Yuan Can says:
====================
This series contains two cleanup patches, remove unused struct.
====================
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Martin KaFai Lau says:
====================
From: Martin KaFai Lau <martin.lau@kernel.org>
The struct_ops is sharing the tracing-trampoline's enter/exit
function which tracks prog->active to avoid recursion. It turns
out the struct_ops bpf prog will hit this prog->active and
unnecessarily skipped running the struct_ops prog. eg. The
'.ssthresh' may run in_task() and then interrupted by softirq
that runs the same '.ssthresh'.
The kernel does not call the tcp-cc's ops in a recursive way,
so this set is to remove the recursion check for struct_ops prog.
v3:
- Clear the bpf_chg_cc_inprogress from the newly cloned tcp_sock
in tcp_create_openreq_child() because the listen sk can
be cloned without lock being held. (Eric Dumazet)
v2:
- v1 [0] turned into a long discussion on a few cases and also
whether it needs to follow the bpf_run_ctx chain if there is
tracing bpf_run_ctx (kprobe/trace/trampoline) running in between.
It is a good signal that it is not obvious enough to reason
about it and needs a tradeoff for a more straight forward approach.
This revision uses one bit out of an existing 1 byte hole
in the tcp_sock. It is in Patch 4.
[0]: https://lore.kernel.org/bpf/20220922225616.3054840-1-kafai@fb.com/T/#md98d40ac5ec295fdadef476c227a3401b2b6b911
====================
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
When a bad bpf prog '.init' calls
bpf_setsockopt(TCP_CONGESTION, "itself"), it will trigger this loop:
.init => bpf_setsockopt(tcp_cc) => .init => bpf_setsockopt(tcp_cc) ...
... => .init => bpf_setsockopt(tcp_cc).
It was prevented by the prog->active counter before but the prog->active
detection cannot be used in struct_ops as explained in the earlier
patch of the set.
In this patch, the second bpf_setsockopt(tcp_cc) is not allowed
in order to break the loop. This is done by using a bit of
an existing 1 byte hole in tcp_sock to check if there is
on-going bpf_setsockopt(TCP_CONGESTION) in this tcp_sock.
Note that this essentially limits only the first '.init' can
call bpf_setsockopt(TCP_CONGESTION) to pick a fallback cc (eg. peer
does not support ECN) and the second '.init' cannot fallback to
another cc. This applies even the second
bpf_setsockopt(TCP_CONGESTION) will not cause a loop.
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20220929070407.965581-5-martin.lau@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
The check on the tcp-cc, "cdg", is done in the bpf_sk_setsockopt which is
used by the bpf_tcp_ca, bpf_lsm, cg_sockopt, and tcp_iter hooks.
However, it is not done for cg sock_ddr, cg sockops, and some of
the bpf_lsm_cgroup hooks.
The tcp-cc "cdg" should have very limited usage. This patch is to
move the "cdg" check to the common sol_tcp_sockopt() so that all
hooks have a consistent behavior. The motivation to make
this check consistent now is because the latter patch will
refactor the bpf_setsockopt(TCP_CONGESTION) into another function,
so it is better to take this chance to refactor this piece
also.
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://lore.kernel.org/r/20220929070407.965581-3-martin.lau@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
The struct_ops prog is to allow using bpf to implement the functions in
a struct (eg. kernel module). The current usage is to implement the
tcp_congestion. The kernel does not call the tcp-cc's ops (ie.
the bpf prog) in a recursive way.
The struct_ops is sharing the tracing-trampoline's enter/exit
function which tracks prog->active to avoid recursion. It is
needed for tracing prog. However, it turns out the struct_ops
bpf prog will hit this prog->active and unnecessarily skipped
running the struct_ops prog. eg. The '.ssthresh' may run in_task()
and then interrupted by softirq that runs the same '.ssthresh'.
Skip running the '.ssthresh' will end up returning random value
to the caller.
The patch adds __bpf_prog_{enter,exit}_struct_ops for the
struct_ops trampoline. They do not track the prog->active
to detect recursion.
