From mboxrd@z Thu Jan 1 00:00:00 1970 Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.remote-ip=209.85.214.171 ARC-Seal: i=1; d=subspace.kernel.org; s=arc20261015; a=rsa-sha256; cv=none; t=1795114999; b=TAep3LStLTOgDLTas0/Ip0XKJBFri1DyKEHuJSp9OjuphpuRi0JOidYL+8LDUpwALLZ9 UwpmssBo+AkliFqRt84zKxeBE7HMMi10yV7vdYHPZtl5znpz0UPqaAmNlKvInc/ZE9S7J LzH4udXJlI5PJu4TnjTX6KcWzYUewy+hVFfZUClU6GFAwKUA8tEE519lVeSXqaHq7v8lm mDhuCyWqifvcNIC2XaZPyn2fMXMvrY13OuIUpqWcYfME5Zy67aBdtMnvWR6BDMLB1N9Ay S1Kg21swbpxe9kq2ACB7Br/vekLncePAJ12RfqLV/OpyLSZkIDcFibC+ttEkaM+Bbrg== ARC-Message-Signature: i=1; d=subspace.kernel.org; s=arc20261015; a=rsa-sha256; c=relaxed/relaxed; t=1795114999; h=DMARC-Filter:Received:DKIM-Signature:X-Google-DKIM-Signature: X-Gm-Message-State:X-Gm-Gg:X-Received:Received:From:To:Cc:Subject: Date:Message-ID:X-Mailer:In-Reply-To:References:MIME-Version: Content-Transfer-Encoding; bh=iP87RkXNbSr52oPH/MbtbcHjsKMutHSyf01JArUpXyc=; b=HzzUm+vuw7ZnRVvvemjQYJjYbNXiSnFusdQ6UqE/kfHULEnD7okU6/Gl1dAsLAPRtKW5 FPxsOeGNhqRog3ZSKDnP0iYSSAem7vx7sYdEghyAn11tae6TCJA67TaoFm0GCxnnsbB1O Ly+Rfo5irq/HLGHJgQDmY6zSrfpmwwD03yswwtffuLKAEeTkImkH8eUHjqLQtMueNNsbe xbygeuC2WXUojYkv7oPxohLnx7ML+MKPJUh84SjToxmmWhDI9HKaU9v9yfxoeLx2diuf5 Ou9IT6JGemfMFInIEsKXQx8eGs39Kv1A4zad5mXt7n8N6HI34nJ5jgGKYLmANobWRAA== ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass header.from=networkplumber.org; spf=pass smtp.mailfrom=networkplumber.org; dkim=pass header.d=networkplumber-org.20260707.gappssmtp.com header.i=@networkplumber-org.20260707.gappssmtp.com header.a=rsa-sha256 header.s=20260707 header.b=Q4dWZsHm; arc=none smtp.remote-ip=209.85.214.171 Received: from mail-pl1-f171.google.com (mail-pl1-f171.google.com [209.85.214.171]) (using TLSv1.3 with cipher TLS_AES_128_GCM_SHA256 (128/128 bits) key-exchange X25519 server-signature ECDSA (prime256v1) server-digest SHA256) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id A127060A2609 for ; Thu, 19 Nov 2026 19:03:18 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.4.2 smtp.subspace.kernel.org A127060A2609 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=quarantine dis=none) header.from=networkplumber.org Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=networkplumber.org Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=networkplumber-org.20260707.gappssmtp.com header.i=@networkplumber-org.20260707.gappssmtp.com header.a=rsa-sha256 header.s=20260707 header.b=Q4dWZsHm Received: by mail-pl1-f171.google.com with SMTP id d9443c01a7336-2f042a34db4so38235ad.0 for ; Thu, 19 Nov 2026 11:03:18 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=networkplumber-org.20260707.gappssmtp.com; s=20260707; t=1795114998; x=1795719798; darn=vger.kernel.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:from:to:cc:subject:date :message-id:reply-to:content-type; bh=iP87RkXNbSr52oPH/MbtbcHjsKMutHSyf01JArUpXyc=; b=Q4dWZsHmHYZ4XBG8CuCrzquhZjxgBP0On0J9q5rC+//RQd49FMNxoRY11EhKVXfmzP P1sOgL240CJM5osmAHOs+AjGhYeA0lurklRyS6DtP7325YVFN43CKs9VpaQZM2zlDnjW pqra5wFtVKjEWt1/VTwGwhb9BO670OjdV4kyDHT5jMmwhE1cqMoZr4rskjhjhkt77FMf dXhNbKopEvhCI3Q/7mc04VTIigYWmkjySAJKS5yZgZ/IVLynhJ1oExhxqjPNZRIy+/ip Ry0ZbrTluoVS6GPBy9kvtt1fGUlZAwVbzGVI6Tf8odjk68Onv/85BvPhMybt2d1A1vqC RDOQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20260707; t=1795114998; x=1795719798; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-gg:x-gm-message-state:from :to:cc:subject:date:message-id:reply-to:content-type; bh=iP87RkXNbSr52oPH/MbtbcHjsKMutHSyf01JArUpXyc=; b=vdDw+2K+GBkKs31TWTsh4pu6V1X8WF+yhGGjUy9hCjK6g/xuh7Vqm9iGEJP/oaryyf az6EGr2sXlyQ9H1KrOXtK3FRpRM4LWBtHl+71RpBhZ7iuNBoQW2q0l9ssa6COWDkfA6P vg3wC89aI3e6+Vkpt5v7Z6cgWTViDUhXkEvPIbe4qnNoiXXPs/xIQgxJHiYDBSEAtrr9 NKqGuRYA0JvC2DjH52mGsSAKuSsKw2w9Tyq5XIzBwvlowPF82VwXouAc4dqcHtn1vXsh cJuqvH7kSXKcKhps3Sjz/Ubd4YqgYayPDFQ0WU8hbiH+cH6kiF64DW/QWAveYgQwOTDc +W/A== X-Gm-Message-State: APuU4Z+47i98cs/+7lULu60Hcrcy461wJ4/KzBzuILoH0JjHCiGneqBH