Re: [RFC PATCH v1.9 12/14] livepatch: create per-task consistency model

[Miroslav Benes]

On Fri, 25 Mar 2016, Josh Poimboeuf wrote:

[...]

> diff --git a/kernel/fork.c b/kernel/fork.c
> index d277e83..27b181e 100644
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -76,6 +76,7 @@
>  #include <linux/compiler.h>
>  #include <linux/sysctl.h>
>  #include <linux/kcov.h>
> +#include <linux/livepatch.h>
>  
>  #include <asm/pgtable.h>
>  #include <asm/pgalloc.h>
> @@ -1615,6 +1616,7 @@ static struct task_struct *copy_process(unsigned long 
> clone_flags,
>       total_forks++;
>       spin_unlock(&current->sighand->siglock);
>       syscall_tracepoint_update(p);
> +     klp_update_task_universe(p);

Shouldn't we copy transition and TIF from the parent? I deal with a race 
in kgraft and the solution seems to be this code exactly at this place in 
copy_process(). I need to think about it.

[...]

> +static void klp_schedule_work(void)
> +{
> +     if (IS_ENABLED(CONFIG_RELIABLE_STACKTRACE))
> +             schedule_delayed_work(&klp_work, round_jiffies_relative(HZ));
> +}

As mentioned in my cover letter reply I'd allow to schedule delayed work 
even for !CONFIG_RELIABLE_STACKTRACE archs and configurations. There is a 
check in klp_try_switch_task() for CONFIG_RELIABLE_STACKTRACE so the 
change should be trivial. The patching could be successful even without 
reliable stack traces especially in combination with a fake signal and 
a syscall boundary migration.

Miroslav