Timer ticks aren't properly serviced while a CPU is in LP2 idle.
Although the Tegra LP2 idle code calls hrtimer_peek_ahead_timers,
because no IRQ regs have been saved, update_process_times is not
called, and thus the timer list is not serviced (and neither is
SMP rebalancing, etc.) This can cause significant delays
scheduling timer-based activity, especially on CPU 1 (which is
not servicing most other IRQs).
Colin Cross suggested a patch based on upstream review feedback
that uses clock notifiers to switch to the "broadcast" clock event
source ("timer0" Tegra timer 3) during LP2, which has a real
interrupt handler defined that calls the clock event handler in
IRQ context, allowing timers to be checked.
Change-Id: Ifa3f4ec662f07dc9636e433f278358f75b65d10c
Signed-off-by: Todd Poynor <toddpoynor@google.com>
The Tegra2 USB controller doesn't properly deal with misaligned DMA
buffers, causing corruption. This is especially prevalent with USB
network adapters, where skbuff alignment is often in the middle of a
4-byte dword.
To avoid this, allocate a temporary buffer for the DMA if the provided
buffer isn't sufficiently aligned.
Signed-off-by: Robert Morell <rmorell@nvidia.com>
Provide optional hooks for the host controller driver to override the
default DMA mapping and unmapping routines. In general, these shouldn't
be necessary unless the host controller has special DMA requirements,
such as alignment contraints. If these are not specified, the
general usb_hcd_(un)?map_urb_for_dma functions will be used instead.
Also, pass the status to unmap_urb_for_dma so it can know whether the
DMA buffer has been overwritten.
Finally, add a flag to be used by these implementations if they
allocated a temporary buffer so it can be freed properly when unmapping.
Signed-off-by: Robert Morell <rmorell@nvidia.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The convention is to prefix symbols exported from the USB HCD core with
"usb_hcd". This change makes unmap_urb_setup_for_dma() and
unmap_urb_for_dma() consistent with that.
Signed-off-by: Robert Morell <rmorell@nvidia.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Looks for ieee registration numver 0x000c03 as per HDMI spec.
Change-Id: I6875b24c66e8754510edabcb4f9ba682a50d6ac1
Signed-off-by: Erik Gilling <konkers@android.com>
Fixed a problem preventing independent setup/teardown
of TX and RX DMAs when setting playback or capture buffer count.
Signed-off-by: Eric Laurent <elaurent@google.com>
instead of fused value. This is required
to meet High Speed USB signaling requirements.
Change-Id: I659b33faa950605ecf040598112e1972047ae7ad
Signed-off-by: Nathan Connell <w14185@motorola.com>
Fix not re-enabling watchdog resume if it was enabled in probe
Add clearing watchdog interrupt in probe
Remove tegra_wdt_set_timeout
Change-Id: I8fdbb6da3eda64a85a73ed85ab979a5ee0261c37
Signed-off-by: Dmitriy Gruzman <dmitriy.gruzman@motorola.com>
while USB is active to eliminate all USB buffer underruns.
Change-Id: I9977224601e715e950284708958be98d37b3e6b1
Signed-off-by: Nathan Connell <w14185@motorola.com>
while USB is active to eliminate all USB buffer underruns.
Change-Id: I7ad8ebfa2d802f91b81839e14fde906da1a0569e
Signed-off-by: Nathan Connell <w14185@motorola.com>
-Add a module param to enable/disable carveout killer
-Fix race condition in code to wait for something to free memory
after firing carveout killer
-Fix the check for current so we always compare task->group_leaders
Change-Id: Ie030978827dce6b0fbbfa1db0d80e4abe59eaa51
Signed-off-by: Rebecca Schultz Zavin <rebecca@android.com>
This fixes a regression where hubs cannot detect new devices once they
have been auto-suspended.
Change-Id: I4b3efcaa9634b9a912060e438527000bbc83dc32
Signed-off-by: Benoit Goby <benoit@android.com>
It is seen that auto-power-gate of MPE causes some issues when long run
camera stress test is run. The issue manifest itself in the form of
MPE sync-point not getting triggered from hardware and hence the caller
who is waiting on it hangs up the system.
Disabling auto-power-gate for MPE seems to resolve this issue.