One exception is when the tcp_congestion's '.init' ops is doing
bpf_setsockopt(TCP_CONGESTION) and then recurs to the same
'.init' ops. This will be addressed in the following patches.
Fixes: ca06f55b90 ("bpf: Add per-program recursion prevention mechanism")
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://lore.kernel.org/r/20220929070407.965581-2-martin.lau@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Wang Yufen says:
====================
Convert some tests to use the preferred ASSERT_* macros instead of the
deprecated CHECK().
====================
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Kui-Feng Lee says:
====================
Allow creating an iterator that loops through resources of one task/thread.
People could only create iterators to loop through all resources of
files, vma, and tasks in the system, even though they were interested in only the
resources of a specific task or process. Passing the additional
parameters, people can now create an iterator to go through all
resources or only the resources of a task.
Major Changes:
- Add new parameters in bpf_iter_link_info to indicate to go through
all tasks or to go through a specific task.
- Change the implementations of BPF iterators of vma, files, and
tasks to allow going through only the resources of a specific task.
- Provide the arguments of parameterized task iterators in
bpf_link_info.
Differences from v10:
- Check pid_alive() to avoid potential errors.
Differences from v9:
- Fix the boundary check of computing page_shift.
- Rewording the reason of checking and returning the same task.
Differences from v8:
- Fix uninitialized variable.
- Avoid redundant work of getting task from pid.
- Change format string to use %u instead of %d.
- Use the value of page_shift to compute correct offset in
bpf_iter_vm_offset.c.
Differences from v7:
- Travel the tasks of a process through task_group linked list
instead of traveling through the whole namespace.
Differences from v6:
- Add part 5 to make bpftool show the value of parameters.
- Change of wording of show_fdinfo() to show pid or tid instead of
always pid.
- Simplify error handling and naming of test cases.
Differences from v5:
- Use user-space tid/pid terminologies in bpf_iter_link_info and
bpf_link_info.
- Fix reference count
- Merge all variants to one 'u32 pid' in internal structs.
(bpf_iter_aux_info and bpf_iter_seq_task_common)
- Compare the result of get_uprobe_offset() with the implementation
with the vma iterators.
- Implement show_fdinfo.
Differences from v4:
- Remove 'type' from bpf_iter_link_info and bpf_link_info.
v10: https://lore.kernel.org/all/20220831181039.2680134-1-kuifeng@fb.com/
v9: https://lore.kernel.org/bpf/20220829192317.486946-1-kuifeng@fb.com/
v8: https://lore.kernel.org/bpf/20220829192317.486946-1-kuifeng@fb.com/
v7: https://lore.kernel.org/bpf/20220826003712.2810158-1-kuifeng@fb.com/
v6: https://lore.kernel.org/bpf/20220819220927.3409575-1-kuifeng@fb.com/
v5: https://lore.kernel.org/bpf/20220811001654.1316689-1-kuifeng@fb.com/
v4: https://lore.kernel.org/bpf/20220809195429.1043220-1-kuifeng@fb.com/
v3: https://lore.kernel.org/bpf/20220809063501.667610-1-kuifeng@fb.com/
v2: https://lore.kernel.org/bpf/20220801232649.2306614-1-kuifeng@fb.com/
v1: https://lore.kernel.org/bpf/20220726051713.840431-1-kuifeng@fb.com/
====================
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Show tid or pid of iterators if giving an argument of tid or pid
For example, the command `bpftool link list` may list following
lines.
1: iter prog 2 target_name bpf_map
2: iter prog 3 target_name bpf_prog
33: iter prog 225 target_name task_file tid 1644
pids test_progs(1644)
Link 33 is a task_file iterator with tid 1644. For now, only targets
of task, task_file and task_vma may be with tid or pid to filter out
tasks other than those belonging to a process (pid) or a thread (tid).
