Re: [PATCH v3 4/4] perf: Use CAP_SYS_ADMIN instead of euid==0 with ftrace

[Mathieu Poirier]

Hi Arnaldo,

On Tue, 13 Aug 2019 at 07:23, Arnaldo Carvalho de Melo
<arnaldo.melo@xxxxxxxxx> wrote:
>
> Em Mon, Aug 12, 2019 at 03:42:17PM -0600, Mathieu Poirier escreveu:
> > On Mon, 12 Aug 2019 at 14:29, Arnaldo Carvalho de Melo
> > <arnaldo.melo@xxxxxxxxx> wrote:
> > >
> > > Em Mon, Aug 12, 2019 at 05:27:06PM -0300, Arnaldo Carvalho de Melo escreveu:
> > > > Em Mon, Aug 12, 2019 at 05:22:51PM -0300, Arnaldo Carvalho de Melo escreveu:
> > > > > Em Wed, Aug 07, 2019 at 10:44:17AM -0400, Igor Lubashev escreveu:
> > > > > > @@ -281,7 +283,7 @@ static int __cmd_ftrace(struct perf_ftrace *ftrace, int argc, const char **argv)
> > > > > >           .events = POLLIN,
> > > > > >   };
> > > > > >
> > > > > > - if (geteuid() != 0) {
> > > > > > + if (!perf_cap__capable(CAP_SYS_ADMIN)) {
> > > > > >           pr_err("ftrace only works for root!\n");
> > > > >
> > > > > I guess we should update the error message too?
> > > > >
> > > >
> > > > I.e. I applied this as a follow up patch:
> > > >
> > > > diff --git a/tools/perf/builtin-ftrace.c b/tools/perf/builtin-ftrace.c
> > > > index 01a5bb58eb04..ba8b65c2f9dc 100644
> > > > --- a/tools/perf/builtin-ftrace.c
> > > > +++ b/tools/perf/builtin-ftrace.c
> > > > @@ -284,7 +284,12 @@ static int __cmd_ftrace(struct perf_ftrace *ftrace, int argc, const char **argv)
> > > >       };
> > > >
> > > >       if (!perf_cap__capable(CAP_SYS_ADMIN)) {
> > > > -             pr_err("ftrace only works for root!\n");
> > > > +             pr_err("ftrace only works for %s!\n",
> > > > +#ifdef HAVE_LIBCAP_SUPPORT
> > > > +             "users with the SYS_ADMIN capability"
> > > > +#else
> > > > +             "root"
> > > > +#endif
> > >
> > >                 );
> > >
> > > :-)
> > >
> > > >               return -1;
> > > >       }
> > > >
> > >
> > > I've pushed the whole set to my tmp.perf/cap branch, please chec
> >
> > Please hold on before moving further - I'm getting a segmentation
> > fault on ARM64 that I'm still trying to figure out.
>
> This is just sitting in my tmp branch, and in my local perf/core branch,
> so that I can test it with the containers, etc.
>
> Is this related to the following fix?

That is the first thing I thought about but no, it has nothing to do
with it.  Patch 3/4 is where the problem shows up.  The code in the
patch is fine, it is the repercussion it has on other part that needs
to be investigated.

Right now I see that kmap->ref_reloc_sym is NULL here [1] when tracing
with anything else than the 'u' option.  I am currently investigating
the problem.

Igor, please see if you can reproduce on QEMU or an ARM64 based platform.

[1] https://elixir.bootlin.com/linux/v5.3-rc4/source/tools/perf/util/event.c#L945

>
> commit 3e70008a6021fffd2cd1614734603ea970773060
> Author: Leo Yan <leo.yan@xxxxxxxxxx>
> Date:   Fri Aug 9 18:47:52 2019 +0800
>
>     perf trace: Fix segmentation fault when access syscall info on arm64
>
>     'perf trace' reports the segmentation fault as below on Arm64:
>
>       # perf trace -e string -e augmented_raw_syscalls.c
>       LLVM: dumping tools/perf/examples/bpf/augmented_raw_syscalls.o
>       perf: Segmentation fault
>       Obtained 12 stack frames.
>       perf(sighandler_dump_stack+0x47) [0xaaaaac96ac87]
>       linux-vdso.so.1(+0x5b7) [0xffffadbeb5b7]
>       /lib/aarch64-linux-gnu/libc.so.6(strlen+0x10) [0xfffface7d5d0]
>       /lib/aarch64-linux-gnu/libc.so.6(_IO_vfprintf+0x1ac7) [0xfffface49f97]
>       /lib/aarch64-linux-gnu/libc.so.6(__vsnprintf_chk+0xc7) [0xffffacedfbe7]
>       perf(scnprintf+0x97) [0xaaaaac9ca3ff]
>       perf(+0x997bb) [0xaaaaac8e37bb]
>       perf(cmd_trace+0x28e7) [0xaaaaac8ec09f]
>       perf(+0xd4a13) [0xaaaaac91ea13]
>       perf(main+0x62f) [0xaaaaac8a147f]
>       /lib/aarch64-linux-gnu/libc.so.6(__libc_start_main+0xe3) [0xfffface22d23]
>       perf(+0x57723) [0xaaaaac8a1723]
>       Segmentation fault
>
>     This issue is introduced by commit 30a910d7d3e0 ("perf trace:
>     Preallocate the syscall table"), it allocates trace->syscalls.table[]
>     array and the element count is 'trace->sctbl->syscalls.nr_entries'; but
>     on Arm64, the system call number is not continuously used; e.g. the
>     syscall maximum id is 436 but the real entries is only 281.
>
>