Commit Graph

73 Commits

Author SHA1 Message Date
Alexey Khoroshilov e6028db014 mei: fix mismatch in mutex unlock-lock in mei_amthif_read()
Users of mei_amthif_read() expect it leaves dev->device_lock held,
while there is a path where mei_amthif_read() unlocks device_lock
and returns -ERESTARTSYS.

The patch move code locking device_lock back before the return.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Cc: Sedat Dilek <sedat.dilek@gmail.com>
Acked-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-01-07 10:28:03 -08:00
Tomas Winkler d692170037 watchdog: mei: avoid oops in watchdog unregister code path
With commit c7d3df3 "mei: use internal watchdog device registration
tracking" will crash the kernel on shutdown path on systems
where ME watchdog is not present.
Since the watchdog was never initialized in such case
the WDOG_UNREGISTERED bit is never set and the system
crashes on access to uninitialized variables down the path.

To solve the issue we query for NULL on watchdog driver driver_data
to check whether the device is registered. This is handled in the
driver and doesn't depend on watchdog core internals.

Cc: Borislav Petkov <bp@alien8.de>
Cc: Wanlong Gao <gaowanlong@cn.fujitsu.com>
Signed-off-by: Jerry Snitselaar <jerry.snitselaar@oracle.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
2012-12-19 22:25:11 +01:00
Bill Pemberton 486a5c28c2 misc: remove use of __devexit
CONFIG_HOTPLUG is going away as an option so __devexit is no
longer needed.

Signed-off-by: Bill Pemberton <wfp5p@virginia.edu>
Cc: "Michał Mirosław" <mirq-linux@rere.qmqm.pl>
Cc: Wolfram Sang <w.sang@pengutronix.de>
Cc: Eric Piel <eric.piel@tremplin-utc.net>
Cc: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-21 12:53:32 -08:00
Bill Pemberton 80c8ae2892 misc: remove use of __devinit
CONFIG_HOTPLUG is going away as an option so __devinit is no longer
needed.

Signed-off-by: Bill Pemberton <wfp5p@virginia.edu>
Cc: "Michał Mirosław" <mirq-linux@rere.qmqm.pl>
Cc: Eric Piel <eric.piel@tremplin-utc.net>
Cc: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-21 12:51:53 -08:00
Bill Pemberton 9306a8b0c2 mei: remove use of __devexit_p
CONFIG_HOTPLUG is going away as an option so __devexit_p is no longer
needed.

Signed-off-by: Bill Pemberton <wfp5p@virginia.edu>
Cc: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-21 12:43:58 -08:00
Samuel Ortiz c1174c0edf mei: Simplify the ME client enumeration code
After enumerating all ME clients we call the client init functions for
all matching UUIDs from a separate context.
This remove the hackish cascading client initialisation process that was
interleaving properties and connection command replies.

Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-21 12:43:58 -08:00
Tomas Winkler be9d87a790 mei: simplify write complete loop in irq handler
extract the common, hence non conditional code
from the if-else statment

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-21 12:43:58 -08:00
Tomas Winkler 9a84d61698 mei: don't mix read and write slots
Do not pass read slots pointer into function
mei_irq_thread_write_handler, the write
slots management is handled internally in the write
handler

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-21 12:43:58 -08:00
Tomas Winkler 24c656e55f mei: streamline amthif write complete function
Rename the function mei_amthif_irq_process_completed
to mei_amthif_irq_write_complete
Remove cl from the parameter list as it
can be extracted from cb block.
Extract the common flow from if statements
and document the logic properly

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-21 12:43:58 -08:00
Tomas Winkler ea3b5fb710 mei: streamline write complete flow function
Rename the function  _mei_irq_thread_cmpl to
mei_irq_thread_write_complete to make clear it deals
with writing. Remove cl from the parameter list as it
can be extracted from cb block.
Extract the common flow from if statements and document
the logic properly

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-21 12:43:58 -08:00
Tomas Winkler 5fb54fb456 mei: use structured buffer for extra write buffer
The structure of the message is static so we don't have
to use and cast the buffer. We can also drop extra_write_index
variable as this information can be extracted directly
from the message header

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-21 12:43:57 -08:00
Tomas Winkler 5bd6471441 mei: compact code for mei bus message creation
1. replace boilerplate code for filling up the bus message header
 with a common wrapper function
2. shorten variable names and use temporal variables
 to save some screen space

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-21 12:43:57 -08:00
Tomas Winkler aeba4a06f2 mei: use the same bus msg for connect and disconnect request
structs hbm_client_connect_request and hbm_client_disconnect_request
have the same layout so we can drop the later

Add kdoc for the request and response structure so it is clear
they can be used for both purposes

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-15 15:34:19 -08:00
Tomas Winkler ff8b2f4e42 mei: use link and unlink terms for connecting ME and HOST client
1. rename mei_me_cl_update_filext to mei_me_cl_link
2. rename mei_remove_client_from_file_list to mei_me_cl_unlink

