Take advantage of the explicit modulemode control to fix
the way parents clocks are managed.
A module must be disabled before any parents are disabled.
That programming model was not possible with the previous
implementation that was considering a modulemode as a leaf
clock node managed by the clock fmwk.
This was leading to bad crash upon disable when the parent
clock was gated before the module completed its transition
to idle.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The RSTCTRL register was accessed using an absolute address.
The usage of hardcoded macros to calculate virtual address from physical
one should be avoided as much as possible.
The usage of an offset will allow future improvement like migration from
the current architecture code toward a module driver.
Update prm_xxx accessors, move definition to the proper header file and
update copyrights.
Change the s16 register offset parameter to u16.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Rajendra Nayak <rnayak@ti.com>
[paul@pwsan.com: use '_prminst_' in function names that are part of the
prminst44xx.c file]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
It is mandatory to wait for a module to be in disabled state before
potentially disabling source clock or re-asserting a reset.
omap_hwmod_idle and omap_hwmod_shutdown does not wait for
the module to be fully idle.
Add a cm_xxx accessor to wait the clkctrl idle status to be disabled.
Fix hwmod_[idle|shutdown] to use this API.
Based on Rajendra's initial patch.
Please note that most interconnects hwmod will return one timeout because
it is impossible for them to be in idle since the processor is accessing
the registers though the interconnect.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Todd Poynor <toddpoynor@google.com>
[paul@pwsan.com: move cpu_is_*() tests to the top of _wait_target_disable();
incorporate some feedback from Todd]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The CLKCTRL register was accessed using an absolute address.
The usage of hardcoded macros to calculate virtual address from physical
one should be avoided as much as possible.
The usage of a offset will allow future improvement like migration from
the current architecture code toward a module driver.
Update cm_xxx accessor, move definition to the proper header file and
update copyrights.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Cc: Todd Poynor <toddpoynor@google.com>
[paul@pwsan.com: renamed 'omap4_cm_' fns to 'omap4_cminst_'; removed empty
fn prototype section from cm44xx.h; incorporated comments from Todd;
documented some functions]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
At boot time, lookup the clkdm_name to get the clkdm
structure pointer for further usage.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The sequence of _ocp_softreset doesn't work for i2c. The i2c module has a
special sequence to reset the module. The sequence is
- Disable the I2C.
- Write to SOFTRESET bit.
- Enable the I2C.
- Poll on the RESETDONE bit.
The sequence is implemented as a function and the i2c_class is updated with
the correct 'reset' pointer. omap_hwmod_softreset function is implemented
which triggers the softreset by writing into sysconfig register. On following
this sequence, i2c module resets properly and timeouts are not seen.
Cc: Rajendra Nayak <rnayak@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Benoit Cousson <b-cousson@ti.com>
Cc: Kevin Hilman <khilman@ti.com>
Signed-off-by: Avinash.H.M <avinashhm@ti.com>
[paul@pwsan.com: combined this patch with a patch to remove
HWMOD_INIT_NO_RESET from the 44xx hwmod flags; change register
offset conditional code to use the IP block revision; minor code
cleanup]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Previously, struct omap_hwmod_dma_info arrays were unterminated; and
users of these arrays used the ARRAY_SIZE() macro to determine the
length of the array. However, ARRAY_SIZE() only works when the array
is in the same scope as the macro user.
So far this hasn't been a problem. However, to reduce duplicated
data, a subsequent patch will move common data to a separate, shared
file. When this is done, ARRAY_SIZE() will no longer be usable.
This patch removes ARRAY_SIZE() usage for struct omap_hwmod_dma_info
arrays and uses a sentinel value (irq == -1) as the array terminator
instead.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Previously, struct omap_hwmod_mpu_irqs arrays were unterminated; and
users of these arrays used the ARRAY_SIZE() macro to determine the
length of the array. However, ARRAY_SIZE() only works when the array
is in the same scope as the macro user.
So far this hasn't been a problem. However, to reduce duplicated
data, a subsequent patch will move common data to a separate, shared
file. When this is done, ARRAY_SIZE() will no longer be usable.
This patch removes ARRAY_SIZE() usage for struct omap_hwmod_mpu_irqs
arrays and uses a sentinel value (irq == -1) as the array terminator
instead.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Previously, struct omap_hwmod_addr_space arrays were unterminated; and
users of these arrays used the ARRAY_SIZE() macro to determine the
length of the array. However, ARRAY_SIZE() only works when the array
is in the same scope as the macro user.
So far this hasn't been a problem. However, to reduce duplicated
data, a subsequent patch will move common data to a separate, shared
file. When this is done, ARRAY_SIZE() will no longer be usable.
This patch removes ARRAY_SIZE() usage for struct omap_hwmod_addr_space
arrays and uses a null structure member as the array terminator
instead.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Move the pr_debug at the top of the function
to trace the entry even if the first test is failing.
That help understanding that we entered the function
but failed in it.
Move the _enable last part out of the test to reduce
indentation and improve readability.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Change the debug into warning to check what IPs are failing.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The HW reset must be de-assert after the clocks are enabled
but before waiting for the target to be ready. Otherwise the
reset might not work properly since the clock is not running
to proceed the reset.
De-assert the reset after _enable_clocks and before
_wait_target_ready.
Re-assert it only when the clocks are disabled.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
It is perfectly valid for some hwmod to not have any
register target address for sysconfig. This is especially
true for interconnect hwmods.
Remove the warning.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The Type 2 type of IPs will not have any enawakeup bit in their sysconfig.
Writing to that bit will instead trigger a softreset.
Check the flag to write this bit only if the module supports it.
Reported-by: Miguel Vadillo <vadillo@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
When calling the shutdown, the module may be already in idle.
Accessing the sysconfig register will then lead to a crash.
In that case, re-enable the module in order to allow the access
to the sysconfig register.
Signed-off-by: Miguel Vadillo <vadillo@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The commit 86009eb326 was adding
the wakeup support for new OMAP4 IPs. This support is incomplete for
busmaster IPs that need as well to use smart-standby with wakeup.
This new standbymode is suported on HSI and USB_HOST_FS for the moment.
