mbox series

[v3,0/3] ftrace: Introduce PERMANENT ftrace_ops flag

Message ID 20191016113316.13415-1-mbenes@suse.cz (mailing list archive)
Headers show
Series ftrace: Introduce PERMANENT ftrace_ops flag | expand

Message

Miroslav Benes Oct. 16, 2019, 11:33 a.m. UTC
Livepatch uses ftrace for redirection to new patched functions. It means
that if ftrace is disabled, all live patched functions are disabled as
well. Toggling global 'ftrace_enabled' sysctl thus affect it directly.
It is not a problem per se, because only administrator can set sysctl
values, but it still may be surprising.

Introduce PERMANENT ftrace_ops flag to amend this. If the
FTRACE_OPS_FL_PERMANENT is set on any ftrace ops, the tracing cannot be
disabled by disabling ftrace_enabled. Equally, a callback with the flag
set cannot be registered if ftrace_enabled is disabled.

v2->v3:
- ftrace_enabled explicitly set to true
- selftest from Joe Lawrence (I just split it to two patches)
- typo fix

v1->v2:
- different logic, proposed by Joe Lawrence

Joe Lawrence (2):
  selftests/livepatch: Make dynamic debug setup and restore generic
  selftests/livepatch: Test interaction with ftrace_enabled

Miroslav Benes (1):
  ftrace: Introduce PERMANENT ftrace_ops flag

 Documentation/trace/ftrace-uses.rst           |  8 +++
 Documentation/trace/ftrace.rst                |  4 +-
 include/linux/ftrace.h                        |  3 +
 kernel/livepatch/patch.c                      |  3 +-
 kernel/trace/ftrace.c                         | 23 ++++++-
 tools/testing/selftests/livepatch/Makefile    |  3 +-
 .../testing/selftests/livepatch/functions.sh  | 34 +++++++---
 .../selftests/livepatch/test-callbacks.sh     |  2 +-
 .../selftests/livepatch/test-ftrace.sh        | 65 +++++++++++++++++++
 .../selftests/livepatch/test-livepatch.sh     |  2 +-
 .../selftests/livepatch/test-shadow-vars.sh   |  2 +-
 11 files changed, 132 insertions(+), 17 deletions(-)
 create mode 100755 tools/testing/selftests/livepatch/test-ftrace.sh

Comments

Steven Rostedt Oct. 16, 2019, 2:58 p.m. UTC | #1
On Wed, 16 Oct 2019 13:33:12 +0200
Miroslav Benes <mbenes@suse.cz> wrote:

> Livepatch uses ftrace for redirection to new patched functions. It means
> that if ftrace is disabled, all live patched functions are disabled as
> well. Toggling global 'ftrace_enabled' sysctl thus affect it directly.
> It is not a problem per se, because only administrator can set sysctl
> values, but it still may be surprising.
> 
> Introduce PERMANENT ftrace_ops flag to amend this. If the
> FTRACE_OPS_FL_PERMANENT is set on any ftrace ops, the tracing cannot be
> disabled by disabling ftrace_enabled. Equally, a callback with the flag
> set cannot be registered if ftrace_enabled is disabled.
> 
> v2->v3:
> - ftrace_enabled explicitly set to true
> - selftest from Joe Lawrence (I just split it to two patches)
> - typo fix
> 
> v1->v2:
> - different logic, proposed by Joe Lawrence
> 
> Joe Lawrence (2):
>   selftests/livepatch: Make dynamic debug setup and restore generic
>   selftests/livepatch: Test interaction with ftrace_enabled
> 
> Miroslav Benes (1):
>   ftrace: Introduce PERMANENT ftrace_ops flag
> 

Would you like me to take all three patches through my tree?

-- Steve
Miroslav Benes Oct. 16, 2019, 3:01 p.m. UTC | #2
On Wed, 16 Oct 2019, Steven Rostedt wrote:

> On Wed, 16 Oct 2019 13:33:12 +0200
> Miroslav Benes <mbenes@suse.cz> wrote:
> 
> > Livepatch uses ftrace for redirection to new patched functions. It means
> > that if ftrace is disabled, all live patched functions are disabled as
> > well. Toggling global 'ftrace_enabled' sysctl thus affect it directly.
> > It is not a problem per se, because only administrator can set sysctl
> > values, but it still may be surprising.
> > 
> > Introduce PERMANENT ftrace_ops flag to amend this. If the
> > FTRACE_OPS_FL_PERMANENT is set on any ftrace ops, the tracing cannot be
> > disabled by disabling ftrace_enabled. Equally, a callback with the flag
> > set cannot be registered if ftrace_enabled is disabled.
> > 
> > v2->v3:
> > - ftrace_enabled explicitly set to true
> > - selftest from Joe Lawrence (I just split it to two patches)
> > - typo fix
> > 
> > v1->v2:
> > - different logic, proposed by Joe Lawrence
> > 
> > Joe Lawrence (2):
> >   selftests/livepatch: Make dynamic debug setup and restore generic
> >   selftests/livepatch: Test interaction with ftrace_enabled
> > 
> > Miroslav Benes (1):
> >   ftrace: Introduce PERMANENT ftrace_ops flag
> > 
> 
> Would you like me to take all three patches through my tree?

I think that would be the easiest, yes.

Thanks
Miroslav