Skip to content

L
linux
Switch branch/tag
  1. 29 Apr, 2020 26 commits
  3. 28 Apr, 2020 4 commits
  5. 27 Apr, 2020 6 commits
  7. 26 Apr, 2020 4 commits
    • Alexei Starovoitov's avatar
      Merge branch 'cloudflare-prog' · f131bd3e
      Alexei Starovoitov authored 
      Lorenz Bauer says:

====================
We've been developing an in-house L4 load balancer based on XDP
and TC for a while. Following Alexei's call for more up-to-date examples of
production BPF in the kernel tree [1], Cloudflare is making this available
under dual GPL-2.0 or BSD 3-clause terms.

The code requires at least v5.3 to function correctly.

1: https://lore.kernel.org/bpf/20200326210719.den5isqxntnoqhmv@ast-mbp/
====================
Signed-off-by: default avatarAlexei Starovoitov <ast@kernel.org>
      f131bd3e
    • Lorenz Bauer's avatar
      selftests/bpf: Add cls_redirect classifier · 23458901
      Lorenz Bauer authored 
      cls_redirect is a TC clsact based replacement for the glb-redirect iptables
module available at [1]. It enables what GitHub calls "second chance"
flows [2], similarly proposed by the Beamer paper [3]. In contrast to
glb-redirect, it also supports migrating UDP flows as long as connected
sockets are used. cls_redirect is in production at Cloudflare, as part of
our own L4 load balancer.

We have modified the encapsulation format slightly from glb-redirect:
glbgue_chained_routing.private_data_type has been repurposed to form a
version field and several flags. Both have been arranged in a way that
a private_data_type value of zero matches the current glb-redirect
behaviour. This means that cls_redirect will understand packets in
glb-redirect format, but not vice versa.

The test suite only covers basic features. For example, cls_redirect will
correctly forward path MTU discovery packets, but this is not exercised.
It is also possible to switch the encapsulation format to GRE on the last
hop, which is also not tested.

There are two major distinctions from glb-redirect: first, cls_redirect
relies on receiving encapsulated packets directly from a router. This is
because we don't have access to the neighbour tables from BPF, yet. See
forward_to_next_hop for details. Second, cls_redirect performs decapsulation
instead of using separate ipip and sit tunnel devices. This
avoids issues with the sit tunnel [4] and makes deploying the classifier
easier: decapsulated packets appear on the same interface, so existing
firewall rules continue to work as expected.

The code base started it's life on v4.19, so there are most likely still
hold overs from old workarounds. In no particular order:

- The function buf_off is required to defeat a clang optimization
  that leads to the verifier rejecting the program due to pointer
  arithmetic in the wrong order.

- The function pkt_parse_ipv6 is force inlined, because it would
  otherwise be rejected due to returning a pointer to stack memory.

- The functions fill_tuple and classify_tcp contain kludges, because
  we've run out of function arguments.

- The logic in general is rather nested, due to verifier restrictions.
  I think this is either because the verifier loses track of constants
  on the stack, or because it can't track enum like variables.

1: https://github.com/github/glb-director/tree/master/src/glb-redirect
2: https://github.com/github/glb-director/blob/master/docs/development/second-chance-design.md
3: https://www.usenix.org/conference/nsdi18/presentation/olteanu
4: https://github.com/github/glb-director/issues/64Signed-off-by: default avatarLorenz Bauer <lmb@cloudflare.com>
Signed-off-by: default avatarAlexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200424185556.7358-2-lmb@cloudflare.com
      23458901
    • Andrii Nakryiko's avatar
      bpf: Make verifier log more relevant by default · 6f8a57cc
      Andrii Nakryiko authored 
      To make BPF verifier verbose log more releavant and easier to use to debug
verification failures, "pop" parts of log that were successfully verified.
This has effect of leaving only verifier logs that correspond to code branches
that lead to verification failure, which in practice should result in much
shorter and more relevant verifier log dumps. This behavior is made the
default behavior and can be overriden to do exhaustive logging by specifying
BPF_LOG_LEVEL2 log level.

Using BPF_LOG_LEVEL2 to disable this behavior is not ideal, because in some
cases it's good to have BPF_LOG_LEVEL2 per-instruction register dump
verbosity, but still have only relevant verifier branches logged. But for this
patch, I didn't want to add any new flags. It might be worth-while to just
rethink how BPF verifier logging is performed and requested and streamline it
a bit. But this trimming of successfully verified branches seems to be useful
and a good default behavior.

To test this, I modified runqslower slightly to introduce read of
uninitialized stack variable. Log (**truncated in the middle** to save many
lines out of this commit message) BEFORE this change:

; int handle__sched_switch(u64 *ctx)
0: (bf) r6 = r1
; struct task_struct *prev = (struct task_struct *)ctx[1];
1: (79) r1 = *(u64 *)(r6 +8)
func 'sched_switch' arg1 has btf_id 151 type STRUCT 'task_struct'
2: (b7) r2 = 0
; struct event event = {};
3: (7b) *(u64 *)(r10 -24) = r2
last_idx 3 first_idx 0
regs=4 stack=0 before 2: (b7) r2 = 0
4: (7b) *(u64 *)(r10 -32) = r2
5: (7b) *(u64 *)(r10 -40) = r2
6: (7b) *(u64 *)(r10 -48) = r2
; if (prev->state == TASK_RUNNING)

[ ... instruction dump from insn #7 through #50 are cut out ... ]