Add the new MSTANDBY_SMART_WKUP flag to mark the IPs that support this
capability.
Enable this new mode when applicable in _enable_wakeup, _disable_wakeup,
_enable_sysc and _idle_sysc.
The omap_hwmod_44xx_data.c will have to be updated to add this new flag.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Djamil Elaidi <d-elaidi@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Fix below compilation warnings.
arch/arm/mach-omap2/omap_hwmod.c: In function 'omap_hwmod_for_each':
arch/arm/mach-omap2/omap_hwmod.c:1631: warning: 'ret' may be used uninitialized in this function
arch/arm/mach-omap2/mux.c: In function 'omap_mux_get_gpio':
arch/arm/mach-omap2/mux.c:917: warning: 'm' may be used uninitialized in this function
Signed-off-by: Govindraj.R <govindraj.raja@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
This avoids going through the list unnecessarily when
idling devices for runtime PM.
Based on an earlier patch by sricharan <r.sricharan@ti.com>.
Signed-off-by: sricharan <r.sricharan@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Add a new clockdomain flag, CLKDM_NO_AUTODEPS, which, when marked on a
clockdomain, will prevent "autodeps" from being associated with the
clockdomain. ("Autodeps" are sleep dependencies and wakeup
dependencies from/to processor modules that are automatically added to
a clockdomain when it is in hardware-supervised idle mode. They are
deprecated -- a relic from the old CDP trees -- but are still in use
for OMAP3.)
Also, prevent the hwmod code from adding or removing initiator
dependencies for clockdomains with this flag set.
This patch should allow others to test which clockdomains actually
still need autodeps.
Thanks to Kevin Hilman <khilman@ti.com> for noting that the original
version should also modify the hwmod code.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@ti.com>
Create a new API that forms a wrapper to _set_module_autoidle()
to modify the AUTOIDLE bit.
This API is intended to be used by drivers that requires direct
manipulation of the AUTOIDLE bits in SYSCONFIG register.
McBSP driver requires autoidle bit to be enabled/disabled while
using sidetone feature.
Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Benoit Cousson <b-cousson@ti.com>
[paul@pwsan.com: restrict the hwmod states that the autoidle bit can be changed
in; changed function name; dropped "int" from "unsigned int long"]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Some boards can't tolerate IP blocks being reset when they are initialized.
Michael Büsch cites a case with the Nokia N810:
http://www.spinics.net/lists/linux-omap/msg47277.html
To allow such boards to continue working normally, allow board file
maintainers to mark IP blocks to prevent them from being reset upon
init. This is done via a hwmod function, omap_hwmod_no_setup_reset().
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Michael Buesch <mb@bu3sch.de>
On OMAP2 and OMAP3 the reset ctrl shift doesn't match the
status bit, as it does on OMAP4, when handling the reset lines.
This patch adds a new member in the reset info structure, so now it
can be added as part of hwmod data, and checked accordingly for
OMAP2 or 3; otherwise, there could be cases when the shift masks
doesn't match both of the registers, and a successful reset might
throw an error message or vice versa.
Signed-off-by: Omar Ramirez Luna <omar.ramirez@ti.com>
[paul@pwsan.com: added a warning if st_shift used on OMAP4; renamed 'r'
variable; improved some documentation]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
_init_clock always returns 0 and does
not propogate the error (in case of failure)
back to the caller, causing _init_clocks to
fail silently.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Acked-by: Benoît Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Autoidle is a single bit, TIOCP_CFG[0], setting on OMAP1/2/3/4 platforms.
In _set_module_autoidle() I am seeing 0x3 value where the mask is computed.
This should be 0x1.
v2:
(1) Modified the subject.
(2) Modified the description with further specific information.
Baseline:
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
Tested Info:
Boot tested on OMAP 1/2/3/4.
Signed-off-by: Tarun Kanti DebBarma <tarun.kanti@ti.com>
Acked-by: Rajendra Nayak <rnayak@ti.com>
Acked-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Add omap_hwmod_setup_one(), which is intended for use early in boot to
selectively setup the hwmods needed for system clocksources and
clockevents, and any other hwmod that is needed in early boot.
omap_hwmod_setup_all() can then be called later in the boot process.
The point is to minimize the amount of code that needs to be run
early.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Kevin Hilman <khilman@ti.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Previously, if a hwmod had already been set up, and the code attempted
to set up the hwmod again, an error would be returned. This is not
really useful behavior if we wish to allow the OMAP core code to setup
the hwmods needed for the Linux clocksources and clockevents before
the rest of the hwmods are setup. So, instead of generating errors,
just ignore the attempt to re-setup the hwmod.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Kevin Hilman <khilman@ti.com>
Move the code that looks for the MPU initiator hwmod to run during
the individual hwmod _register() function. (Previously, it ran after
all hwmods were registered in the omap_hwmod_late_init() function.)
This is done so code can late-initialize a few individual hwmods --
for example, for the system timer -- before the entire set of hwmods is
initialized later in boot via omap_hwmod_late_init().
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Kevin Hilman <khilman@ti.com>
Rename omap_hwmod_init() to omap_hwmod_register(). Rename
omap_hwmod_late_init() to omap_hwmod_setup_all(). Also change all of
the callers to reflect the new names. While here, update some
copyrights.
Suggested by Tony Lindgren <tony@atomide.com>.
N.B. The comment in mach-omap2/serial.c may no longer be correct, given
recent changes in init order.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Kevin Hilman <khilman@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
There's no longer any reason why we should prevent multiple
calls to omap_hwmod_init(). It is now simply used to register an
array of hwmods.
This should allow a subset of hwmods (e.g., hwmods
handling the system clocksource and clockevents) to be registered
earlier than the remaining mass of hwmods.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Kevin Hilman <khilman@ti.com>
Adds support for resource API to get address space info other than just MPU.
The drivers can now use platform_get_resource_byname() to get resource of
type 'IORESOURCE_MEM' by name.
Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Acked-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Allow hwmod state changes to mux pads based on the state changes.
