Change all instances of EXPORT_SYMBOL() in the core EDAC module to
EXPORT_SYMBOL_GPL().
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

- Add entries to MAINTAINERS list for EDAC-E752X, EDAC-E7XXX, and
  EDAC-R82600 chipset drivers

- Fix MAINTAINERS entry for EDAC-CORE so it uses tabs rather than
  spaces to indent.  This is consistent with how the other entries are
  formatted.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Patch from Dave Jiang <djiang@mvista.com>: Fix EDAC e752x driver so it
outputs sysbus-specific error message when sysbus error detected.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Fix spelling errors in EDAC documentation.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Cosmetic indentation/formatting cleanup for EDAC code.  Make sure we
are using tabs rather than spaces to indent, etc.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Fix EDAC code so EXPORT_SYMBOL comes after the function that is being
exported.  This is to maintain consistency with the rest of the kernel.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

- Add x86 dependency in drivers/edac/Kconfig for all current
  platform-specific modules.

- Add PCI dependency to Radisys 82600 driver
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

- Fix code so we always hold mem_ctls_mutex while we are stepping
  through the list of mem_ctl_info structures.  Otherwise bad things
  may happen if one task is stepping through the list while another
  task is modifying it.  We may eventually want to use reference
  counting to manage the mem_ctl_info structures.  In the meantime we
  may as well fix this bug.

- Don't disable interrupts while we are walking the list of
  mem_ctl_info structures in check_mc_devices().  This is unnecessary.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

- After we unregister a kobject, wait for our kobject release method
  to call complete().  This causes us to wait until the kobject
  reference count reaches 0.  Otherwise, a task accessing the EDAC
  sysfs interface can hold the reference count above 0 until after the
  EDAC module has been unloaded.  When the reference count finally
  drops to 0, this will result in an attempt to call our release
  method inside the EDAC module after the module has already been
  unloaded.

  This isn't the best fix, since a process can get stuck sleeping forever
  uninterruptibly if the user does the following:

      rmmod my_module < /sys/my_sysfs/file

  I'll go back and implement a better fix later.  However this should
  be ok for now.

- Call edac_remove_sysfs_mci_device() from edac_mc_del_mc() rather
  than from edac_mc_free().  Since edac_mc_add_mc() calls
  edac_create_sysfs_mci_device(), edac_mc_del_mc() should call
  edac_remove_sysfs_mci_device().
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

- Remove calls to kobject_init().  These are unnecessary because
  kobject_register() calls kobject_init().

- Remove extra calls to kobject_put().  When we call
  kobject_unregister(), this releases our reference to the kobject.
  The extra calls to kobject_put() may cause the reference count to
  drop to 0 while a kobject is still in use.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This is part 2 of a 2-part patch set.

Fix edac_mc_add_mc() so it cleans up properly if call to
edac_create_sysfs_mci_device() fails.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This is part 1 of a 2-part patch set.  The code changes are split into
two parts to make the patches more readable.

Move complete_mc_list_del() and del_mc_from_global_list() so we can
call del_mc_from_global_list() from edac_mc_add_mc() without forward
declarations.  Perhaps using forward declarations would be better?
I'm doing things this way because the rest of the code is missing
them.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Fix xxx_probe1() functions so they call xxx_get_error_info() functions
to clear initial errors.  This is simpler and cleaner than duplicating
the low-level code for accessing PCI config space.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Fix minor logic bug in e7xxx_remove_one().
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

- Fix i82875p_probe1() so it calls pci_get_device() instead of
  pci_find_device().
- Fix i82875p_probe1() so it cleans up properly on failure.
- Fix i82875p_init() so it cleans up properly on failure.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

- Fix i82860_init() so it cleans up properly on failure.
- Fix i82860_exit() so it cleans up properly.
- Fix typo in comment (i.e. www.redhat.com.com).
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

- Add ctl_dev field to "struct e752x_dev_info".  Then we can eliminate
  ugly switch statement from e752x_probe1().