Oh9NrwjEFiOafDe5uQifuk7EFrw1vsuZOcyn5F7fmIlpQiqHVf231+zUIkH6pYNxFyjyECK6utg 9kerHo6Q= X-Gm-Gg: AZFdTZcvRWpExT74uhzmyvVxG0PbpgGkAs4jDFg73QBTS37EBdV+jlHy6LrvrnGcQhy 3V803MqIQWBUa+eFvdoWqml3leJrhqJO2wUfX3HAyvGrROdMPcVy+NlLbt3WSwtlXDWUcGeKDkf hmYK1rSukc2+7oiSmx8WdpizA1AajCGUCsoh1YDRdRgRj2rf8WFboIIOH1viL4ur8yfFKFEjVfE Wc8OhmdyX/q3kcABPjWkbU0dNQdjCaNn/ryfvNWXEwPKZG7FMHs5fV98R1BhXqP4hGZ6QIayjk6 gFkHGOSJrKrHp3AmzrUpC6ORSaL63fWb27mM6Zk6NvpB6wW52LjKJAvD4GCpfYNe0mnR8IqNQy1 aqo1XTJ9kQFxzFHLGbxHgt2jJyTS7w4wdHcW5g1IAq51mOzSK5L8uC77UiVN9megAuGfOQnqKGC xLWyrshf+dYfYBD4GCcktbXKAKt7QZx+rZCmXA1QJLQjtMqQ6o6xeimRcSapadPov4VNeW1hJpj FbyN/txyMMunks4/b8UzSlK4U04qjPRNoJINfjp18iRxBH0b+clzEAi5D8EwncVyOyi61hsxpFD Tnb0v3PvwmghnY4kLVzdloMNxCxHOLN+DRFVNKpXpk9/ShUs/jFF+4gP7+bq3s66ZYNT X-Received: by 2002:a17:902:da91:b0:2ef:f6dd:87f7 with SMTP id d9443c01a7336-2f031e1aec9mr14246245ad.7.1795114997687; Thu, 19 Nov 2026 11:03:17 -0800 (PST) Received: from phoenix.lan (204-195-112-43.wavecable.com. [204.195.112.43]) by smtp.gmail.com with ESMTPSA id d9443c01a7336-2f03f0de00asm3750225ad.35.2026.11.19.11.03.16 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Thu, 19 Nov 2026 11:03:17 -0800 (PST) From: Stephen Hemminger To: netdev@vger.kernel.org Cc: dsahern@kernel.org, Stephen Hemminger , Jamal Hadi Salim , Thomas Graf , Daniel Borkmann , Jiri Pirko , Alexei Starovoitov Subject: [PATCH iproute2-next v2 3/7] tc: remove eBPF map export and exec bpf import Date: Thu, 19 Nov 2026 11:02:38 -0800 Message-ID: <20261119190312.578885-4-stephen@networkplumber.org> X-Mailer: git-send-email 2.53.0 In-Reply-To: <20261119190312.578885-1-stephen@networkplumber.org> References: <20261119190312.578885-1-stephen@networkplumber.org> X-Mailing-List: netdev@vger.kernel.org List-Id: List-Post: List-Help: List-Subscribe: List-Unsubscribe: List-Owner: Precedence: list MIME-Version: 1.0 Content-Transfer-Encoding: 8bit The export option of tc bpf filters and actions sends the map file descriptors to whatever process is bound to the given Unix socket. The socket owner is not checked, so the maps can end up with an unrelated process that happens to be bound to that path. The path cannot be made safe without redesigning the trust model around it, so remove the export option, the exec bpf import and run commands, and the TC_BPF_UDS environment variable. Default builds have used libbpf for years and send nothing, so they lose nothing here. Builds configured with LIBBPF_FORCE=off did have a working import/run, and it was the only in-tree way to hand maps to a helper without pinning them; those builds lose that. Pin the maps in bpffs instead and have the helper open them by path, or use bpftool to pass them along. Fixes: 6256f8c9e45f ("tc, bpf: finalize eBPF support for cls and act front-end") Reported-by: Anthropic OSS scanner (ANT-2026-5S9IO0L4) Assisted-by: Claude:claude-opus-5-5 Signed-off-by: Stephen Hemminger --- bash-completion/tc | 18 +--- include/bpf_scm.h | 77 ---------------- include/bpf_util.h | 18 +--- lib/bpf_legacy.c | 217 +-------------------------------------------- man/man8/tc-bpf.8 | 109 +++-------------------- tc/e_bpf.c | 157 ++++++-------------------------- tc/f_bpf.c | 12 +-- tc/m_bpf.c | 13 +-- 8 files changed, 48 insertions(+), 573 deletions(-) delete mode 100644 include/bpf_scm.h diff --git a/bash-completion/tc b/bash-completion/tc index c18288a3..dfbe5624 100644 --- a/bash-completion/tc +++ b/bash-completion/tc @@ -205,7 +205,7 @@ _tc_direct_complete() ;; # BPF options - used for filters, actions, and exec - export|bytecode|bytecode-file|object-file) + bytecode|bytecode-file|object-file) _filedir return 0 ;; @@ -230,10 +230,6 @@ _tc_direct_complete() fi return 0 ;; - import|run) - _filedir - return 0 - ;; type) COMPREPLY+=( $( compgen -W 'cls act' -- "$cur" ) ) return 0 @@ -387,10 +383,7 @@ _tc_qdisc_options() _tc_bpf_options() { [[ ${words[${#words[@]}-3]} == object-file ]] && \ - _tc_once_attr 