Add four tests to tailcalls selftest explicitly named
"tailcall_bpf2bpf_X" as their purpose is to validate that combination
of tailcalls with bpf2bpf calls are working properly.
These tests also validate LD_ABS from subprograms.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
LD_[ABS|IND] instructions may return from the function early. bpf_tail_call
pseudo instruction is either fallthrough or return. Allow them in the
subprograms only when subprograms are BTF annotated and have scalar return
types. Allow ld_abs and tail_call in the main program even if it calls into
subprograms. In the past that was not ok to do for ld_abs, since it was JITed
with special exit sequence. Since bpf_gen_ld_abs() was introduced the ld_abs
looks like normal exit insn from JIT point of view, so it's safe to allow them
in the main program.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Relax verifier's restriction that was meant to forbid tailcall usage
when subprog count was higher than 1.
Also, do not max out the stack depth of program that utilizes tailcalls.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
This commit serves two things:
1) it optimizes BPF prologue/epilogue generation
2) it makes possible to have tailcalls within BPF subprogram
Both points are related to each other since without 1), 2) could not be
achieved.
In [1], Alexei says:
"The prologue will look like:
nop5
xor eax,eax // two new bytes if bpf_tail_call() is used in this
// function
push rbp
mov rbp, rsp
sub rsp, rounded_stack_depth
push rax // zero init tail_call counter
variable number of push rbx,r13,r14,r15
Then bpf_tail_call will pop variable number rbx,..
and final 'pop rax'
Then 'add rsp, size_of_current_stack_frame'
jmp to next function and skip over 'nop5; xor eax,eax; push rpb; mov
rbp, rsp'
This way new function will set its own stack size and will init tail
call
counter with whatever value the parent had.
If next function doesn't use bpf_tail_call it won't have 'xor eax,eax'.
Instead it would need to have 'nop2' in there."
Implement that suggestion.
Since the layout of stack is changed, tail call counter handling can not
rely anymore on popping it to rbx just like it have been handled for
constant prologue case and later overwrite of rbx with actual value of
rbx pushed to stack. Therefore, let's use one of the register (%rcx) that
is considered to be volatile/caller-saved and pop the value of tail call
counter in there in the epilogue.
Drop the BUILD_BUG_ON in emit_prologue and in
emit_bpf_tail_call_indirect where instruction layout is not constant
anymore.
Introduce new poke target, 'tailcall_bypass' to poke descriptor that is
dedicated for skipping the register pops and stack unwind that are
generated right before the actual jump to target program.
For case when the target program is not present, BPF program will skip
the pop instructions and nop5 dedicated for jmpq $target. An example of
such state when only R6 of callee saved registers is used by program:
ffffffffc0513aa1: e9 0e 00 00 00 jmpq 0xffffffffc0513ab4
ffffffffc0513aa6: 5b pop %rbx
ffffffffc0513aa7: 58 pop %rax
ffffffffc0513aa8: 48 81 c4 00 00 00 00 add $0x0,%rsp
ffffffffc0513aaf: 0f 1f 44 00 00 nopl 0x0(%rax,%rax,1)
ffffffffc0513ab4: 48 89 df mov %rbx,%rdi
When target program is inserted, the jump that was there to skip
pops/nop5 will become the nop5, so CPU will go over pops and do the
actual tailcall.
One might ask why there simply can not be pushes after the nop5?
In the following example snippet:
ffffffffc037030c: 48 89 fb mov %rdi,%rbx
(...)
ffffffffc0370332: 5b pop %rbx
ffffffffc0370333: 58 pop %rax
ffffffffc0370334: 48 81 c4 00 00 00 00 add $0x0,%rsp
ffffffffc037033b: 0f 1f 44 00 00 nopl 0x0(%rax,%rax,1)
ffffffffc0370340: 48 81 ec 00 00 00 00 sub $0x0,%rsp
ffffffffc0370347: 50 push %rax
ffffffffc0370348: 53 push %rbx
ffffffffc0370349: 48 89 df mov %rbx,%rdi
ffffffffc037034c: e8 f7 21 00 00 callq 0xffffffffc0372548
There is the bpf2bpf call (at ffffffffc037034c) right after the tailcall
and jump target is not present. ctx is in %rbx register and BPF
subprogram that we will call into on ffffffffc037034c is relying on it,
e.g. it will pick ctx from there. Such code layout is therefore broken
as we would overwrite the content of %rbx with the value that was pushed
on the prologue. That is the reason for the 'bypass' approach.
Special care needs to be taken during the install/update/remove of
tailcall target. In case when target program is not present, the CPU
must not execute the pop instructions that precede the tailcall.
