Re: [PATCH] uprobes: don't enable/disable signle step if the user did it

[Oleg Nesterov <oleg@redhat.com>]

On 07/27, Sebastian Andrzej Siewior wrote:
>
> On 07/26/2012 07:31 PM, Oleg Nesterov wrote:
>> Well. I agree, this needs changes. To begin with, uprobe should avoid
>> user_enable_single_step() which does access_process_vm(). And I suspect
>> uprobes have the problems with TIF_FORCED_TF logic.
>
> Why? Shouldn't wee keep the trap flag if the instruction on which we
> placed the uprobe activates it?

Yes. But user_enable_single_step() is not the right interface.

>> But I am not sure about this patch...
>>
>> On 07/26, Sebastian Andrzej Siewior wrote:
>>>
>>> @@ -1528,7 +1528,10 @@ static void handle_swbp(struct pt_regs *regs)
>>>
>>>   	utask->state = UTASK_SSTEP;
>>>   	if (!pre_ssout(uprobe, regs, bp_vaddr)) {
>>> -		user_enable_single_step(current);
>>> +		if (test_tsk_thread_flag(current, TIF_SINGLESTEP))
>>> +			uprobe->flags |= UPROBE_USER_SSTEP;
>>> +		else
>>> +			user_enable_single_step(current);
>>
>> This is x86 specific, TIF_SINGLESTEP is not defined on every arch.
>
> It is not defined on every arch but I wouldn't say it is 86 specific.
> From the architectures which have user_enable_single_step() defined I
> see

But we do not need TIF_SINGLESTEP. At all. Again, this is ptrace thing
connected to user_enable_single_step().

Sebastian, I am sorry for being terse, I'll write another email later.

Oleg.