bpf: Add bpf_ktime_get_real_ns
Subsystems (12)
PATCH v2 1/2 bpf: Add bpf_ktime_get_real_ns
From: Bimmy Pujari <[email protected]>
The existing bpf helper functions to get timestamp return the time elapsed since system boot. This timestamp is not particularly useful where epoch timestamp is required or more than one server is involved and time sync is required. Instead, you want to use CLOCK_REALTIME, which provides epoch timestamp.
Hence add bfp_ktime_get_real_ns() based around CLOCK_REALTIME.
--- drivers/media/rc/bpf-lirc.c | 2 ++ include/linux/bpf.h | 1 + include/uapi/linux/bpf.h | 8 ++++++++ kernel/bpf/core.c | 1 + kernel/bpf/helpers.c | 13 +++++++++++++ kernel/trace/bpf_trace.c | 2 ++ tools/include/uapi/linux/bpf.h | 8 ++++++++ 7 files changed, 35 insertions(+)
drivers/media/rc/bpf-lirc.c +2 -0
@@ -103,6 +103,8 @@ lirc_mode2_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)return &bpf_map_peek_elem_proto;case BPF_FUNC_ktime_get_ns:return &bpf_ktime_get_ns_proto; +case BPF_FUNC_ktime_get_real_ns: +return &bpf_ktime_get_real_ns_proto;case BPF_FUNC_ktime_get_boot_ns:return &bpf_ktime_get_boot_ns_proto;case BPF_FUNC_tail_call:
include/linux/bpf.h +1 -0
@@ -1756,6 +1756,7 @@ extern const struct bpf_func_proto bpf_get_smp_processor_id_proto;extern const struct bpf_func_proto bpf_get_numa_node_id_proto;extern const struct bpf_func_proto bpf_tail_call_proto;extern const struct bpf_func_proto bpf_ktime_get_ns_proto; +extern const struct bpf_func_proto bpf_ktime_get_real_ns_proto;extern const struct bpf_func_proto bpf_ktime_get_boot_ns_proto;extern const struct bpf_func_proto bpf_get_current_pid_tgid_proto;extern const struct bpf_func_proto bpf_get_current_uid_gid_proto;
include/uapi/linux/bpf.h +8 -0
@@ -3586,6 +3586,13 @@ union bpf_attr {* the data in *dst*. This is a wrapper of **copy_from_user**\ (). * Return * 0 on success, or a negative error in case of failure. +* +* u64 bpf_ktime_get_real_ns(void) +* Description +* Return the real time in nanoseconds. +* See: **clock_gettime**\ (**CLOCK_REALTIME**) +* Return +* Current *ktime*. */ #define __BPF_FUNC_MAPPER (FN ) \ FN(unspec), \
@@ -3737,6 +3744,7 @@ union bpf_attr {FN (inode_storage_delete), \FN (d_path), \FN (copy_from_user), \ +FN (ktime_get_real_ns), \/* */ /* integer value in 'imm' field of BPF_CALL instruction selects which helper
kernel/bpf/core.c +1 -0
@@ -2207,6 +2207,7 @@ const struct bpf_func_proto bpf_get_prandom_u32_proto __weak;const struct bpf_func_proto bpf_get_smp_processor_id_proto__weak ;const struct bpf_func_proto bpf_get_numa_node_id_proto__weak ;const struct bpf_func_proto bpf_ktime_get_ns_proto__weak ; +const struct bpf_func_proto bpf_ktime_get_real_ns_proto__weak ;const struct bpf_func_proto bpf_ktime_get_boot_ns_proto__weak ;const struct bpf_func_proto bpf_get_current_pid_tgid_proto__weak ;
kernel/bpf/helpers.c +13 -0