By default, only enable and disable the pads. In some rare cases
dynamic remuxing for the idles states is needed, this can be done
by passing the enable, idle, and off pads from board-*.c file along
with OMAP_DEVICE_PAD_REMUX flag.
Thanks to Paul Walmsley <paul@booyaka.com> for the comments on the
hwmod related changes.
Signed-off-by: Tony Lindgren <tony@atomide.com>
Implement OMAP PM layer omap_pm_get_dev_context_loss_count() API by
creating similar APIs at the omap_device and omap_hwmod levels. The
omap_hwmod level call is the layer with access to the powerdomain
core, so it is the place where the powerdomain is queried to get the
context loss count.
The new APIs return an unsigned value that can wrap as the
context-loss count grows. However, the wrapping is not important as
the role of this function is to determine context loss by checking for
any difference in subsequent calls to this function.
Note that these APIs at each level can return zero when no context
loss is detected, or on errors. This is to avoid returning error
codes which could potentially be mistaken for large context loss
counters.
NOTE: only works for devices which have been converted to use
omap_device/omap_hwmod.
Longer term, we could possibly remove this API from the OMAP PM layer,
and instead directly use the omap_device level API.
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The new OMAP4 IPs introduced a new idle mode named smart-idle with wakeup.
This new idlemode replaces the enawakeup for the new IPs but seems to
coexist as well for some legacy IPs (UART, GPIO, MCSPI...)
Add the new SIDLE_SMART_WKUP flag to mark the IPs that support this
capability.
The omap_hwmod_44xx_data.c will have to be updated to add this new flag.
Enable this new mode when applicable in _enable_wakeup, _enable_sysc and
_idle_sysc.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Tested-by: Sebastien Guiriec <s-guiriec@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Rajendra Nayak <rnayak@ti.com>
In cases where a module (hwmod) does not become accesible on enabling
the main clocks (can happen if there are external clocks needed
for the module to become accesible), make sure the clocks are not
left enabled.
This ensures that when the requisite external dependencies are met
a omap_hwmod_enable and omap_hwmod_idle/shutdown would rightly enable
and disable clocks using clk framework. Leaving the clocks enabled in
the error case causes additional usecounting at the clock framework
level leaving the clock enabled forever.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
The hwmod list will be built are init time and never
be modified at runtime. There is no need anymore to protect
the list from concurrent accesses using a mutex.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
_register, _find_mpu_port_index and _find_mpu_rt_base are static APIs
that will be used only during the omap_hwmod initialization phase.
There is no need to keep them for runtime.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Do not allow omap_hwmod_register to be used outside the core
hwmod code. An omap_hwmod should be registered only at init time.
Remove the omap_hwmod_unregister that is not used today since the
hwmod list will be built once at init time and never be modified
at runtime.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
In the omap_hwmod core, most of the SYSCONFIG register helper
functions do not directly write the register, but instead just modify
a value passed in.
This patch converts the _enable_wakeup() and _disable_wakeup() helper
functions to take a value argument and only modify it instead of
actually writing the register. This makes the wakeup helpers
consistent with the other helper functions and avoids unintentional
problems like the following.
This problem was found after discovering that GPIO wakeups were no
longer functional. The root cause was that the ENAWAKEUP bit of the
SYSCONFIG register was being unintentionaly overwritten, leaving
wakeups disabled after the following two commits were combined:
commit: 9980ce53c9
OMAP: hwmod: Enable module wakeup if in smartidle
commit: 78f26e872f
OMAP: hwmod: Set autoidle after smartidle during _sysc_enable
There resulting in code in _enable_sysc() was this:
/*
* XXX The clock framework should handle this, by
* calling into this code. But this must wait until the
* clock structures are tagged with omap_hwmod entries
*/
if ((oh->flags & HWMOD_SET_DEFAULT_CLOCKACT) &&
(sf & SYSC_HAS_CLOCKACTIVITY))
_set_clockactivity(oh, oh->class->sysc->clockact, &v);
_write_sysconfig(v, oh);
so here, 'v' has wakeups disabled.
/* If slave is in SMARTIDLE, also enable wakeup */
if ((sf & SYSC_HAS_SIDLEMODE) && !(oh->flags & HWMOD_SWSUP_SIDLE))
_enable_wakeup(oh);
Here wakeup is enabled in the SYSCONFIG register (but 'v' is not updated)
/*
* Set the autoidle bit only after setting the smartidle bit
* Setting this will not have any impact on the other modules.
*/
if (sf & SYSC_HAS_AUTOIDLE) {
idlemode = (oh->flags & HWMOD_NO_OCP_AUTOIDLE) ?
0 : 1;
_set_module_autoidle(oh, idlemode, &v);
_write_sysconfig(v, oh);
}
And here, SYSCONFIG is updated again using 'v', which does not have
wakeups enabled, resulting in ENAWAKEUP being cleared.
Special thanks to Benoit Cousson for pointing out that wakeups were
supposed to be automatically enabled when a hwmod is enabled, and thus
helping target the root cause of this problem.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
The OMAP powerdomain code and data is all OMAP2+-specific. This seems
unlikely to change any time soon. Move plat-omap/include/plat/powerdomain.h
to mach-omap2/powerdomain.h. The primary point of doing this is to remove
the temptation for unrelated upper-layer code to access powerdomain code
and data directly.
As part of this process, remove the references to powerdomain data
from the GPIO "driver" and the OMAP PM no-op layer, both in plat-omap.
Change the DSPBridge code to point to the new location for the
powerdomain headers. The DSPBridge code should not be including the
powerdomain headers; these should be removed.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Omar Ramirez Luna <omar.ramirez@ti.com>
Cc: Felipe Contreras <felipe.contreras@gmail.com>
Cc: Greg Kroah-Hartman <greg@kroah.com>
The OMAP clockdomain code and data is all OMAP2+-specific. This seems
unlikely to change any time soon. Move plat-omap/include/plat/clockdomain.h
to mach-omap2/clockdomain.h. The primary point of doing this is to remove
the temptation for unrelated upper-layer code to access clockdomain code
and data directly.