To address that, the following states can be defined:
A nop, unwind, nop
B nop, unwind, tail
C skip, unwind, nop
D skip, unwind, tail
A is forbidden (lead to incorrectness). The state transitions between
tailcall install/update/remove will work as follows:
First install tail call f: C->D->B(f)
* poke the tailcall, after that get rid of the skip
Update tail call f to f': B(f)->B(f')
* poke the tailcall (poke->tailcall_target) and do NOT touch the
poke->tailcall_bypass
Remove tail call: B(f')->C(f')
* poke->tailcall_bypass is poked back to jump, then we wait the RCU
grace period so that other programs will finish its execution and
after that we are safe to remove the poke->tailcall_target
Install new tail call (f''): C(f')->D(f'')->B(f'').
* same as first step
This way CPU can never be exposed to "unwind, tail" state.
Last but not least, when tailcalls get mixed with bpf2bpf calls, it
would be possible to encounter the endless loop due to clearing the
tailcall counter if for example we would use the tailcall3-like from BPF
selftests program that would be subprogram-based, meaning the tailcall
would be present within the BPF subprogram.
This test, broken down to particular steps, would do:
entry -> set tailcall counter to 0, bump it by 1, tailcall to func0
func0 -> call subprog_tail
(we are NOT skipping the first 11 bytes of prologue and this subprogram
has a tailcall, therefore we clear the counter...)
subprog -> do the same thing as entry
and then loop forever.
To address this, the idea is to go through the call chain of bpf2bpf progs
and look for a tailcall presence throughout whole chain. If we saw a single
tail call then each node in this call chain needs to be marked as a subprog
that can reach the tailcall. We would later feed the JIT with this info
and:
- set eax to 0 only when tailcall is reachable and this is the entry prog
- if tailcall is reachable but there's no tailcall in insns of currently
JITed prog then push rax anyway, so that it will be possible to
propagate further down the call chain
- finally if tailcall is reachable, then we need to precede the 'call'
insn with mov rax, [rbp - (stack_depth + 8)]
Tail call related cases from test_verifier kselftest are also working
fine. Sample BPF programs that utilize tail calls (sockex3, tracex5)
work properly as well.
[1]: https://lore.kernel.org/bpf/20200517043227.2gpq22ifoq37ogst@ast-mbp.dhcp.thefacebook.com/
Suggested-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Protect against potential stack overflow that might happen when bpf2bpf
calls get combined with tailcalls. Limit the caller's stack depth for
such case down to 256 so that the worst case scenario would result in 8k
stack size (32 which is tailcall limit * 256 = 8k).
Suggested-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reflect the actual purpose of poke->ip and rename it to
poke->tailcall_target so that it will not the be confused with another
poke target that will be introduced in next commit.
While at it, do the same thing with poke->ip_stable - rename it to
poke->tailcall_target_stable.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Previously, there was no need for poke descriptors being present in
subprogram's bpf_prog_aux struct since tailcalls were simply not allowed
in them. Each subprog is JITed independently so in order to enable
JITing subprograms that use tailcalls, do the following:
- in fixup_bpf_calls() store the index of tailcall insn onto the generated
poke descriptor,
- in case when insn patching occurs, adjust the tailcall insn idx from
bpf_patch_insn_data,
- then in jit_subprogs() check whether the given poke descriptor belongs
to the current subprog by checking if that previously stored absolute
index of tail call insn is in the scope of the insns of given subprog,
- update the insn->imm with new poke descriptor slot so that while JITing
the proper poke descriptor will be grabbed
This way each of the main program's poke descriptors are distributed
across the subprograms poke descriptor array, so main program's
descriptors can be untracked out of the prog array map.
Add also subprog's aux struct to the BPF map poke_progs list by calling
on it map_poke_track().
In case of any error, call the map_poke_untrack() on subprog's aux
structs that have already been registered to prog array map.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Currently, %rax is used to store the jump target when BPF program is
emitting the retpoline instructions that are handling the indirect
tailcall.
There is a plan to use %rax for different purpose, which is storing the
tail call counter. In order to preserve this value across the tailcalls,
adjust the BPF indirect tailcalls so that the target program will reside
in %rcx and teach the retpoline instructions about new location of jump
target.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Merge 183 tests from test_btf into test_progs framework to be exercised
regularly. All the test_btf tests that were moved are modeled as proper
sub-tests in test_progs framework for ease of debugging and reporting.
No functional or behavioral changes were intended, I tried to preserve
original behavior as much as possible. E.g., `test_progs -v` will activate
"always_log" flag to emit BTF validation log.
