This patch provides OMAP4 Syslink release notes for
L24.0.9 release.
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>
Signed-off-by: Suman Anna <s-anna@ti.com>

This patch adds notify dependancies for Tesla
bridgedriver
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

This patch provides OMAP4 Syslink release notes for
L24.0.8 release.
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>
Signed-off-by: Suman Anna <s-anna@ti.com>

This patch fixes coding style in tesla driver
sources.
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

This patch fixes minor checkpatch warnings.
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

This patch fixes coding style and compilation
warnings.
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

Use kernel bit operation functions instead of
custom defined.
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

This patch cleans up tesla header files
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

Fixed typo in Makefile to build dspbridge
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

Intial Bridge changes for new tree These are the bridge changes for OMAP4 and Notify Support. Require cleanup.
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>

Provide build support for tesla in omap4 setup
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>

Port 3430 bridge files for tesla support. Also disables
PM.

This patch adds tesla driver to teslabridge tree
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>

Fixed the bug where the L4 peripheral address table was getting
corrupted after calling mmu module's function to update TLB entry.
this resulted in the consecutive runs failure as wrong addresses
gets programmed due to L4 peripheral table corruption.
Signed-off-by: C A Subramaniam <subramaniam.ca@ti.com>
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>

In sysmgr_ioctl, the execution of CMD_SYSMGR_STOPCALLBACK does not set the
return value in cargs.api_status.  This caused an error sometimes because the
status was uninitialized.  Now it sets the status to 0 just like for the cases
of CMD_SYSMGR_LOADCALLBACK and CMD_SYSMGR_STARTCALLBACK.  Note that since the
platform callbacks return void, we must set the api_status manually.
Signed-off-by: Simon Que <sque@ti.com>

In proc4430, ducati_destroy() is called from both proc4430_destroy() and
proc4430_stop().  Since ducati_setup() is called from proc4430_start(), the
call to ducati_destroy() in proc4430_destroy() is invalid.  It generates a
crash.
Signed-off-by: Simon Que <sque@ti.com>

This fixes the bug in hw mmu module where the
table entries that has memory segment information
was getting corrupted after first run. This bug
causes the consecutive runs to fail.
This patch also fixes the memory leak associated with
not calling iounmap and fixes a compile warning as one
of the function is unused.
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>

Failed statements are printed in the IPC module xxx_destroy() functions
only after checking if the status value indicates an error condition. This
is to avoid confusion by the existing trace statement word 'failed'. This
statement was being printed if the xxx_destroy jumps to exit label because
the module's reference count is not yet zero.
Signed-off-by: Suman Anna <s-anna@ti.com>

This patch modifies the memory addresses in sysmgr to be in
line with the new memory map changes required to use the
TLB entries for Ducati.
Signed-off-by: C A Subramaniam <subramaniam.ca@ti.com>
Signed-off-by: Suman Anna <s-anna@ti.com>

This fixes the 16MB TLB issue.

Signed-off-by: Ramesh Gupta G<grgupta@ti.com>
Signed-off-by: C A Subramaniam <subramaniam.ca@ti.com>

This patch programs all the required memory regions for Ducati
as TLBs to provide a sw solution for the hw bug in SDC. A total
of 49MB of physical memory is mapped, and the physical address is
assumed to begin at 0x88000000.
Signed-off-by: Suman Anna <s-anna@ti.com>

This patch de-couples SysManager calls from Proc module.
Added new SysManger calls that get invoked from user
space.
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>

This patch for the notify module fixes the following issues:

	- Random signal of exit event given to the user
	  thread in the read ioctl call.
	- Memory leak because of the absence of
	  iounmap() call.
Signed-off-by: C A Subramaniam <subramaniam.ca@ti.com>