DSPBridge also uses the clockdomain headers for some reason, so,
modify it also. The DSPBridge code should not be including the
clockdomain headers; these should be removed.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Omar Ramirez Luna <omar.ramirez@ti.com>
Cc: Felipe Contreras <felipe.contreras@gmail.com>
Cc: Greg Kroah-Hartman <greg@kroah.com>
Tested-by: Rajendra Nayak <rnayak@ti.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
In preparation for adding OMAP4-specific PRCM accessor/mutator
functions, split the existing OMAP2/3 PRCM code into OMAP2/3-specific
files. Most of what was in mach-omap2/{cm,prm}.{c,h} has now been
moved into mach-omap2/{cm,prm}2xxx_3xxx.{c,h}, since it was
OMAP2xxx/3xxx-specific.
This process also requires the #includes in each of these files to be
changed to reference the new file name. As part of doing so, add some
comments into plat-omap/sram.c and plat-omap/mcbsp.c, which use
"sideways includes", to indicate that these users of the PRM/CM includes
should not be doing so.
Thanks to Felipe Contreras <felipe.contreras@gmail.com> for comments on this
patch.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Jarkko Nikula <jhnikula@gmail.com>
Cc: Peter Ujfalusi <peter.ujfalusi@nokia.com>
Cc: Liam Girdwood <lrg@slimlogic.co.uk>
Cc: Omar Ramirez Luna <omar.ramirez@ti.com>
Acked-by: Omar Ramirez Luna <omar.ramirez@ti.com>
Cc: Felipe Contreras <felipe.contreras@gmail.com>
Acked-by: Felipe Contreras <felipe.contreras@gmail.com>
Cc: Greg Kroah-Hartman <greg@kroah.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Reviewed-by: Kevin Hilman <khilman@deeprootsystems.com>
Tested-by: Kevin Hilman <khilman@deeprootsystems.com>
Tested-by: Rajendra Nayak <rnayak@ti.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Split the existing cm44xx.h file into cm1_44xx.h and cm2_44xx.h files
so they match their underlying OMAP hardware modules. Add clockdomain
offset information.
Add header files for the MPU local PRCM, prcm_mpu44xx.h, and for the
SCRM, scrm44xx.h. SCRM register offsets still need to be added; TI
should do this.
Move the "_MOD" macros out of the prcm-common.h header file, into the
header file of the hardware module that they belong to. For example,
OMAP4430_PRM_*_MOD macros have been moved into the prm44xx.h header.
Adjust #includes of all files that used the old PRCM header file names
to point to the new filenames.
The autogeneration scripts have been updated accordingly.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Reviewed-by: Kevin Hilman <khilman@deeprootsystems.com>
Tested-by: Kevin Hilman <khilman@deeprootsystems.com>
Tested-by: Rajendra Nayak <rnayak@ti.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Do not skip the sysc programming in the hmwod framework based
on the cached value alone, since at times the module might have lost
context (due to the Powerdomain in which the module belongs
transitions to either Open Switch RET or OFF).
Identifying if a module has lost context requires atleast one
register read, and since a register read has more latency than
a write, it makes sense to do a blind write always.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Acked-by: Kevin Hilman <khilman@deeprootsystems.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoit Cousson <b-cousson@ti.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Change the per-hwmod mutex to a spinlock. (The per-hwmod lock
serializes most post-initialization hwmod operations such as enable,
idle, and shutdown.) Spinlocks are needed, because in some cases,
hwmods must be enabled from timer interrupt disabled-context, such as
an ISR. The current use-case that is driving this is the OMAP GPIO
block ISR: it can trigger interrupts even with its clocks disabled,
but these clocks are needed for register accesses in the ISR to succeed.
This patch also effectively reverts commit
848240223c - this patch makes
_omap_hwmod_enable() and _omap_hwmod_init() static, renames them back
to _enable() and _idle(), and changes their callers to call the
spinlocking versions. Previously, since omap_hwmod_{enable,init}()
attempted to take mutexes, these functions could not be called while
the timer interrupt was disabled; but now that the functions use
spinlocks and save and restore the IRQ state, it is appropriate to
call them directly.
Kevin Hilman <khilman@deeprootsystems.com> originally proposed this
patch - thanks Kevin.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Benoît Cousson <b-cousson@ti.com>
The standard omap_hwmod.c _reset() code relies on an IP block's
OCP_SYSCONFIG.SOFTRESET register bit to reset the IP block. This
works for most IP blocks on the chip, but unfortunately not all. For
example, initiator-only IP blocks often don't have any MPU-accessible
OCP-header registers, and therefore the MPU can't write to any
OCP_SYSCONFIG registers in that block. Other IP blocks, such as the
IVA and I2C, require a specialized reset sequence.
Since we need to be able to reset these IP blocks as well, allow
custom IP block reset functions to be passed into the hwmod code via a
per-hwmod-class reset function pointer, struct omap_hwmod_class.reset.
If .reset is non-null, then the hwmod _reset() code will call the custom
function instead of the standard OCP SOFTRESET-based code.
As part of this change, rename most of the existing _reset() function
code to _ocp_softreset(), to indicate more clearly that it does not work
for all cases.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Paul Hunt <hunt@ti.com>
Cc: Stanley Liu <stanley_liu@ti.com>
Allow board files and OMAP core code to control the state that some or
all of the hwmods end up in at the end of _setup() (called by
omap_hwmod_late_init() ). Reimplement the old skip_setup_idle code in
terms of this new postsetup state code.
There are two use-cases for this patch: the !CONFIG_PM_RUNTIME case,
in which all IP blocks should stay enabled after _setup() finishes;
and the MPU watchdog case, in which the watchdog IP block should enter
idle if watchdog coverage of kernel initialization is desired, and
should be disabled otherwise.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Charulatha Varadarajan <charu@ti.com>
Some OMAP IP blocks, such as the watchdog timers, cannot be completely
shut down via the standard hwmod shutdown mechanism. This patch
enables the hwmod data files to supply a pointer to a custom
pre-shutdown function via the struct omap_hwmod_class.pre_shutdown
function pointer. If the struct omap_hwmod_class.pre_shutdown
function pointer is non-null, the function will be executed before the
existing hwmod shutdown code runs.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Some modules which have 16bit registers can cause imprecise
aborts if a __raw_readl/writel is used to read/write 32 bits.
