[PATCH RFC v3 1/5] AArch64: Add single-step and breakpoint handler hooks

[sandeepa.prabhu@linaro.org (Sandeepa Prabhu)]

On 3 October 2013 22:23, Will Deacon <will.deacon@arm.com> wrote:
> On Tue, Oct 01, 2013 at 04:57:56PM +0100, Sandeepa Prabhu wrote:
>> AArch64 Single Steping and Breakpoint debug exceptions will be
>> used by multiple debug framworks like kprobes & kgdb.
>>
>> This patch implements the hooks for those frameworks to register
>> their own handlers for handling breakpoint and single step events.
>>
>> Reworked the debug exception handler in entry.S: do_dbg to route
>> software breakpoint (BRK64) exception to do_debug_exception()
>>
>> Signed-off-by: Sandeepa Prabhu <sandeepa.prabhu@linaro.org>
>> Signed-off-by: Deepak Saxena <dsaxena@linaro.org>
>> ---
>>  arch/arm64/include/asm/debug-monitors.h | 23 +++++++++
>>  arch/arm64/kernel/debug-monitors.c      | 85 +++++++++++++++++++++++++++++++--
>>  arch/arm64/kernel/entry.S               |  2 +
>>  3 files changed, 107 insertions(+), 3 deletions(-)
>>
>> diff --git a/arch/arm64/include/asm/debug-monitors.h b/arch/arm64/include/asm/debug-monitors.h
>> index a2232d0..8e354b3 100644
>> --- a/arch/arm64/include/asm/debug-monitors.h
>> +++ b/arch/arm64/include/asm/debug-monitors.h
>> @@ -16,6 +16,8 @@
>>  #ifndef __ASM_DEBUG_MONITORS_H
>>  #define __ASM_DEBUG_MONITORS_H
>>
>> +#include <linux/rculist.h>
>> +
>>  #ifdef __KERNEL__
>>
>>  #define      DBG_ESR_EVT(x)          (((x) >> 27) & 0x7)
>> @@ -62,6 +64,27 @@ struct task_struct;
>>
>>  #define DBG_ARCH_ID_RESERVED 0       /* In case of ptrace ABI updates. */
>>
>> +#define DEBUG_HOOK_HANDLED   0
>> +#define DEBUG_HOOK_ERROR     1
>
> Cosmetic: we use DBG vs DEBUG in the rest of this header.
Ok, I'll change it to DBG_*

>
>> +struct step_hook {
>> +     struct list_head node;
>> +     int (*fn)(struct pt_regs *regs, unsigned int esr);
>> +};
>> +
>> +void register_step_hook(struct step_hook *hook);
>> +void unregister_step_hook(struct step_hook *hook);
>> +
>> +struct break_hook {
>> +     struct list_head node;
>> +     u32 esr_val;
>> +     u32 esr_mask;
>> +     int (*fn)(struct pt_regs *regs, unsigned int esr);
>> +};
>> +
>> +void register_break_hook(struct break_hook *hook);
>> +void unregister_break_hook(struct break_hook *hook);
>> +
>>  u8 debug_monitors_arch(void);
>>
>>  void enable_debug_monitors(enum debug_el el);
>> diff --git a/arch/arm64/kernel/debug-monitors.c b/arch/arm64/kernel/debug-monitors.c
>> index cbfacf7..fbbf824 100644
>> --- a/arch/arm64/kernel/debug-monitors.c
>> +++ b/arch/arm64/kernel/debug-monitors.c
>> @@ -188,6 +188,43 @@ static void clear_regs_spsr_ss(struct pt_regs *regs)
>>       regs->pstate = spsr;
>>  }
>>
>> +/* EL1 Single Step Handler hooks */
>> +static LIST_HEAD(step_hook);
>> +
>> +void register_step_hook(struct step_hook *hook)
>> +{
>> +     list_add_rcu(&hook->node, &step_hook);
>> +}
>
> This isn't safe against concurrent registrations. Why don't you use an
> rwlock instead? Then you take the writer lock here...
>
>> +/*
>> + * Call registered single step handers
>> + * There is no Syndrome info to check for determining the handler.
>> + * So we call all the registered handlers, until the right handler is
>> + * found which returns zero.
>> + */
>> +static int call_step_hook(struct pt_regs *regs, unsigned int esr)
>> +{
>> +     struct step_hook *hook;
>> +     int retval = DEBUG_HOOK_ERROR;
>> +
>> +     rcu_read_lock();
>
> ... and the reader lock here.
Hmm, rwlock sounds good, there wont be lock contention when concurrent
handlers on different CPU. I will change it to rwlocks,
can be used for call_break_hook as well instead of normal spin-lock to
reduce contention.

