public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* [PATCH]kprobe handler discard user space trap
@ 2006-02-28  7:23 bibo,mao
  2006-02-28 11:34 ` Prasanna S Panchamukhi
  0 siblings, 1 reply; 4+ messages in thread
From: bibo,mao @ 2006-02-28  7:23 UTC (permalink / raw)
  To: Andrew Morton
  Cc: linux-kernel, Ananth N Mavinakayanahalli, Keshavamurthy, Anil S,
	Prasanna S Panchamukhi, hiramatu

Currently kprobe handler traps only happen in kernel space, so function
kprobe_exceptions_notify should skip traps which happen in user space.
This patch modifies this, and it is based on 2.6.16-rc4.

Signed-off-by: bibo mao <bibo.mao@intel.com>

diff -Nruap a/arch/i386/kernel/kprobes.c b/arch/i386/kernel/kprobes.c
--- a/arch/i386/kernel/kprobes.c    2006-02-25 17:08:52.000000000 +0800
+++ b/arch/i386/kernel/kprobes.c    2006-03-01 10:37:50.000000000 +0800
@@ -463,6 +463,9 @@ int __kprobes kprobe_exceptions_notify(s
      struct die_args *args = (struct die_args *)data;
      int ret = NOTIFY_DONE;

+    if (user_mode(args->regs))
+        return ret;
+
      switch (val) {
      case DIE_INT3:
          if (kprobe_handler(args->regs))
diff -Nruap a/arch/ia64/kernel/kprobes.c b/arch/ia64/kernel/kprobes.c
--- a/arch/ia64/kernel/kprobes.c    2006-02-25 17:08:53.000000000 +0800
+++ b/arch/ia64/kernel/kprobes.c    2006-03-01 10:39:15.000000000 +0800
@@ -740,6 +740,9 @@ int __kprobes kprobe_exceptions_notify(s
      struct die_args *args = (struct die_args *)data;
      int ret = NOTIFY_DONE;

+    if (user_mode(args->regs))
+        return ret;
+
      switch(val) {
      case DIE_BREAK:
          /* err is break number from ia64_bad_break() */
diff -Nruap a/arch/powerpc/kernel/kprobes.c b/arch/powerpc/kernel/kprobes.c
--- a/arch/powerpc/kernel/kprobes.c    2006-02-25 17:08:52.000000000 +0800
+++ b/arch/powerpc/kernel/kprobes.c    2006-03-01 10:39:53.000000000 +0800
@@ -397,6 +397,9 @@ int __kprobes kprobe_exceptions_notify(s
      struct die_args *args = (struct die_args *)data;
      int ret = NOTIFY_DONE;

+    if (user_mode(args->regs))
+        return ret;
+
      switch (val) {
      case DIE_BPT:
          if (kprobe_handler(args->regs))
diff -Nruap a/arch/sparc64/kernel/kprobes.c b/arch/sparc64/kernel/kprobes.c
--- a/arch/sparc64/kernel/kprobes.c    2006-02-25 17:08:52.000000000 +0800
+++ b/arch/sparc64/kernel/kprobes.c    2006-03-01 10:40:16.000000000 +0800
@@ -324,6 +324,9 @@ int __kprobes kprobe_exceptions_notify(s
      struct die_args *args = (struct die_args *)data;
      int ret = NOTIFY_DONE;

+    if (user_mode(args->regs))
+        return ret;
+
      switch (val) {
      case DIE_DEBUG:
          if (kprobe_handler(args->regs))
diff -Nruap a/arch/x86_64/kernel/kprobes.c b/arch/x86_64/kernel/kprobes.c
--- a/arch/x86_64/kernel/kprobes.c    2006-02-25 17:08:52.000000000 +0800
+++ b/arch/x86_64/kernel/kprobes.c    2006-03-01 10:38:48.000000000 +0800
@@ -601,6 +601,9 @@ int __kprobes kprobe_exceptions_notify(s
      struct die_args *args = (struct die_args *)data;
      int ret = NOTIFY_DONE;

+    if (user_mode(args->regs))
+        return ret;
+
      switch (val) {
      case DIE_INT3:
          if (kprobe_handler(args->regs))