@@ -155,6 +155,17 @@ const struct bpf_func_proto bpf_ktime_get_ns_proto = { .ret_type = RET_INTEGER, }; +BPF_CALL_0 (bpf_ktime_get_real_ns) +{ +/* NMI safe access to clock realtime */ +return ktime_get_real_fast_ns (); +} + +const struct bpf_func_proto bpf_ktime_get_real_ns_proto = { + .func = bpf_ktime_get_real_ns, + .gpl_only = true, + .ret_type =RET_INTEGER , +};BPF_CALL_0 (bpf_ktime_get_boot_ns) {/* NMI safe access to clock boottime */
@@ -655,6 +666,8 @@ bpf_base_func_proto(enum bpf_func_id func_id)return &bpf_tail_call_proto;case BPF_FUNC_ktime_get_ns:return &bpf_ktime_get_ns_proto; +case BPF_FUNC_ktime_get_real_ns: +return &bpf_ktime_get_real_ns_proto;case BPF_FUNC_ktime_get_boot_ns:return &bpf_ktime_get_boot_ns_proto;case BPF_FUNC_ringbuf_output:
kernel/trace/bpf_trace.c +2 -0
@@ -1165,6 +1165,8 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)return &bpf_map_peek_elem_proto;case BPF_FUNC_ktime_get_ns:return &bpf_ktime_get_ns_proto; +case BPF_FUNC_ktime_get_real_ns: +return &bpf_ktime_get_real_ns_proto;case BPF_FUNC_ktime_get_boot_ns:return &bpf_ktime_get_boot_ns_proto;case BPF_FUNC_tail_call:
tools/include/uapi/linux/bpf.h +8 -0
@@ -3586,6 +3586,13 @@ union bpf_attr {* the data in *dst*. This is a wrapper of **copy_from_user**\ (). * Return * 0 on success, or a negative error in case of failure. +* +* u64 bpf_ktime_get_real_ns(void) +* Description +* Return the real time in nanoseconds. +* See: **clock_gettime**\ (**CLOCK_REALTIME**) +* Return +* Current *ktime*. */ #define __BPF_FUNC_MAPPER (FN ) \ FN(unspec), \
@@ -3737,6 +3744,7 @@ union bpf_attr {FN (inode_storage_delete), \FN (d_path), \FN (copy_from_user), \ +FN (ktime_get_real_ns), \/* */ /* integer value in 'imm' field of BPF_CALL instruction selects which helper
Signature · 1 line
2.17.1
PATCH v2 2/2 selftests/bpf: Verifying real time helper function
From: Bimmy Pujari <[email protected]>
Test xdping measures RTT from xdp using monotonic time helper. Extending xdping test to use real time helper function in order to verify this helper function.
--- .../testing/selftests/bpf/progs/xdping_kern.c | 183 +---------------- .../testing/selftests/bpf/progs/xdping_kern.h | 193 ++++++++++++++++++ .../bpf/progs/xdping_realtime_kern.c | 4 + tools/testing/selftests/bpf/test_xdping.sh | 44 +++- 4 files changed, 235 insertions(+), 189 deletions(-) create mode 100644 tools/testing/selftests/bpf/progs/xdping_kern.h create mode 100644 tools/testing/selftests/bpf/progs/xdping_realtime_kern.c
tools/testing/selftests/bpf/progs/xdping_kern.c +2 -181
@@ -1,184 +1,5 @@// SPDX-License-Identifier: GPL-2.0 -/* Copyright (c) 2019, Oracle and/or its affiliates. All rights reserved. */ +#undef REALTIME -#define KBUILD_MODNAME "foo" -#include <stddef.h> -#include <string.h> -#include <linux/bpf.h> -#include <linux/icmp.h> -#include <linux/in.h> -#include <linux/if_ether.h> -#include <linux/if_packet.h> -#include <linux/if_vlan.h> -#include <linux/ip.h> +#include "xdping_kern.h" -#include <bpf/bpf_helpers.h> -#include <bpf/bpf_endian.h> - -#include "xdping.h" - -struct { - __uint(type, BPF_MAP_TYPE_HASH); - __uint(max_entries, 256); - __type(key, __u32); - __type(value, struct pinginfo); -} ping_map SEC(".maps"); - -static __always_inline void swap_src_dst_mac(void *data) -{ -unsigned short *p = data; -unsigned short dst[3 ]; - - dst[0 ] = p[0 ]; - dst[1 ] = p[1 ]; - dst[2 ] = p[2 ]; - p[0 ] = p[3 ]; - p[1 ] = p[4 ]; - p[2 ] = p[5 ]; - p[3 ] = dst[0 ]; - p[4 ] = dst[1 ]; - p[5 ] = dst[2 ]; -} - -static __always_inline __u16 csum_fold_helper(__wsum sum) -{ - sum = (sum &0xffff ) + (sum >>16 ); -return ~((sum &0xffff ) + (sum >>16 )); -} - -static __always_inline __u16 ipv4_csum(void *data_start, int data_size) -{ -__wsum sum; - - sum =bpf_csum_diff (0 ,0 , data_start, data_size,0 ); -return csum_fold_helper (sum); -} - -#define ICMP_ECHO_LEN 64 - -static __always_inline int icmp_check(struct xdp_md *ctx, int type) -{ -void *data_end = (void *)(long )ctx->data_end; -void *data = (void *)(long )ctx->data; -struct ethhdr *eth = data; -struct icmphdr *icmph; -struct iphdr *iph; - -if (data +sizeof (*eth) +sizeof (*iph) +ICMP_ECHO_LEN > data_end) -return XDP_PASS ; - -if (eth->h_proto !=bpf_htons (ETH_P_IP )) -return XDP_PASS ; - - iph = data +sizeof (*eth); - -if (iph->protocol !=IPPROTO_ICMP ) -return XDP_PASS ; - -if (bpf_ntohs (iph->tot_len) -sizeof (*iph) !=ICMP_ECHO_LEN ) -return XDP_PASS ; - - icmph = data +sizeof (*eth) +sizeof (*iph); - -if (icmph->type != type) -return XDP_PASS ; - -return XDP_TX ; -} - -SEC("xdpclient") -int xdping_client (struct xdp_md *ctx) -{ -void *data_end = (void *)(long )ctx->data_end; -void *data = (void *)(long )ctx->data; -struct pinginfo *pinginfo =NULL ; -struct ethhdr *eth = data; -struct icmphdr *icmph; -struct iphdr *iph; -__u64 recvtime; -__be32 raddr; -__be16 seq; -int ret; -__u8 i; - - ret =icmp_check (ctx,ICMP_ECHOREPLY ); - -if (ret !=XDP_TX ) -return ret; - - iph = data +sizeof (*eth); - icmph = data +sizeof (*eth) +sizeof (*iph); - raddr = iph->saddr; - -/* Record time reply received. */ - recvtime =bpf_ktime_get_ns (); - pinginfo =bpf_map_lookup_elem (&ping_map, &raddr); -if (!pinginfo || pinginfo->seq != icmph->un.echo.sequence) -return XDP_PASS ; - -if (pinginfo->start) { -#pragma clang loop unroll(full) -for (i =0 ; i <XDPING_MAX_COUNT ; i++) { -if (pinginfo->times[i] ==0 ) -break ; - } -/* verifier is fussy here... */ -if (i <XDPING_MAX_COUNT ) { - pinginfo->times[i] = recvtime - - pinginfo->start; - pinginfo->start =0 ; - i++; - } -/* No more space for values? */ -if (i == pinginfo->count || i ==XDPING_MAX_COUNT ) -return XDP_PASS ; - } - -/* Now convert reply back into echo request. */ -swap_src_dst_mac (data); - iph->saddr = iph->daddr; - iph->daddr = raddr; - icmph->type =ICMP_ECHO ; - seq =bpf_htons (bpf_ntohs (icmph->un.echo.sequence) +1 ); - icmph->un.echo.sequence = seq; - icmph->checksum =0 ; - icmph->checksum =ipv4_csum (icmph,ICMP_ECHO_LEN ); - - pinginfo->seq = seq; - pinginfo->start =bpf_ktime_get_ns (); - -return XDP_TX ; -} - -SEC("xdpserver") -int xdping_server (struct xdp_md *ctx) -{ -void *data_end = (void *)(long )ctx->data_end; -void *data = (void *)(long )ctx->data; -struct ethhdr *eth = data; -struct icmphdr *icmph; -struct iphdr *iph; -__be32 raddr; -int ret; - - ret =icmp_check (ctx,ICMP_ECHO ); - -if (ret !