = irq_to_desc(ipi); + ipi_nmi_id = ipi; + } +} -- 2.34.1

2 7

[PATCH OLK-6.6] memcg: support ksm merge any mode per cgroup
by Nanyong Sun 22 Dec '23

22 Dec '23

hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I8OIQR ---------------------------------------------------------------------- Add control file "memory.ksm" to enable ksm per cgroup. Echo to 1 will set all tasks currently in the cgroup to ksm merge any mode, which means ksm gets enabled for all vma's of a process. Meanwhile echo to 0 will disable ksm for them and unmerge the merged pages. Cat the file will show the above state and ksm related profits of this cgroup. Signed-off-by: Nanyong Sun <sunnanyong(a)huawei.com> --- .../admin-guide/cgroup-v1/memory.rst | 1 + mm/memcontrol.c | 110 +++++++++++++++++- 2 files changed, 109 insertions(+), 2 deletions(-) diff --git a/Documentation/admin-guide/cgroup-v1/memory.rst b/Documentation/admin-guide/cgroup-v1/memory.rst index ff456871bf4b..3fdb48435e8e 100644 --- a/Documentation/admin-guide/cgroup-v1/memory.rst +++ b/Documentation/admin-guide/cgroup-v1/memory.rst @@ -109,6 +109,7 @@ Brief summary of control files. memory.kmem.tcp.failcnt show the number of tcp buf memory usage hits limits memory.kmem.tcp.max_usage_in_bytes show max tcp buf memory usage recorded + memory.ksm set/show ksm merge any mode ==================================== ========================================== 1. History diff --git a/mm/memcontrol.c b/mm/memcontrol.c index 2d9a873e5522..30cafc2f22c1 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -73,6 +73,7 @@ #include <linux/uaccess.h> #include <trace/events/vmscan.h> +#include <linux/ksm.h> struct cgroup_subsys memory_cgrp_subsys __read_mostly; EXPORT_SYMBOL(memory_cgrp_subsys); @@ -242,10 +243,15 @@ enum res_type { iter != NULL; \ iter = mem_cgroup_iter(NULL, iter, NULL)) +static inline bool __task_is_dying(struct task_struct *task) +{ + return tsk_is_oom_victim(task) || fatal_signal_pending(task) || + (task->flags & PF_EXITING); +} + static inline bool task_is_dying(void) { - return tsk_is_oom_victim(current) || fatal_signal_pending(current) || - (current->flags & PF_EXITING); + return __task_is_dying(current); } /* Some nice accessors for the vmpressure. */ @@ -5100,6 +5106,98 @@ static int mem_cgroup_slab_show(struct seq_file *m, void *p) } #endif +#ifdef CONFIG_KSM +static int memcg_set_ksm_for_tasks(struct mem_cgroup *memcg, bool enable) +{ + struct task_struct *task; + struct mm_struct *mm; + struct css_task_iter it; + int ret = 0; + + css_task_iter_start(&memcg->css, CSS_TASK_ITER_PROCS, &it); + while (!ret && (task = css_task_iter_next(&it))) { + if (__task_is_dying(task)) + continue; + + mm = get_task_mm(task); + if (!mm) + continue; + + if (mmap_write_lock_killable(mm)) { + mmput(mm); + continue; + } + + if (enable) + ret = ksm_enable_merge_any(mm); + else + ret = ksm_disable_merge_any(mm); + + mmap_write_unlock(mm); + mmput(mm); + } + css_task_iter_end(&it); + + return ret; +} + +static int memory_ksm_show(struct seq_file *m, void *v) +{ + unsigned long ksm_merging_pages = 0; + unsigned long ksm_rmap_items = 0; + long ksm_process_profits = 0; + unsigned int tasks = 0; + struct task_struct *task; + struct mm_struct *mm; + struct css_task_iter it; + struct mem_cgroup *memcg = mem_cgroup_from_seq(m); + + css_task_iter_start(&memcg->css, CSS_TASK_ITER_PROCS, &it); + while ((task = css_task_iter_next(&it))) { + mm = get_task_mm(task); + if (!mm) + continue; + + if (test_bit(MMF_VM_MERGE_ANY, &mm->flags)) + tasks++; + + ksm_rmap_items += mm->ksm_rmap_items; + ksm_merging_pages += mm->ksm_merging_pages; + ksm_process_profits += ksm_process_profit(mm); + mmput(mm); + } + css_task_iter_end(&it); + + seq_printf(m, "merge any tasks: %u\n", tasks); + seq_printf(m, "ksm_rmap_items %lu\n", ksm_rmap_items); + seq_printf(m, "ksm_merging_pages %lu\n", ksm_merging_pages); + seq_printf(m, "ksm_process_profits %ld\n", ksm_process_profits); + return 0; +} + +static ssize_t memory_ksm_write(struct kernfs_open_file *of, char *buf, + size_t nbytes, loff_t off) +{ + bool enable; + int err; + struct mem_cgroup *memcg = mem_cgroup_from_css(of_css(of)); + + buf = strstrip(buf); + if (!buf) + return -EINVAL; + + err = kstrtobool(buf, &enable); + if (err) + return err; + + err = memcg_set_ksm_for_tasks(memcg, enable); + if (err) + return err; + + return nbytes; +} +#endif /* CONFIG_KSM */ + static int memory_stat_show(struct seq_file *m, void *v); #ifdef CONFIG_MEMCG_V1_RECLAIM @@ -5337,6 +5435,14 @@ static struct cftype mem_cgroup_legacy_files[] = { .name = "reclaim", .write = memory_reclaim, }, +#endif +#ifdef CONFIG_KSM + { + .name = "ksm", + .flags = CFTYPE_NOT_ON_ROOT, + .write = memory_ksm_write, + .seq_show = memory_ksm_show, + }, #endif { }, /* terminate */ }; -- 2.25.1

2 1

[PATCH OLK-6.6 0/3] Print rootfs and tmpfs files charged by memcg
by Jinjiang Tu 22 Dec '23

22 Dec '23

Support to print rootfs files and tmpfs files that having pages charged in given memory cgroup. The files infomations can be printed through interface "memory.memfs_files_info" or printed when OOM is triggered. Jinjiang Tu (1): fs: move {lock, unlock}_mount_hash to fs/mount.h Liu Shixin (2): mm/memcg_memfs_info: show files that having pages charged in mem_cgroup config: enable CONFIG_MEMCG_MEMFS_INFO by default Documentation/vm/memcg_memfs_info.rst | 40 ++++ arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + fs/mount.h | 10 + fs/namespace.c | 10 - include/linux/memcg_memfs_info.h | 23 ++ init/Kconfig | 10 + mm/Makefile | 1 + mm/memcg_memfs_info.c | 318 +++++++++++++++++++++++++ mm/memcontrol.c | 12 + 10 files changed, 416 insertions(+), 10 deletions(-) create mode 100644 Documentation/vm/memcg_memfs_info.rst create mode 100644 include/linux/memcg_memfs_info.h create mode 100644 mm/memcg_memfs_info.c -- 2.25.1

2 4

2024

2023

2022

2021

2020

2019

Kernel