51: (b7) r2 = 16
52: (85) call bpf_get_current_comm#16
last_idx 52 first_idx 42
regs=4 stack=0 before 51: (b7) r2 = 16
; bpf_perf_event_output(ctx, &events, BPF_F_CURRENT_CPU,
53: (bf) r1 = r6
54: (18) r2 = 0xffff8881f3868800
56: (18) r3 = 0xffffffff
58: (bf) r4 = r7
59: (b7) r5 = 32
60: (85) call bpf_perf_event_output#25
last_idx 60 first_idx 53
regs=20 stack=0 before 59: (b7) r5 = 32
61: (bf) r2 = r10
; event.pid = pid;
62: (07) r2 += -16
; bpf_map_delete_elem(&start, &pid);
63: (18) r1 = 0xffff8881f3868000
65: (85) call bpf_map_delete_elem#3
; }
66: (b7) r0 = 0
67: (95) exit

from 44 to 66: safe

from 34 to 66: safe

from 11 to 28: R1_w=inv0 R2_w=inv0 R6_w=ctx(id=0,off=0,imm=0) R10=fp0 fp-8=mmmm???? fp-24_w=00000000 fp-32_w=00000000 fp-40_w=00000000 fp-48_w=00000000
; bpf_map_update_elem(&start, &pid, &ts, 0);
28: (bf) r2 = r10
;
29: (07) r2 += -16
; tsp = bpf_map_lookup_elem(&start, &pid);
30: (18) r1 = 0xffff8881f3868000
32: (85) call bpf_map_lookup_elem#1
invalid indirect read from stack off -16+0 size 4
processed 65 insns (limit 1000000) max_states_per_insn 1 total_states 5 peak_states 5 mark_read 4

Notice how there is a successful code path from instruction 0 through 67, few
successfully verified jumps (44->66, 34->66), and only after that 11->28 jump
plus error on instruction #32.

AFTER this change (full verifier log, **no truncation**):

; int handle__sched_switch(u64 *ctx)
0: (bf) r6 = r1
; struct task_struct *prev = (struct task_struct *)ctx[1];
1: (79) r1 = *(u64 *)(r6 +8)
func 'sched_switch' arg1 has btf_id 151 type STRUCT 'task_struct'
2: (b7) r2 = 0
; struct event event = {};
3: (7b) *(u64 *)(r10 -24) = r2
last_idx 3 first_idx 0
regs=4 stack=0 before 2: (b7) r2 = 0
4: (7b) *(u64 *)(r10 -32) = r2
5: (7b) *(u64 *)(r10 -40) = r2
6: (7b) *(u64 *)(r10 -48) = r2
; if (prev->state == TASK_RUNNING)
7: (79) r2 = *(u64 *)(r1 +16)
; if (prev->state == TASK_RUNNING)
8: (55) if r2 != 0x0 goto pc+19
 R1_w=ptr_task_struct(id=0,off=0,imm=0) R2_w=inv0 R6_w=ctx(id=0,off=0,imm=0) R10=fp0 fp-24_w=00000000 fp-32_w=00000000 fp-40_w=00000000 fp-48_w=00000000
; trace_enqueue(prev->tgid, prev->pid);
9: (61) r1 = *(u32 *)(r1 +1184)
10: (63) *(u32 *)(r10 -4) = r1
; if (!pid || (targ_pid && targ_pid != pid))
11: (15) if r1 == 0x0 goto pc+16

from 11 to 28: R1_w=inv0 R2_w=inv0 R6_w=ctx(id=0,off=0,imm=0) R10=fp0 fp-8=mmmm???? fp-24_w=00000000 fp-32_w=00000000 fp-40_w=00000000 fp-48_w=00000000
; bpf_map_update_elem(&start, &pid, &ts, 0);
28: (bf) r2 = r10
;
29: (07) r2 += -16
; tsp = bpf_map_lookup_elem(&start, &pid);
30: (18) r1 = 0xffff8881db3ce800
32: (85) call bpf_map_lookup_elem#1
invalid indirect read from stack off -16+0 size 4
processed 65 insns (limit 1000000) max_states_per_insn 1 total_states 5 peak_states 5 mark_read 4

Notice how in this case, there are 0-11 instructions + jump from 11 to
28 is recorded + 28-32 instructions with error on insn #32.

test_verifier test runner was updated to specify BPF_LOG_LEVEL2 for
VERBOSE_ACCEPT expected result due to potentially "incomplete" success verbose
log at BPF_LOG_LEVEL1.

On success, verbose log will only have a summary of number of processed
instructions, etc, but no branch tracing log. Having just a last succesful
branch tracing seemed weird and confusing. Having small and clean summary log
in success case seems quite logical and nice, though.
Signed-off-by: default avatarAndrii Nakryiko <andriin@fb.com>
Signed-off-by: default avatarAlexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200423195850.1259827-1-andriin@fb.com
      6f8a57cc
    • Maciej Żenczykowski's avatar
      bpf: add bpf_ktime_get_boot_ns() · 71d19214
      Maciej Żenczykowski authored 
      On a device like a cellphone which is constantly suspending
and resuming CLOCK_MONOTONIC is not particularly useful for
keeping track of or reacting to external network events.
Instead you want to use CLOCK_BOOTTIME.

Hence add bpf_ktime_get_boot_ns() as a mirror of bpf_ktime_get_ns()
based around CLOCK_BOOTTIME instead of CLOCK_MONOTONIC.
Signed-off-by: default avatarMaciej Żenczykowski <maze@google.com>
Signed-off-by: default avatarAlexei Starovoitov <ast@kernel.org>
      71d19214
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7