=XDP_TX ) -return ret; - - iph = data +sizeof (*eth); - icmph = data +sizeof (*eth) +sizeof (*iph); - raddr = iph->saddr; - -/* Now convert request into echo reply. */ -swap_src_dst_mac (data); - iph->saddr = iph->daddr; - iph->daddr = raddr; - icmph->type =ICMP_ECHOREPLY ; - icmph->checksum =0 ; - icmph->checksum =ipv4_csum (icmph,ICMP_ECHO_LEN ); - -return XDP_TX ; -} - -char _license[] SEC("license") = "GPL";
tools/testing/selftests/bpf/progs/xdping_kern.h new file +193 -0
193 lines not on this page · raw message
tools/testing/selftests/bpf/progs/xdping_realtime_kern.c new file +4 -0
@@ -0,0 +1,4 @@ +// SPDX-License-Identifier: GPL-2.0 +#define REALTIME + +#include "xdping_kern.h"
tools/testing/selftests/bpf/test_xdping.sh +36 -8
74 lines not on this page · raw message
Signature · 1 line
2.17.1
Quoted 92 lines · On Wed, Sep 23, 2020 at 5:03 PM <[email protected]> wrote:
From: Bimmy Pujari <[email protected]>
The existing bpf helper functions to get timestamp return the time elapsed since system boot. This timestamp is not particularly useful where epoch timestamp is required or more than one server is involved and time sync is required. Instead, you want to use
CLOCK_REALTIME, which provides epoch timestamp.
Hence addbfp_ktime_get_real_ns()based aroundCLOCK_REALTIME.Signed-off-by Ashkan Nikravesh <[email protected]>Signed-off-by Bimmy Pujari <[email protected]>--- drivers/media/rc/bpf-lirc.c | 2 ++ include/linux/bpf.h | 1 + include/uapi/linux/bpf.h | 8 ++++++++ kernel/bpf/core.c | 1 + kernel/bpf/helpers.c | 13 +++++++++++++ kernel/trace/bpf_trace.c | 2 ++ tools/include/uapi/linux/bpf.h | 8 ++++++++ 7 files changed, 35 insertions(+)drivers/media/rc/bpf-lirc.c +2
-0@@ -103,6 +103,8 @@ lirc_mode2_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)return &bpf_map_peek_elem_proto;case BPF_FUNC_ktime_get_ns:return &bpf_ktime_get_ns_proto; +case BPF_FUNC_ktime_get_real_ns: +return &bpf_ktime_get_real_ns_proto;case BPF_FUNC_ktime_get_boot_ns:return &bpf_ktime_get_boot_ns_proto;case BPF_FUNC_tail_call:include/linux/bpf.h +1
-0@@ -1756,6 +1756,7 @@ extern const struct bpf_func_proto bpf_get_smp_processor_id_proto;extern const struct bpf_func_proto bpf_get_numa_node_id_proto;extern const struct bpf_func_proto bpf_tail_call_proto;extern const struct bpf_func_proto bpf_ktime_get_ns_proto; +extern const struct bpf_func_proto bpf_ktime_get_real_ns_proto;extern const struct bpf_func_proto bpf_ktime_get_boot_ns_proto;extern const struct bpf_func_proto bpf_get_current_pid_tgid_proto;extern const struct bpf_func_proto bpf_get_current_uid_gid_proto;include/uapi/linux/bpf.h +8