Add an additional flag to identify modules which have such
hard requirement, and handle it in the hwmod framework.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Acked-by: Paul Walmsley <paul@pwsan.com>
Tested-by: Kevin Hilman <khilman@deeprootsystems.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
OMAP USBOTG module has a requirement to set the autoidle bit only after
setting smartidle bit. Modified the _sys_enable api to set the smartidle
first and then the autoidle bit. Setting this will not have any impact on the
other modules.
Signed-off-by: Hema HK <hemahk@ti.com>
Signed-off-by: Partha Basak <p-basak2@ti.com>
Cc: Felipe Balbi <balbi@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Acked-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
This patch:
- adds more documentation to the hwmod code
- fixes some documentation typos elsewhere in the file
- changes the _sysc_*() function names to appear in (verb, noun) order,
to match the rest of the function names.
This patch should not result in any functional change.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Benoît Cousson <b-cousson@ti.com>
If a module's OCP slave port is programmed to be in smartidle,
its also necessary that they have module level wakeup enabled.
Update _sysc_enable in hwmod framework to do this.
The thread "[PATCH 7/8] : Hwmod api changes" archived here:
http://www.mail-archive.com/linux-omap@vger.kernel.org/msg34212.html
has additional technical information on the rationale of this patch.
Sergei Shtylyov <sshtylyov@mvista.com> identified an indentation
problem with this patch - thanks, Sergei.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Partha Basak <p-basak2@ti.com>
Signed-off-by: Benoît Cousson <b-cousson@ti.com>
[paul@pwsan.com: revised patch description]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Sergei Shtylyov <sshtylyov@mvista.com>
Some modules (like GPIO, DSS...) require optionals clock to be enabled
in order to complete the sofreset properly.
Add a HWMOD_CONTROL_OPT_CLKS_IN_RESET flag to force all optional clocks
to be enabled before reset. Disabled them once the reset is done.
TODO:
For the moment it is very hard to understand from the HW spec, which
optional clock is needed and which one is not. So the current approach
will enable all the optional clocks.
Paul proposed a much finer approach that will allow to tag only the needed
clock in the optional clock table. This might be doable as soon as we have
a clear understanding of these dependencies.
Reported-by: Partha Basak <p-basak2@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
In OMAP3 a specific SYSSTATUS register was used to get the softreset status.
Starting in OMAP4, some IPs does not have SYSSTATUS register and instead
use the SYSC softreset bit to provide the status.
Other cases might exist:
- Some IPs like McBSP does have a softreset control but no reset status.
- Some IPs that represent subsystem, like the DSS, can contains
a reset status without softreset control. The status is the aggregation
of all the sub modules reset status.
- Add a new flag (SYSC_HAS_RESET_STATUS) to identify the new programming model
and replace the previous SYSS_MISSING, that was used to flag IP with
softreset control but without the SYSSTATUS register, with a specific
SYSS_HAS_RESET_STATUS flag.
- MCSPI and MMC contains both programming models, so the legacy one
will be prevented by removing the syss offset field that become useless.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Expose an hardreset API from hwmod in order to assert / deassert all the
individual reset lines that belong to an hwmod. This API is needed by
some of the more complicated processor drivers, e.g., DSP/Bridge,
Syslink, etc.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Force the softreset of every IPs during the _setup phase.
IPs that cannot support softreset or that should not
be reset must set the HWMOD_INIT_NO_RESET flag in the
hwmod struct.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Tested-by: Kevin Hilman <khilman@deeprootsystems.com>
Most processor IPs does have a hardreset signal controlled by the PRM.
This is different of the softreset used for local IP reset from the
SYSCONFIG register.
The granularity can be much finer than orginal HWMOD, for ex, the IVA
hwmod contains 3 reset lines, the IPU 3 as well, the DSP 2...
Since this granularity is needed by the driver, we have to ensure
than one hwmod exist for each hardreset line.
- Store reset lines as hwmod resources that a driver can query by name like
an irq or sdma line.
- Add two functions for asserting / deasserting reset lines in hwmods
processor that require manual reset control.
- Add one functions to get the current reset state.
- If an hwmod contains only one line, an automatic assertion / de-assertion
is done.
-> de-assert the hardreset line only during enable from disable transition
-> assert the hardreset line only during shutdown
Note: The hwmods with hardreset line and HWMOD_INIT_NO_RESET flag must be
kept in INITIALIZED state.
They can be properly enabled only if the hardreset line is de-asserted
before.
For information here is the list of IPs with HW reset control
on an OMAP4430 device:
RM_DSP_RSTCTRL
1,1,'RST2','RW','1','DSP - MMU, cache and slave interface reset control'
0,0,'RST1','RW','1','DSP - DSP reset control'
RM_IVA_RSTCTRL
2,2,'RST3','RW','1','IVA logic and SL2 reset control'
1,1,'RST2','RW','1','IVA Sequencer2 reset control'
0,0,'RST1','RW','1','IVA sequencer1 reset control'
RM_IPU_RSTCTRL
2,2,'RST3','RW','1','IPU MMU and CACHE interface reset control.'
1,1,'RST2','RW','1','IPU Cortex M3 CPU2 reset control.'
0,0,'RST1','RW','1','IPU Cortex M3 CPU1 reset control.'
PRM_RSTCTRL
1,1,'RST_GLOBAL_COLD_SW','RW','0','Global COLD software reset control.'
0,0,'RST_GLOBAL_WARM_SW','RW','0','Global WARM software reset control.'
RM_CPU0_CPU0_RSTCTRL
RM_CPU1_CPU1_RSTCTRL
0,0,'RST','RW','0','Cortex A9 CPU0&1 warm local reset control'
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
[paul@pwsan.com: made the hardreset functions static; moved the register
twiddling into prm*.c functions in previous patches; changed the
function names to conform with hwmod practice]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Currently omap_hwmod_mutex is being used to protect both the list
access/modification and concurrent access to hwmod functions. This
patch separates these two types of locking.
