[Bug kprobes/2476] New: Switch Kprobes inline functions to __kprobes?

[Bug kprobes/2476] New: Switch Kprobes inline functions to __kprobes?

[jkenisto at us dot ibm dot com]

In LKML on 3/20/2006, Andrew Morton said:
> +static inline int uprobe_fault_handler(struct pt_regs *regs, int trapnr)

If, for some reason, the compiler decides to not inline this function then
you have a hunk of code which isn't marked __kprobes, but it should be.

I'd suggest that you remove all inlining from this code and add the
appropriate section markers.

Or I guess you could use __always_inline, but I'm not sure that it's really
worth the fuss and obscurity of doing that.

All kprobes-related code should be audited for this problem.

-- 
           Summary: Switch Kprobes inline functions to __kprobes?
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: kprobes
        AssignedTo: systemtap at sources dot redhat dot com
        ReportedBy: jkenisto at us dot ibm dot com


http://sourceware.org/bugzilla/show_bug.cgi?id=2476

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug kprobes/2476] Switch Kprobes inline functions to __kprobes?

[prasanna at in dot ibm dot com]

------- Additional Comments From prasanna at in dot ibm dot com  2006-03-23 04:56 -------
Jim,

Earlier kprobes used a spin_lock() to serialize kprobe execution and hence to
avoid deadlocks waiting for spin_lock(), kprobes section was introduced.
Also this kprobes section was useful to avoid probes on other routines which 
caused system crash.

Now since we have RCU based implementation, we need to re-evaluate if actually
__kprobes section is required for all the routines used by kprobes. In such a
case mark only those routines where probes on them would be unsafe.

Thanks
Prasanna

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=2476

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug kprobes/2476] Switch Kprobes inline functions to __kprobes?

[prasanna at in dot ibm dot com]

------- Additional Comments From prasanna at in dot ibm dot com  2006-03-25 09:51 -------
Created an attachment (id=941)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=941&action=view)
Removes inlined functions and adds them to kprobes section

I have removed the inlined kprobes functions and added them to kprobes section.

Please let me know if you have any issues.

Thanks
Prasanna

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=2476

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug kprobes/2476] Switch Kprobes inline functions to __kprobes?

[prasanna at in dot ibm dot com]

------- Additional Comments From prasanna at in dot ibm dot com  2006-04-07 14:12 -------
Created an attachment (id=959)
 --> (http://sourceware.org/bugzilla/attachment.cgi?id=959&action=view)
Patche port to all architectures

This tar ball contains patches ported to all architectures to remove inline for
kprobes routines and add __kprobes, to prevent probes on all kprobes routines.

Thanks
Prasanna

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=2476

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug kprobes/2476] Switch Kprobes inline functions to __kprobes?

[prasanna at in dot ibm dot com]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |anil dot s dot keshavamurthy
                   |                            |at intel dot com


http://sourceware.org/bugzilla/show_bug.cgi?id=2476

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug kprobes/2476] Switch Kprobes inline functions to __kprobes?

[prasanna at in dot ibm dot com]

------- Additional Comments From prasanna at in dot ibm dot com  2006-04-08 07:43 -------
Anil,

Can you please get this patch tested on ia64.
Please let know if you have any issues.

Thanks
Prasanna

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |WAITING


http://sourceware.org/bugzilla/show_bug.cgi?id=2476

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug kprobes/2476] Switch Kprobes inline functions to __kprobes?

[bibo dot mao at intel dot com]

------- Additional Comments From bibo dot mao at intel dot com  2006-04-10 03:24 -------
This patch works my IA64 box, but I have one question, almost all function in 
arch/i386/kernel/kprobe.c is prefixed with __kprobes. currently I think that 
the whole kernel path for KPROBE_REENTER routine should not be probed, and I do 
know whether there are other cases which should not be probed also.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=2476

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

[Bug kprobes/2476] Switch Kprobes inline functions to __kprobes?

[prasanna at in dot ibm dot com]

------- Additional Comments From prasanna at in dot ibm dot com  2006-04-10 04:30 -------
Subject: Re:  Switch Kprobes inline functions to __kprobes?

On Mon, Apr 10, 2006 at 03:24:09AM -0000, bibo dot mao at intel dot com wrote:
> 
> ------- Additional Comments From bibo dot mao at intel dot com  2006-04-10 03:24 -------
> This patch works my IA64 box, but I have one question, almost all function in 
> arch/i386/kernel/kprobe.c is prefixed with __kprobes. currently I think that 
> the whole kernel path for KPROBE_REENTER routine should not be probed, and I do 
> know whether there are other cases which should not be probed also.

True, apart from KPROBE_REENTER, probes on routines from interrupt
handler to the kprobes_handler() can cause recursion.

Thanks
Prasanna


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=2476

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Re: [Bug kprobes/2476] Switch Kprobes inline functions to __kprobes?

[Prasanna S Panchamukhi]

On Mon, Apr 10, 2006 at 03:24:09AM -0000, bibo dot mao at intel dot com wrote:
> 
> ------- Additional Comments From bibo dot mao at intel dot com  2006-04-10 03:24 -------
> This patch works my IA64 box, but I have one question, almost all function in 
> arch/i386/kernel/kprobe.c is prefixed with __kprobes. currently I think that 
> the whole kernel path for KPROBE_REENTER routine should not be probed, and I do 
> know whether there are other cases which should not be probed also.

True, apart from KPROBE_REENTER, probes on routines from interrupt
handler to the kprobes_handler() can cause recursion.

Thanks
Prasanna
-- 
Prasanna S Panchamukhi
Linux Technology Center
India Software Labs, IBM Bangalore
Email: prasanna@in.ibm.com
Ph: 91-80-51776329

[Bug kprobes/2476] Switch Kprobes inline functions to __kprobes?

[prasanna at in dot ibm dot com]

------- Additional Comments From prasanna at in dot ibm dot com  2006-04-13 16:16 -------
patches are submitted to lkml and currently there patches are in -mm tree.

Thanks
Prasanna

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
         Resolution|                            |FIXED


http://sourceware.org/bugzilla/show_bug.cgi?id=2476

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.