Xmon data-breakpoint feature is broken.
Whenever there is a watchpoint match occurs, hw_breakpoint_handler will
be called by do_break via notifier chains mechanism. If watchpoint is
registered by xmon, hw_breakpoint_handler won't find any associated
perf_event and returns immediately with NOTIFY_STOP. Similarly, do_break
also returns without notifying to xmon.
Solve this by returning NOTIFY_DONE when hw_breakpoint_handler does not
find any perf_event associated with matched watchpoint.
Signed-off-by: Ravi Bangoria <[email protected]>
---
arch/powerpc/kernel/hw_breakpoint.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/arch/powerpc/kernel/hw_breakpoint.c b/arch/powerpc/kernel/hw_breakpoint.c
index 03d089b..469d86d 100644
--- a/arch/powerpc/kernel/hw_breakpoint.c
+++ b/arch/powerpc/kernel/hw_breakpoint.c
@@ -228,8 +228,10 @@ int hw_breakpoint_handler(struct die_args *args)
rcu_read_lock();
bp = __this_cpu_read(bp_per_reg);
- if (!bp)
+ if (!bp) {
+ rc = NOTIFY_DONE;
goto out;
+ }
info = counter_arch_bp(bp);
/*
--
1.8.3.1
Ravi Bangoria <[email protected]> writes:
> Xmon data-breakpoint feature is broken.
>
> Whenever there is a watchpoint match occurs, hw_breakpoint_handler will
> be called by do_break via notifier chains mechanism. If watchpoint is
> registered by xmon, hw_breakpoint_handler won't find any associated
> perf_event and returns immediately with NOTIFY_STOP. Similarly, do_break
> also returns without notifying to xmon.
>
> Solve this by returning NOTIFY_DONE when hw_breakpoint_handler does not
> find any perf_event associated with matched watchpoint.
.. rather than NOTIFY_STOP, which tells the core code to continue
calling the other breakpoint handlers including the xmon one.
Right?
Also any idea when we broke this?
cheers
Thanks Michael,
On Tuesday 22 November 2016 05:03 PM, Michael Ellerman wrote:
> Ravi Bangoria <[email protected]> writes:
>
>> Xmon data-breakpoint feature is broken.
>>
>> Whenever there is a watchpoint match occurs, hw_breakpoint_handler will
>> be called by do_break via notifier chains mechanism. If watchpoint is
>> registered by xmon, hw_breakpoint_handler won't find any associated
>> perf_event and returns immediately with NOTIFY_STOP. Similarly, do_break
>> also returns without notifying to xmon.
>>
>> Solve this by returning NOTIFY_DONE when hw_breakpoint_handler does not
>> find any perf_event associated with matched watchpoint.
> .. rather than NOTIFY_STOP, which tells the core code to continue
> calling the other breakpoint handlers including the xmon one.
>
> Right?
Yes.
> Also any idea when we broke this?
Hmm, not sure exactly. The code is same since it was merged in 2010 when
support for hw_breakpoint was added for server processor.
-Ravi
> cheers
>