Age | Commit message (Collapse) | Author |
|
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
HW_PERF_EVENTS is dependent on !ARCH_OMAP3 unfortunately...
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
Re-apply the following patch. It was partially reverted by a merge
with mutex addition.
This fix is required because, as expected, the concerned code generates
an error with gcc4.6 (but not with gcc4.5!).
OMAP2+: mailbox: fix lookups for multiple mailboxes
The pointer math in omap_mbox_get() is not quite right, and leads to
passing NULL to strcmp() when searching for an mbox that is not first
in the list.
Convert to using array indexing as is done in all the other functions
which walk the mbox list.
Tested on OMAP2420/n810, OMAP3630/zoom3, OMAP4430/Blaze
Signed-off-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Conflicts:
arch/arm/plat-omap/mailbox.c
Signed-off-by: Sebastien Jan <s-jan@ti.com>
|
|
Signed-off-by: Sebastien Jan <s-jan@ti.com>
|
|
ARM errata 727915 for PL310 has been updated to include a new
workaround required for PL310 r2p0 for l2x0_flush_all, which also
affects l2x0_clean_all in my testing. For r2p0, clean or flush
each set/way individually. For r3p0 or greater, use the debug
register for cleaning and flushing.
Requires exporting the cache_id, sets and ways detected in the
init function for later use.
Change-Id: I215055cbe5dc7e4e8184fb2befc4aff672ef0a12
Signed-off-by: Colin Cross <ccross@android.com>
|
|
Ducati reset procedure fails in HW managed (HW_AUTO) mode.
The failure trace is:
omap_hwmod: ipu_c0: failed to hardreset
Signed-off-by: Sebastien Jan <s-jan@ti.com>
|
|
The return status for the userspace ProcMgr_start and ProcMgr_stop
API is not returned properly. This causes the user-space to ignore
and proceed with the IPC synchronization sequence even if the
processor has not been put into a proper device state (asserting
or deasserting Reset).
Change-Id: Icfa9fbba3e7753809a3934b961afe15016ebfd70
Signed-off-by: Suman Anna <s-anna@ti.com>
|
|
Signed-off-by: Sebastien Jan <s-jan@ti.com>
|
|
Upstream hwspinlocks implementation conflicts with legacy one.
For now, deactivate the upstream version and use the legacy one.
Signed-off-by: Sebastien Jan <s-jan@ti.com>
|
|
This reverts commit 9a84fc90be096d69e376a31f1c8cf15fde9b9299.
|
|
Add support for reading the MAC address from the system registers if there
is an EEPROM present. This involves caching the KS_CCR register for later
use (will also be useful for ETHTOOL support) and adding a print to say
that there is an EEPROM present.
Signed-off-by: Abraham Arce <x0066660@ti.com>
|
|
Signed-off-by: Andy Green <andy.green@linaro.org>
|
|
The ipu device is missing from the omap_ipupm_data. This
results in a null device pointer being used when applying
constraints. This change provides an api which can be used
to return a device when queried by name. This is to support
initializing omap_ipupm_data with a device.
Change-Id: I490e61db6361652aa0fd63ba51e98af937eb800e
Signed-off-by: Bryan Honza <honza@ti.com>
|
|
For each mailbox interrupt,process both cores(sysm3 and appm3) ipc stack.
This is to fix the issue of message being dropped, when both cores tries
to put the message at the same time. For example both sysm3 and appm3 are
trying to write into MBX and appm3 sees that there is already a message
pending by SYSM3.In this case appm3 will not interrupt A9.But A9 will only
process sysm3 ipc stack.So to take care of this,process both cores ipc stack
for each MBX interrupt.
Change-Id: I209db12d54204818f6c4b24aab162263bbd53d3b
Signed-off-by: Subin K G <subin.kg@ti.com>
|
|
Currenlty ipc_attach is whiling inside the kernel till
it become success.This has been changed to a retry for
500ms, after that it will report failure to userspace.
This change is to avoid rebooting of the board, in case
if ipc_attach fails.
Change-Id: Ie47043c54858140e198af7354b9441232a117760
Signed-off-by: Subin K G <subin.kg@ti.com>
|
|
rproc_get_state was returning omap_device status instead of
the current rproc status. This mismatch was leading to some
unhandled states like when the remote proc is hibernated. This
was causing users-space registration for PROC_START events to
timeout, due to the mismatch states.
Change-Id: I90601d7d69be73f0ca3031dbe0d1254d545c8bf4
Signed-off-by: Juan Gutierrez <jgutierrez@ti.com>
|
|
The is_cached parameter is not initialized prior to calling
platform_mem_unmap from proc4430_detach. This change sets the
is_cached parameter to false to match the value used when mapped
so that the memory entries are now unmapped consistently.
