[v4,0/4] Add ftrace direct call for arm64

Message

Florent Revest April 3, 2023, 11:20 a.m. UTC

This series adds ftrace direct call support to arm64.
This makes BPF tracing programs (fentry/fexit/fmod_ret/lsm) work on arm64.

It is meant to be taken by the arm64 tree but it depends on the
trace-direct-v6.3-rc3 tag of the linux-trace tree:
  git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace.git
That tag was created by Steven Rostedt so the arm64 tree can pull the prior work
this depends on. [1]

Thanks to the ftrace refactoring under that tag, an ftrace_ops backing a ftrace
direct call will only ever point to *one* direct call. This means we can look up
the direct called trampoline address stored in the ops from the ftrace_caller
trampoline in the case when the destination would be out of reach of a BL
instruction at the ftrace callsite. This fixes limitations of previous attempts
such as [2].

This series has been tested on arm64 with:
1- CONFIG_FTRACE_SELFTEST
2- samples/ftrace/*.ko (cf: patch 3)
3- tools/testing/selftests/bpf/test_progs (cf: patch 4)

Changes since v3 [3]:
- Added "BTI C" instructions at the beginning of each ftrace direct call sample
- Added Mark Rutland's Acked-by to patch 2

1: https://lore.kernel.org/all/ZB2Nl7fzpHoq5V20@FVFF77S0Q05N/
2: https://lore.kernel.org/all/[e-mail]/
3: https://lore.kernel.org/bpf/[e-mail]/

Florent Revest (4):
  arm64: ftrace: Add direct call support
  arm64: ftrace: Simplify get_ftrace_plt
  arm64: ftrace: Add direct call trampoline samples support
  selftests/bpf: Update the tests deny list on aarch64

 arch/arm64/Kconfig                           |  6 ++
 arch/arm64/include/asm/ftrace.h              | 22 +++++
 arch/arm64/kernel/asm-offsets.c              |  6 ++
 arch/arm64/kernel/entry-ftrace.S             | 90 ++++++++++++++++----
 arch/arm64/kernel/ftrace.c                   | 46 +++++++---
 samples/ftrace/ftrace-direct-modify.c        | 32 +++++++
 samples/ftrace/ftrace-direct-multi-modify.c  | 36 ++++++++
 samples/ftrace/ftrace-direct-multi.c         | 22 +++++
 samples/ftrace/ftrace-direct-too.c           | 25 ++++++
 samples/ftrace/ftrace-direct.c               | 23 +++++
 tools/testing/selftests/bpf/DENYLIST.aarch64 | 82 ++----------------
 11 files changed, 288 insertions(+), 102 deletions(-)

Comments

Florent Revest April 3, 2023, 11:25 a.m. UTC | #1

On Mon, Apr 3, 2023 at 1:21 PM Florent Revest <[e-mail]> wrote:
>
> This series adds ftrace direct call support to arm64.
> This makes BPF tracing programs (fentry/fexit/fmod_ret/lsm) work on arm64.
>
> It is meant to be taken by the arm64 tree but it depends on the
> trace-direct-v6.3-rc3 tag of the linux-trace tree:
>   git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace.git
> That tag was created by Steven Rostedt so the arm64 tree can pull the prior work
> this depends on. [1]
>
> Thanks to the ftrace refactoring under that tag, an ftrace_ops backing a ftrace
> direct call will only ever point to *one* direct call. This means we can look up
> the direct called trampoline address stored in the ops from the ftrace_caller
> trampoline in the case when the destination would be out of reach of a BL
> instruction at the ftrace callsite. This fixes limitations of previous attempts
> such as [2].
>
> This series has been tested on arm64 with:
> 1- CONFIG_FTRACE_SELFTEST
> 2- samples/ftrace/*.ko (cf: patch 3)
> 3- tools/testing/selftests/bpf/test_progs (cf: patch 4)
>
> Changes since v3 [3]:
> - Added "BTI C" instructions at the beginning of each ftrace direct call sample

Ugh, I am an idiot (let's just blame Mondays!) and didn't actually
amend this change, I'm sending the series again as a v5 and this time
with the change actually folded in... Please ignore this v4, sorry for
the noise! :|