- Remove code from e752x_probe1() that clears initial PCI bus parity
  errors.  The core EDAC module already does this.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Eliminate unnecessary calls to pci_dev_get() and pci_dev_put() from
amd76x driver.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Perform the following name substitutions on all source files:

    sed 's/BS_MOD_STR/EDAC_MOD_STR/g'
    sed 's/bs_thread_info/edac_thread_info/g'
    sed 's/bs_thread/edac_thread/g'
    sed 's/bs_xstr/edac_xstr/g'
    sed 's/bs_str/edac_str/g'

The names that start with BS_ or bs_ are artifacts of when the code
was called "bluesmoke".
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This implements the following idea:

On Monday 30 January 2006 19:22, Eric W. Biederman wrote:
> One piece missing from this conversation is the issue that we need errors
> in a uniform format.  That is why edac_mc has helper functions.
>
> However there will always be errors that don't fit any particular model.
> Could we add a edac_printk(dev, );  That is similar to dev_printk but
> prints out an EDAC header and the device on which the error was found?
> Letting the rest of the string be user specified.
>
> For actual control that interface may be to blunt, but at least for people
> looking in the logs it allows all of the errors to be detected and
> harvested.
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This patch was originally posted by Christoph Hellwig (see
http://lkml.org/lkml/2006/2/14/331):

"Christoph Hellwig" <hch@lst.de> wrote:
> Use the kthread_ API instead of opencoding lots of hairy code for kernel
> thread creation and teardown, including tasklist_lock abuse.
>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NDavid S. Peterson <dsp@llnl.gov>
Cc: <dave_peterson@pobox.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

This fixes coverity id #2.  the if (i==0) is pretty useless, since we
assing i=0, just the line before.
Signed-off-by: NEric Sesterhenn <snakebyte@gmx.de>
Cc: Karsten Keil <kkeil@suse.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

And: Tilman Schmidt <tilman@imap.cc>

This patch adds the connection-specific module "usb_gigaset", the hardware
driver for Gigaset base stations connected via the M105 USB DECT adapter.  It
contains the code for handling probe/disconnect, AT command/response
transmission, and call setup and termination, as well as handling asynchronous
data transfers, PPP framing, byte stuffing, and flow control.
Signed-off-by: NHansjoerg Lipp <hjlipp@web.de>
Signed-off-by: NTilman Schmidt <tilman@imap.cc>
Cc: Karsten Keil <kkeil@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

And: Tilman Schmidt <tilman@imap.cc>

This patch adds the payload data handler for the connection-specific module
"bas_gigaset".  It contains the code for handling isochronous data transfers,
HDLC framing and flow control.
Signed-off-by: NHansjoerg Lipp <hjlipp@web.de>
Signed-off-by: NTilman Schmidt <tilman@imap.cc>
Cc: Karsten Keil <kkeil@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

And: Tilman Schmidt <tilman@imap.cc>

This patch adds the main source file of the connection-specific module
"bas_gigaset", the hardware driver for Gigaset base stations connected
directly to the computer via USB.  It contains the code for handling
probe/disconnect, AT command/response transmission, and call setup and
termination.
Signed-off-by: NHansjoerg Lipp <hjlipp@web.de>
Signed-off-by: NTilman Schmidt <tilman@imap.cc>
Cc: Karsten Keil <kkeil@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

And: Tilman Schmidt <tilman@imap.cc>

This patch adds the procfs interface to the gigaset module.  The procfs
interface provides access to status information and statistics about the
Gigaset devices.  If the drivers are built with the debugging option it also
allows to change the amount of debugging output on the fly.
Signed-off-by: NHansjoerg Lipp <hjlipp@web.de>
Signed-off-by: NTilman Schmidt <tilman@imap.cc>
Cc: Karsten Keil <kkeil@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

