The current Makefile compiles the cpuidle34xx.c and cpuidle44xx.c files
even if the cpuidle option is not set in the kernel.
This patch fixes this by creating a section in the Makefile where these
files are compiled only if the CONFIG_CPU_IDLE option is set.
This modification breaks an implicit dependency between CPU_IDLE and PM as
they belong to the same block in the Makefile. This is fixed in the Kconfig
by selecting explicitely PM is CPU_IDLE is set.
The linux coding style recommend to use no-op functions in the headers
when the subsystem is disabled instead of adding big section in C files.
This patch fix this also.
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Reviewed-by: Jean Pihet <j-pihet@ti.com>
Reviewed-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
prcm_setup_regs() blindly accesses IVA bits
in the PRM and calls omap3_iva_idle() which
does more IVA related register accesses.
Only do this if the IVA hardware actually
exists.
Signed-off-by: Mark A. Greer <mgreer@animalcreek.com>
Signed-off-by: Kevin Hilman <khilman@ti.com>
Remove prm_clkdm and cm_clkdm and allow hwmods to have no clockdomain.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
The following commit did not care about the error handling path:
commit c1a7b32a14
KVM: Avoid wasting pages for small lpage_info arrays
If memory allocation fails, vfree() will be called with the address
returned by kzalloc(). This patch fixes this issue.
Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp>
Signed-off-by: Avi Kivity <avi@redhat.com>
At the moment we call kvmppc_pin_guest_page() in kvmppc_update_vpa()
with two spinlocks held: the vcore lock and the vcpu->vpa_update_lock.
This is not good, since kvmppc_pin_guest_page() calls down_read() and
get_user_pages_fast(), both of which can sleep. This bug was introduced
in 2e25aa5f ("KVM: PPC: Book3S HV: Make virtual processor area
registration more robust").
This arranges to drop those spinlocks before calling
kvmppc_pin_guest_page() and re-take them afterwards. Dropping the
vcore lock in kvmppc_run_core() means we have to set the vcore_state
field to VCORE_RUNNING before we drop the lock, so that other vcpus
won't try to run this vcore.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Acked-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This solves a section mismatch warning. I hadn't noticed this before,
because my compiler was inlining tegra_cpu_reset_handler_enable() inside
tegra_cpu_reset_handler_init(), which is already __init, but I switched
compilers and it stopped doing that.
Cc: <stable@kernel.org> # v3.4
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Olof Johansson <olof@lixom.net>
Remove the existing alias for pad_fck, prcm_fck from the clock data and
add them as opt_clks to the hwmod data.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Remove the existing alias for pad_fck, prcm_fck from the clock data and
add them as opt_clks to the hwmod data.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Remove the existing alias for pad_fck, prcm_fck from the clock data and
add them as opt_clks to the hwmod data.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Provide an interface for a driver to call SCM functions to
set a boot address and boot mode.
Signed-off-by: Omar Ramirez Luna <omar.luna@linaro.org>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
SCM contains boot addr and boot mode registers to control
other processors on different OMAP versions. It controls the
boot address and mode for DSP based subsystems like: IVA 2.1
(OMAP2430), IVA 2.2 (OMAP3) and DSP (OMAP4).
If contained within SCM registers, when a processor is
booting it uses BOOTADDR to start running the code at that
location. BOOTMOD register specifies a different set of
modes for the processor to execute when booting (from direct,
idle, self-loop, user and default).
Since there was no offset associated with OMAP4, this patch
defines it.
Signed-off-by: Omar Ramirez Luna <omar.luna@linaro.org>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
There are certain fields inside 'struct dpll_data' which are
included under ARCH_OMAP3 and ARCH_OMAP4 option, which makes it
difficult to use it for new devices like, am33xx, ti81xx, etc...
So remove the ifdef completely, this will add few fields to the struct
unused, but it improves readability and maintainability of the code.