^ permalink raw reply	[flat|nested] 4+ messages in thread
* RE: [PATCH]kprobe handler discard user space trap
@ 2006-03-02  6:36 Zhang, Yanmin
  0 siblings, 0 replies; 4+ messages in thread
From: Zhang, Yanmin @ 2006-03-02  6:36 UTC (permalink / raw)
  To: bibo mao, prasanna
  Cc: Mao, Bibo, Andrew Morton, linux-kernel,
	Ananth N Mavinakayanahalli, Keshavamurthy, Anil S, hiramatu

>>-----Original Message-----
>>From: linux-kernel-owner@vger.kernel.org [mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of bibo mao
>>Sent: 2006年3月1日 14:23
>>To: prasanna@in.ibm.com
>>Cc: Mao, Bibo; Andrew Morton; linux-kernel@vger.kernel.org; Ananth N Mavinakayanahalli; Keshavamurthy, Anil S;
>>hiramatu@sdl.hitachi.co.jp
>>Subject: Re: [PATCH]kprobe handler discard user space trap
>>
>>This patch removes code in kprobe_handler() function which calculates
>>user space int3 trap address in i386 architecture. And this patch is
>>based on kprobe-handler-discard-user-space-trap.patch, against
>>2.6.16-rc5-mm1.
>>
>>Signed-off-by: bibo mao <bibo.mao@intel.com>
>>
>>--- a/arch/i386/kernel/kprobes.c	2006-03-01 12:48:54.000000000 +0800
>>+++ b/arch/i386/kernel/kprobes.c	2006-03-01 14:08:12.000000000 +0800
>>@@ -203,7 +203,7 @@ static int __kprobes kprobe_handler(stru
>>  {
>>  	struct kprobe *p;
>>  	int ret = 0;
>>-	kprobe_opcode_t *addr = NULL;
>>+	kprobe_opcode_t *addr = (kprobe_opcode_t *)(regs->eip -
>>sizeof(kprobe_opcode_t));
>>  	unsigned long *lp;
>>  	struct kprobe_ctlblk *kcb;
>>  #ifdef CONFIG_PREEMPT
>>@@ -217,17 +217,6 @@ static int __kprobes kprobe_handler(stru
>>  	preempt_disable();
>>  	kcb = get_kprobe_ctlblk();
>>
>>-	/* Check if the application is using LDT entry for its code segment and
>>-	 * calculate the address by reading the base address from the LDT entry.
>>-	 */
>>-	if ((regs->xcs & 4) && (current->mm)) {
>>-		lp = (unsigned long *) ((unsigned long)((regs->xcs >> 3) * 8)
>>-					+ (char *) current->mm->context.ldt);
>>-		addr = (kprobe_opcode_t *) (get_desc_base(lp) + regs->eip -
>>-						sizeof(kprobe_opcode_t));
>>-	} else {
>>-		addr = (kprobe_opcode_t *)(regs->eip - sizeof(kprobe_opcode_t));
>>-	}
>>  	/* Check we're not actually recursing */
>>  	if (kprobe_running()) {
>>  		p = get_kprobe(addr);
>>
>>Prasanna S Panchamukhi wrote:
>>> On Tue, Feb 28, 2006 at 03:23:12PM +0800, bibo,mao wrote:
>>>> Currently kprobe handler traps only happen in kernel space, so function
>>>> kprobe_exceptions_notify should skip traps which happen in user space.
>>>> This patch modifies this, and it is based on 2.6.16-rc4.
>>>
>>> You need to remove the code which calculates the
>>> user space address in kprobe_handler() and also you need
>>> to remove the code that checks for VM86 in i386, since
>>> your patch check if user/vm86 at the top level.
>>>
>>> Thanks
>>> Prasanna
>>>
>>>> Signed-off-by: bibo mao <bibo.mao@intel.com>
>>>>
>>>> diff -Nruap a/arch/i386/kernel/kprobes.c b/arch/i386/kernel/kprobes.c
>>>> --- a/arch/i386/kernel/kprobes.c    2006-02-25 17:08:52.000000000 +0800
>>>> +++ b/arch/i386/kernel/kprobes.c    2006-03-01 10:37:50.000000000 +0800
>>>> @@ -463,6 +463,9 @@ int __kprobes kprobe_exceptions_notify(s
>>>>      struct die_args *args = (struct die_args *)data;
>>>>      int ret = NOTIFY_DONE;