And: Tilman Schmidt <tilman@imap.cc>

This patch adds the tty interface to the gigaset module.  The tty interface
provides direct access to the AT command set of the Gigaset devices.
Signed-off-by: NHansjoerg Lipp <hjlipp@web.de>
Signed-off-by: NTilman Schmidt <tilman@imap.cc>
Cc: Karsten Keil <kkeil@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

And: Tilman Schmidt <tilman@imap.cc>

This patch adds the isdn4linux subsystem interface to the gigaset module.  The
isdn4linux subsystem interface handles requests from and notifications to the
isdn4linux subsystem.
Signed-off-by: NHansjoerg Lipp <hjlipp@web.de>
Signed-off-by: NTilman Schmidt <tilman@imap.cc>
Cc: Karsten Keil <kkeil@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

And: Tilman Schmidt <tilman@imap.cc>

This patch adds the event layer to the gigaset module.  The event layer
serializes events from hardware, userspace, and other kernel subsystems.
Signed-off-by: NHansjoerg Lipp <hjlipp@web.de>
Signed-off-by: NTilman Schmidt <tilman@imap.cc>
Cc: Karsten Keil <kkeil@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

And: Tilman Schmidt <tilman@imap.cc>

This patch adds the common include file for the Siemens Gigaset drivers,
providing definitions used by all of the Gigaset ISDN driver source files.  It
also adds the main source file of the gigaset module which manages common
functions not specific to the type of connection to the device.
Signed-off-by: NHansjoerg Lipp <hjlipp@web.de>
Signed-off-by: NTilman Schmidt <tilman@imap.cc>
Cc: Karsten Keil <kkeil@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

And: Tilman Schmidt <tilman@imap.cc>

The following patches add drivers for the Siemens Gigaset 3070 family of ISDN
DECT PABXes connected via USB, either directly or over a DECT link using a
Gigaset M105 or compatible DECT data adapter.  The devices are integrated as
ISDN adapters within the isdn4linux framework, supporting incoming and
outgoing voice and data connections, and also as tty devices providing access
to device specific AT commands.

Supported devices include models 3070, 3075, 4170, 4175, SX205, SX255, and
SX353 from the Siemens Gigaset product family, as well as the technically
identical models 45isdn and 721X from the Deutsche Telekom Sinus series.
Supported DECT adapters are the Gigaset M105 data and the technically
identical Gigaset USB Adapter DECT, Sinus 45 data 2, and Sinus 721 data (but
not the Gigaset M34 and Sinus 702 data which advertise themselves as CDC-ACM
devices).

These drivers have been developed over the last four years within the
SourceForge project http://sourceforge.net/projects/gigaset307x/.  They are
being used successfully in several installations for dial-in Internet access
and for voice call switching with Asterisk.

This is our second attempt at submitting these drivers, taking into account
the comments we received to our first submission on 2005-12-11.

The patch set adds three kernel modules:

- a common module "gigaset" encapsulating the common logic for
  controlling the PABX and the interfaces to userspace and the
  isdn4linux subsystem.

- a connection-specific module "bas_gigaset" which handles
  communication with the PABX over a direct USB connection.

- a connection-specific module "usb_gigaset" which does the same
  for a DECT connection using the Gigaset M105 USB DECT adapter.

We also have a module "ser_gigaset" which supports the Gigaset M101 RS232 DECT
adapter, but we didn't judge it fit for inclusion in the kernel, as it does
direct programming of a i8250 serial port.  It should probably be rewritten as
a serial line discipline but so far we lack the neccessary knowledge about
writing a line discipline for that.

The drivers have been working with kernel releases 2.2 and 2.4 as well as 2.6,
and although we took efforts to remove the compatibility code for this
submission, it probably still shows in places.  Please make allowances.

This patch:

Prepare the kernel build infrastructure for addition of the Gigaset ISDN
drivers.  It creates a Makefile and Kconfig file for the Gigaset driver and
hooks them into those of the isdn4linux subsystem.  It also adds a MAINTAINERS
entry for the driver.