First, omap_hwmod_mutex is used only to protect access and
modification of omap_hwmod_list. Also cleaned up some comments
referring to this mutex that are no longer needed.
Then, for protecting concurrent access to hwmod functions, use a
per-hwmod mutex. This protects concurrent access to a single hwmod,
but would allow concurrent access to different hwmods.
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
[paul@pwsan.com: added structure documentation; changed mutex variable
name]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The reset function wrongly used the state flag as a bit mask and was trying
to re-enable after a reset.
hwmod is still enabled for the PRCM point of view after a softreset
so there is no need to re-enable.
Remove the state check from omap_hwmod_reset since the _reset
function is checking that as well and in addition can generate
a warning
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
[b-cousson@ti.com: remove the wrong test, remove the re-enable]
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Tested-by: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Rajendra Nayak <rnayak@ti.com>
The disable function was disabling clocks and dependencies
from both enable and idle state. Since idle function is already
disabling both, an enable -> idle -> disable sequence will
try to disable twice the clocks and thus generate a
"Trying disable clock XXX with 0 usecount" warning.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Tested-by: Kevin Hilman <khilman@deeprootsystems.com>
The dma request line attribute was named dma channel, which leads
to confusion with the real dma channel definition.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Tested-by: Kevin Hilman <khilman@deeprootsystems.com>
Update some minor documentation issues and update copyright for
omap_device/omap_hwmod code.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Add omap_device_get_mpu_rt_va(). This is intended to be used by
device drivers (currently, via a struct platform_data function
pointer) to retrieve their corresponding device's virtual base address
that the MPU should use to access the device. This is needed because
the omap_hwmod code does its own ioremap(), in order to gain access to
the module's OCP_SYSCONFIG register.
Add omap_hwmod_get_mpu_rt_va(). omap_device_get_mpu_rt_va() calls this
function to do the real work.
While here, rename struct omap_hwmod._rt_va to struct
omap_hwmod._mpu_rt_va, to reinforce that it refers to the MPU's
register target virtual address base (as opposed to, for example, the
L3's).
In the future, this belongs as a function in an omap_bus, so it is not
necessary to call this through a platform_data function pointer.
The use-case for this function was originally presented by Santosh
Shilimkar <santosh.shilimkar@ti.com>.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
On kernels that don't use the omap_device_enable() calls to enable
devices, leave all on-chip devices enabled in hwmod _setup().
Otherwise, accesses to those devices are likely to fail, crashing the
system. It's expected that kernels built without CONFIG_PM_RUNTIME
will be the primary use-case for this. This functionality is
controlled by adding an extra parameter to omap_hwmod_late_init().
This patch is based on the patch "OMAP: hwmod: don't auto-disable
hwmod when !CONFIG_PM_RUNTIME" by Kevin Hilman
<khilman@deeprootsystems.com>.
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Some hwmods may need to be idled/enabled in atomic context, so
non-locking versions of these functions are required.
Most users should not need these and usage of theses should be
controlled to understand why access is being done in atomic context.
For this reason, the non-locking functions are only exposed at the
hwmod level and not at the omap-device level.
The use-case that led to the need for the non-locking versions is
hwmods that are enabled/idled from within the core idle/suspend path.
Since interrupts are already disabled here, the mutex-based locking in
hwmod can sleep and will cause potential deadlocks.
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
As reported by Sergei, a couple of braces were missing after
the WARN removal patch.
[07/22] OMAP: hwmod: Replace WARN by pr_warning if clock lookup failed
https://patchwork.kernel.org/patch/100756/
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
[paul@pwsan.com: fixed patch description per Anand's E-mail]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Sergei Shtylyov <sshtylyov@mvista.com>
Cc: Anand Gadiyar <gadiyar@ti.com>
Add some missing credits for people who have contributed significant features
or fixes.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Tero Kristo <tero.kristo@nokia.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Thara Gopinath <thara@ti.com>
Most of the clock nodes belong to a clock domain, but it is perfectly valid
to have clock without clock domain.
Root clocks for example does not belong to any clock domain.
Keep the warning but reduce the verbosity.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
In the lastest OMAP4 hwmod data file, the _hwmod was removed
in order to save some memory space and because it does not
bring a lot.
The same cleanup will be have to done for other hwmods in
OMAP2 & 3 data files.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Rajendra Nayak <rnayak@ti.com>
During the _init_clocks phase, the iteration is stopped but the
status is still change from _HWMOD_STATE_REGISTERED to
_HWMOD_STATE_CLKS_INITED.
Since the _setup phase will be done nevertheless, it might be
better to keep initializing the others clocks nodes and just
keep the warning.
It is much easier to debug when a important number of clocks
name are wrong during the early debug phase of a new platform.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The WARN is a little bit too verbose and is not providing
usefull information in that case.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The previous clock API was returning a standard linux error code in
case of failure. This is not the case anymore with the new
omap_clk_get_by_name API. A NULL value means that the clock node
does not exist.
Replace all the IS_ERR check by a !clk check.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The iteration is currently done on the omap_hwmod_ocp_if pointer
and not on the table pointer that reference them.
It worked most of the time because the structure are contiguous in
memory.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Some initiator modules in OMAP2 & 3 does not have IDLEST bit,
in that case we cannot detect the module readiness by
polling that bit and must exist the function immediately
assuming that the module is ready.
The previous flag was affected to the OCP interface. While it is
technically true that the idlest is related to the L4 slave
interface of the module, the PRCM status belong to the module.
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The return of the omap4_cm_wait_module_ready function is checked
in order to avoid accessing the sysconfig register if the module is
not in the correct state.
In that case the _setup will exit without trying to reset
using sysconfig.
For the moment a warning is printed. A proper management of fclk
and module reset will have to be done in order to init correctly
the problematic IPs listed below.
<4>omap_hwmod: ivahd: cannot be enabled (3)
<4>omap_hwmod: iss: cannot be enabled (3)
<4>omap_hwmod: tesla: cannot be enabled (3)
<4>omap_hwmod: sdma: cannot be enabled (3)
<4>omap_hwmod: sl2: cannot be enabled (3)
<4>omap_hwmod: sad2d: cannot be enabled (3)
<4>omap_hwmod: ducati: cannot be enabled (3)
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
These files include linux/bootmem.h without using anything from this
file; remove the unnecessary include.