Added the following changes
	- Removed array that was used to track ducati
	  driver objects.
	- Removed array introduced in syslink 2-0-0-12
	  migration to track notify driver.
	- Introduced a list implementation for tracking
	  notify objects.
	- Removed the index variable introduced in syslink
	  2-0-0-12 migration.
	- Re-organized notify_ducatidrv_create() function
	  call.
	- Added support for multiple creates to be called
	  with same driver name.
Signed-off-by: C A Subramaniam <subramaniam.ca@ti.com>

This patch adds the changes to Proc start and
Proc stop based on the validation on Zebu.
Proc start function is extended to enable M3
clock and program Ducati MMU. Proc Stop is extended
to put Cores in reset and disable M3 clock, and the
option to disable individual Core and MMU.
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>

platform_stop_callback now reads the config info for each module, just like
platform_start_callback does.  The problem was that it didn't read it before,
nor did it save the config info from platform_start_callback for each Ducati
core separately.  Thus only the AppM3 config was saved, and then used for SysM3
cleanup.
Signed-off-by: Simon Que <sque@ti.com>

This patch fixes the bug in notify drv read where
the read thread wouldn't get blocked if the event
list is empty. This bug resulted in board freeze
in Zebu enviornment because the read thread was
blockign all other threads from getting scheduled.

Signed-off-by: Hari Kanigeri<h-kanigeri2@ti.com>

Fix the casting compile warning in platform destroy.
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>

Added probe and remove functions, added platform_device_add,
platform_device_alloc calls and also added few dev_dbgs to
demo the usage of dev_dbg.
Signed-off-by: Imtiaz Ahmed <x0116819@ti.com>

This patch fixes the issues found in..
	<1> notify_ducati.c: we were not unmapping
		"driver_obj->ctrl_ptr". This is done using
		unmap_ducati_virt_mem() implemented now in
		ducatienabler.
	<2> ducatienabler.c: implemented unmap_ducati_virt_mem()
		to iounmap() the "shm_virt_addr" we get calling
		ioremap() in get_ducati_virt_mem().
	<3> ducatienabler.h: added unmap_ducati_virt_mem().
	<4> notify_ducatidriver.h: added unmap_ducati_virt_mem().
Signed-off-by: Jayan John <x00jayan@ti.com>

This patch fixes an issue found in notify_dispatcher.c where we
were not calling iounmap() on "mailbx_hw_config.mbox_linear_addr"
which we had obtained using ioremap() earlier in ntfy_disp_init().
Signed-off-by: Jayan John <x00jayan@ti.com>

This patch fixes an issue found in platform.c where we were not
unlocking "platform_mem_state.gate" in case of a ioremap() failure.
Signed-off-by: Jayan John <x00jayan@ti.com>

This patch fixes the issues found in..
	<1> platform.c:
		- calling platform_mem_unmap() on all addreses
                	on which we have called platform_mem_map().
		- calling proc_mgr_detach for SysM3 and AppM3.
	<2> -proc4430.c:
		- proc4430_detach() fixed to unmap all reqd.
			addresses.
		- cleanup
Signed-off-by: Jayan John <x00jayan@ti.com>

When a heapbuf object is created, there are two memory buffers allocated -- one
for the handle and one for the object itself.  That is in _heapbuf_create.  In
heapbuf_delete, both are freed.  But in heapbuf_close, only the handle is freed.
The object pointer is not freed and caused a memory leak.  This patch adds a
call to free the object.
Signed-off-by: Simon Que <sque@ti.com>

In the function notify_ducatidrv_unregister_event, it finds the event listener
and deletes the list element containing that listener, but did not free the
listener object itself, which was kmalloced in notify_ducatidrv_register_event.
The listener is now being freed.
Signed-off-by: Simon Que <sque@ti.com>

sysmgr_destroy was not cleaning up platform_mem.  As a result, memory allocated
by platform_mem_setup was not freed by platform_mem_destroy.  This has been
fixed.  Also, platform_mem_destroy was incorrectly freeing the allocated gate
handle.  It now frees it correctly.
Signed-off-by: Simon Que <sque@ti.com>