'section export' - [[ ${words[${#words[@]}-5]} == object-file ]] && \ - [[ ${words[${#words[@]}-3]} =~ (section|export) ]] && \ - _tc_once_attr 'section export' + _tc_once_attr 'section' _tc_one_of_list 'bytecode bytecode-file object-file object-pinned' _tc_once_attr 'verbose index direct-action action classid' return 0 @@ -524,11 +517,6 @@ _tc_action_options() _tc_exec_options() { case $1 in - import) - [[ ${words[${#words[@]}-3]} == import ]] && \ - _tc_once_attr 'run' - return 0 - ;; graft) COMPREPLY+=( $( compgen -W 'key type' -- "$cur" ) ) [[ ${words[${#words[@]}-3]} == object-file ]] && \ @@ -772,7 +760,7 @@ _tc() exec) case $subcmd in bpf) - local excmd exwd EXEC_KIND=' import debug graft ' + local excmd exwd EXEC_KIND=' debug graft ' for ((exwd=$subcword; exwd < ${#words[@]}-1; exwd++)); do if [[ $EXEC_KIND =~ ' '${words[exwd]}' ' ]]; then excmd=${words[exwd]} diff --git a/include/bpf_scm.h b/include/bpf_scm.h deleted file mode 100644 index 669f0538..00000000 --- a/include/bpf_scm.h +++ /dev/null @@ -1,77 +0,0 @@ -/* SPDX-License-Identifier: GPL-2.0 */ -#ifndef __BPF_SCM__ -#define __BPF_SCM__ - -#include -#include -#include - -#include "utils.h" -#include "bpf_elf.h" - -#define BPF_SCM_AUX_VER 1 -#define BPF_SCM_MAX_FDS ELF_MAX_MAPS -#define BPF_SCM_MSG_SIZE 1024 - -struct bpf_elf_st { - dev_t st_dev; - ino_t st_ino; -}; - -struct bpf_map_aux { - unsigned short uds_ver; - unsigned short num_ent; - char obj_name[64]; - struct bpf_elf_st obj_st; - struct bpf_elf_map ent[BPF_SCM_MAX_FDS]; -}; - -struct bpf_map_set_msg { - struct msghdr hdr; - struct iovec iov; - char msg_buf[BPF_SCM_MSG_SIZE]; - struct bpf_map_aux aux; -}; - -static inline int *bpf_map_set_init(struct bpf_map_set_msg *msg, - struct sockaddr_un *addr, - unsigned int addr_len) -{ - const unsigned int cmsg_ctl_len = sizeof(int) * BPF_SCM_MAX_FDS; - struct cmsghdr *cmsg; - - msg->iov.iov_base = &msg->aux; - msg->iov.iov_len = sizeof(msg->aux); - - msg->hdr.msg_iov = &msg->iov; - msg->hdr.msg_iovlen = 1; - - msg->hdr.msg_name = (struct sockaddr *)addr; - msg->hdr.msg_namelen = addr_len; - - BUILD_BUG_ON(sizeof(msg->msg_buf) < cmsg_ctl_len); - msg->hdr.msg_control = &msg->msg_buf; - msg->hdr.msg_controllen = cmsg_ctl_len; - - cmsg = CMSG_FIRSTHDR(&msg->hdr); - cmsg->cmsg_len = msg->hdr.msg_controllen; - cmsg->cmsg_level = SOL_SOCKET; - cmsg->cmsg_type = SCM_RIGHTS; - - return (int *)CMSG_DATA(cmsg); -} - -static inline void bpf_map_set_init_single(struct bpf_map_set_msg *msg, - int num) -{ - struct cmsghdr *cmsg; - - msg->hdr.msg_controllen = CMSG_LEN(sizeof(int) * num); - msg->iov.iov_len = offsetof(struct bpf_map_aux, ent) + - sizeof(struct bpf_elf_map) * num; - - cmsg = CMSG_FIRSTHDR(&msg->hdr); - cmsg->cmsg_len = msg->hdr.msg_controllen; -} - -#endif /* __BPF_SCM__ */ diff --git a/include/bpf_util.h b/include/bpf_util.h index e1b8d327..78c38e80 100644 --- a/include/bpf_util.h +++ b/include/bpf_util.h @@ -16,9 +16,8 @@ #include #include "utils.h" -#include "bpf_scm.h" +#include "bpf_elf.h" -#define BPF_ENV_UDS "TC_BPF_UDS" #define BPF_ENV_MNT "TC_BPF_MNT" #ifndef BPF_MAX_LOG @@ -60,7 +59,6 @@ struct bpf_cfg_in { const char *object; const char *section; const char *prog_name; - const char *uds; enum bpf_prog_type type; enum bpf_mode mode; __u32 ifindex; @@ -282,9 +280,6 @@ int bpf_dump_prog_info(FILE *f, uint32_t id); int bpf(int cmd, union bpf_attr *attr, unsigned int size); #ifdef HAVE_ELF -int bpf_send_map_fds(const char *path, const char *obj); -int bpf_recv_map_fds(const char *path, int *fds, struct bpf_map_aux *aux, - unsigned int entries); #ifdef HAVE_LIBBPF int iproute2_bpf_elf_ctx_init(struct bpf_cfg_in *cfg); int iproute2_bpf_fetch_ancillary(void); @@ -296,17 +291,6 @@ int iproute2_find_map_name_by_id(unsigned int map_id, char *name); int iproute2_load_libbpf(struct bpf_cfg_in *cfg); #endif /* HAVE_LIBBPF */ #else -static inline int bpf_send_map_fds(const char *path, const char *obj) -{ - return 0; -} - -static inline int bpf_recv_map_fds(const char *path, int *fds, - struct bpf_map_aux *aux, - unsigned int entries) -{ - return -1; -} #ifdef HAVE_LIBBPF static inline int iproute2_load_libbpf(struct bpf_cfg_in *cfg) { diff --git a/lib/bpf_legacy.c b/lib/bpf_legacy.c index 5b8ac47d..f9844fd4 100644 --- a/lib/bpf_legacy.c +++ b/lib/bpf_legacy.c @@ -42,13 +42,11 @@ #include "bpf_util.h" #include "bpf_elf.h" -#include "bpf_scm.h" struct bpf_prog_meta { const char *type; const char *subdir; const char *section; - bool may_uds_export; }; static const enum bpf_prog_type __bpf_types[] = { @@ -65,13 +63,11 @@ static const struct bpf_prog_meta __bpf_prog_meta[] = { .type = "cls", .subdir = "tc", .section = ELF_SECTION_CLASSIFIER, - .may_uds_export = true, }, [BPF_PROG_TYPE_SCHED_ACT] = { .type = "act", .subdir = "tc", .section = ELF_SECTION_ACTION, - .may_uds_export = true, }, [BPF_PROG_TYPE_XDP] = { .type = "xdp", @@ -827,7 +823,7 @@ static int bpf_obj_pinned(const char *pathname, enum bpf_prog_type type) static int bpf_do_parse(struct bpf_cfg_in *cfg, const bool *opt_tbl) { - const char *file, *section, *uds_name, *prog_name; + const char *file, *section, *prog_name; bool verbose = false; int i, ret, argc; char **argv; @@ -858,7 +854,7 @@ static int bpf_do_parse(struct bpf_cfg_in *cfg, const bool *opt_tbl) } NEXT_ARG(); - file = section = uds_name = prog_name = NULL; + file = section = prog_name = NULL; if (cfg->mode == EBPF_OBJECT || cfg->mode == EBPF_PINNED) { file = *argv; NEXT_ARG_FWD(); @@ -901,16 +897,6 @@ static int bpf_do_parse(struct bpf_cfg_in *cfg, const bool *opt_tbl) NEXT_ARG_FWD(); } - if (__bpf_prog_meta[cfg->type].may_uds_export) { - uds_name = getenv(BPF_ENV_UDS); - if (argc > 0 && !uds_name && - matches(*argv, "export") == 0) { - NEXT_ARG(); - uds_name = *argv; - NEXT_ARG_FWD(); - } - } - if (argc > 0 && matches(*argv, "verbose") == 0) { verbose = true; NEXT_ARG_FWD(); @@ -934,7 +920,6 @@ static int bpf_do_parse(struct bpf_cfg_in *cfg, const bool *opt_tbl) cfg->object = file; cfg->section = section; - cfg->uds = uds_name; cfg->argc = argc; cfg->argv = argv; cfg->verbose = verbose; @@ -1179,7 +1164,6 @@ struct bpf_elf_ctx { enum bpf_prog_type type; __u32 ifindex; bool verbose; - struct bpf_elf_st stat; struct bpf_hash_entry *ht[256]; char *log; size_t log_size; @@ -1191,13 +1175,6 @@ struct bpf_elf_sec_data { const char *sec_name; }; -struct bpf_map_data { - int *fds; - const char *obj; - struct bpf_elf_st *st; - struct bpf_elf_map *ent; -}; - static bool bpf_log_has_data(struct bpf_elf_ctx *ctx) { return ctx->log && ctx->log[0]; @@ -2647,24 +2624,6 @@ static int bpf_fill_prog_arrays(struct bpf_elf_ctx *ctx) return 0; } -static void bpf_save_finfo(struct bpf_elf_ctx *ctx) -{ - struct stat st; - int ret; - - memset(&ctx->stat, 0, sizeof(ctx->stat)); - - ret = fstat(ctx->obj_fd, &st); - if (ret < 0) { - fprintf(stderr, "Stat of elf file failed: %s\n", - strerror(errno)); - return; - } - - ctx->stat.st_dev = st.st_dev; - ctx->stat.st_ino = st.st_ino; -} - static int bpf_read_pin_mapping(FILE *fp, uint32_t *id, char *path) { char buff[PATH_MAX]; @@ -2877,7 +2836,6 @@ static int bpf_elf_ctx_init(struct bpf_elf_ctx *ctx, const char *pathname, goto out_free; } - bpf_save_finfo(ctx); bpf_hash_init(ctx, CONF_ETC_DIR "/bpf_pinning"); if (ret == -ENOENT) ret = bpf_hash_init(ctx, CONF_USR_DIR "/bpf_pinning"); @@ -2892,19 +2850,6 @@ out_fd: return ret; } -static int bpf_maps_count(struct bpf_elf_ctx *ctx) -{ - int i, count = 0; - - for (i = 0; i < ARRAY_SIZE(ctx->map_fds); i++) { - if (!ctx->map_fds[i]) - break; - count++; - } - - return count; -} - static void bpf_maps_teardown(struct bpf_elf_ctx *ctx) { int i; @@ -2975,164 +2920,6 @@ out: return fd; } -static int -bpf_map_set_send(int fd, struct sockaddr_un *addr, unsigned int addr_len, - const struct bpf_map_data *aux, unsigned int entries) -{ - struct bpf_map_set_msg msg = { - .aux.uds_ver = BPF_SCM_AUX_VER, - .aux.num_ent = entries, - }; - int *cmsg_buf, min_fd; - char *amsg_buf; - int i; - - strlcpy(msg.aux.obj_name, aux->obj, sizeof(msg.aux.obj_name)); - memcpy(&msg.aux.obj_st, aux->st, sizeof(msg.aux.obj_st)); - - cmsg_buf = bpf_map_set_init(&msg, addr, addr_len); - amsg_buf = (char *)msg.aux.ent; - - for (i = 0; i < entries; i += min_fd) { - int ret; - - min_fd = min(BPF_SCM_MAX_FDS * 1U, entries - i); - bpf_map_set_init_single(&msg, min_fd); - - memcpy(cmsg_buf, &aux->fds[i], sizeof(aux->fds[0]) * min_fd); - memcpy(amsg_buf, &aux->ent[i], sizeof(aux->ent[0]) * min_fd); - - ret = sendmsg(fd, &msg.hdr, 0); - if (ret <= 0) - return ret ? : -1; - } - - return 0; -} - -static int -bpf_map_set_recv(int fd, int *fds, struct bpf_map_aux *aux, - unsigned int entries) -{ - struct bpf_map_set_msg msg; - int *cmsg_buf, min_fd; - char *amsg_buf, *mmsg_buf; - unsigned int needed = 1; - int i; - - cmsg_buf = bpf_map_set_init(&msg, NULL, 0); - amsg_buf = (char *)msg.aux.ent; - mmsg_buf = (char *)&msg.aux; - - for (i = 0; i < min(entries, needed); i += min_fd) { - struct cmsghdr *cmsg; - int ret; - - min_fd = min(entries, entries - i); - bpf_map_set_init_single(&msg, min_fd); - - ret = recvmsg(fd, &msg.hdr, 0); - if (ret <= 0) - return ret ? : -1; - - cmsg = CMSG_FIRSTHDR(&msg.hdr); - if (!cmsg || cmsg->cmsg_type != SCM_RIGHTS) - return -EINVAL; - if (msg.hdr.msg_flags & MSG_CTRUNC) - return -EIO; - if (msg.aux.uds_ver != BPF_SCM_AUX_VER) - return -ENOSYS; - - min_fd = (cmsg->cmsg_len - sizeof(*cmsg)) / sizeof(fd); - if (min_fd > entries || min_fd <= 0) - return -EINVAL; - - memcpy(&fds[i], cmsg_buf, sizeof(fds[0]) * min_fd); - memcpy(&aux->ent[i], amsg_buf, sizeof(aux->ent[0]) * min_fd); - memcpy(aux, mmsg_buf, offsetof(struct bpf_map_aux, ent)); - - needed = aux->num_ent; - } - - return 0; -} - -int bpf_send_map_fds(const char *path, const char *obj) -{ - struct bpf_elf_ctx *ctx = &__ctx; - struct sockaddr_un addr = { .sun_family = AF_UNIX }; - struct bpf_map_data bpf_aux = { - .fds = ctx->map_fds, - .ent = ctx->maps, - .st = &ctx->stat, - .obj = obj, - }; - int fd, ret = -1; - - fd = socket(AF_UNIX, SOCK_DGRAM, 0); - if (fd < 0) { - fprintf(stderr, "Cannot open socket: %s\n", - strerror(errno)); - goto out; - } - - strlcpy(addr.sun_path, path, sizeof(addr.sun_path)); - - ret = connect(fd, (struct sockaddr *)&addr, sizeof(addr)); - if (ret < 0) { - fprintf(stderr, "Cannot connect to %s: %s\n", - path, strerror(errno)); - goto out; - } - - ret = bpf_map_set_send(fd, &addr, sizeof(addr), &bpf_aux, - bpf_maps_count(ctx)); - if (ret < 0) - fprintf(stderr, "Cannot send fds to %s: %s\n", - path, strerror(errno)); - - bpf_maps_teardown(ctx); -out: - if (fd >= 0) - close(fd); - return ret; -} - -int bpf_recv_map_fds(const char *path, int *fds, struct bpf_map_aux *aux, - unsigned int entries) -{ - struct sockaddr_un addr = { .sun_family = AF_UNIX }; - int fd, ret = -1; - - fd = socket(AF_UNIX, SOCK_DGRAM, 0); - if (fd < 0) { - fprintf(stderr, "Cannot open socket: %s\n", - strerror(errno)); - goto out; - } - - strlcpy(addr.sun_path, path, sizeof(addr.sun_path)); - - ret = bind(fd, (struct sockaddr *)&addr, sizeof(addr)); - if (ret < 0) { - fprintf(stderr, "Cannot bind to socket: %s\n", - strerror(errno)); - goto out; - } - - ret = bpf_map_set_recv(fd, fds, aux, entries); - if (ret < 0) - fprintf(stderr, "Cannot recv fds from %s: %s\n", - path, strerror(errno)); - - unlink(addr.sun_path); - -out: - if (fd >= 0) - close(fd); - return ret; -} - #ifdef HAVE_LIBBPF /* The following functions are wrapper functions for libbpf code to be * compatible with the legacy format. So all the functions have prefix diff --git a/man/man8/tc-bpf.8 b/man/man8/tc-bpf.8 index 01230ce6..0e38e69b 100644 --- a/man/man8/tc-bpf.8 +++ b/man/man8/tc-bpf.8 @@ -10,8 +10,6 @@ queueing disciplines OBJ_FILE ] [ .B section CLS_NAME ] [ -.B export -UDS_FILE ] [ .B verbose ] [ .B direct-action @@ -35,8 +33,6 @@ CLASSID ] OBJ_FILE ] [ .B section CLS_NAME ] [ -.B export -UDS_FILE ] [ .B verbose ] @@ -158,17 +154,6 @@ that a single object file can contain multiple classifier and actions, the corresponding section name needs to be specified, if it differs from the defaults. -.SS export -points to a Unix domain socket file. In case the eBPF object file also -contains a section named "maps" with eBPF map specifications, then the -map file descriptors can be handed off via the Unix domain socket to -an eBPF "agent" herding all descriptors after tc lifetime. This can be -some third party application implementing the IPC counterpart for the -import, that uses them for calling into -.B bpf(2) -system call to read out or update eBPF map data from user space, for -example, for monitoring purposes or to push down new policies. - .SS verbose if set, it will dump the eBPF verifier output, even if loading the eBPF program was successful. By default, only on error, the verifier log is @@ -424,90 +409,20 @@ example: The advantage of this is that the classifier and the two actions can then share eBPF maps with each other, if implemented in the programs. -In order to access eBPF maps from user space beyond -.B tc(8) -setup lifetime, the ownership can be transferred to an eBPF agent via -Unix domain sockets. There are two possibilities for implementing this: - -.B 1) -implementation of an own eBPF agent that takes care of setting up -the Unix domain socket and implementing the protocol that +eBPF maps can be accessed from user space beyond .B tc(8) -dictates. A code example of this can be found inside the iproute2 -source package under: -.B examples/bpf/ - -.B 2) -use -.B tc exec -for transferring the eBPF map file descriptors through a Unix domain -socket, and spawning an application such as -.B sh(1) -\&. This approach's advantage is that tc will place the file descriptors -into the environment and thus make them available just like stdin, stdout, -stderr file descriptors, meaning, in case user applications run from within -this fd-owner shell, they can terminate and restart without losing eBPF -maps file descriptors. Example invocation with the previous classifier and -action mixture: - -.in +4n -.B tc exec bpf imp /tmp/bpf -.br -.B tc filter add dev em1 parent 1: bpf obj bpf.o exp /tmp/bpf flowid 1:1 \e -.br -.in +25n -.B action bpf obj bpf.o sec action-mark \e -.br -.B action bpf obj bpf.o sec action-rand ok -.in -25n -.in -4n - -Assuming that eBPF maps are shared with classifier and actions, it's -enough to export them once, for example, from within the classifier -or action command. tc will setup all eBPF map file descriptors at the -time when the object file is first parsed. - -When a shell has been spawned, the environment will have a couple of -eBPF related variables. BPF_NUM_MAPS provides the total number of maps -that have been transferred over the Unix domain socket. BPF_MAP's -value is the file descriptor number that can be accessed in eBPF agent -applications, in other words, it can directly be used as the file -descriptor value for the +setup lifetime by pinning them in the BPF file system. Maps declared +with the +.B pinning +attribute set to +.B LIBBPF_PIN_BY_NAME +are pinned under +.B /sys/fs/bpf/tc/globals/ +and can be inspected or updated with +.B bpftool(8) +or any program using the .B bpf(2) -system call to retrieve or alter eBPF map values. denotes the -identifier of the eBPF map. It corresponds to the -.B id -member of -.B struct bpf_elf_map -\& from the tc eBPF map specification. - -The environment in this example looks as follows: - -.in +4n -.nf -.sp -sh# env | grep BPF - BPF_NUM_MAPS=3 - BPF_MAP1=6 - BPF_MAP0=5 - BPF_MAP2=7 -sh# ls -la /proc/self/fd - [...] - lrwx------. 1 root root 64 Apr 14 16:46 5 -> anon_inode:bpf-map - lrwx------. 1 root root 64 Apr 14 16:46 6 -> anon_inode:bpf-map - lrwx------. 1 root root 64 Apr 14 16:46 7 -> anon_inode:bpf-map -sh# my_bpf_agent -.fi -.in - -eBPF agents are very useful in that they can prepopulate eBPF maps from -user space, monitor statistics via maps and based on that feedback, for -example, rewrite classids in eBPF map values during runtime. Given that eBPF -agents are implemented as normal applications, they can also dynamically -receive traffic control policies from external controllers and thus push -them down into eBPF maps to dynamically adapt to network conditions. Moreover, -eBPF maps can also be shared with other eBPF program types (e.g. tracing), -thus very powerful combination can therefore be implemented. +system call. .SS eBPF PROGRAMMING diff --git a/tc/e_bpf.c b/tc/e_bpf.c index cca853f9..391a3554 100644 --- a/tc/e_bpf.c +++ b/tc/e_bpf.c @@ -6,167 +6,60 @@ */ #include -#include #include "utils.h" #include "tc_util.h" #include "bpf_util.h" -#include "bpf_elf.h" -#include "bpf_scm.h" - -#define BPF_DEFAULT_CMD "/bin/sh" - -static char *argv_default[] = { BPF_DEFAULT_CMD, NULL }; static void explain(void) { fprintf(stderr, - "Usage: ... bpf [ import UDS_FILE ] [ run CMD ]\n" - " ... bpf [ debug ]\n" + "Usage: ... bpf [ debug ]\n" " ... bpf [ graft MAP_FILE ] [ key KEY ]\n" " `... [ object-file OBJ_FILE ] [ type TYPE ] [ section NAME ] [ verbose ]\n" " `... [ object-pinned PROG_FILE ]\n" "\n" - "Where UDS_FILE provides the name of a unix domain socket file\n" - "to import eBPF maps and the optional CMD denotes the command\n" - "to be executed (default: \'%s\').\n" "Where MAP_FILE points to a pinned map, OBJ_FILE to an object file\n" "and PROG_FILE to a pinned program. TYPE can be {cls, act}, where\n" "\'cls\' is default. KEY is optional and can be inferred from the\n" - "section name, otherwise it needs to be provided.\n", - BPF_DEFAULT_CMD); -} - -static int bpf_num_env_entries(void) -{ - char **envp; - int num; - - for (num = 0, envp = environ; *envp != NULL; envp++) - num++; - return num; + "section name, otherwise it needs to be provided.\n"); } static int parse_bpf(const struct exec_util *eu, int argc, char **argv) { - char **argv_run = argv_default, **envp_run, *tmp; - int ret, i, env_old, env_num, env_map; - const char *bpf_uds_name = NULL; - int fds[BPF_SCM_MAX_FDS] = {}; - struct bpf_map_aux aux = {}; - if (argc == 0) return 0; - while (argc > 0) { - if (matches(*argv, "run") == 0) { - NEXT_ARG(); - argv_run = argv; - break; - } else if (matches(*argv, "import") == 0) { - NEXT_ARG(); - bpf_uds_name = *argv; - } else if (matches(*argv, "debug") == 0 || - matches(*argv, "dbg") == 0) { - if (bpf_trace_pipe()) - fprintf(stderr, - "No trace pipe, tracefs not mounted?\n"); - return -1; - } else if (matches(*argv, "graft") == 0) { - const char *bpf_map_path; - bool has_key = false; - uint32_t key; - - NEXT_ARG(); - bpf_map_path = *argv; + if (matches(*argv, "debug") == 0 || + matches(*argv, "dbg") == 0) { + if (bpf_trace_pipe()) + fprintf(stderr, + "No trace pipe, tracefs not mounted?\n"); + return -1; + } else if (matches(*argv, "graft") == 0) { + const char *bpf_map_path; + bool has_key = false; + uint32_t key; + + NEXT_ARG(); + bpf_map_path = *argv; + NEXT_ARG(); + if (matches(*argv, "key") == 0) { NEXT_ARG(); - if (matches(*argv, "key") == 0) { - NEXT_ARG(); - if (get_unsigned(&key, *argv, 0)) { - fprintf(stderr, "Illegal \"key\"\n"); - return -1; - } - has_key = true; - NEXT_ARG(); + if (get_unsigned(&key, *argv, 0)) { + fprintf(stderr, "Illegal \"key\"\n"); + return -1; } - return bpf_graft_map(bpf_map_path, has_key ? - &key : NULL, argc, argv); - } else { - explain(); - return -1; + has_key = true; + NEXT_ARG(); } - - NEXT_ARG_FWD(); + return bpf_graft_map(bpf_map_path, has_key ? + &key : NULL, argc, argv); } - if (!bpf_uds_name) { - fprintf(stderr, "bpf: No import parameter provided!\n"); - explain(); - return -1; - } - - if (argv_run != argv_default && argc == 0) { - fprintf(stderr, "bpf: No run command provided!