Change-Id: I7690d317093cd82c13c9894e8cbc0fb2432dff02
Signed-off-by: Bryan Honza <honza@ti.com>
|
|
atomic_cmpmask_and_set() is operating on an uninitialized attach_count
which can pass the cmpmask as if it has been initialized
resulting in the cmpmask_and_set() not initializing the attach_count.
For instance, if the uninitialized attach_count happens to be
0xffffffff, atomic_cmpmask_and_set() will not change the attach_count.
This attach_count is allocated on the heap.
Change-Id: Iaadc3f093287d965ff4d00f8d3f9c63ce596a6d4
Signed-off-by: Bryan Honza <honza@ti.com>
|
|
If we disable SYSLINK_RECOVERY option in drivers->syslink
->SYSLINK_RECOVERY, build breaks with below errors
drivers/built-in.o: In function `devh_notification_handler':
drivers/dsp/syslink/devh/44xx/devh44xx.c:103: undefined reference to
`ipc_recover_schedule'
drivers/built-in.o: In function `devh44xx_wdt_ipc_notifier_call':
drivers/dsp/syslink/devh/44xx/devh44xx.c:244: undefined reference to
`ipc_recover_schedule'
make: *** [.tmp_vmlinux1] Error 1
Fixed by moving ipc_revover_schedule function under SYSLINK_RECOVERY Flag.
Change-Id: Idb4de38bbe1e2586eebac9f76150616ea93d56f3
Signed-off-by: Shivananda Hebbar <x0hebbar@ti.com>
|
|
Store the pid of process who call the iommu open in it's private
data so at the time of release it can be used to send death notific-
ation instead of using current pid who calls the release.
Change-Id: I7008a593b19fe1cef16f82874ffd9ad1e17910a6
Signed-off-by: Subin K G <subin.kg@ti.com>
Signed-off-by: Fernando Guzman Lugo <fernando.lugo@ti.com>
|
|
Store the pid of process who call the ipc_open in it's private
data so at the time of release it can use this to send death
notification instead of using current pid whenever ipc release
has been called.
Change-Id: I86429112bd9668869ceeae43195309be2311ab5d
Signed-off-by: Subin K G <subin.kg@ti.com>
Signed-off-by: Fernando Guzman Lugo <fernando.lugo@ti.com>
|
|
Instead of using the current pid use the pid passed to
the devh.This is the required since the process who opens
the driver handle (iommu, tiler, ipc),may not the one
releasing the handle.And in turn devh will be giving the
wrong pid to ipu pm.
Change-Id: I153ef85cbc0b5d671b82cfa5f07bbb1e99c39531
Signed-off-by: Subin K G <subin.kg@ti.com>
|
|
There were parts not protected when register/unregister single
events.
Change-Id: I53d99659b91abbbcef66117bca7c37bf73e7aa97
Signed-off-by: Fernando Guzman Lugo <fernando.lugo@ti.com>
|
|
Add protection againts adding/removing notify drivers_handles.
Change-Id: I53f391d97b9b115ac64d007cc0a0d8ef7cc80a3b
Signed-off-by: Fernando Guzman Lugo <fernando.lugo@ti.com>
|
|
if you don't unregister the callback for notify events first, the
transportshm object can be dereferenced in the callback causing
a kerner panic.
Change-Id: I0f55e2b5070bec7c617392b3080914c0372f98bf
Signed-off-by: Fernando Guzman Lugo <fernando.lugo@ti.com>
|
|
A constraint for CORE domain is requested by the IOMMU driver
before releasing the Ducati's RST3, and is released when the
IOMMU is being disabled.
This ensures that the CORE domain does not go into Retention
after releasing RST3, thereby eliminating any context loss in
the Ducati Unicache registers due to CORE domain OSWR transitions.
For runtime PM, the constraint is relaxed after the IOMMU driver
has saved its context when Ducati is either self-hibernating or
being suspended, and the constraint is requested again before
restoring the context for ducati resume.
Change-Id: Ib4042226d2529be141d69917bdb211ee6f30bd22
Signed-off-by: Suman Anna <s-anna@ti.com>
Signed-off-by: Miguel Vadillo <vadillo@ti.com>
Signed-off-by: Juan Gutierrez <jgutierrez@ti.com>
Signed-off-by: Paul Hunt <hunt@ti.com>
|
|
Signed-off-by: Sebastien Jan <s-jan@ti.com>
|
|
Signed-off-by: Sebastien Jan <s-jan@ti.com>
|
|
Signed-off-by: Sebastien Jan <s-jan@ti.com>
|
|
Revert "ARM: dma-mapping: remove dmac_clean_range and dmac_inv_range"
This reverts commit 702b94bff3c50542a6e4ab9a4f4cef093262fe65
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>
|
|
This patch adds vanilla l-o 2.6.38 specific changes. The patch might get
redundant once the OMAP_PM specific changes/ baseport changes go in.