Code style, documenation, and usage of both function is updated

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-15 15:34:19 -08:00
Tomas Winkler 744f0f2f42 mei: extract amthif specific code from mei_poll to mei_amthif_poll
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-15 15:34:18 -08:00
Tomas Winkler a562d5c25a mei: move amthif specific release code to amithif
Move amthif code part into separate function
mei_amthif_release.
Also helper functions mei_clear_list and mei_clear_lists
are moved along

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-15 15:34:18 -08:00
Tomas Winkler 4b8960b492 mei: rename enum mei_cb_major_types to enum mei_cb_file_ops
1. Rename mei_cb_major_types to more understandable mei_cb_file_ops
2. Rename member struct mei_cl_cb of this type to simple 'fop_type'
3. Add kernel doc for the type

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-15 15:34:18 -08:00
Tomas Winkler db3ed43185 mei: use type struct mei_cl *cl instead of void in struct mei_cb
We can use correct type 'struct mei_cl' instead of
'void *' for file_private in the struct mei_cb
as there is no other type assigned to this member of the structure

We rename the member from file_private to cl

Remove about 10 lines of declarations of temporary variables
used for type casting

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-15 15:34:18 -08:00
Tomas Winkler e773efc405 mei: amthif: prefix cb list with amthif
amthif cb list were prefixed with amthi_ instead
if amthif.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-15 15:34:18 -08:00
Tomas Winkler ab5c4a56d4 mei: move amthif specific code from mei_write to mei_amthif_write
For sake of amthif consolidation move amthif specific code from
mei_write to mei_amthif_write

The original mei_amthif_write to mei_amthif_send_cmd
as this function deals with sending single command
while mei_amthif_write is interface function called from
the main driver which in turns calls mei_amthif_send_cmd

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-01 12:31:19 -07:00
Tomas Winkler c7d3df354d mei: use internal watchdog device registration tracking
remove bool wd_interface_reg as watchdog device already
keeps track of its registration

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-01 12:31:19 -07:00
Tomas Winkler 15ea19105b mei: mei_clear_list: kill file_temp
file_temp is used only once, so there is no any benefit of creating
a temporary variable

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-01 12:31:19 -07:00
Tomas Winkler 19838fb853 mei: extract AMTHI functions into the amthif.c file
Move AMT Host Interface functions into the new amthif.c file.
All functions has now common prefix: mei_amthif_

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-01 12:31:19 -07:00
Tomas Winkler 3870c3206b mei: normalize timeouts definitions
1. The hardware book defines timeouts in seconds
 so we stick to this and define the wrapper function
 mei_secs_to_jiffies  around  msecs_to_jiffies
 to use be used instead multiplying by HZ

2. We add name space prefix MEI_ to all timer defines

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-01 12:31:19 -07:00
Tomas Winkler b0d0cf77e7 mei: mei_write: revamp function flow
Use goto statement for error handling instead of deeper if constructs
and rename label 'unlock_dev' to more appropriate 'err'

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-11-01 12:31:17 -07:00
Tomas Winkler 664df38b3c mei: use mei_io_cb_ warppers also for control flows
move the mei_io_cb_ wrappers to to iorw.c for global use
and use them also for handling control flows

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-10-24 15:37:31 -07:00
Tomas Winkler 601a1efa63 mei: rename mei_free_cb_private to mei_io_cb_free
1. cb_private was an old name that we depriacated in earlier
cleanups

2. we also group the funcion declaration with other _io_
functions

3. Don't check cb for NULL as mei_io_cb_free is NULL safe

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-10-24 15:37:08 -07:00
Tomas Winkler fb601adb35 mei: kill usless struct mei_io_list
kill useless mei_io_list list wrapper and use directly
struct mei_cl_cb mei_cb which was its only member for managing io queues

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-10-24 15:36:39 -07:00
Tomas Winkler 75f0ee1559 mei: mei_write: revamp error path handling
1. unify common amt and regular error path and use it early in the
  function
2. fix indentation
3. propagate error code directly from copy_from_user
4. print out errors using dev_err instead of dev_dbg

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-10-24 15:35:03 -07:00
Tomas Winkler 33d28c9205 mei: add allocation and initialization wrappers for io callback
mei_io_cb_init - allocat and initializate mei_cl_cb
mei_io_cb_alloc_req/resp_buf are separate function as buffers
are not always needed

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-10-24 15:35:03 -07:00
Tomas Winkler ebb108ef93 mei: rename mei_cl_cb.information to mei_cl_cb.buf_idx
rename 'information' member of the struct mei_cl_cb to
more self-descriptive 'buf_idx'

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-10-24 15:35:03 -07:00
Greg Kroah-Hartman 65929215d8 char/misc: remove CONFIG_EXPERIMENTAL dependencies
As discussed at the kernel summit this year, CONFIG_EXPERIMENTAL means
nothing, so let's get rid of it.