The only difference is in reducing the max_entries limit for pretty-printing
tests from (128 * 1024) to just 128 to reduce tests running time without
reducing the coverage.
Example test run:
$ sudo ./test_progs -n 8
...
#8 btf:OK
Summary: 1/183 PASSED, 0 SKIPPED, 0 FAILED
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200916004819.3767489-1-andriin@fb.com
Stanislav Fomichev says:
====================
Currently, if a user wants to store arbitrary metadata for an eBPF
program, for example, the program build commit hash or version, they
could store it in a map, and conveniently libbpf uses .data section to
populate an internal map. However, if the program does not actually
reference the map, then the map would be de-refcounted and freed.
This patch set introduces a new syscall BPF_PROG_BIND_MAP to add a map
to a program's used_maps, even if the program instructions does not
reference the map.
libbpf is extended to always BPF_PROG_BIND_MAP .rodata section so the
metadata is kept in place.
bpftool is also extended to print metadata in the 'bpftool prog' list.
The variable is considered metadata if it starts with the
magic 'bpf_metadata_' prefix; everything after the prefix is the
metadata name.
An example use of this would be BPF C file declaring:
volatile const char bpf_metadata_commit_hash[] SEC(".rodata") = "abcdef123456";
and bpftool would emit:
$ bpftool prog
[...]
metadata:
commit_hash = "abcdef123456"
v6 changes:
* libbpf: drop FEAT_GLOBAL_DATA from probe_prog_bind_map (Andrii Nakryiko)
* bpftool: combine find_metadata_map_id & find_metadata;
drops extra bpf_map_get_fd_by_id and bpf_map_get_fd_by_id (Andrii Nakryiko)
* bpftool: use strncmp instead of strstr (Andrii Nakryiko)
* bpftool: memset(map_info) and extra empty line (Andrii Nakryiko)
v5 changes:
* selftest: verify that prog holds rodata (Andrii Nakryiko)
* selftest: use volatile for metadata (Andrii Nakryiko)
* bpftool: use sizeof in BPF_METADATA_PREFIX_LEN (Andrii Nakryiko)
* bpftool: new find_metadata that does map lookup (Andrii Nakryiko)
* libbpf: don't generalize probe_create_global_data (Andrii Nakryiko)
* libbpf: use OPTS_VALID in bpf_prog_bind_map (Andrii Nakryiko)
* libbpf: keep LIBBPF_0.2.0 sorted (Andrii Nakryiko)
v4 changes:
* Don't return EEXIST from syscall if already bound (Andrii Nakryiko)
* Removed --metadata argument (Andrii Nakryiko)
* Removed custom .metadata section (Alexei Starovoitov)
* Addressed Andrii's suggestions about btf helpers and vsi (Andrii Nakryiko)
* Moved bpf_prog_find_metadata into bpftool (Alexei Starovoitov)
v3 changes:
* API changes for bpf_prog_find_metadata (Toke Høiland-Jørgensen)
v2 changes:
* Made struct bpf_prog_bind_opts in libbpf so flags is optional.
* Deduped probe_kern_global_data and probe_prog_bind_map into a common
helper.
* Added comment regarding why EEXIST is ignored in libbpf bind map.
* Froze all LIBBPF_MAP_METADATA internal maps.
* Moved bpf_prog_bind_map into new LIBBPF_0.1.1 in libbpf.map.
* Added p_err() calls on error cases in bpftool show_prog_metadata.
* Reverse christmas tree coding style in bpftool show_prog_metadata.
* Made bpftool gen skeleton recognize .metadata as an internal map and
generate datasec definition in skeleton.
* Added C test using skeleton to see asset that the metadata is what we
expect and rebinding causes EEXIST.
v1 changes:
* Fixed a few missing unlocks, and missing close while iterating map fds.
* Move mutex initialization to right after prog aux allocation, and mutex
destroy to right after prog aux free.
* s/ADD_MAP/BIND_MAP/
* Use mutex only instead of RCU to protect the used_map array & count.
Cc: YiFei Zhu <zhuyifei1999@gmail.com>
====================
Acked-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Dump metadata in the 'bpftool prog' list if it's present.
For some formatting some BTF code is put directly in the
metadata dumping. Sanity checks on the map and the kind of the btf_type
to make sure we are actually dumping what we are expecting.
A helper jsonw_reset is added to json writer so we can reuse the same
json writer without having extraneous commas.
Sample output:
$ bpftool prog
6: cgroup_skb name prog tag bcf7977d3b93787c gpl
[...]
btf_id 4
metadata:
a = "foo"
b = 1
$ bpftool prog --json --pretty
[{
"id": 6,
[...]