This patch depends on patches 2 to 9 of the present set, as without the actual
source files, activating the options added here will cause the kernel build to
fail.
Signed-off-by: NHansjoerg Lipp <hjlipp@web.de>
Signed-off-by: NTilman Schmidt <tilman@imap.cc>
Cc: Karsten Keil <kkeil@suse.de>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Provide proper kprobes fault handling, if a user-specified pre/post handlers
tries to access user address space, through copy_from_user(), get_user() etc.

The user-specified fault handler gets called only if the fault occurs while
executing user-specified handlers.  In such a case user-specified handler is
allowed to fix it first, later if the user-specifed fault handler does not fix
it, we try to fix it by calling fix_exception().

The user-specified handler will not be called if the fault happens when single
stepping the original instruction, instead we reset the current probe and
allow the system page fault handler to fix it up.

I could not test this patch for sparc64.
Signed-off-by: NPrasanna S Panchamukhi <prasanna@in.ibm.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Provide proper kprobes fault handling, if a user-specified pre/post handlers
tries to access user address space, through copy_from_user(), get_user() etc.

The user-specified fault handler gets called only if the fault occurs while
executing user-specified handlers.  In such a case user-specified handler is
allowed to fix it first, later if the user-specifed fault handler does not fix
it, we try to fix it by calling fix_exception().

The user-specified handler will not be called if the fault happens when single
stepping the original instruction, instead we reset the current probe and
allow the system page fault handler to fix it up.
Signed-off-by: NPrasanna S Panchamukhi <prasanna@in.ibm.com>
Acked-by: Anil S Keshavamurthy<anil.s.keshavamurthy@intel.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Provide proper kprobes fault handling, if a user-specified pre/post handlers
tries to access user address space, through copy_from_user(), get_user() etc.

The user-specified fault handler gets called only if the fault occurs while
executing user-specified handlers.  In such a case user-specified handler is
allowed to fix it first, later if the user-specifed fault handler does not fix
it, we try to fix it by calling fix_exception().

The user-specified handler will not be called if the fault happens when single
stepping the original instruction, instead we reset the current probe and
allow the system page fault handler to fix it up.
Signed-off-by: NPrasanna S Panchamukhi <prasanna@in.ibm.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Provide proper kprobes fault handling, if a user-specified pre/post handlers
tries to access user address space, through copy_from_user(), get_user() etc.

The user-specified fault handler gets called only if the fault occurs while
executing user-specified handlers.  In such a case user-specified handler is
allowed to fix it first, later if the user-specifed fault handler does not fix
it, we try to fix it by calling fix_exception().

The user-specified handler will not be called if the fault happens when single
stepping the original instruction, instead we reset the current probe and
allow the system page fault handler to fix it up.
Signed-off-by: NPrasanna S Panchamukhi <prasanna@in.ibm.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Provide proper kprobes fault handling, if a user-specified pre/post handlers
tries to access user address space, through copy_from_user(), get_user() etc.

The user-specified fault handler gets called only if the fault occurs while
executing user-specified handlers.  In such a case user-specified handler is
allowed to fix it first, later if the user-specifed fault handler does not fix
it, we try to fix it by calling fix_exception().

The user-specified handler will not be called if the fault happens when single
stepping the original instruction, instead we reset the current probe and
allow the system page fault handler to fix it up.
Signed-off-by: NPrasanna S Panchamukhi <prasanna@in.ibm.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

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: Nbibo mao <bibo.mao@intel.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: "Keshavamurthy, Anil S" <anil.s.keshavamurthy@intel.com>
Cc: <hiramatu@sdl.hitachi.co.jp>
Signed-off-by: NPrasanna S Panchamukhi <prasanna@in.ibm.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

When kretprobe probes the schedule() function, if the probed process exits
then schedule() will never return, so some kretprobe instances will never
be recycled.