Signed-off-by: Kevin Hilman <khilman@ti.com>
[hvaibhav@ti.com: wrote detailed changelog]
Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: R Sricharan <r.sricharan@ti.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
* topic/sebastian-devinit-fixups:
scripts/modpost: check for bad references in .pci.fixups area
sh/PCI: move fixup hooks from __init to __devinit
powerpc/PCI: move fixup hooks from __init to __devinit
frv/PCI: move fixup hooks from __init to __devinit
arm/PCI: move fixup hooks from __init to __devinit
alpha/PCI: move fixup hooks from __init to __devinit
PCI: move fixup hooks from __init to __devinit
x86/PCI: move fixup hooks from __init to __devinit
Rather than use runtime cpu_is* checking inside _init_clkdm, initialize
SoC specific function pointer at init time.
Signed-off-by: Kevin Hilman <khilman@ti.com>
[paul@pwsan.com: convert to use soc_ops function pointers; remove second para
from commit message since soc_ops function pointers are now set during hwmod
layer init]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Rather than using cpu_is* checking at runtime, initialize SoC specific
function pointers for the various hard reset functions at init time.
Signed-off-by: Kevin Hilman <khilman@ti.com>
[paul@pwsan.com: convert to use soc_ops function pointers; add kerneldoc]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Rather than using cpu_is* checking at runtime, initialize an SoC specific
function pointer for wait_target_ready().
While here, downgrade the BUG() to a WARN_ON() so it gives a noisy
warning instead of causing a kernel panic.
Signed-off-by: Kevin Hilman <khilman@ti.com>
[paul@pwsan.com: convert to use soc_ops function pointers; add kerneldoc;
move soc_ops functions to their own section in the code; integrated
the _wait_target_ready() function with the OMAP2/OMAP4 variants;
renamed the wait_module_ready field to wait_target_ready]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
_omap4_wait_target_disable() is called only from inside _omap4_disable_module()
which is already protected by SoC specific checks. Remove the cpu_is check
here.
Signed-off-by: Kevin Hilman <khilman@ti.com>
The enable/disable module functions are specific to SoCs with
OMAP4-class PRCM. Rather than use cpu_is* checks at runtime inside
the enable/disable module functions, use cpu_is at init time to
initialize function pointers only for SoCs that need them.
NOTE: the cpu_is* check for _enable_module was different than
the one for _disable_module, and this patch uses
cpu_is_omap44xx() for both.
Signed-off-by: Kevin Hilman <khilman@ti.com>
[paul@pwsan.com: moved soc_ops function pointers to be per-kernel rather than
per-hwmod since they do not vary by hwmod; added kerneldoc]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
_enable_module is specific to SoCs with PRCM interfaces similar to
that of the OMAP4, so rename it to be consistent with the
corresponding _omap4_disable_module.
Signed-off-by: Kevin Hilman <khilman@ti.com>
[paul@pwsan.com: tweaked commit message]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
AM33XX PRCM module consists of various clockdomains, in all
total we have 18 clockdomains available, with following
controlling options,
- SW Sleep: sw forced sleep transition
- SW Wakeup: sw forced wakeup transition
This patch adds all available clockdomain data, respective
clockdomain operations for AM33XX family of device, and also
integrates it into existing OMAP framework.
Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
Signed-off-by: Afzal Mohammed <afzal@ti.com>
Signed-off-by: Vaibhav Bedia <vaibhav.bedia@ti.com>
Cc: Kevin Hilman <khilman@ti.com>
Cc: Rajendra Nayak <rnayak@ti.com>
CC: Tony Lindgren <tony@atomide.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Benoit Cousson <b-cousson@ti.com>
[paul@pwsan.com: removed CLKDM_NO_AUTODEPS from clockdomain flags, removed
unnecessary .clktrctrl_offs field; updated for 3.5]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Add offset & mask fields to struct powerdomain
In case of AM33xx family of devices, there is no consistency between
PWRSTCTRL & PWRSTST register offsers in PRM space, for example -
PRM_XXX PWRSTCTRL PWRSTST
=======================================
PRM_PER_MOD: 0x0C, 0x08
PRM_WKUP_MOD: 0x04, 0x08
PRM_MPU_MOD: 0x00, 0x04
PRM_DEVICE_MOD: NA, NA
And also, there is no consistency between bit-offsets inside
PWRSTCTRL & PWRSTST register, for example -
PRM_XXX LOGICRET MEMON MEMRET
=======================================
GFX_PWRCTRL: 2, 17, 6
PER_PWRCTRL: 3, 25, 29
MPU_PWRCTRL: 2, 18, 22
WKUP_PWRCTRL: 3, NA, NA
This means, we need to maintain and pass on all this information
in powerdomain handle; so adding fields for,
- PWRSTCTRL/ST register offset
- Logic retention state mask
- mem_on/ret/pwrst/retst mask
Currently, this fields is only applicable and used for AM33XX devices.
Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
Cc: Benoit Cousson <b-cousson@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Kevin Hilman <khilman@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Rajendra Nayak <rnayak@ti.com>
[paul@pwsan.com: this patch is a combination of "Add offset & mask fields to
struct powerdomain" and the powerdomain portions of "ARM: OMAP3+: am33xx:
Add powerdomain & PRM support"; updated for 3.5]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
As far as PRM/CM/PRCM modules are concerned, AM33XX device is
different than OMAP3 and OMAP4 architectures; so similar to
PRM implementation, handle AM33XX CM separately.
This patch introduces AM33XX CM module low-level api's, used and
required by omap clockdomain and hwmod framework.
Please note that cm-regbits-33xx.h (register bit field offset)
and cm33xx.h (register addr offset) files are mostly auto generated.
Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
Signed-off-by: Afzal Mohammed <afzal@ti.com>
Signed-off-by: Vaibhav Bedia <vaibhav.bedia@ti.com>
Cc: Kevin Hilman <khilman@ti.com>
Cc: Rajendra Nayak <rnayak@ti.com>
CC: Tony Lindgren <tony@atomide.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Benoit Cousson <b-cousson@ti.com>
[paul@pwsan.com: split the hwmod code changes in this patch into a separate
patch; updated for 3.5]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
As far as PRM/CM/PRCM modules are concerned, AM33XX device is
different than OMAP3 and OMAP4 architectures; so we need to handle it
separately. This patch adds support for the PRM APIs required for
AM33XX device.
Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
Signed-off-by: Afzal Mohammed <afzal@ti.com>
Cc: Benoit Cousson <b-cousson@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Kevin Hilman <khilman@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Rajendra Nayak <rnayak@ti.com>
[paul@pwsan.com: separated the PRM parts of "ARM: OMAP3+: am33xx: Add
powerdomain & PRM support" into this patch; fixed Makefile prm33xx.o
location; cleaned up some checkpatch violations; updated for 3.5]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Instruction bundles are always little-endian, even when running in
big-endian mode. I missed this internal bug fix when cherry-picking
the big-endian code to return to the community.
Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
These dummy headers are unused. Commit
6ea96e111e ("ARM: davinci: add back dummy
header files") only added them as a temporary measure. They can be
removed now.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Sekhar Nori <nsekhar@ti.com>
This is a preparatory patch for the KVM/ARM implementation. KVM/ARM will use
the KVM_IRQ_LINE ioctl, which is currently conditional on
__KVM_HAVE_IOAPIC, but ARM obviously doesn't have any IOAPIC support and we
need a separate define.
Signed-off-by: Christoffer Dall <c.dall@virtualopensystems.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
This reverts commit d509835e32. That commit
breaks support for the generic pass-through mode in the driver for formats,
not natively supported by it. Besides due to a merge conflict it also breaks
driver compilation:
drivers/media/video/mx2_camera.c: In function 'mx2_camera_set_bus_param':
drivers/media/video/mx2_camera.c:937: error: 'pixfmt' undeclared (first use in this function)
drivers/media/video/mx2_camera.c:937: error: (Each undeclared identifier is reported only once
drivers/media/video/mx2_camera.c:937: error: for each function it appears in.)
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Cc: Javier Martin <javier.martin@vista-silicon.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This patch adds the generic Intel uncore PMU support, including helper
functions that add/delete uncore events, a hrtimer that periodically
polls the counters to avoid overflow and code that places all events
for a particular socket onto a single cpu.
The code design is based on the structure of Sandy Bridge-EP's uncore
subsystem, which consists of a variety of components, each component
contains one or more "boxes".