-0@@ -3586,6 +3586,13 @@ union bpf_attr {* the data in *dst*. This is a wrapper of **copy_from_user**\ (). * Return * 0 on success, or a negative error in case of failure. +* +* u64 bpf_ktime_get_real_ns(void) +* Description +* Return the real time in nanoseconds. +* See: **clock_gettime**\ (**CLOCK_REALTIME**) +* Return +* Current *ktime*. */ #define __BPF_FUNC_MAPPER (FN ) \ FN(unspec), \@@ -3737,6 +3744,7 @@ union bpf_attr {FN (inode_storage_delete), \FN (d_path), \FN (copy_from_user), \ +FN (ktime_get_real_ns), \/* */ /* integer value in 'imm' field of BPF_CALL instruction selects which helper kernel/bpf/core.c +1
-0@@ -2207,6 +2207,7 @@ const struct bpf_func_proto bpf_get_prandom_u32_proto __weak;const struct bpf_func_proto bpf_get_smp_processor_id_proto__weak ;const struct bpf_func_proto bpf_get_numa_node_id_proto__weak ;const struct bpf_func_proto bpf_ktime_get_ns_proto__weak ; +const struct bpf_func_proto bpf_ktime_get_real_ns_proto__weak ;const struct bpf_func_proto bpf_ktime_get_boot_ns_proto__weak ;const struct bpf_func_proto bpf_get_current_pid_tgid_proto__weak ;kernel/bpf/helpers.c +0
-0@@ -155,6 +155,17 @@ const struct bpf_func_proto bpf_ktime_get_ns_proto = { .ret_type = RET_INTEGER, };
kernel/bpf/helpers.c const struct bpf_func_proto bpf_ktime_get_ns_proto = { · line 166
Quoted 10 lines
kernel/bpf/helpers.c +9
-0+BPF_CALL_0 (bpf_ktime_get_real_ns) +{ +/* NMI safe access to clock realtime */ +return ktime_get_real_fast_ns (); +} + +const struct bpf_func_proto bpf_ktime_get_real_ns_proto = { + .func = bpf_ktime_get_real_ns, + .gpl_only = true,
imho should be false, this is normally accessible to userspace code via syscall, no reason why it should be gpl only for bpf
Quoted 19 lines
kernel/bpf/helpers.c +4
-0+ .ret_type = RET_INTEGER, +};BPF_CALL_0 (bpf_ktime_get_boot_ns) {/* NMI safe access to clock boottime */ @@ -655,6 +666,8 @@ bpf_base_func_proto(enum bpf_func_id func_id)return &bpf_tail_call_proto;case BPF_FUNC_ktime_get_ns:return &bpf_ktime_get_ns_proto; +case BPF_FUNC_ktime_get_real_ns: +return &bpf_ktime_get_real_ns_proto;case BPF_FUNC_ktime_get_boot_ns:return &bpf_ktime_get_boot_ns_proto;case BPF_FUNC_ringbuf_output:kernel/trace/bpf_trace.c +0
-0
kernel/trace/bpf_trace.c bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog) · line 1171
Quoted 7 lines
kernel/trace/bpf_trace.c +2
-0@@ -1165,6 +1165,8 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)return &bpf_map_peek_elem_proto;case BPF_FUNC_ktime_get_ns:return &bpf_ktime_get_ns_proto; +case BPF_FUNC_ktime_get_real_ns: +return &bpf_ktime_get_real_ns_proto;case BPF_FUNC_ktime_get_boot_ns:return &bpf_ktime_get_boot_ns_proto;
should probably be here to stay alpha sorted more or less (also applies to other places)
Quoted 28 lines
kernel/trace/bpf_trace.c +0
-0case BPF_FUNC_tail_call:tools/include/uapi/linux/bpf.h +8
