Re: [PATCH] Kprobes: Oops! in unregister_kprobe()

[Jim Keniston <jkenisto at us dot ibm dot com>]

On Mon, 2005-04-11 at 07:57, Prasanna S Panchamukhi wrote:
> Hi,
> 
> Please find the patch below to fix Oops! in unregister_kprobe().
> Please let me know if you any issues.
> 
> Thanks
> Prasanna
> 
> 
> kernel oops! when unregister_kprobe() is called on a non-registered
> kprobe. This patch fixes the above problem by checking if the probe exists
> before unregistering.
> 
> Signed-off-by: Prasanna S Panchamukhi <prasanna@in.ibm.com>
> 
> 
> ---
> 
>  linux-2.6.12-rc2-prasanna/kernel/kprobes.c |    6 +++++-
>  1 files changed, 5 insertions(+), 1 deletion(-)
> 
> diff -puN kernel/kprobes.c~kprobes-unregister-oops-fix kernel/kprobes.c
> --- linux-2.6.12-rc2/kernel/kprobes.c~kprobes-unregister-oops-fix	2005-04-11 17:23:34.000000000 +0530
> +++ linux-2.6.12-rc2-prasanna/kernel/kprobes.c	2005-04-11 17:32:50.000000000 +0530
> @@ -110,13 +110,17 @@ rm_kprobe:
>  void unregister_kprobe(struct kprobe *p)
>  {
>  	unsigned long flags;
> -	arch_remove_kprobe(p);
>  	spin_lock_irqsave(&kprobe_lock, flags);
> +	if (!get_kprobe(p->addr)) {
> +		spin_unlock_irqrestore(&kprobe_lock, flags);
> +		return;
> +	}
>  	*p->addr = p->opcode;
>  	hlist_del(&p->hlist);
>  	flush_icache_range((unsigned long) p->addr,
>  			   (unsigned long) p->addr + sizeof(kprobe_opcode_t));
>  	spin_unlock_irqrestore(&kprobe_lock, flags);
> +	arch_remove_kprobe(p);
>  }

This bug is a byproduct of the x86_64 wart described in
http://sourceware.org/ml/systemtap/2005-q2/msg00059.html.

Jim