\n"); - explain(); - return -1; - } - - ret = bpf_recv_map_fds(bpf_uds_name, fds, &aux, ARRAY_SIZE(fds)); - if (ret < 0) { - fprintf(stderr, "bpf: Could not receive fds!\n"); - return -1; - } - - if (aux.num_ent == 0) { - envp_run = environ; - goto out; - } - - env_old = bpf_num_env_entries(); - env_num = env_old + aux.num_ent + 2; - env_map = env_old + 1; - - envp_run = malloc(sizeof(*envp_run) * env_num); - if (!envp_run) { - fprintf(stderr, "bpf: No memory left to allocate env!\n"); - goto err; - } - - for (i = 0; i < env_old; i++) - envp_run[i] = environ[i]; - - ret = asprintf(&tmp, "BPF_NUM_MAPS=%u", aux.num_ent); - if (ret < 0) - goto err_free; - - envp_run[env_old] = tmp; - - for (i = env_map; i < env_num - 1; i++) { - ret = asprintf(&tmp, "BPF_MAP%u=%u", - aux.ent[i - env_map].id, - fds[i - env_map]); - if (ret < 0) - goto err_free_env; - - envp_run[i] = tmp; - } - - envp_run[env_num - 1] = NULL; -out: - ret = execvpe(argv_run[0], argv_run, envp_run); - free(envp_run); - return ret; - -err_free_env: - for (--i; i >= env_old; i--) - free(envp_run[i]); -err_free: - free(envp_run); -err: - for (i = 0; i < aux.num_ent; i++) - close(fds[i]); + explain(); return -1; } diff --git a/tc/f_bpf.c b/tc/f_bpf.c index 50fe01ca..a4be332d 100644 --- a/tc/f_bpf.c +++ b/tc/f_bpf.c @@ -27,7 +27,7 @@ static void explain(void) " bytecode-file FILE\n" "\n" "eBPF use case:\n" - " object-file FILE [ section CLS_NAME ] [ export UDS_FILE ]" + " object-file FILE [ section CLS_NAME ]" " [ verbose ] [ direct-action ] [ skip_hw | skip_sw ]\n" " object-pinned FILE [ direct-action ] [ skip_hw | skip_sw ]\n" "\n" @@ -74,7 +74,6 @@ static const struct bpf_cfg_ops bpf_cb_ops = { static int bpf_parse_opt(const struct filter_util *qu, char *handle, int argc, char **argv, struct nlmsghdr *n) { - const char *bpf_obj = NULL, *bpf_uds_name = NULL; struct tcmsg *t = NLMSG_DATA(n); unsigned int bpf_gen_flags = 0; unsigned int bpf_flags = 0; @@ -82,7 +81,6 @@ static int bpf_parse_opt(const struct filter_util *qu, char *handle, bool seen_run = false; bool skip_sw = false; struct rtattr *tail; - int ret = 0; if (handle) { if (get_u32(&t->tcm_handle, handle, 0)) { @@ -117,9 +115,6 @@ opt_bpf: argc = cfg.argc; argv = cfg.argv; - - bpf_obj = cfg.object; - bpf_uds_name = cfg.uds; } else if (matches(*argv, "classid") == 0 || matches(*argv, "flowid") == 0) { unsigned int classid; @@ -181,10 +176,7 @@ opt_bpf: tail->rta_len = (((void *)n) + n->nlmsg_len) - (void *)tail; - if (bpf_uds_name) - ret = bpf_send_map_fds(bpf_uds_name, bpf_obj); - - return ret; + return 0; } static int bpf_print_opt(const struct filter_util *qu, FILE *f, diff --git a/tc/m_bpf.c b/tc/m_bpf.c index 83b21b57..e893756c 100644 --- a/tc/m_bpf.c +++ b/tc/m_bpf.c @@ -29,7 +29,7 @@ static void explain(void) " bytecode-file FILE\n" "\n" "eBPF use case:\n" - " object-file FILE [ section ACT_NAME ] [ export UDS_FILE ]" + " object-file FILE [ section ACT_NAME ]" " [ verbose ]\n" " object-pinned FILE\n" "\n" @@ -72,12 +72,11 @@ static const struct bpf_cfg_ops bpf_cb_ops = { static int bpf_parse_opt(const struct action_util *a, int *ptr_argc, char ***ptr_argv, int tca_id, struct nlmsghdr *n) { - const char *bpf_obj = NULL, *bpf_uds_name = NULL; struct tc_act_bpf parm = {}; struct bpf_cfg_in cfg = {}; bool seen_run = false; struct rtattr *tail; - int argc, ret = 0; + int argc; char **argv; argv = *ptr_argv; @@ -107,9 +106,6 @@ opt_bpf: argc = cfg.argc; argv = cfg.argv; - - bpf_obj = cfg.object; - bpf_uds_name = cfg.uds; } else if (matches(*argv, "help") == 0) { explain(); return -1; @@ -142,13 +138,10 @@ opt_bpf: addattr_l(n, MAX_MSG, TCA_ACT_BPF_PARMS, &parm, sizeof(parm)); addattr_nest_end(n, tail); - if (bpf_uds_name) - ret = bpf_send_map_fds(bpf_uds_name, bpf_obj); - *ptr_argc = argc; *ptr_argv = argv; - return ret; + return 0; } static int bpf_print_opt(const struct action_util *au, FILE *f, struct rtattr *arg) -- 2.53.0