-1@@ -3586,6 +3586,13 @@ union bpf_attr {* the data in *dst*. This is a wrapper of **copy_from_user**\ (). * Return * 0 on success, or a negative error in case of failure. +* +* u64 bpf_ktime_get_real_ns(void) +* Description +* Return the real time in nanoseconds. +* See: **clock_gettime**\ (**CLOCK_REALTIME**) +* Return +* Current *ktime*. */ #define __BPF_FUNC_MAPPER (FN ) \ FN(unspec), \@@ -3737,6 +3744,7 @@ union bpf_attr {FN (inode_storage_delete), \FN (d_path), \FN (copy_from_user), \ +FN (ktime_get_real_ns), \/* */ /* integer value in 'imm' field of BPF_CALL instruction selects which helper --
Quoted 2 lines
2.17.1
Maciej Żenczykowski, Kernel Networking Developer @ Google
Quoted 8 lines · Maciej Żenczykowski
Quoted 8 lines
kernel/bpf/helpers.c +0
-0@@ -155,6 +155,17 @@ const struct bpf_func_proto bpf_ktime_get_ns_proto = { .ret_type = RET_INTEGER, };
Quoted 13 lines
kernel/bpf/helpers.c const struct bpf_func_proto bpf_ktime_get_ns_proto = { · line 166
Quoted 9 lines
kernel/bpf/helpers.c +9
-0+BPF_CALL_0 (bpf_ktime_get_real_ns) +{ +/* NMI safe access to clock realtime */ +return ktime_get_real_fast_ns (); +} + +const struct bpf_func_proto bpf_ktime_get_real_ns_proto = { + .func = bpf_ktime_get_real_ns, + .gpl_only = true,imho should be false, this is normally accessible to userspace code via syscall, no reason why it should be gpl only for bpf
agreed, no reason for the bpf hook to be gpl_only.
Glad to see the v2 of this patch; I wondered what happened to this helper.
-----Original Message-----
From: David Ahern <[email protected]>
Sent: Wednesday, September 23, 2020 7:19 PM
To: Maciej Żenczykowski <[email protected]>; Pujari, Bimmy <[email protected]>
Cc: bpf <[email protected]>; Linux NetDev <[email protected]>; [email protected]; Alexei Starovoitov <[email protected]>; Daniel Borkmann <[email protected]>; Martin Lau <[email protected]>; Nikravesh, Ashkan <[email protected]>
Subject: Re: [PATCH bpf-next v2 1/2] bpf: Add bpf_ktime_get_real_ns
Quoted 8 lines · Maciej Żenczykowski
Quoted 8 lines
kernel/bpf/helpers.c index renamed from kernel/bpf/helpers.c +0
-0
5cc7425ee476..776ff58f969d100644kernel/bpf/helpers.c +0
-0@@ -155,6 +155,17 @@ const struct bpf_func_proto bpf_ktime_get_ns_proto = { .ret_type = RET_INTEGER, };
Quoted 12 lines
kernel/bpf/helpers.c const struct bpf_func_proto bpf_ktime_get_ns_proto = { · line 166
Quoted 8 lines
kernel/bpf/helpers.c +8
-0+BPF_CALL_0 (bpf_ktime_get_real_ns) +{ +/* NMI safe access to clock realtime */ +return ktime_get_real_fast_ns (); } + +const struct bpf_func_proto bpf_ktime_get_real_ns_proto = { + .func = bpf_ktime_get_real_ns, + .gpl_only = true,imho should be false, this is normally accessible to userspace code via syscall, no reason why it should be gpl only for bpf
agreed, no reason for the bpf hook to be gpl_only.
Glad to see the v2 of this patch; I wondered what happened to this helper.
BP: Thanks Maciej & Ahern for taking time to look at it. I will update the changes suggested shortly.