| Message ID | 20220621070116.307221-1-jthinz@mailbox.tu-berlin.de (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 6dc7a0baf1a70b7d22662d38481824c14ddd80c5 |
| Delegated to: | BPF |
| Headers | show |
| Series | [bpf-next] selftests/bpf: Fix rare segfault in sock_fields prog test | expand |
On 6/21/22 9:01 AM, Jörn-Thorben Hinz wrote: > test_sock_fields__detach() got called with a null pointer here when one > of the CHECKs or ASSERTs up to the test_sock_fields__open_and_load() > call resulted in a jump to the "done" label. > > A skeletons *__detach() is not safe to call with a null pointer, though. > This led to a segfault. > > Go the easy route and only call test_sock_fields__destroy() which is > null-pointer safe and includes detaching. > > Came across this while looking[1] to introduce the usage of > bpf_tcp_helpers.h (included in progs/test_sock_fields.c) together with > vmlinux.h. > > [1] https://lore.kernel.org/bpf/629bc069dd807d7ac646f836e9dca28bbc1108e2.camel@mailbox.tu-berlin.de/ > > Fixes: 8f50f16ff39d ("selftests/bpf: Extend verifier and bpf_sock tests for dst_port loads") > Signed-off-by: Jörn-Thorben Hinz <jthinz@mailbox.tu-berlin.de> > --- > tools/testing/selftests/bpf/prog_tests/sock_fields.c | 1 - > 1 file changed, 1 deletion(-) > > diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > index 9d211b5c22c4..7d23166c77af 100644 > --- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > +++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > @@ -394,7 +394,6 @@ void serial_test_sock_fields(void) > test(); > > done: > - test_sock_fields__detach(skel); > test_sock_fields__destroy(skel); > if (child_cg_fd >= 0) > close(child_cg_fd); > Great catch! I think we have similar detach & destroy pattern in a number of places in selftests. Should we rather just move the label, like: diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c b/tools/testing/selftests/bpf/prog_tests/sock_fields.c index 9d211b5c22c4..e8a947241e37 100644 --- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c +++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c @@ -393,8 +393,8 @@ void serial_test_sock_fields(void) test(); -done: test_sock_fields__detach(skel); +done: test_sock_fields__destroy(skel); if (child_cg_fd >= 0) close(child_cg_fd);
On Tue, Jun 21, 2022 at 07:00 PM +02, Daniel Borkmann wrote: > On 6/21/22 9:01 AM, Jörn-Thorben Hinz wrote: >> test_sock_fields__detach() got called with a null pointer here when one >> of the CHECKs or ASSERTs up to the test_sock_fields__open_and_load() >> call resulted in a jump to the "done" label. >> A skeletons *__detach() is not safe to call with a null pointer, though. >> This led to a segfault. >> Go the easy route and only call test_sock_fields__destroy() which is >> null-pointer safe and includes detaching. >> Came across this while looking[1] to introduce the usage of >> bpf_tcp_helpers.h (included in progs/test_sock_fields.c) together with >> vmlinux.h. >> [1] >> https://lore.kernel.org/bpf/629bc069dd807d7ac646f836e9dca28bbc1108e2.camel@mailbox.tu-berlin.de/ >> Fixes: 8f50f16ff39d ("selftests/bpf: Extend verifier and bpf_sock tests for >> dst_port loads") >> Signed-off-by: Jörn-Thorben Hinz <jthinz@mailbox.tu-berlin.de> >> --- >> tools/testing/selftests/bpf/prog_tests/sock_fields.c | 1 - >> 1 file changed, 1 deletion(-) >> diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c >> b/tools/testing/selftests/bpf/prog_tests/sock_fields.c >> index 9d211b5c22c4..7d23166c77af 100644 >> --- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c >> +++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c >> @@ -394,7 +394,6 @@ void serial_test_sock_fields(void) >> test(); >> done: >> - test_sock_fields__detach(skel); >> test_sock_fields__destroy(skel); >> if (child_cg_fd >= 0) >> close(child_cg_fd); >> > > Great catch! I think we have similar detach & destroy pattern in a number > of places in selftests. > > Should we rather just move the label, like: > > diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > index 9d211b5c22c4..e8a947241e37 100644 > --- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > +++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > @@ -393,8 +393,8 @@ void serial_test_sock_fields(void) > > test(); > > -done: > test_sock_fields__detach(skel); > +done: > test_sock_fields__destroy(skel); > if (child_cg_fd >= 0) > close(child_cg_fd); *__destroy() will call bpf_object__detach_skeleton(), so it LGTM. Thanks for the fix. Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