"btf_id": 4,
"metadata": {
"a": "foo",
"b": 1
}
}
]
Signed-off-by: YiFei Zhu <zhuyifei@google.com>
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Cc: YiFei Zhu <zhuyifei1999@gmail.com>
Link: https://lore.kernel.org/bpf/20200915234543.3220146-5-sdf@google.com
When syncing latest libbpf repo to bcc, ubuntu 16.04 (4.4.0 LTS kernel)
failed compilation for xsk.c:
In file included from /tmp/debuild.0jkauG/bcc/src/cc/libbpf/src/xsk.c:23:0:
/tmp/debuild.0jkauG/bcc/src/cc/libbpf/src/xsk.c: In function ‘xsk_get_ctx’:
/tmp/debuild.0jkauG/bcc/src/cc/libbpf/include/linux/list.h:81:9: warning: implicit
declaration of function ‘container_of’ [-Wimplicit-function-declaration]
container_of(ptr, type, member)
^
/tmp/debuild.0jkauG/bcc/src/cc/libbpf/include/linux/list.h:83:9: note: in expansion
of macro ‘list_entry’
list_entry((ptr)->next, type, member)
...
src/cc/CMakeFiles/bpf-static.dir/build.make:209: recipe for target
'src/cc/CMakeFiles/bpf-static.dir/libbpf/src/xsk.c.o' failed
Commit 2f6324a393 ("libbpf: Support shared umems between queues and devices")
added include file <linux/list.h>, which uses macro "container_of".
xsk.c file also includes <linux/ethtool.h> before <linux/list.h>.
In a more recent distro kernel, <linux/ethtool.h> includes <linux/kernel.h>
which contains the macro definition for "container_of". So compilation is all fine.
But in ubuntu 16.04 kernel, <linux/ethtool.h> does not contain <linux/kernel.h>
which caused the above compilation error.
Let explicitly add <linux/kernel.h> in xsk.c to avoid compilation error
in old distro's.
Fixes: 2f6324a393 ("libbpf: Support shared umems between queues and devices")
Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Acked-by: Song Liu <songliubraving@fb.com>
Link: https://lore.kernel.org/bpf/20200914223210.1831262-1-yhs@fb.com
When building bpf selftests like
make -C tools/testing/selftests/bpf -j20
I hit the following errors:
...
GEN /net-next/tools/testing/selftests/bpf/tools/build/bpftool/Documentation/bpftool-gen.8
<stdin>:75: (WARNING/2) Block quote ends without a blank line; unexpected unindent.
<stdin>:71: (WARNING/2) Literal block ends without a blank line; unexpected unindent.
<stdin>:85: (WARNING/2) Literal block ends without a blank line; unexpected unindent.
<stdin>:57: (WARNING/2) Block quote ends without a blank line; unexpected unindent.
<stdin>:66: (WARNING/2) Literal block ends without a blank line; unexpected unindent.
<stdin>:109: (WARNING/2) Literal block ends without a blank line; unexpected unindent.
<stdin>:175: (WARNING/2) Literal block ends without a blank line; unexpected unindent.
<stdin>:273: (WARNING/2) Literal block ends without a blank line; unexpected unindent.
make[1]: *** [/net-next/tools/testing/selftests/bpf/tools/build/bpftool/Documentation/bpftool-perf.8] Error 12
make[1]: *** Waiting for unfinished jobs....
make[1]: *** [/net-next/tools/testing/selftests/bpf/tools/build/bpftool/Documentation/bpftool-iter.8] Error 12
make[1]: *** [/net-next/tools/testing/selftests/bpf/tools/build/bpftool/Documentation/bpftool-struct_ops.8] Error 12
...
I am using:
-bash-4.4$ rst2man --version
rst2man (Docutils 0.11 [repository], Python 2.7.5, on linux2)
-bash-4.4$
The Makefile generated final .rst file (e.g., bpftool-cgroup.rst) looks like
...
ID AttachType AttachFlags Name
\n SEE ALSO\n========\n\t**bpf**\ (2),\n\t**bpf-helpers**\
(7),\n\t**bpftool**\ (8),\n\t**bpftool-btf**\
(8),\n\t**bpftool-feature**\ (8),\n\t**bpftool-gen**\
(8),\n\t**bpftool-iter**\ (8),\n\t**bpftool-link**\
(8),\n\t**bpftool-map**\ (8),\n\t**bpftool-net**\
(8),\n\t**bpftool-perf**\ (8),\n\t**bpftool-prog**\
(8),\n\t**bpftool-struct_ops**\ (8)\n
The rst2man generated .8 file looks like
Literal block ends without a blank line; unexpected unindent.
