lkml.org 
[lkml]   [2020]   [Nov]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v5 14/21] kprobes: Remove NMI context check
On Mon, 2 Nov 2020 09:27:26 -0500
Steven Rostedt <rostedt@goodmis.org> wrote:

>
> [ Peter Z, please take a look a this ]
>
> On Mon, 2 Nov 2020 16:02:34 +0900
> Masami Hiramatsu <mhiramat@kernel.org> wrote:
>
> > >From 509b27efef8c7dbf56cab2e812916d6cd778c745 Mon Sep 17 00:00:00 2001
> > From: Masami Hiramatsu <mhiramat@kernel.org>
> > Date: Mon, 2 Nov 2020 15:37:28 +0900
> > Subject: [PATCH] kprobes: Disable lockdep for kprobe busy area
> >
> > Since the code area in between kprobe_busy_begin()/end() prohibits
> > other kprobs to call probe handlers, we can avoid inconsitent
> > locks there. But lockdep doesn't know that, so it warns rp->lock
> > or kretprobe_table_lock.
> >
> > To supress those false-positive errors, disable lockdep while
> > kprobe_busy is set.
> >
> > Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
> > ---
> > kernel/kprobes.c | 2 ++
> > 1 file changed, 2 insertions(+)
> >
> > diff --git a/kernel/kprobes.c b/kernel/kprobes.c
> > index 8a12a25fa40d..c7196e583600 100644
> > --- a/kernel/kprobes.c
> > +++ b/kernel/kprobes.c
> > @@ -1295,10 +1295,12 @@ void kprobe_busy_begin(void)
> > __this_cpu_write(current_kprobe, &kprobe_busy);
> > kcb = get_kprobe_ctlblk();
> > kcb->kprobe_status = KPROBE_HIT_ACTIVE;
> > + lockdep_off();
> > }
> >
> > void kprobe_busy_end(void)
> > {
> > + lockdep_on();
> > __this_cpu_write(current_kprobe, NULL);
> > preempt_enable();
> > }
> > --
>
> No, this is not the correct workaround (too big of a hammer). You could do
> the following:
>
> From 4139d9c8437b0bd2262e989ca4eb0a83b7e7bb72 Mon Sep 17 00:00:00 2001
> From: "Steven Rostedt (VMware)" <rostedt@goodmis.org>
> Date: Mon, 2 Nov 2020 09:17:49 -0500
> Subject: [PATCH] kprobes: Tell lockdep about kprobe nesting
>
> Since the kprobe handlers have protection that prohibits other handlers from
> executing in other contexts (like if an NMI comes in while processing a
> kprobe, and executes the same kprobe, it will get fail with a "busy"
> return). Lockdep is unaware of this protection. Use lockdep's nesting api to
> differentiate between locks taken in NMI context and other context to
> supress the false warnings.

Ah, OK. This looks good to me.

BTW, in_nmi() in pre_handler_kretprobe() always be true because
now int3 is treated as an NMI. So you can always pass 1 there.

Acked-by: Masami Hiramatsu <mhiramat@kernel.org>

Thank you,

>
> Link: https://lore.kernel.org/r/20201102160234.fa0ae70915ad9e2b21c08b85@kernel.org
>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
> ---
> kernel/kprobes.c | 24 ++++++++++++++++++++----
> 1 file changed, 20 insertions(+), 4 deletions(-)
>
> diff --git a/kernel/kprobes.c b/kernel/kprobes.c
> index 8a12a25fa40d..ccb285867059 100644
> --- a/kernel/kprobes.c
> +++ b/kernel/kprobes.c
> @@ -1249,7 +1249,12 @@ __acquires(hlist_lock)
>
> *head = &kretprobe_inst_table[hash];
> hlist_lock = kretprobe_table_lock_ptr(hash);
> - raw_spin_lock_irqsave(hlist_lock, *flags);
> + /*
> + * Nested is a workaround that will soon not be needed.
> + * There's other protections that make sure the same lock
> + * is not taken on the same CPU that lockdep is unaware of.
> + */
> + raw_spin_lock_irqsave_nested(hlist_lock, *flags, !!in_nmi());
> }
> NOKPROBE_SYMBOL(kretprobe_hash_lock);
>
> @@ -1258,7 +1263,12 @@ static void kretprobe_table_lock(unsigned long hash,
> __acquires(hlist_lock)
> {
> raw_spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash);
> - raw_spin_lock_irqsave(hlist_lock, *flags);
> + /*
> + * Nested is a workaround that will soon not be needed.
> + * There's other protections that make sure the same lock
> + * is not taken on the same CPU that lockdep is unaware of.
> + */
> + raw_spin_lock_irqsave_nested(hlist_lock, *flags, !!in_nmi());
> }
> NOKPROBE_SYMBOL(kretprobe_table_lock);
>
> @@ -2025,10 +2035,16 @@ static int pre_handler_kretprobe(struct kprobe *p, struct pt_regs *regs)
> struct kretprobe *rp = container_of(p, struct kretprobe, kp);
> unsigned long hash, flags = 0;
> struct kretprobe_instance *ri;
> + int nmi = !!in_nmi();
>
> /* TODO: consider to only swap the RA after the last pre_handler fired */
> hash = hash_ptr(current, KPROBE_HASH_BITS);
> - raw_spin_lock_irqsave(&rp->lock, flags);
> + /*
> + * Nested is a workaround that will soon not be needed.
> + * There's other protections that make sure the same lock
> + * is not taken on the same CPU that lockdep is unaware of.
> + */
> + raw_spin_lock_irqsave_nested(&rp->lock, flags, nmi);
> if (!hlist_empty(&rp->free_instances)) {
> ri = hlist_entry(rp->free_instances.first,
> struct kretprobe_instance, hlist);
> @@ -2039,7 +2055,7 @@ static int pre_handler_kretprobe(struct kprobe *p, struct pt_regs *regs)
> ri->task = current;
>
> if (rp->entry_handler && rp->entry_handler(ri, regs)) {
> - raw_spin_lock_irqsave(&rp->lock, flags);
> + raw_spin_lock_irqsave_nested(&rp->lock, flags, nmi);
> hlist_add_head(&ri->hlist, &rp->free_instances);
> raw_spin_unlock_irqrestore(&rp->lock, flags);
> return 0;
> --
> 2.25.4
>


--
Masami Hiramatsu <mhiramat@kernel.org>

\
 
 \ /
  Last update: 2020-11-03 06:41    [W:0.101 / U:0.920 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site