Check the return value of kfifo_out ti avoid a warning
Do not use any constraint req/rel since the fwk is not
ready yet.
Signed-off-by: Miguel Vadillo <vadillo@ti.com>
Signed-off-by: Subramaniam C.A <subramaniam.ca@ti.com>
Signed-off-by: Cris Jansson <cjansson@ti.com>
|
|
After an iommufault iss and fdif need to be reset
for the proper reuse of both.
The proper api to call is _softreset() but for that
sysc struct in hwmod needs to be enabled but that part
is not ready.
This is hack until the proper fix is done in sysc and
the proper api can be used to softreset.
Change-Id: I7568308bcd49697946a08a1b03b6a846dd0c9f5d
Signed-off-by: Miguel Vadillo <vadillo@ti.com>
Signed-off-by: Juan Gutierrez <jgutierrez@ti.com>
|
|
This patch adds hwmod definitions for all devices that are used/
requested by Syslink/ipu-pm.
The devices are:
--sl2if iss ipu mailbox hwspinlock fdif
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Miguel Vadillo <vadillo@ti.com>
Signed-off-by: Simon Que <sque@ti.com>
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
|
|
This patch uses unlocked_ioctl instead of ioctl in all SysLink drivers.
The support for ioctl was removed since BKL is not to be used anymore.
Signed-off-by: Subramaniam C.A <subramaniam.ca@ti.com>
|
|
This patch uses unlocked_ioctl instead of ioctl. The support for
ioctl was removed since BKL is not to be used anymore
Signed-off-by: Subramaniam C.A <subramaniam.ca@ti.com>
|
|
This patch enables hibernation by default and also adds build support
for iommu-debug.The patch is a port of these additions from 2.6.35 omap
android kernel.
Signed-off-by: Miguel Vadillo <vadillo@ti.com>
Signed-off-by: Leed Aguilar <leed.aguilar@ti.com>
Signed-off-by: C A Subramaniam <subramaniam.ca@ti.com>
|
|
This patch adds support to build the remoteproc, ipudev and
hwspinlock code for OMAP4. This was absent in the linux-omap
2.6.38 code.
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>
Signed-off-by: Suman Anna <s-anna@ti.com>
Signed-off-by: Ramesh Gupta G <grgupta@ti.com>
Signed-off-by: Simon Que <sque@ti.com>
Signed-off-by: Paul Hunt <hunt@ti.com>
Signed-off-by: Subramaniam C.A <subramaniam.ca@ti.com>
|
|
After recovery the mechanism to release all the resources
and constraints was not taking in to consideration the IPU
that is created behind the scenes to handle the sys and app
constraints and as a consequence all the constraints attached
to it were not released leading to not hit retention in idle path.
That IPU behind scenes is a special rcb slot that only has constraints;
the fix is taking care of this case and releases only the constraints.
If the IPU is explicitly requested then it will be consider as a normal
resource.
Change-Id: I49a7ab05912bcc4de4e61c223ba0f4dd4c7a2d83
Signed-off-by: Miguel Vadillo <vadillo@ti.com>
Signed-off-by: Juan Gutierrez <jgutierrez@ti.com>
Signed-off-by: Paul Hunt <hunt@ti.com>
|
|
The unregister mmu fault ioctl was wrongly calling the register_mmu_fault
function instead of unregister_mmu_fault function.
Change-Id: I85acae8eebadff4e75927a4454fce5d61cbe216e
Signed-off-by: Hari Kanigeri <h-kanigeri2@ti.com>
|
|
Initilaize the local err to 0 in map_iovm_area
Change-Id: I070562c89a662c95fe501d89ddea2e1af42a11e2
Signed-off-by: Axel Haslam <axelhaslam@ti.com>
|
|
In iommu_save_tlb_entries and iommu_restore_tlb_entries moved initialization
of the io table entry to after the check for the obj validity.
Change-Id: I54ba196327422269f7ee64d00db707b7fedfda73
Signed-off-by: Axel Haslam <axelhaslam@ti.com>
|
|
In the function user_to_device_map
-Initialized the local w to 0
-Check the vma pointer and return if NULL
-In error condition verify that gen_pool is valid before
freeing the memory
Change-Id: I7c3fa08dcb561564cc3e45bdc2b20d404bdbb514
Signed-off-by: Axel Haslam <axelhaslam@ti.com>
|