Cc: Kees Cook <keescook@chromium.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Tomas Winkler <tomas.winkler@intel.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Cc: Paul Bolle <pebolle@tiscali.nl>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Anatolij Gustschin <agust@denx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-09-18 16:14:53 +01:00
Tomas Winkler a4136b49fb mei: don't print buffer as a string
non readable junk was printed to the logs
we will add proper buffer dumping mechanism later if needed

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-09-10 15:51:15 -07:00
Tomas Winkler f060939d7c mei: struct mei_message_data doesn't have to be packed
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-09-10 15:51:15 -07:00
Tomas Winkler 1b8129479a mei: add error messages for open count errors
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-09-10 15:51:15 -07:00
Tomas Winkler 9af514232e mei: add lynx point pci device ids
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-09-05 13:45:56 -07:00
Tomas Winkler 1e2776c3af mei: fix max number of open handles
There was internal confusion in wether bus message
clinet (0) is counted in or not

The bitmap me_clients_map that accomodate
was initialized w/o it (255) but later on it
the clinet 0 was reserved

Thus were able to open only 252 instead of 253 clients

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-09-05 13:45:56 -07:00
Tomas Winkler daed6b5e78 mei: rename struct pci_dev *mei_device to mei_pdev
1. rename mei_device variable to mei_pdev to remove
confusion with type 'struct mei_device'
2. mei_pdev no longer need to be gloabal so make it static
 and remove the declaration from the header file

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-09-05 13:45:56 -07:00
Tomas Winkler 09649a85ad mei: wd: use watchdog_set/get_drvdata for passing mei_device
use watchdog_set/get_drvdata for passing mei_device
to watchdog_ops handlers instead of using global mei_device

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-08-16 09:51:02 -07:00
Tomas Winkler c216fdeb2e mei: wd: decouple and revamp watchdog state machine
Before ME watchdog was exported through standard watchdog interface
it was closed and started together with the mei device.

The major issue is that closing ME watchdog disabled also MEI device,
to fix this the watchdog state machine has to be independent from MEI
state machine.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-08-16 09:51:02 -07:00
Tomas Winkler 248ffdf7c9 mei: wd: rename watchdog constants to be more descriptive
1. rename defines to more be descriptive
2. remove duplicated defines from interface.h
3. add common prefix MEI_

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-08-16 09:51:02 -07:00
Tomas Winkler c8df72920c mei: wd: add option WDIOF_SETTIMEOUT
According watchdog-kernel-api.txt WDIOF_SETTIMEOUT
should be set if the driver supplies set_timeout function

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-08-16 09:51:02 -07:00
Tomas Winkler 9bb3a5897e mei: fix device stall after wd is stopped
After watchdog was disabled the driver would stall
due to wrong calculation of credits reduction

The cat&paste bug was introduced in the commit
7bdf72d3d8
mei: introduce mei_data2slots wrapper

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-08-16 09:50:17 -07:00
Tomas Winkler b210d7506f mei: name space for mei device state
1. add MEI_DEV_ prefix for mei device state enums
2. rename mei_state to dev_state
3. add constant to string translation for debug purposes

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-08-16 09:43:58 -07:00
Tomas Winkler 6ddf3aea42 mei: style : reformat PCI device IDs
1. reformat PCI ids list in hw.h for better readability
2. update some code and brand names

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-08-16 09:43:58 -07:00
Tomas Winkler 068c0ae966 mei: use KBUILD_MODNAME when allocating resources from the OS
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-08-16 09:43:58 -07:00
Tomas Winkler 9a123f1983 mei: add mei_quirk_probe function
The main purpose of this function is to exclude ME devices
without support for MEI/HECI interface from binding

Currently affected systems are C600/X79 based servers
that expose PCI device even though it doesn't supported ME Interface.
MEI driver accessing such nonfunctional device can corrupt
the system.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-08-16 09:40:28 -07:00
Tomas Winkler 07b509b794 mei: revamp me client search function
me client search functions returns index
into me_client array according me client id
or me client uuid.

1. Add common prefix for the functions mei_me_cl_<>
2. create new function mei_me_cl_by_id that wraps open
 coded loops scattered over the code
3. rename mei_find_me_client_index to  mei_me_cl_by_uuid
4. rename mei_find_me_client_update_filext  to
 mei_me_cl_update_filext and updates its parameter names

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-08-16 09:40:28 -07:00
Tomas Winkler 6078188e2b mei: use module_pci_driver
use module_pci_driver() macro to wrap standard
pci module registration into a single line

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-19 15:33:30 -07:00
Greg Kroah-Hartman 5b1f6d81e4 Merge 3.5-rc7 into char-misc-next.
This lets us pick up the mei driver changes that we need in order to
handle future merge issues.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-18 06:22:45 -07:00