Re: [PATCH RESEND v8 0/9] LoongArch: Add ftrace support

[Jeff Xie]

On Sat, Nov 19, 2022 at 1:21 PM Huacai Chen <chenhuacai@xxxxxxxxxx> wrote:
>
> Hi, Qing,
>
> I have applied this series to
> https://github.com/loongson/linux/commits/loongarch-next with some
> slight modifications, please test that branch to see if everything is
> correct.

I found the call trace relative issue for the branch:
https://github.com/loongson/linux/commits/loongarch-next

test01:

[root@loongarch tracing]# echo do_sys_openat2 > ./set_ftrace_filter
[root@loongarch tracing]# echo 1 > ./options/func_stack_trace
[root@loongarch tracing]# echo function > ./current_tracer
[root@loongarch tracing]# cat ./trace
# tracer: function
#
# entries-in-buffer/entries-written: 6/6   #P:1
#
#                                _-----=> irqs-off/BH-disabled
#                               / _----=> need-resched
#                              | / _---=> hardirq/softirq
#                              || / _--=> preempt-depth
#                              ||| / _-=> migrate-disable
#                              |||| /     delay
#           TASK-PID     CPU#  |||||  TIMESTAMP  FUNCTION
#              | |         |   |||||     |         |
              sh-84      [000] .....    37.784924: do_sys_openat2 <-sys_openat
              sh-84      [000] .....    37.786382: <stack trace>
 => sys_openat
 => do_syscall
 => handle_syscall
 => 0                  // abnormal output

test02:

[root@loongarch tracing]# echo do_sys_openat2:stacktrace > ./set_ftrace_filter
[root@loongarch tracing]# cat ./trace
# tracer: nop
#
# entries-in-buffer/entries-written: 3/3   #P:1
#
#                                _-----=> irqs-off/BH-disabled
#                               / _----=> need-resched
#                              | / _---=> hardirq/softirq
#                              || / _--=> preempt-depth
#                              ||| / _-=> migrate-disable
#                              |||| /     delay
#           TASK-PID     CPU#  |||||  TIMESTAMP  FUNCTION
#              | |         |   |||||     |         |
              sh-84      [000] .n.2.   107.946986: <stack trace>
 => do_syscall
 => handle_syscall
 => 0xd9ad812ae
              sh-84      [000] ...2.   110.584354: <stack trace>
 => do_syscall
 => handle_syscall
 => 0xd9ad812ae    // abnormal output

test03:

[root@loongarch tracing]# echo 1 > ./options/stacktrace
[root@loongarch tracing]# echo 1 > ./events/sched/sched_switch/enable
[root@loongarch tracing]# cat ./trace
...
<idle>-0       [000] d..2.    27.981495: sched_switch:
prev_comm=swapper/0 prev_pid=0 prev_prio=120 prev_state=R ==>
next_comm=rcu_preempt next
_pid=15 next_prio=120
          <idle>-0       [000] d..2.    27.981508: <stack trace>
 => __schedule
 => schedule_idle
 => cpu_startup_entry
 => kernel_init
 => 0x0074706d   // abnormal output