.sp
n SEEALSOn========nt**bpf**(2),nt**bpf\-helpers**(7),nt**bpftool**(8),nt**bpftool\-btf**(8),nt**
bpftool\-feature**(8),nt**bpftool\-gen**(8),nt**bpftool\-iter**(8),nt**bpftool\-link**(8),nt**
bpftool\-map**(8),nt**bpftool\-net**(8),nt**bpftool\-perf**(8),nt**bpftool\-prog**(8),nt**
bpftool\-struct_ops**(8)n
Looks like that particular version of rst2man prefers to have actual new line
instead of \n.
Since `echo -e` may not be available in some environment, let us use `printf`.
Format string "%b" is used for `printf` to ensure all escape characters are
interpretted properly.
Fixes: 18841da981 ("tools: bpftool: Automate generation for "SEE ALSO" sections in man pages")
Suggested-by: Andrii Nakryiko <andrii.nakryiko@gmail.com>
Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Song Liu <songliubraving@fb.com>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Cc: Quentin Monnet <quentin@isovalent.com>
Link: https://lore.kernel.org/bpf/20200914183110.999906-1-yhs@fb.com
Fix a possible deadlock in the l2fwd application in xdpsock that can
occur when there is no space in the Tx ring. There are two ways to get
the kernel to consume entries in the Tx ring: calling sendto() to make
it send packets and freeing entries from the completion ring, as the
kernel will not send a packet if there is no space for it to add a
completion entry in the completion ring. The Tx loop in l2fwd only
used to call sendto(). This patches adds cleaning the completion ring
in that loop.
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/1599726666-8431-3-git-send-email-magnus.karlsson@gmail.com
Fix the sending of a single packet (or small burst) in xdpsock when
executing in copy mode. Currently, the l2fwd application in xdpsock
only transmits the packets after a batch of them has been received,
which might be confusing if you only send one packet and expect that
it is returned pronto. Fix this by calling sendto() more often and add
a comment in the code that states that this can be optimized if
needed.
Reported-by: Tirthendu Sarkar <tirthendu.sarkar@intel.com>
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/1599726666-8431-2-git-send-email-magnus.karlsson@gmail.com
In order to branch around tail calls (due to out-of-bounds index,
exceeding tail call count or missing tail call target), JIT uses
label[0] field, which contains the address of the instruction following
the tail call. When there are multiple tail calls, label[0] value comes
from handling of a previous tail call, which is incorrect.
Fix by getting rid of label array and resolving the label address
locally: for all 3 branches that jump to it, emit 0 offsets at the
beginning, and then backpatch them with the correct value.
Also, do not use the long jump infrastructure: the tail call sequence
is known to be short, so make all 3 jumps short.
Fixes: 6651ee070b ("s390/bpf: implement bpf_tail_call() helper")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200909232141.3099367-1-iii@linux.ibm.com
Neal Cardwell says:
====================
This patch series reorganizes TCP congestion control initialization so that if
EBPF code called by tcp_init_transfer() sets the congestion control algorithm
by calling setsockopt(TCP_CONGESTION) then the TCP stack initializes the
congestion control module immediately, instead of having tcp_init_transfer()
later initialize the congestion control module.
This increases flexibility for the EBPF code that runs at connection
establishment time, and simplifies the code.
This has the following benefits:
(1) This allows CC module customizations made by the EBPF called in
tcp_init_transfer() to persist, and not be wiped out by a later
call to tcp_init_congestion_control() in tcp_init_transfer().
(2) Does not flip the order of EBPF and CC init, to avoid causing bugs
for existing code upstream that depends on the current order.
(3) Does not cause 2 initializations for for CC in the case where the
EBPF called in tcp_init_transfer() wants to set the CC to a new CC
algorithm.
(4) Allows follow-on simplifications to the code in net/core/filter.c
and net/ipv4/tcp_cong.c, which currently both have some complexity
to special-case CC initialization to avoid double CC
initialization if EBPF sets the CC.
changes in v2:
o rebase onto bpf-next
o add another follow-on simplification suggested by Martin KaFai Lau:
"tcp: simplify tcp_set_congestion_control() load=false case"
changes in v3:
o no change in commits
o resent patch series from @gmail.com, since mail from ncardwell@google.com
stopped being accepted at netdev@vger.kernel.org mid-way through processing
the v2 patch series (between patches 2 and 3), confusing patchwork about
which patches belonged to the v2 patch series
====================
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Ilya Leoshkevich
Maciej Fijalkowski
Alexei Starovoitov
Andrii Nakryiko
YiFei Zhu
Yonghong Song
Magnus Karlsson