mbox series

[v2,0/3] bpf: switch to new usercopy helpers

Message ID 20191016004138.24845-1-christian.brauner@ubuntu.com
Headers show
Series bpf: switch to new usercopy helpers | expand

Message

Christian Brauner Oct. 16, 2019, 12:41 a.m. UTC
Hey everyone,

In v5.4-rc2 we added two new helpers check_zeroed_user() and
copy_struct_from_user() including selftests (cf. [1]). It is a generic
interface designed to copy a struct from userspace. The helpers will be
especially useful for structs versioned by size of which we have quite a
few.

The most obvious benefit is that this helper lets us get rid of
duplicate code. We've already switched over sched_setattr(), perf_event_open(),
and clone3(). More importantly it will also help to ensure that users
implementing versioning-by-size end up with the same core semantics.

This point is especially crucial since we have at least one case where
versioning-by-size is used but with slighly different semantics:
sched_setattr(), perf_event_open(), and clone3() all do do similar
checks to copy_struct_from_user() while rt_sigprocmask(2) always rejects
differently-sized struct arguments.

This little series switches over bpf codepaths that have hand-rolled
implementations of these helpers.

Thanks!
Christian

/* v1 */
Link: https://lore.kernel.org/r/20191009160907.10981-1-christian.brauner@ubuntu.com

/* v2 */
- rebase onto bpf-next

/* Reference */
[1]: f5a1a536fa14 ("lib: introduce copy_struct_from_user() helper")

Christian Brauner (3):
  bpf: use check_zeroed_user() in bpf_check_uarg_tail_zero()
  bpf: use copy_struct_from_user() in bpf_prog_get_info_by_fd()
  bpf: use copy_struct_from_user() in bpf() syscall

 kernel/bpf/syscall.c | 46 +++++++++++++++++---------------------------
 1 file changed, 18 insertions(+), 28 deletions(-)

Comments

Christian Brauner Oct. 16, 2019, 12:48 a.m. UTC | #1
On Wed, Oct 16, 2019 at 02:41:35AM +0200, Christian Brauner wrote:
> Hey everyone,
> 
> In v5.4-rc2 we added two new helpers check_zeroed_user() and
> copy_struct_from_user() including selftests (cf. [1]). It is a generic
> interface designed to copy a struct from userspace. The helpers will be
> especially useful for structs versioned by size of which we have quite a
> few.
> 
> The most obvious benefit is that this helper lets us get rid of
> duplicate code. We've already switched over sched_setattr(), perf_event_open(),
> and clone3(). More importantly it will also help to ensure that users
> implementing versioning-by-size end up with the same core semantics.
> 
> This point is especially crucial since we have at least one case where
> versioning-by-size is used but with slighly different semantics:
> sched_setattr(), perf_event_open(), and clone3() all do do similar
> checks to copy_struct_from_user() while rt_sigprocmask(2) always rejects
> differently-sized struct arguments.
> 
> This little series switches over bpf codepaths that have hand-rolled
> implementations of these helpers.
> 
> Thanks!
> Christian
> 
> /* v1 */
> Link: https://lore.kernel.org/r/20191009160907.10981-1-christian.brauner@ubuntu.com
> 
> /* v2 */
> - rebase onto bpf-next
> 
> /* Reference */
> [1]: f5a1a536fa14 ("lib: introduce copy_struct_from_user() helper")

Alexei, instead of applying the series you can also just pull from me:

The following changes since commit 5bc60de50dfea235634fdf38cbc992fb968d113b:

  selftests: bpf: Don't try to read files without read permission (2019-10-15 16:27:25 -0700)

are available in the Git repository at:

  git@gitolite.kernel.org:pub/scm/linux/kernel/git/brauner/linux tags/bpf-copy-struct-from-user

for you to fetch changes up to da1699b959d182bb161be3ffc17eab063b2aedd2:

  bpf: use copy_struct_from_user() in bpf() syscall (2019-10-16 02:35:11 +0200)

----------------------------------------------------------------
bpf-copy-struct-from-user

----------------------------------------------------------------
Christian Brauner (3):
      bpf: use check_zeroed_user() in bpf_check_uarg_tail_zero()
      bpf: use copy_struct_from_user() in bpf_prog_get_info_by_fd()
      bpf: use copy_struct_from_user() in bpf() syscall

 kernel/bpf/syscall.c | 46 ++++++++++++++++++----------------------------
 1 file changed, 18 insertions(+), 28 deletions(-)
Alexei Starovoitov Oct. 16, 2019, 2:14 a.m. UTC | #2
On Tue, Oct 15, 2019 at 5:41 PM Christian Brauner
<christian.brauner@ubuntu.com> wrote:
>
> Hey everyone,
>
> In v5.4-rc2 we added two new helpers check_zeroed_user() and
> copy_struct_from_user() including selftests (cf. [1]). It is a generic
> interface designed to copy a struct from userspace. The helpers will be
> especially useful for structs versioned by size of which we have quite a
> few.

Was it tested?
Either your conversion is incorrect or that generic helper is broken.
./test_progs -n 2
and
./test_btf
are catching the bug:
BTF prog info raw test[8] (line_info (No subprog. zero tailing
line_info): do_test_info_raw:6205:FAIL prog_fd:-1
expected_prog_load_failure:0 errno:7
nonzero tailing record in line_infoprocessed 0 insns (limit 1000000)
max_states_per_insn 0 total_states 0 peak_states 0 mark_read 0
Christian Brauner Oct. 16, 2019, 3:31 a.m. UTC | #3
On Tue, Oct 15, 2019 at 07:14:42PM -0700, Alexei Starovoitov wrote:
> On Tue, Oct 15, 2019 at 5:41 PM Christian Brauner
> <christian.brauner@ubuntu.com> wrote:
> >
> > Hey everyone,
> >
> > In v5.4-rc2 we added two new helpers check_zeroed_user() and
> > copy_struct_from_user() including selftests (cf. [1]). It is a generic
> > interface designed to copy a struct from userspace. The helpers will be
> > especially useful for structs versioned by size of which we have quite a
> > few.
> 
> Was it tested?
> Either your conversion is incorrect or that generic helper is broken.
> ./test_progs -n 2
> and
> ./test_btf
> are catching the bug:
> BTF prog info raw test[8] (line_info (No subprog. zero tailing
> line_info): do_test_info_raw:6205:FAIL prog_fd:-1
> expected_prog_load_failure:0 errno:7
> nonzero tailing record in line_infoprocessed 0 insns (limit 1000000)
> max_states_per_insn 0 total_states 0 peak_states 0 mark_read 0

Ugh, I misrememberd what the helper I helped design returns. The fix is:

diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 5db9887a8f4c..0920593eacd0 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -78,11 +78,8 @@ int bpf_check_uarg_tail_zero(void __user *uaddr,
                return 0;

        err = check_zeroed_user(uaddr + expected_size, rest);
-       if (err < 0)
-               return err;
-
-       if (err)
-               return -E2BIG;
+       if (err <= 0)
+               return err ?: -E2BIG;

        return 0;
 }

aka check_zeroed_user() returns 0 if non-zero bytes are present, 1 if no
non-zero bytes were present, and -errno on error.

I'll send a fixed version. The tests pass for me with this.

Christian