>
>> +     list_for_each_entry_rcu(hook, &step_hook, node) {
>> +             retval = hook->fn(regs, esr);
>> +             if (retval == DEBUG_HOOK_HANDLED)
>> +                     break;
>> +     }
>> +
>> +     rcu_read_unlock();
>> +
>> +     return retval;
>> +}
>> +
>>  static int single_step_handler(unsigned long addr, unsigned int esr,
>>                              struct pt_regs *regs)
>>  {
>> @@ -215,8 +252,11 @@ static int single_step_handler(unsigned long addr, unsigned int esr,
>>                */
>>               user_rewind_single_step(current);
>>       } else {
>> -             /* TODO: route to KGDB */
>> -             pr_warning("Unexpected kernel single-step exception at EL1\n");
>> +             /* Call single step handlers for kgdb/kprobes */
>
> Useless comment.
I will re-frame, how about simple "Call registered single step hook functions" ?
>
>> +             if (call_step_hook(regs, esr) == DEBUG_HOOK_HANDLED)
>> +                     return 0;
>> +
>> +             pr_warn("unexpected single step exception at %lx!\n", addr);
>
> Why have you reworded this warning?
oops, mistake it was debug change to print addr, revert it in next version.
>
>>               /*
>>                * Re-enable stepping since we know that we will be
>>                * returning to regs.
>> @@ -227,11 +267,50 @@ static int single_step_handler(unsigned long addr, unsigned int esr,
>>       return 0;
>>  }
>>
>> +
>> +static LIST_HEAD(break_hook);
>> +static DEFINE_RAW_SPINLOCK(break_hook_lock);
>> +
>> +void register_break_hook(struct break_hook *hook)
>> +{
>> +     raw_spin_lock(&break_hook_lock);
>> +     list_add(&hook->node, &break_hook);
>> +     raw_spin_unlock(&break_hook_lock);
>> +}
>> +
>> +void unregister_break_hook(struct break_hook *hook)
>> +{
>> +     raw_spin_lock(&break_hook_lock);
>> +     list_del(&hook->node);
>> +     raw_spin_unlock(&break_hook_lock);
>> +}
>> +
>> +static int call_break_hook(struct pt_regs *regs, unsigned int esr)
>> +{
>> +     struct break_hook *hook;
>> +     int (*fn)(struct pt_regs *regs, unsigned int esr) = NULL;
>> +
>> +     raw_spin_lock(&break_hook_lock);
>> +     list_for_each_entry(hook, &break_hook, node)
>> +             if ((esr & hook->esr_mask) == hook->esr_val)
>> +                     fn = hook->fn;
>> +     raw_spin_unlock(&break_hook_lock);
>> +
>> +     return fn ? fn(regs, esr) : DEBUG_HOOK_ERROR;
>> +}
>> +
>>  static int brk_handler(unsigned long addr, unsigned int esr,
>>                      struct pt_regs *regs)
>>  {
>>       siginfo_t info;
>>
>> +     /* call single step handlers for kgdb/kprobes */
>> +     if (call_break_hook(regs, esr) == DEBUG_HOOK_HANDLED)
>> +             return 0;
>> +
>> +     pr_warn("unexpected brk exception at %llx, esr=0x%x\n",
>> +                     instruction_pointer(regs), esr);
>
> %lx for the pc.
Hmm, shall correct it in v4.
>
> Will