Signed-off-by: Kui-Feng Lee <kuifeng@fb.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Quentin Monnet <quentin@isovalent.com>
Acked-by: Yonghong Song <yhs@fb.com>
Acked-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://lore.kernel.org/bpf/20220926184957.208194-6-kuifeng@fb.com
Show information of iterators in the respective files under
/proc/<pid>/fdinfo/.
For example, for a task file iterator with 1723 as the value of tid
parameter, its fdinfo would look like the following lines.
pos: 0
flags: 02000000
mnt_id: 14
ino: 38
link_type: iter
link_id: 51
prog_tag: a590ac96db22b825
prog_id: 299
target_name: task_file
task_type: TID
tid: 1723
This patch add the last three fields. task_type is the type of the
task parameter. TID means the iterator visit only the thread
specified by tid. The value of tid in the above example is 1723. For
the case of PID task_type, it means the iterator visits only threads
of a process and will show the pid value of the process instead of a
tid.
Signed-off-by: Kui-Feng Lee <kuifeng@fb.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Yonghong Song <yhs@fb.com>
Acked-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://lore.kernel.org/bpf/20220926184957.208194-4-kuifeng@fb.com
Allow creating an iterator that loops through resources of one
thread/process.
People could only create iterators to loop through all resources of
files, vma, and tasks in the system, even though they were interested
in only the resources of a specific task or process. Passing the
additional parameters, people can now create an iterator to go
through all resources or only the resources of a task.
Signed-off-by: Kui-Feng Lee <kuifeng@fb.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Yonghong Song <yhs@fb.com>
Acked-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://lore.kernel.org/bpf/20220926184957.208194-2-kuifeng@fb.com
Drop the requirement for system-wide kernel UAPI headers to provide full
struct btf_enum64 definition. This is an unexpected requirement that
slipped in libbpf 1.0 and put unnecessary pressure ([0]) on users to have
a bleeding-edge kernel UAPI header from unreleased Linux 6.0.
To achieve this, we forward declare struct btf_enum64. But that's not
enough as there is btf_enum64_value() helper that expects to know the
layout of struct btf_enum64. So we get a bit creative with
reinterpreting memory layout as array of __u32 and accesing lo32/hi32
fields as array elements. Alternative way would be to have a local
pointer variable for anonymous struct with exactly the same layout as
struct btf_enum64, but that gets us into C++ compiler errors complaining
about invalid type casts. So play it safe, if ugly.
[0] Closes: https://github.com/libbpf/libbpf/issues/562
Fixes: d90ec262b3 ("libbpf: Add enum64 support for btf_dump")
Reported-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Link: https://lore.kernel.org/bpf/20220927042940.147185-1-andrii@kernel.org
When running rootless with special capabilities like:
FOWNER / DAC_OVERRIDE / DAC_READ_SEARCH
The "access" API will not make the proper check if there is really
access to a file or not.
>From the access man page:
"
The check is done using the calling process's real UID and GID, rather
than the effective IDs as is done when actually attempting an operation
(e.g., open(2)) on the file. Similarly, for the root user, the check
uses the set of permitted capabilities rather than the set of effective
capabilities; ***and for non-root users, the check uses an empty set of
capabilities.***
"
What that means is that for non-root user the access API will not do the
proper validation if the process really has permission to a file or not.
To resolve this this patch replaces all the access API calls with
faccessat with AT_EACCESS flag.
Signed-off-by: Jon Doron <jond@wiz.io>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20220925070431.1313680-1-arilou@gmail.com
Song Liu says:
====================
Changes v1 => v2:
1. Update arch_prepare_bpf_dispatcher to use a RO image and a RW buffer.
(Alexei) Note: I haven't found an existing test to cover this part, so
this part was tested manually (comparing the generated dispatcher is
the same).
Jeff Layton reported CPA W^X warning linux-next [1]. It turns out to be
W^X issue with bpf trampoline and bpf dispatcher. Fix these by:
1. Use bpf_prog_pack for bpf_dispatcher;
2. Set memory permission properly with bpf trampoline.
[1] https://lore.kernel.org/lkml/c84cc27c1a5031a003039748c3c099732a718aec.camel@kernel.org/
====================
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Mark the trampoline as RO+X after arch_prepare_bpf_trampoline, so that
the trampoine follows W^X rule strictly. This will turn off warnings like
CPA refuse W^X violation: 8000000000000163 -> 0000000000000163 range: ...
Also remove bpf_jit_alloc_exec_page(), since it is not used any more.
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20220926184739.3512547-3-song@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Allocate bpf_dispatcher with bpf_prog_pack_alloc so that bpf_dispatcher
can share pages with bpf programs.
arch_prepare_bpf_dispatcher() is updated to provide a RW buffer as working
area for arch code to write to.
This also fixes CPA W^X warnning like:
CPA refuse W^X violation: 8000000000000163 -> 0000000000000163 range: ...
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20220926184739.3512547-2-song@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Jiri Olsa says:
====================
Martynas reported bpf_get_func_ip returning +4 address when
CONFIG_X86_KERNEL_IBT option is enabled and I found there are
some failing bpf tests when this option is enabled.
The CONFIG_X86_KERNEL_IBT option adds endbr instruction at the
function entry, so the idea is to 'fix' entry ip for kprobe_multi
and trampoline probes, because they are placed on the function
entry.
v5 changes:
- updated uapi/linux/bpf.h headers with comment for
bpf_get_func_ip returning 0 [Andrii]
- added acks
v4 changes:
- used get_kernel_nofault to read previous instruction [Peter]
- used movabs instruction in trampoline comment [Peter]
- renamed fentry_ip argument in kprobe_multi_link_handler [Peter]
v3 changes:
- using 'unused' bpf function to get IBT config option
into selftest skeleton
- rebased to current bpf-next/master
- added ack/review from Masami
v2 changes:
- change kprobes get_func_ip to return zero for kprobes
attached within the function body [Andrii]
- detect IBT config and properly test kprobe with offset
[Andrii]
v1 changes:
- read previous instruction in kprobe_multi link handler
and adjust entry_ip for CONFIG_X86_KERNEL_IBT option
- split first patch into 2 separate changes
- update changelogs
====================
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
With CONFIG_X86_KERNEL_IBT enabled the test for kprobe with offset
won't work because of the extra endbr instruction.
As suggested by Andrii adding CONFIG_X86_KERNEL_IBT detection
and using appropriate offset value based on that.
Also removing test7 program, because it does the same as test6.
Suggested-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/r/20220926153340.1621984-7-jolsa@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Changing return value of kprobe's version of bpf_get_func_ip
to return zero if the attach address is not on the function's
entry point.
For kprobes attached in the middle of the function we can't easily
get to the function address especially now with the CONFIG_X86_KERNEL_IBT
support.
If user cares about current IP for kprobes attached within the
function body, they can get it with PT_REGS_IP(ctx).
Suggested-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Martynas Pumputis <m@lambda.lt>
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/r/20220926153340.1621984-6-jolsa@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Martynas reported bpf_get_func_ip returning +4 address when
CONFIG_X86_KERNEL_IBT option is enabled.
When CONFIG_X86_KERNEL_IBT is enabled we'll have endbr instruction
at the function entry, which screws return value of bpf_get_func_ip()
helper that should return the function address.
There's short term workaround for kprobe_multi bpf program made by
Alexei [1], but we need this fixup also for bpf_get_attach_cookie,
that returns cookie based on the entry_ip value.
Moving the fixup in the fprobe handler, so both bpf_get_func_ip
and bpf_get_attach_cookie get expected function address when
CONFIG_X86_KERNEL_IBT option is enabled.
Also renaming kprobe_multi_link_handler entry_ip argument to fentry_ip
so it's clearer this is an ftrace __fentry__ ip.
[1] commit 7f0059b58f ("selftests/bpf: Fix kprobe_multi test.")
Cc: Peter Zijlstra <peterz@infradead.org>
Reported-by: Martynas Pumputis <m@lambda.lt>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Link: https://lore.kernel.org/r/20220926153340.1621984-5-jolsa@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