>
> Huacai
>
> On Wed, Nov 16, 2022 at 4:23 PM Qing Zhang <zhangqing@xxxxxxxxxxx> wrote:
> >
> > This patch series to support basic and dynamic ftrace.
> >
> > 1) -pg
> > Use `-pg` makes stub like a child function `void _mcount(void *ra)`.
> > Thus, it can be seen store RA and open stack before `call _mcount`.
> > Find `open stack` at first, and then find `store RA`.
> >
> > 2) -fpatchable-function-entry=2
> > The compiler has inserted 2 NOPs before the regular function prologue.
> > T series registers are available and safe because of LoongArch psABI.
> >
> > At runtime, replace nop with bl to enable ftrace call and replace bl with
> > nop to disable ftrace call. The bl requires us to save the original RA value,
> > so here it saves RA at t0.
> > details are:
> >
> > | Compiled   |       Disabled         |        Enabled         |
> > +------------+------------------------+------------------------+
> > | nop        | move     t0, ra        | move     t0, ra        |
> > | nop        | nop                    | bl      ftrace_caller  |
> > | func_body  | func_body              | func_body              |
> >
> > The RA value will be recovered by ftrace_regs_entry, and restored into RA
> > before returning to the regular function prologue. When a function is not
> > being traced, the move t0, ra is not harmful.
> >
> > performs a series of startup tests on ftrace and The test cases in selftests
> > has passed on LoongArch.
> >
> > Changes in v2:
> >  - Remove patch "LoongArch: ftrace: Add CALLER_ADDRx macros" there are other
> >    better ways
> >  Suggested by Steve:
> >  - Add HAVE_DYNAMIC_FTRACE_WITH_ARGS support (6/10)
> >  Suggested by Jinyang:
> >  - Change addu16id to lu12iw and Adjust module_finalize return value (7/10)
> >  - Use the "jr" pseudo-instruction where applicable (1/10)
> >  - Use the "la.pcrel" instead of "la" (3/10)
> >
> > Changes in v3:
> >  Reported by Jeff:
> >  - Fix unwind state when option func_stack_trace (10/10)
> >
> > Changes in v4:
> >  - No comments. Just resend the series.
> >  - Rebased onto v6.0.0-rc4.
> >
> > Changes in v5:
> >  - Modify the indentation of Kconfig and small changes
> >
> > Changes in v6:
> >  Suggested by Huacai:
> >  - Adjusting the patch Sequence
> >  - renamed mcount-dyn.S for consistency
> >
> > Changes in v7:
> >  - delete redefinition
> >
> > Changes in v8:
> >  - remove useless macro judgment and modify the return location
> >  - move some code to Patch-3
> >
> > Qing Zhang (9):
> >   LoongArch/ftrace: Add basic support
> >   LoongArch/ftrace: Add recordmcount support
> >   LoongArch/ftrace: Add dynamic function tracer support
> >   LoongArch/ftrace: Add dynamic function graph tracer support
> >   LoongArch/ftrace: Add HAVE_DYNAMIC_FTRACE_WITH_REGS support
> >   LoongArch/ftrace: Add HAVE_DYNAMIC_FTRACE_WITH_ARGS support
> >   LoongArch/ftrace: Add HAVE_FUNCTION_GRAPH_RET_ADDR_PTR support
> >   LoongArch: modules/ftrace: Initialize PLT at load time
> >   LoongArch: Enable CONFIG_KALLSYMS_ALL and CONFIG_DEBUG_FS
> >
> >  arch/loongarch/Kconfig                     |   7 +
> >  arch/loongarch/Makefile                    |   5 +
> >  arch/loongarch/configs/loongson3_defconfig |   2 +
> >  arch/loongarch/include/asm/ftrace.h        |  59 +++++
> >  arch/loongarch/include/asm/inst.h          |  15 ++
> >  arch/loongarch/include/asm/module.h        |   5 +-
> >  arch/loongarch/include/asm/module.lds.h    |   1 +
> >  arch/loongarch/include/asm/unwind.h        |   3 +-
> >  arch/loongarch/kernel/Makefile             |  13 +
> >  arch/loongarch/kernel/ftrace.c             |  74 ++++++
> >  arch/loongarch/kernel/ftrace_dyn.c         | 264 +++++++++++++++++++++
> >  arch/loongarch/kernel/inst.c               | 127 ++++++++++
> >  arch/loongarch/kernel/mcount.S             |  94 ++++++++
> >  arch/loongarch/kernel/mcount_dyn.S         | 154 ++++++++++++
> >  arch/loongarch/kernel/module-sections.c    |  11 +
> >  arch/loongarch/kernel/module.c             |  21 ++
> >  arch/loongarch/kernel/unwind_guess.c       |   4 +-
> >  arch/loongarch/kernel/unwind_prologue.c    |  46 +++-
> >  scripts/recordmcount.c                     |  23 ++
> >  19 files changed, 918 insertions(+), 10 deletions(-)
> >  create mode 100644 arch/loongarch/include/asm/ftrace.h
> >  create mode 100644 arch/loongarch/kernel/ftrace.c
> >  create mode 100644 arch/loongarch/kernel/ftrace_dyn.c
> >  create mode 100644 arch/loongarch/kernel/mcount.S
> >  create mode 100644 arch/loongarch/kernel/mcount_dyn.S
> >
> > --
> > 2.36.0
> >
> >


-- 
Thanks,
JeffXie