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
This patch adds check for presence of clockdomain structure in the API
omap_hwmod_get_pwrdm before trying to access the powerdomain structure.
This will prevent unnecessary crashing of the system in case of a
clock node with out an associated clockdomain.
Signed-off-by: Thara Gopinath <thara@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Add support for categorizing and iterating over hardware IP blocks by
the "class" of the IP block. The class is the type of the IP block:
e.g., "timer", "timer1ms", etc. Move the OCP_SYSCONFIG/SYSSTATUS data
from the struct omap_hwmod into the struct omap_hwmod_class, since
it's expected to stay consistent for each class. While here, fix some
comments.
The hwmod_class structures in this patch were designed and proposed by
Benoît Cousson <b-cousson@ti.com> and were refined in a discussion
between Thara Gopinath <thara@ti.com>, Kevin Hilman
<khilman@deeprootsystems.com>, and myself.
This patch uses WARN() lines that are longer than 80 characters, as
Kevin noted a broader lkml consensus to increase greppability by
keeping the messages all on one line.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Benoît Cousson <b-cousson@ti.com>
Cc: Thara Gopinath <thara@ti.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
The OMAP hwmod core code is intended to use SoC IP block description
structures that are autogenerated from TI's OMAP hardware database.
Currently the hwmod code uses clkdev device + connection addressing to
identify clocks. This causes problems in the hwmod autogeneration
process, since the TI hardware database doesn't use platform_device or
clkdev addressing; it uses a single clock signal name string, which
tends to bear some resemblance to what is used in the OMAP TRMs. This
patch converts the hwmod code and existing data to use omap_clk_get_by_name(),
introduced in the previous patch.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
In OMAP3 Some modules like Smartreflex do not have the regular sysconfig
register.Instead clockactivity bits are part of another register at a
different bit position than the usual bit positions 8 and 9.
In OMAP4, a new scheme is available due to the new protocol
between the PRCM and the IPs. Depending of the scheme, the SYSCONFIG
bitfields position will be different.
The IP_REVISION register should be at offset 0x00.
It should contain a SCHEME field. From this we can determine whether
the IP follows legacy scheme or the new scheme.
31:30 SCHEME Used to distinguish between old scheme and current.
Read 0x0: Legacy protocol.
Read 0x1: New PRCM protocol defined for new OMAP4 IPs
For legacy IP
13:12 MIDLEMODE
11:8 CLOCKACTIVITY
6 EMUSOFT
5 EMUFREE
4:3 SIDLEMODE
2 ENAWAKEUP
1 SOFTRESET
0 AUTOIDLE
For new OMAP4 IP's, the bit position in SYSCONFIG is (for simple target):
5:4 STANDBYMODE (Ex MIDLEMODE)
3:2 IDLEMODE (Ex SIDLEMODE)
1 FREEEMU (Ex EMUFREE)
0 SOFTRESET
Unfortunately In OMAP4 also some IPs will not follow any of these
two schemes. This is the case at least for McASP, SmartReflex
and some security IPs.
This patch introduces a new field sysc_fields in omap_hwmod_sysconfig which
can be used by the hwmod structures to specify the offsets for the
sysconfig register of the IP.Also two static structures
omap_hwmod_sysc_type1 and omap_hwmod_sysc_type2 are defined
which can be used directly to populate the sysc_fields if the IP follows
legacy or new OMAP4 scheme. If the IP follows none of these two schemes
a new omap_hwmod_sysc_fields structure has to be defined and
passed as part of omap_hwmod_sysconfig.
Signed-off-by: Thara Gopinath <thara@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Some HW blocks have errata which requires specific slave idle mode
under certain conditions.
This patch adds an hwmod API to allow setting slave idlemode
ensuring that any SYSCONFIG register updates go through hwmod.
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Move clockdomain wakeup dependency and sleep dependency data
structures from the powerdomain layer to the clockdomain layer, where
they belong. These dependencies were originally placed in the
powerdomain layer due to unclear documentation; however, it is clear
now that these dependencies are between clockdomains. For OMAP2/3,
this is not such a big problem, but for OMAP4 this needs to be fixed.
Thanks to Benoît Cousson <b-cousson@ti.com> for his advice on this
patch.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
In the current implementation the sysconfig value is read into
_sysc_cache once and an actual update to the sysconfig register
happens only if the new value paased is differnt from the one in _sysc_cache.
_sysc_cache is updated only if _HWMOD_SYSCONFIG_LOADED is not set.
This can lead to the follwing issue if off mode is enabled in modules
which employs "always-retore" mechanism of context save and restore.
a. The module sets the sysconfig register through omap_device_enable.
Here _sysc_cache is updated with the value written to the sysconfig
register and left.
b. The power domain containig the module enters off mode and the
module context is lost.
c. The module in use becomes active and calls omap_device_enable to
enable itself. Here a read of sysconfig register does not happen
as _HWMOD_SYSCONFIG_LOADED flag is set. The value to be written
to the sysconfig register will be same as the one written in step a.
Since _sysc_cache reflects the previous written value an update
of the sysconfig register does not happen.
This means in modules which employs "always-restore" mechanism
after off , the sysconfig regsiters will never get updated.
This patch introduces a flag SYSC_NO_CACHE which if set ensures that the
sysconfig register is always read into _sysc_cache before an update is
attempted.
This flags need to be set only by modules which does not do a context save
but re-initializes the registers every time the module is accessed. This
includes modules like i2c, smartreflex etc.
Signed-off-by: Thara Gopinath <thara@ti.com>
[paul@pwsan.com: tweaked to apply on a different head, added flag comment]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
WARN if a clock/hwmod is missing a clockdomain association since
resulting hwmod will not be able to correctly enable/disable clocks.
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Earlier, the hwmod code had considered the OCP_SYSCONFIG.CLOCKACTIVITY
bits to be incremental power saving bits, controlling internal IP
block clock gates. This was a misapprehension. The CLOCKACTIVITY
bits are used to indicate, in advance, which clocks will be cut when
the module acknowledges an idle request. This enables the IP block to
take whatever action is necessary to complete any in-progress work
before asserting its IdleAck.
In the current Linux-OMAP code, this implies that the clock framework
should be changing module CLOCKACTIVITY bits as module clocks are enabled
and disabled. We don't do that yet, but in the future, we should.
This must wait until the clock tree is annotated with omap_hwmod pointers
(or vice-versa). In the meantime, drop most of the hwmod code that
controls CLOCKACTIVITY bits to avoid confusion.
This patch has benefited from many illuminating discussions with (in
alphabetical order) Benoît Cousson <b-cousson@ti.com>, Rajendra Nayak
<rnayak@ti.com>, and Sebastien Sabatier <s-sabatier1@ti.com>.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Cc: Sebastien Sabatier <s-sabatier1@ti.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Replace the existing u8 array of module MPU IRQ lines with a struct
that includes a name - similar to the existing struct
omap_hwmod_dma_info. Device drivers can then use
platform_get_resource_byname() to retrieve specific IRQs without nasty
dependencies on array ordering.
Thanks to Benoît Cousson <b-cousson@ti.com> and Kevin Hilman
<khilman@deeprootsystems.com> for feedback on this approach.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
This patch fills in the OCP_SYSCONFIG.AUTOIDLE handling in the OMAP
hwmod code.
After this patch, the hwmod code will set the module AUTOIDLE bit
(generally <module>.OCP_SYSCONFIG.AUTOIDLE) to 1 by default upon
enable. If the hwmod flag HWMOD_NO_OCP_AUTOIDLE is set, AUTOIDLE will
be set to 0 upon enable. Upon module disable, AUTOIDLE will be set to
1.
Enabling module autoidle should save some power. The only reason to
not set the OCP_SYSCONFIG.AUTOIDLE bit is if there is a bug in the
module RTL, e.g., the MPUINTC block on OMAP3.
Comments from Kevin Hilman <khilman@deeprootsystems.com> inspired this patch,
and Kevin tested an earlier version of this patch.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Tested-by: Kevin Hilman <khilman@deeprootsystems.com>
Reprogram the module's OCP_SYSCONFIG register after module reset (SOFTRESET
= 1). This may not be needed, but the definition of the reset performed by
the SOFTRESET bit is unclear.
Kevin Hilman <khilman@deeprootsystems.com> tested an earlier version of
this patch.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Tested-by: Kevin Hilman <khilman@deeprootsystems.com>
Fix loop bailout off-by-one bugs reported by Juha Leppänen
<juha_motorsportcom@luukku.com>.
This second version incorporates comments from Russell King
<linux@arm.linux.org.uk>. A new macro, 'omap_test_timeout', has
been created, with cleaner code, and existing code has been converted
to use it.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Juha Leppänen <juha_motorsportcom@luukku.com>
Cc: Russell King <linux@arm.linux.org.uk>
Move the remaining headers under plat-omap/include/mach
to plat-omap/include/plat. Also search and replace the
files using these headers to include using the right path.
This was done with:
#!/bin/bash
mach_dir_old="arch/arm/plat-omap/include/mach"
plat_dir_new="arch/arm/plat-omap/include/plat"
headers=$(cd $mach_dir_old && ls *.h)
omap_dirs="arch/arm/*omap*/ \
drivers/video/omap \
sound/soc/omap"
other_files="drivers/leds/leds-ams-delta.c \
drivers/mfd/menelaus.c \
drivers/mfd/twl4030-core.c \
drivers/mtd/nand/ams-delta.c"
for header in $headers; do
old="#include <mach\/$header"
new="#include <plat\/$header"
for dir in $omap_dirs; do
find $dir -type f -name \*.[chS] | \
xargs sed -i "s/$old/$new/"
done
find drivers/ -type f -name \*omap*.[chS] | \
xargs sed -i "s/$old/$new/"
for file in $other_files; do
sed -i "s/$old/$new/" $file
done
done
for header in $(ls $mach_dir_old/*.h); do
git mv $header $plat_dir_new/
done
Signed-off-by: Tony Lindgren <tony@atomide.com>
Connect the omap_hwmod code to the kernel boot. Create some basic
interconnect and device structures for OMAP2/3 chips.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
OMAP SoCs can be considered a collection of hardware IP blocks
connected by various interconnects. The bus topology and device
integration data is somewhat more complex than platform_device can
encode. This patch creates code and structures to manage information
about OMAP on-chip devices ("hardware modules") and their integration
to the rest of the chip. Hardware module data is intended to be
generated dynamically from the TI hardware database for the OMAP4
chips and beyond, easing Linux support for new chip variants.
This code currently:
- resets and configures all hardware modules upon startup, reducing bootloader
dependencies;
- provides hooks for Linux driver model code to enable, idle, and shutdown
hardware modules (forthcoming patch);
- waits for hardware modules to leave idle once their clocks
are enabled and OCP_SYSCONFIG bits are set appropriately.
- provides a means to pass arbitrary IP block configuration data (e.g.,
FIFO size) to the device driver (via the dev_attr void pointer)
In the future this code is intended to:
- estimate interconnect bandwidth and latency characteristics to
ensure constraints are satisfied during DVFS
- provide *GRPSEL bit data to the powerdomain code
- handle pin/ball muxing for devices
- generate IO mapping information dynamically
- supply device firewall configuration data
- provide hardware module data to other on-chip coprocessor software
- allow the removal of the "disable unused clocks" code in the OMAP2/3
clock code
This patch represents a collaborative effort involving many people from TI,
Nokia, and the Linux-OMAP community.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoit Cousson <b-cousson@ti.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Cc: Vikram Pandita <vikram.pandita@ti.com>
Cc: Sakari Poussa <sakari.poussa@nokia.com>
Cc: Anand Sawant <sawant@ti.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Eric Thomas <ethomas@ti.com>
Cc: Richard Woodruff <r-woodruff2@ti.com>
Benoit Cousson