In this patch the parent process will recycle retprobe instances of the
probed function and there will be no memory leak of kretprobe instances.
Signed-off-by: Nbibo mao <bibo.mao@intel.com>
Cc: Masami Hiramatsu <hiramatu@sdl.hitachi.co.jp>
Cc: Prasanna S Panchamukhi <prasanna@in.ibm.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

In normal operation, kretprobe makes a target function return to trampoline
code.  A kprobe (called trampoline_probe) has been inserted in the trampoline
code.  When the kernel hits this kprobe, it calls kretprobe's handler and it
returns to the original return address.

Kretprobe-booster removes the trampoline_probe.  It allows the trampoline code
to call kretprobe's handler directly instead of invoking kprobe.  The
trampoline code returns to the original return address.

(changelog from Chuck Ebbert <76306.1226@compuserve.com> - thanks ;))
Signed-off-by: NMasami Hiramatsu <hiramatu@sdl.hitachi.co.jp>
Cc: Prasanna S Panchamukhi <prasanna@in.ibm.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Chuck Ebbert <76306.1226@compuserve.com>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>

Current kprobe copies the original instruction at the probe point and replaces
it with a breakpoint instruction (int3).  When the kernel hits the probe
point, kprobe handler is invoked.  And the copied instruction is single-step
executed on the copied buffer (not on the original address) by kprobe.  After
that, the kprobe checks registers and modify it (if need) as if the
instructions was executed on the original address.

My proposal is based on the fact there are many instructions which do NOT
require the register modification after the single-step execution.  When the
copied instruction is a kind of them, kprobe just jumps back to the next
instruction after single-step execution.  If so, why don't we execute those
instructions directly?

With kprobe-booster patch, kprobes will execute a copied instruction directly
and (if need) jump back to original code.  This direct execution is executed
when the kprobe don't have both post_handler and break_handler, and the copied
instruction can be executed directly.

I sorted instructions which can be executed directly or not;

- Call instructions are NG(can not be executed directly).
  We should correct the return address pushed into top of stack.
- Indirect instructions except for absolute indirect-jumps
  are NG. Those instructions changes EIP randomly. We should
  check EIP and correct it.
- Instructions that change EIP beyond the range of the
  instruction buffer are NG.
- Instructions that change EIP to tail 5 bytes of the
  instruction buffer (it is the size of a jump instruction).
  We must write a jump instruction which backs to original
  kernel code in the instruction buffer.
- Break point instruction is NG. We should not touch EIP and
  pass to other handlers.
- Absolute direct/indirect jumps are OK.- Conditional Jumps are NG.
- Halt and software-interruptions are NG. Because it will stay on
  the instruction buffer of kprobes.
- Prefixes are NG.
- Unknown/reserved opcode is NG.
- Other 1 byte instructions are OK. But those instructions need a
  jump back code.
- 2 bytes instructions are mapped sparsely. So, in this release,
  this patch don't boost those instructions.

>From Intel's IA-32 opcode map described in IA-32 Intel Architecture Software
Developer's Manual Vol.2 B, I determined that following opcodes are not
boostable.

- 0FH (2byte escape)
- 70H - 7FH (Jump on condition)
- 9AH (Call) and 9CH (Pushf)
- C0H-C1H (Grp 2: includes reserved opcode)
- C6H-C7H (Grp11: includes reserved opcode)
- CCH-CEH (Software-interrupt)
- D0H-D3H (Grp2: includes reserved opcode)
- D6H (Reserved)
- D8H-DFH (Coprocessor)
- E0H-E3H (loop/conditional jump)
- E8H (Call)
- F0H-F3H (Prefixes and reserved)
- F4H (Halt)
- F6H-F7H (Grp3: includes reserved opcode)
- FEH-FFH(Grp4,5: includes reserved opcode)

Kprobe-booster checks whether target instruction can be boosted (can be
executed directly) at arch_copy_kprobe() function.  If the target instruction
can be boosted, it clears "boostable" flag.  If not, it sets "boostable" flag
-1.  This is disabled status.  In resume_execution() function, If "boostable"
flag is cleared, kprobe-booster measures the size of the target instruction
and sets "boostable" flag 1.