Original work of Jay Cheng <jacheng@nvidia.com>
Signed-off-by: Mayuresh Kulkarni <mkulkarni@nvidia.com>
Change-Id: Iaef9f959cbc00dd715352ca637955cd2bea6f355
-Modify the carveout killer to only kill tasks with lower priorities
than the one that's trying to allocate
-After delivering a sigkill to a task, wait for something to exit and
cleanup before retrying the allocation
Change-Id: If62b6ed008a73fc3c347ff26735a83eee284909e
Signed-off-by: Rebecca Schultz Zavin <rebecca@android.com>
The hardware aes engine is required for key management for
HDCP, but since it's performance is bad, we don't want it
enabled for dm-crypt. This hack changes the cra_name field
so dm-crypt won't find a match, but leaves the cra_driver_name
field unchanged to the device file interface will still find it.
Signed-off-by: Ken Sumrall <ksumrall@android.com>
- enable/disable clocks only once per request
- create a thread for handling the driver's work queue
- always set vde clocks to the max before processing a request
Change-Id: I935e5523e9e913c93705cc694f8a475d212c15ce
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>
No need to maintain a reference to the task struct if the client
is a kernel thread. In this case just set the task to NULL.
Change-Id: Ica4785388932f6b298eeb0da04b78b0e1cdc3a44
Signed-off-by: Rebecca Schultz Zavin <rebecca@android.com>
avp does not use the crypto engine inside the vde block,
so it is ok if we do not reset the entire block before
any crypto operation. this would save more than 100us per crypto
operation.
Change-Id: I4e6f4940c5b6e95498b6f7a62498fbf956e25254
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>
- Acquire the DMA request spinlock in dma_tx_complete_callback()
(this was omitted)
- allow_suspend() was called by mistake in start_recording_if_necessary()
every time it was executed which could cause the wake lock to be
released during audio capture.
Signed-off-by: Eric Laurent <elaurent@google.com>
Signed-off-by: Iliyan Malchev <malchev@google.com>
When picking clock divider values, the clock framework picks
the closest frequency that is lower than the requested
frequency. If the value from a clock divider rounds down,
and then the new rounded down frequency is requested, it
will get rounded down again, resulting in a frequency two
steps lower than the original requested frequency.
Fix the problem by rounding up when calculating the frequency
coming out of a clock divider, so if that frequency is
requested again, the same divider value will be picked.
Change-Id: Ieaf74448f67d91aeb7ba08226e48c092d8afaa2b
Signed-off-by: Colin Cross <ccross@android.com>
Overlays can get their internal state corrupted during and underflow
condition. The only way to fix this state is to reset the DC. If we get
4 consecutive frames with underflows, assume we're hosed and reset.
Change-Id: Icdf61517837c8570b8de35f585075de08aa35fe7
Signed-off-by: Erik Gilling <konkers@android.com>
Cc: Michael I. Gold <gold@nvidia.com>
Add a config option to enable the watchdog driver from probe
Reduce the default timeout to 30 seconds
Read the reset reason and print a messge if the last reboot was
due to watchdog reset
Change-Id: I7500ef236089b06ffcff3d8e7d86a5b9060b59cb
Signed-off-by: Rebecca Schultz Zavin <rebecca@android.com>
HDMI was not working after LP0 because the plld were being reset
during suspend. plls and pllu were also not being saved. Add
all three to tegra_clk_suspend.
The lock time for plld is 1000 us, so increase the delay after
setting the PLLs.
Add a BUG_ON to ensure the size of the suspend context area is
correct.
Originally fixed by Mayuresh Kulkarni.
Original-author: Mayuresh Kulkarni <mkulkarni@nvidia.com>
Change-Id: I50a3e994c6e3cab5989aa7a8e26e7a2eb66b6dfb
Signed-off-by: Colin Cross <ccross@android.com>
This change attempts to reclaim carveout memory by killing
other carveout users when an allocation fails. Processes
are killed in order of priority from lowest to highest, and then
from largest to smallest users.
Change-Id: Iee8a6f36269bc8165d691000a153dbf9f4337775
Signed-off-by: Rebecca Schultz Zavin <rebecca@android.com>
Increase the max_frequency entries for clocks that can be driven from pll_a
to match the fastest pll_a table entry.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Colin Cross <ccross@android.com>
