Re: [Fwd: Re: [PATCH] Return probe]

[Prasanna S Panchamukhi <prasanna at in dot ibm dot com>]

> Here's the new retprobe patch, this patch should apply to 
> linix-2.6.12-rc2-mm2 + Anandth's multihadler (the one posted on LKML).
> 

Jim, Hien, 

This patch looks good to me, just one more comment.

>--- linux-2.6.12-rc2-mm2-mpapr15.org/arch/i386/kernel/entry.S   2005-04-11 13:54:37.000000000 -0700
>+++ linux-2.6.12-rc2-mm2-mprobe15apr/arch/i386/kernel/entry.S   2005-04-27 16:02:52.970687262 -0700
>@@ -997,3 +997,31 @@
>        .long sys_vperfctr_read
>
> syscall_table_size=(.-sys_call_table)
>+
>+#ifdef CONFIG_KPROBES
>+/*
>+ * For function-return probes, init_kprobes() establishes a probepoint
>+ * here. When a retprobed function returns, this probe is hit and
>+ * trampoline_probe_handler() runs, calling the kretprobe's handler.
>+ * 16 nop since we copy MAX_INSN_SIZE into p->ainsn.insn.
>+ */
>+ENTRY(kretprobe_trampoline)
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+       nop
>+/* NOT REACHED */
>+#endif
>+
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
	You need not modify entry.S file. You can use inline assembly and 
	move this kretprobe_trampoline() to arch/i386/kernel/kprobes.c file.

Also there are lots of references to __builtin_return_address() in the kernel.
 If a kretprobe is placed on some routine() and that routine calls 
__builtin_return_address(), then __builtin_return_address() will return 
incorrect value, since kretprobe modifies the retaddress stored on the stack. 
Is this limitation acceptable ?

Thanks
Prasanna

-- 

Prasanna S Panchamukhi
Linux Technology Center
India Software Labs, IBM Bangalore
Ph: 91-80-25044636
<prasanna@in.ibm.com>