(Tooling support follows in the next patches.)
Signed-off-by: Zheng Yan <zheng.z.yan@intel.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1339741902-8449-6-git-send-email-zheng.z.yan@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The RDPMC index calculation is wrong for AMD family 15h
(X86_FEATURE_ PERFCTR_CORE set). This leads to a #GP when
accessing the counter:
Pid: 2237, comm: syslog-ng Not tainted 3.5.0-rc1-perf-x86_64-standard-g130ff90 #135 AMD Pike/Pike
RIP: 0010:[<ffffffff8100dc33>] [<ffffffff8100dc33>] x86_perf_event_update+0x27/0x66
While the msr address offset is (index << 1) we must use index to
select the correct rdpmc.
Signed-off-by: Robert Richter <robert.richter@amd.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Vince Weaver <vweaver1@eecs.utk.edu>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Currently dummy voltage domain data is being created
in order to succeed boot process, nothing has been done
w.r.t actual hardware (voltage control).
Also, hook up the AM33XX voltage domain to OMAP framework.
Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
Signed-off-by: Afzal Mohammed <afzal@ti.com>
Cc: Benoit Cousson <b-cousson@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Kevin Hilman <khilman@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Rajendra Nayak <rnayak@ti.com>
Acked-by: Kevin Hilman <khilman@ti.com>
[paul@pwsan.com: updated for 3.5]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Define AM33XX control register, in order to allow access to
control register address space, also add CONTROL_SEC_CLK_CTRL
register offset; both are required in clock tree data,
for wdt0 and timer0 clock source select configuration.
CONTROL.SEC_CLK_CTRL register is provided to select/configure
clock input for WDT0 and TIMER0.
Signed-off-by: Vaibhav Hiremath <hvaibhav@ti.com>
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Tony Lindgren <tony@atomide.com>
[paul@pwsan.com: added include of plat/am33xx.h to fix build break;
added AM33XX_CONTROL_STATUS bitfields that will be needed for the clock
tree; fixed some control.h whitespace problems while here]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Commit a530257240 (OMAP: Add debugfs
node to show the summary of all clocks) introduced clock summary,
however, we are interested in seeing snapshot of the clock state, not
in dynamically changing clock configurations as the data provided by
clock summary will then be useless for debugging configuration
issues. So, hold the common lock when dumping the clock summary.
Cc: Paul Walmsley <paul@pwsan.com>
Cc: Tony Lindgren <tony@atomide.com>
Signed-off-by: Todd Poynor <toddpoynor@google.com>
[nm@ti.com: added commit message]
Signed-off-by: Nishanth Menon <nm@ti.com>
[paul@pwsan.com: minor edits to commit message]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
v3.5-rc1 fails to build when DT and iconnect is enabled because of
this now none existant include file.
Also remove the other two SPI include files, which are not needed
with the move to DT.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Acked-by: Arnaud Patard <arnaud.patard@rtp-net.org>
Signed-off-by: Olof Johansson <olof@lixom.net>
With the revamped realmode trampoline code, it is trivial to extend
support for reboot=bios to x86-64. Furthermore, while we are at it,
remove the restriction that only we can only override the reboot CPU
on 32 bits.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Link: http://lkml.kernel.org/n/tip-jopx7y6g6dbcx4tpal8q0jlr@git.kernel.org
Pull ARM SoC fixes from Olof Johansson:
"A bit larger set than usual, unfortunately -- I've been sitting on
them longer than I meant to so it's really more like 2 -rc pull
requests in one, volume-wise.