>>>>
>>>> +    if (user_mode(args->regs))
>>>> +        return ret;
>>>> +
>>>>      switch (val) {
>>>>      case DIE_INT3:
>>>>          if (kprobe_handler(args->regs))
>>>> diff -Nruap a/arch/ia64/kernel/kprobes.c b/arch/ia64/kernel/kprobes.c
>>>> --- a/arch/ia64/kernel/kprobes.c    2006-02-25 17:08:53.000000000 +0800
>>>> +++ b/arch/ia64/kernel/kprobes.c    2006-03-01 10:39:15.000000000 +0800
>>>> @@ -740,6 +740,9 @@ int __kprobes kprobe_exceptions_notify(s
>>>>      struct die_args *args = (struct die_args *)data;
>>>>      int ret = NOTIFY_DONE;
>>>>
>>>> +    if (user_mode(args->regs))
>>>> +        return ret;
>>>> +
>>>>      switch(val) {
>>>>      case DIE_BREAK:
>>>>          /* err is break number from ia64_bad_break() */
>>>> diff -Nruap a/arch/powerpc/kernel/kprobes.c b/arch/powerpc/kernel/kprobes.c
>>>> --- a/arch/powerpc/kernel/kprobes.c    2006-02-25 17:08:52.000000000 +0800
>>>> +++ b/arch/powerpc/kernel/kprobes.c    2006-03-01 10:39:53.000000000 +0800
>>>> @@ -397,6 +397,9 @@ int __kprobes kprobe_exceptions_notify(s
>>>>      struct die_args *args = (struct die_args *)data;
>>>>      int ret = NOTIFY_DONE;
>>>>
>>>> +    if (user_mode(args->regs))
>>>> +        return ret;
>>>> +
>>>>      switch (val) {
>>>>      case DIE_BPT:
>>>>          if (kprobe_handler(args->regs))
>>>> diff -Nruap a/arch/sparc64/kernel/kprobes.c b/arch/sparc64/kernel/kprobes.c
>>>> --- a/arch/sparc64/kernel/kprobes.c    2006-02-25 17:08:52.000000000 +0800
>>>> +++ b/arch/sparc64/kernel/kprobes.c    2006-03-01 10:40:16.000000000 +0800
>>>> @@ -324,6 +324,9 @@ int __kprobes kprobe_exceptions_notify(s
>>>>      struct die_args *args = (struct die_args *)data;
>>>>      int ret = NOTIFY_DONE;
>>>>
>>>> +    if (user_mode(args->regs))
>>>> +        return ret;
>>>> +
>>>>      switch (val) {
>>>>      case DIE_DEBUG:
>>>>          if (kprobe_handler(args->regs))
>>>> diff -Nruap a/arch/x86_64/kernel/kprobes.c b/arch/x86_64/kernel/kprobes.c
>>>> --- a/arch/x86_64/kernel/kprobes.c    2006-02-25 17:08:52.000000000 +0800
>>>> +++ b/arch/x86_64/kernel/kprobes.c    2006-03-01 10:38:48.000000000 +0800
>>>> @@ -601,6 +601,9 @@ int __kprobes kprobe_exceptions_notify(s
>>>>      struct die_args *args = (struct die_args *)data;
>>>>      int ret = NOTIFY_DONE;
>>>>
>>>> +    if (user_mode(args->regs))
>>>> +        return ret;
>>>> +
>>>>      switch (val) {
>>>>      case DIE_INT3:
>>>>          if (kprobe_handler(args->regs))
>>>
When my ia64 box reboots with kernel 2.6.16-rc5-mm1, kernel hanged in kprobe_exceptions_notify.
args->regs doesn't always point to a valid address. It might be NULL. machine_restart calls notify_die 
with parameter regs=NULL. Pls. check args->regs before using user_mode.

Pls. also check if other archs have the same problem.


^ permalink raw reply	[flat|nested] 4+ messages in thread
end of thread, other threads:[~2006-03-02  6:37 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-02-28  7:23 [PATCH]kprobe handler discard user space trap bibo,mao
2006-02-28 11:34 ` Prasanna S Panchamukhi
2006-03-01  6:23   ` bibo mao
  -- strict thread matches above, loose matches on Subject: below --
2006-03-02  6:36 Zhang, Yanmin

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox