Re: [PATCH bpf-next v9 0/5] Parameterize task iterators.

[Kui-Feng Lee <kuifeng@fb.com>]

On Wed, 2022-08-31 at 13:32 +0200, Jiri Olsa wrote:
> On Tue, Aug 30, 2022 at 07:37:39PM -0700, Kui-Feng Lee wrote:
> > 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.
> 
> hi,
> I'm getting bpf_iter/vma_offset fail:
> 
> test_task_vma_offset_common:PASS:bpf_iter_vma_offset__open_and_load 0
> nsec
> test_task_vma_offset_common:PASS:attach_iter 0 nsec
> test_task_vma_offset_common:PASS:create_iter 0 nsec
> test_task_vma_offset_common:PASS:strcmp 0 nsec
> test_task_vma_offset_common:FAIL:offset unexpected offset: actual
> 257222 != expected 203974
> test_task_vma_offset_common:PASS:unique_tgid_count 0 nsec
> test_task_vma_offset_common:PASS:bpf_iter_vma_offset__open_and_load 0
> nsec
> test_task_vma_offset_common:PASS:attach_iter 0 nsec
> test_task_vma_offset_common:PASS:create_iter 0 nsec
> test_task_vma_offset_common:PASS:strcmp 0 nsec
> test_task_vma_offset_common:FAIL:offset unexpected offset: actual
> 257222 != expected 203974
> test_task_vma_offset_common:PASS:unique_tgid_count 0 nsec
> test_task_vma_offset_common:PASS:bpf_iter_vma_offset__open_and_load 0
> nsec
> test_task_vma_offset_common:PASS:attach_iter 0 nsec
> test_task_vma_offset_common:PASS:create_iter 0 nsec
> test_task_vma_offset_common:PASS:strcmp 0 nsec
> test_task_vma_offset_common:FAIL:offset unexpected offset: actual
> 257222 != expected 203974
> test_task_vma_offset_common:PASS:unique_tgid_count 0 nsec
> #11/38   bpf_iter/vma_offset:FAIL
> 
> jirka

I could not reproduce it.  However, I found it should be an incorrect
boundary check of computing page_shift.  On my device, it happened to
be loaded with an offset 0 from the head of the file, so it passed the
test case accidentally.

Anyway, I will fix it.

> 
> > 
> > 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.
> > 
> > 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/
> > 
> > Kui-Feng Lee (5):
> >   bpf: Parameterize task iterators.
> >   bpf: Handle bpf_link_info for the parameterized task BPF
> > iterators.
> >   bpf: Handle show_fdinfo for the parameterized task BPF iterators
> >   selftests/bpf: Test parameterized task BPF iterators.
> >   bpftool: Show parameters of BPF task iterators.
> > 
> >  include/linux/bpf.h                           |  25 ++
> >  include/uapi/linux/bpf.h                      |  10 +
> >  kernel/bpf/task_iter.c                        | 227 ++++++++++++--
> >  tools/bpf/bpftool/link.c                      |  19 ++
> >  tools/include/uapi/linux/bpf.h                |  10 +
> >  .../selftests/bpf/prog_tests/bpf_iter.c       | 282
> > ++++++++++++++++--
> >  .../selftests/bpf/prog_tests/btf_dump.c       |   2 +-
> >  .../selftests/bpf/progs/bpf_iter_task.c       |   9 +
> >  .../selftests/bpf/progs/bpf_iter_task_file.c  |   9 +-
> >  .../selftests/bpf/progs/bpf_iter_task_vma.c   |   7 +-
> >  .../selftests/bpf/progs/bpf_iter_vma_offset.c |  37 +++
> >  11 files changed, 591 insertions(+), 46 deletions(-)
> >  create mode 100644
> > tools/testing/selftests/bpf/progs/bpf_iter_vma_offset.c
> > 
> > -- 
> > 2.30.2
> >