On Tue, 2022-06-21 at 19:00 +0200, Daniel Borkmann wrote: > On 6/21/22 9:01 AM, Jörn-Thorben Hinz wrote: > > test_sock_fields__detach() got called with a null pointer here when > > one > > of the CHECKs or ASSERTs up to the > > test_sock_fields__open_and_load() > > call resulted in a jump to the "done" label. > > > > A skeletons *__detach() is not safe to call with a null pointer, > > though. > > This led to a segfault. > > > > Go the easy route and only call test_sock_fields__destroy() which > > is > > null-pointer safe and includes detaching. > > > > Came across this while looking[1] to introduce the usage of > > bpf_tcp_helpers.h (included in progs/test_sock_fields.c) together > > with > > vmlinux.h. > > > > [1] > > https://lore.kernel.org/bpf/629bc069dd807d7ac646f836e9dca28bbc1108e2.camel@mailbox.tu-berlin.de/ > > > > Fixes: 8f50f16ff39d ("selftests/bpf: Extend verifier and bpf_sock > > tests for dst_port loads") > > Signed-off-by: Jörn-Thorben Hinz <jthinz@mailbox.tu-berlin.de> > > --- > > tools/testing/selftests/bpf/prog_tests/sock_fields.c | 1 - > > 1 file changed, 1 deletion(-) > > > > diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > index 9d211b5c22c4..7d23166c77af 100644 > > --- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > +++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > @@ -394,7 +394,6 @@ void serial_test_sock_fields(void) > > test(); > > > > done: > > - test_sock_fields__detach(skel); > > test_sock_fields__destroy(skel); > > if (child_cg_fd >= 0) > > close(child_cg_fd); > > > > Great catch! I think we have similar detach & destroy pattern in a > number > of places in selftests. Did a quick grep for other __detach(skel) calls yesterday. I didn’t find similar places that were too obviously problematic. > > Should we rather just move the label, like: Sure, if you would prefer that? Let me know. Since this test—unlike others—does not attach the skel twice (like prog_tests/test_lsm.c), or reads/asserts values from the skel’s data sections between detach and destroy (like prog_tests/timer.c), my thought was to just let __destroy() do all the clean-up. > > diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > index 9d211b5c22c4..e8a947241e37 100644 > --- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > +++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > @@ -393,8 +393,8 @@ void serial_test_sock_fields(void) > > test(); > > -done: > test_sock_fields__detach(skel); > +done: > test_sock_fields__destroy(skel); > if (child_cg_fd >= 0) > close(child_cg_fd);
Jörn-Thorben Hinz wrote: > test_sock_fields__detach() got called with a null pointer here when one > of the CHECKs or ASSERTs up to the test_sock_fields__open_and_load() > call resulted in a jump to the "done" label. > > A skeletons *__detach() is not safe to call with a null pointer, though. > This led to a segfault. > > Go the easy route and only call test_sock_fields__destroy() which is > null-pointer safe and includes detaching. > > Came across this while looking[1] to introduce the usage of > bpf_tcp_helpers.h (included in progs/test_sock_fields.c) together with > vmlinux.h. > > [1] https://lore.kernel.org/bpf/629bc069dd807d7ac646f836e9dca28bbc1108e2.camel@mailbox.tu-berlin.de/ > > Fixes: 8f50f16ff39d ("selftests/bpf: Extend verifier and bpf_sock tests for dst_port loads") > Signed-off-by: Jörn-Thorben Hinz <jthinz@mailbox.tu-berlin.de> > --- > tools/testing/selftests/bpf/prog_tests/sock_fields.c | 1 - > 1 file changed, 1 deletion(-) > > diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > index 9d211b5c22c4..7d23166c77af 100644 > --- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > +++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > @@ -394,7 +394,6 @@ void serial_test_sock_fields(void) > test(); > > done: > - test_sock_fields__detach(skel); > test_sock_fields__destroy(skel); > if (child_cg_fd >= 0) > close(child_cg_fd); > -- > 2.30.2 > But we should still call __detach(skel) after the !skel check is done I assume. So rather than remove it should add a new label and jump to that, done: test_sock_fields__detach(); done_no_skel: test_sock_fields__destroy()
On Tue, 2022-06-21 at 12:54 -0700, John Fastabend wrote: > Jörn-Thorben Hinz wrote: > > test_sock_fields__detach() got called with a null pointer here when > > one > > of the CHECKs or ASSERTs up to the > > test_sock_fields__open_and_load() > > call resulted in a jump to the "done" label. > > > > A skeletons *__detach() is not safe to call with a null pointer, > > though. > > This led to a segfault. > > > > Go the easy route and only call test_sock_fields__destroy() which > > is > > null-pointer safe and includes detaching. > > > > Came across this while looking[1] to introduce the usage of > > bpf_tcp_helpers.h (included in progs/test_sock_fields.c) together > > with > > vmlinux.h. > > > > [1] > > https://lore.kernel.org/bpf/629bc069dd807d7ac646f836e9dca28bbc1108e2.camel@mailbox.tu-berlin.de/ > > > > Fixes: 8f50f16ff39d ("selftests/bpf: Extend verifier and bpf_sock > > tests for dst_port loads") > > Signed-off-by: Jörn-Thorben Hinz <jthinz@mailbox.tu-berlin.de> > > --- > > tools/testing/selftests/bpf/prog_tests/sock_fields.c | 1 - > > 1 file changed, 1 deletion(-) > > > > diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > index 9d211b5c22c4..7d23166c77af 100644 > > --- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > +++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > @@ -394,7 +394,6 @@ void serial_test_sock_fields(void) > > test(); > > > > done: > > - test_sock_fields__detach(skel); > > test_sock_fields__destroy(skel); > > if (child_cg_fd >= 0) > > close(child_cg_fd); > > -- > > 2.30.2 > > > > But we should still call __detach(skel) after the !skel check > is done I assume. If I’m not mistaken, that’s not necessary for a proper clean-up. It should be more of a stylistic question. See the parallel message from Daniel (and replies). test_sock_fields__detach() directly translates to bpf_object__detach_skeleton(). test_sock_fields__destroy() basically translates to bpf_object__destroy_skeleton(), including a null-ptr check. But bpf_object__destroy_skeleton() calls bpf_object__detach_skeleton() as its first step. So calling __detach()/__detach_skeleton() explicitly and separately is not necessary for a clean exit, if not otherwise required. > So rather than remove it should add a new label > and jump to that, > > > done: > test_sock_fields__detach(); > done_no_skel: > test_sock_fields__destroy()
Jörn-Thorben Hinz wrote: > On Tue, 2022-06-21 at 12:54 -0700, John Fastabend wrote: > > Jörn-Thorben Hinz wrote: > > > test_sock_fields__detach() got called with a null pointer here when > > > one > > > of the CHECKs or ASSERTs up to the > > > test_sock_fields__open_and_load() > > > call resulted in a jump to the "done" label. > > > > > > A skeletons *__detach() is not safe to call with a null pointer, > > > though. > > > This led to a segfault. > > > > > > Go the easy route and only call test_sock_fields__destroy() which > > > is > > > null-pointer safe and includes detaching. > > > > > > Came across this while looking[1] to introduce the usage of > > > bpf_tcp_helpers.h (included in progs/test_sock_fields.c) together > > > with > > > vmlinux.h. > > > > > > [1] > > > https://lore.kernel.org/bpf/629bc069dd807d7ac646f836e9dca28bbc1108e2.camel@mailbox.tu-berlin.de/ > > > > > > Fixes: 8f50f16ff39d ("selftests/bpf: Extend verifier and bpf_sock > > > tests for dst_port loads") > > > Signed-off-by: Jörn-Thorben Hinz <jthinz@mailbox.tu-berlin.de> > > > --- > > > tools/testing/selftests/bpf/prog_tests/sock_fields.c | 1 - > > > 1 file changed, 1 deletion(-) > > > > > > diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > > b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > > index 9d211b5c22c4..7d23166c77af 100644 > > > --- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > > +++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c > > > @@ -394,7 +394,6 @@ void serial_test_sock_fields(void) > > > test(); > > > > > > done: > > > - test_sock_fields__detach(skel); > > > test_sock_fields__destroy(skel); > > > if (child_cg_fd >= 0) > > > close(child_cg_fd); > > > -- > > > 2.30.2 > > > > > > > But we should still call __detach(skel) after the !skel check > > is done I assume. > If I’m not mistaken, that’s not necessary for a proper clean-up. It > should be more of a stylistic question. See the parallel message from > Daniel (and replies). > > test_sock_fields__detach() directly translates to > bpf_object__detach_skeleton(). test_sock_fields__destroy() basically > translates to bpf_object__destroy_skeleton(), including a null-ptr > check. > > But bpf_object__destroy_skeleton() calls bpf_object__detach_skeleton() > as its first step. So calling __detach()/__detach_skeleton() explicitly > and separately is not necessary for a clean exit, if not otherwise > required. Seems to be the case nice catch. I'm OK with it as is then. Acked-by: John Fastabend <john.fastabend@gmail.com> > > > > So rather than remove it should add a new label > > and jump to that, > > > > > > done: > > test_sock_fields__detach(); > > done_no_skel: > > test_sock_fields__destroy() > >
On Tue, Jun 21, 2022 at 09:01:16AM +0200, Jörn-Thorben Hinz wrote: > Go the easy route and only call test_sock_fields__destroy() which is > null-pointer safe and includes detaching. Reviewed-by: Martin KaFai Lau <kafai@fb.com>
Hello: This patch was applied to bpf/bpf-next.git (master) by Andrii Nakryiko <andrii@kernel.org>: On Tue, 21 Jun 2022 09:01:16 +0200 you wrote: > test_sock_fields__detach() got called with a null pointer here when one > of the CHECKs or ASSERTs up to the test_sock_fields__open_and_load() > call resulted in a jump to the "done" label. > > A skeletons *__detach() is not safe to call with a null pointer, though. > This led to a segfault. > > [...] Here is the summary with links: - [bpf-next] selftests/bpf: Fix rare segfault in sock_fields prog test https://git.kernel.org/bpf/bpf-next/c/6dc7a0baf1a7 You are awesome, thank you!
diff --git a/tools/testing/selftests/bpf/prog_tests/sock_fields.c b/tools/testing/selftests/bpf/prog_tests/sock_fields.c index 9d211b5c22c4..7d23166c77af 100644 --- a/tools/testing/selftests/bpf/prog_tests/sock_fields.c +++ b/tools/testing/selftests/bpf/prog_tests/sock_fields.c @@ -394,7 +394,6 @@ void serial_test_sock_fields(void) test(); done: - test_sock_fields__detach(skel); test_sock_fields__destroy(skel); if (child_cg_fd >= 0) close(child_cg_fd);
test_sock_fields__detach() got called with a null pointer here when one of the CHECKs or ASSERTs up to the test_sock_fields__open_and_load() call resulted in a jump to the "done" label. A skeletons *__detach() is not safe to call with a null pointer, though. This led to a segfault. Go the easy route and only call test_sock_fields__destroy() which is null-pointer safe and includes detaching. Came across this while looking[1] to introduce the usage of bpf_tcp_helpers.h (included in progs/test_sock_fields.c) together with vmlinux.h. [1] https://lore.kernel.org/bpf/629bc069dd807d7ac646f836e9dca28bbc1108e2.camel@mailbox.tu-berlin.de/ Fixes: 8f50f16ff39d ("selftests/bpf: Extend verifier and bpf_sock tests for dst_port loads") Signed-off-by: Jörn-Thorben Hinz <jthinz@mailbox.tu-berlin.de> --- tools/testing/selftests/bpf/prog_tests/sock_fields.c | 1 - 1 file changed, 1 deletion(-)