Nearly everything is fixes for fallout from the merge window, or other
fixes for bugs. The one exception is the highbank L2-enablement
patch, but it was contained enough that I picked it up anyway:
- i.MX fixes, mostly for clock and pinctrl changes
- OMAP fixes, mostly PM-related
- A patch to enable L2 on highbank
- A couple of fixes for PXA, Kirkwood, Versatile"
* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (30 commits)
ARM: Kirkwood: Fix clk problems modular ethernet driver
arm: versatile: fix and enable PCI I/O space
ARM: highbank: Add smc calls to enable/disable the L2
ARM i.MX imx21ads: Fix overlapping static i/o mappings
ARM: imx6: exit coherency when shutting down a cpu
ARM: mx51: Add pinctrl_provide_dummies()
ARM: mx31: Add pinctrl_provide_dummies()
ARM: OMAP2+: Fix compile for CONFIG_TIDSPBRIDGE platform init code
ARM: OMAP3: Fix omap3_l3_block_irq warning when CONFIG_BUG is not set
ARM: OMAP: Fix MMC_OMAP build when only MMC_OMAP_HS is selected
OMAP2+: UART: Add mechanism to probe uart pins and configure rx wakeup
ARM: mmp: fix missing cascade_irq in irq handler
ARM: dts: update memory size on brownstone
ARM i.MX27 Visstrim M10: fix gpio handling.
ARM i.MX53: Fix PLL4 base address
ARM i.MX pllv2: make round_rate accurate
ARM i.MX pllv2: use standard register set unconditionally
ARM: OMAP: Fix lis3lv02d accelerometer to use gpio_to_irq
ARM: imx: only call l2x0_init if it's available
ARM: imx: only specify i2c device type once
...
When the ethernet driver was built as a module, it would lock the
machine when loaded. At boot the ethernet clks are unused, so get
turned off. Later, when the module is loaded, the probe function
would access the hardware before the clock was restarted, and the
machine would lock. It has also been determined that when the clk is
turned off, the interface forgets its MAC address, which for most
systems, is set by the boot loader.
When the machine setup file creates a platform device for the
interface, prepare and enable the clock for the interface. This will
ensure it is not turned off. However, if the setup file only
instantiates one platform device, the other will have its clk
disabled, thus maybe saving a little power.
Report-by: Simon Baatz <gmbnomis@gmail.com>
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Tested-by: Simon Baatz <gmbnomis@gmail.com>
Signed-off-by: Olof Johansson <olof@lixom.net>
With commit 4d5fc58dbe (ARM: remove bunch of now unused
mach/io.h files), the I/O space setup was completely broken on
versatile. This patch fixes that and prepares for further
I/O space clean-up.
I/O space handling on the versatile platform is currently
broken in multiple ways. Most importantly, the ports do
not get mapped into the virtual address space at all.
Also, there is some amount of confusion between PCI I/O
space and other statically mapped MMIO registers in the
platform code:
* The __io_address() macro that is used to access the
platform register maps to the same __io macro that gets
used for I/O space.
* The IO_SPACE_LIMIT is set to a value that is much larger
than the total available space.
* The I/O resource of the PCI bus is set to the physical
address of the mapping, which is way outside of the
actual I/O space limit as well as the address range that
gets decoded by traditional PCI cards.
* No attempt is made to stay outside of the ISA port range
that some device drivers try access.
* No resource gets requested as a child of ioport_resource,
but an IORESOURCE_IO type mapping gets requested
as a child of iomem_resource.
This patch attempts to correct all of the above. This makes
it possible to use virtio-pci based virtual devices as well
as actual PCI cards including those with legacy ISA port
ranges like VGA.
Some of the issues seem to be duplicated on other platforms.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
[rob: update to 3.5-rc2 and io.h cleanup related changes]
Signed-off-by: Rob Herring <rob.herring@calxeda.com>
Tested-by: Robert Schwebel <r.schwebel@pengutronix.de>
Signed-off-by: Olof Johansson <olof@lixom.net>
Pull arch/tile update from Chris Metcalf:
"This one-line bug fix unbreaks glibc robust mutexes (among other
things no doubt), from code merged in during the 3.5 merge window but
which we had been running internally at Tilera for almost a year."
* 'stable' of git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile:
tile: fix bug in get_user() for 4-byte values
The definition of 32-bit values in the 64-bit tilegx architecture is that
they should be sign-extended regardless of whether they are considered
signed or unsigned by the compiler. Accordingly, we need to use an
"ld4s" rather than "ld4u" to load and sign-extend for get_user().
This fixes glibc bug 14238 (see http://sourceware.org/bugzilla),
introduced during the 3.5 merge window.
Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