In kprobe_handler(), kprobe checks the "boostable" flag.  If the flag is 1, it
resets current kprobe and executes instruction buffer directly instead of
single stepping.

When unregistering a boosted kprobe, it calls synchronize_sched()
after "int3" is removed. So we can ensure followings after
the synchronize_sched() called.
- interrupt handlers are finished on all CPUs.
- instruction buffer is not executed on all CPUs.
And we can release the boosted kprobe safely.

And also, on preemptible kernel, the booster is not enabled where the kernel
preemption is enabled.  So, there are no preempted threads on the instruction
buffer.

The description of kretprobe-booster:
====================================

In the normal operation, kretprobe make a target function return to trampoline
code.  And a kprobe (called trampoline_probe) have been inserted at the
trampoline code.  When the kernel hits this kprobe, it calls kretprobe's
handler and it returns to original return address.

Kretprobe-booster patch removes the trampoline_probe.  It allows the
trampoline code to call kretprobe's handler directly instead of invoking
kprobe.  And tranpoline code returns to original return address.

This new trampoline code stores and restores registers, so the kretprobe
handler is still able to access those registers.

Current kprobe has about 1.3 usec/probe(*) overhead, and kprobe-booster patch
reduces it to 0.6 usec/probe(*).  Also current kretprobe has about 2.0
usec/probe(*) overhead.  Kprobe-booster patch reduces it to 1.3 usec/probe(*),
and the combination of both kprobe-booster patch and kretprobe-booster patch
reduces it to 0.9 usec/probe(*).

I expect the combination of both patches can reduce half of a probing
overhead.

Performance numbers strongly depend on the processor model.

Andrew Morton wrote:
> These preempt tricks look rather nasty.  Can you please describe what the
> problem is, precisely?  And how this code avoids it?  Perhaps we can find
> something cleaner.

The problem is how to remove the copied instructions of the
kprobe *safely* on the preemptable kernel (CONFIG_PREEMPT=y).

Kprobes basically executes the following actions;

(1)int3
(2)preempt_disable()
(3)kprobe_prehandler()
(4)copied instructioin(single step)
(5)kprobe_posthandler()
(6)preempt_enable()
(7)return to the original code

During the execution of copied instruction, preemption is
disabled (from step (2) to (6)).
When unregistering the probes, Kprobe waits for RCU
quiescent state by using synchronize_sched() after removing
int3 instruction.
Thus we can ensure the copied instruction is not executed.

On the other hand, kprobe-booster executes the following actions;

(1)int3
(2)preempt_disable()
(3)kprobe_prehandler()
(4)preempt_enable()             <-- this one is added by my patch
(5)copied instruction(direct execution)
(6)jmp back to the original code

The problem is that we have no way to prevent preemption on
step (5) or (6). We cannot call preempt_disable() after step (6),
because there are no rooms to do that. Thus, some other
processes may be preempted at step(5) or (6) on preemptable kernel.
And I couldn't find the easy way to ensure that other processes'
stack do *not* have the address of them. (I thought some way
to do that, but those are very costly.)

So currently, I simply boost the kprobe only when the probe
point is already preemption disabled.

> Also, the patch adds a preempt_enable() but I don't see a corresponding
> preempt_disable().  Am I missing something?

It is corresponding to the preempt_disable() in the top of
kprobe_handler().
I copied the code of kprobe_handler() here:

static int __kprobes kprobe_handler(struct pt_regs *regs)
{
        struct kprobe *p;
        int ret = 0;
        kprobe_opcode_t *addr = NULL;
        unsigned long *lp;
        struct kprobe_ctlblk *kcb;

        /*
         * We don't want to be preempted for the entire
         * duration of kprobe processing
         */
        preempt_disable();             <-- HERE
        kcb = get_kprobe_ctlblk();
Signed-off-by: NMasami Hiramatsu <hiramatu@sdl.hitachi.co.jp>
Cc: Prasanna S Panchamukhi <prasanna@in.ibm.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NLinus Torvalds <torvalds@osdl.org>