[PATCH] Transparently handle <.symbol> lookup for kprobes

[PATCH] Transparently handle <.symbol> lookup for kprobes

[Srinivasa Ds]

[-- Attachment #1: Type: text/plain, Size: 623 bytes --]

When data symbols are not present in kernel image, user needs to add
dot(".") before function name explicitly, that he wants to probe in kprobe 
module on ppc64.

for ex:-
When data symbols are missing on ppc64,
====================
[root@llm27lp1 ~]# cat /proc/kallsyms | grep do_fork
c00000000006283c T .do_fork
==============================
User needs add "." to "do_fork"

kp.symbol_name = ".do_fork";
============================

This makes kprobe modules unportable. Below patch fixes the problem.


Signed-off-by: Srinivasa Ds <srinivasa@in.ibm.com>
Signed-off-by: Ananth N Mavinakayanahalli <ananth@in.ibm.com>



[-- Attachment #2: final.patch --]
[-- Type: text/x-diff, Size: 730 bytes --]

---
 include/asm-powerpc/kprobes.h |    6 ++++++
 1 file changed, 6 insertions(+)

Index: linux-2.6.21-rc7/include/asm-powerpc/kprobes.h
===================================================================
--- linux-2.6.21-rc7.orig/include/asm-powerpc/kprobes.h
+++ linux-2.6.21-rc7/include/asm-powerpc/kprobes.h
@@ -64,6 +64,12 @@ typedef unsigned int kprobe_opcode_t;
 				addr = *(kprobe_opcode_t **)addr;	\
 		} else if (name[0] != '.')				\
 			addr = *(kprobe_opcode_t **)addr;		\
+	} else {							\
+		char dot_name[KSYM_NAME_LEN+1];				\
+		dot_name[0] = '.';					\
+		dot_name[1] = '\0';					\
+		strncat(dot_name, name, KSYM_NAME_LEN);			\
+		addr = (kprobe_opcode_t *)kallsyms_lookup_name(dot_name); \
 	}								\
 }
 

Re: [PATCH] Transparently handle <.symbol> lookup for kprobes

[Paul Mackerras]

Srinivasa Ds writes:

> +	} else {							\
> +		char dot_name[KSYM_NAME_LEN+1];				\
> +		dot_name[0] = '.';					\
> +		dot_name[1] = '\0';					\
> +		strncat(dot_name, name, KSYM_NAME_LEN);			\

Assuming the kernel strncat works like the userspace one does, there
is a possibility that dot_name[] won't be properly null-terminated
here.  If strlen(name) >= KSYM_NAME_LEN-1, then strncat will set
dot_name[KSYM_NAME_LEN-1] to something non-null and won't touch
dot_name[KSYM_NAME_LEN].

Paul.

Re: [PATCH] Transparently handle <.symbol> lookup for kprobes

[Srinivasa Ds]

Paul Mackerras wrote:
> Srinivasa Ds writes:
> 
>> +	} else {							\
>> +		char dot_name[KSYM_NAME_LEN+1];				\
>> +		dot_name[0] = '.';					\
>> +		dot_name[1] = '\0';					\
>> +		strncat(dot_name, name, KSYM_NAME_LEN);			\
> 
> Assuming the kernel strncat works like the userspace one does, there
> is a possibility that dot_name[] won't be properly null-terminated
> here.  If strlen(name) >= KSYM_NAME_LEN-1, then strncat will set
> dot_name[KSYM_NAME_LEN-1] to something non-null and won't touch
> dot_name[KSYM_NAME_LEN].

Irrespective of length of the string, kernel implementation of
strncat(lib/string.c) ensures that last character of string is set to
null. So dot_name[] is always null terminated.

========================
char *strncat(char *dest, const char *src, size_t count)
{
        char *tmp = dest;

        if (count) {
                while (*dest)
                        dest++;
                while ((*dest++ = *src++) != 0) {
                        if (--count == 0) {
                                *dest = '\0';
                                break;
                        }
                }
        }
        return tmp;
}
EXPORT_SYMBOL(strncat);
===============================

Is this OK then ??


Thanks
 Srinivasa DS

> 
> Paul.