Gscan capabilities were updated with new capabilities supported
by the device. While at it, simplify the firmware support
conditional and move both conditions into the WARN() to make it
easier to undertand and use the unlikely() for both.
Signed-off-by: Ayala Beker <ayala.beker@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The firmware/hardware only supports checking AES-CMAC on RX, not
using it on TX. For station mode this is fine, since it's the only
thing it will ever do. For AP mode, it never receives such frames,
but must be able to transmit them. This is currently broken since
we try to enable them for hardware crypto (for RX only) and then
treat them as TX_CMD_SEC_EXT, leading to FIFO underruns during TX
so the frames never go out to the air.
To fix this, simply use software on TX in AP (and IBSS) mode.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Newer firmware versions put different data in the memory
which is read by the driver upon firmware crash. Just
change the variable names in the code and the name of the
data in the log that we print withouth any functional
change.
On older firmware, there will be a mismatch between the
names that are printed and the content itself, but that's
harmless.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
When trying to reach high Rx throughput of more than 500Mbps on
a device with a relatively weak CPU (Atom x5-Z8500), CPU utilization
may become a bottleneck. Analysis showed that we are looping in
iwl_pcie_rx_handle for very long periods which led to starvation
of other threads (iwl_pcie_rx_handle runs with _bh disabled).
We were handling Rx and allocating new buffers and the new buffers
were ready quickly enough to be available before we had finished
handling all the buffers available in the hardware. As a
consequence, we called iwl_pcie_rxq_restock to refill the hardware
with the new buffers, and start again handling new buffers without
exiting the function. Since we read the hardware pointer again when
we goto restart, new buffers were handled immediately instead of
exiting the function.
This patch avoids refilling RBs inside rx handling loop, unless an
emergency situation is reached. It also doesn't read the hardware
pointer again unless we are in an emergency (unlikely) case.
This significantly reduce the maximal time we spend in
iwl_pcie_rx_handle with _bh disabled.
Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
iwl_mvm_tcool_get_cur_state is the function that returns the
cooling state index to the sysfs handler. This function returns
mvm->cooling_dev.cur_state but that variable was set to the
budget and not the cooling state index. Fix that.
Add a missing blank line while at it.
Signed-off-by: Chaya Rachel Ivgi <chaya.rachel.ivgi@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
We need to track the next packet that we will reclaim in
order to know when the Tx queues are empty. This is useful
when we open or tear down an A-MPDU session which requires
to switch queue.
The next packet being reclaimed is identified by its WiFi
sequence number and this is relevant only when we use QoS.
QoS NDPs do have a TID but have a meaningless sequence
number. The spec mandates the receiver to ignore the
sequence number in this case, allowing the transmitter to
put any sequence number. Our implementation leaves it 0.
When we reclaim a QoS NDP, we can't update the next_relcaim
counter since the sequence number of the QoS NDP itself is
invalid.
We used to update the next_reclaim based on the sequence
number of the QoS NDP which reset it to 1 (0 + 1) and
because of this, we never knew when the queue got empty.
This had to sad consequence to stuck the A-MPDU state
machine in a transient state.
To fix this, don't update next_reclaim when we reclaim
a QoS NDP.
Alesya saw this bug when testing u-APSD. Because the
A-MPDU state machine was stuck in EMPTYING_DELBA, we
updated mac80211 that we still have frames for that
station when it got back to sleep. mac80211 then wrongly
set the TIM bit in the beacon and requested to release
non-existent frames from the A-MPDU queue. This led to
a situation where the client was trying to poll frames
but we had no frames to send.
Reported-by: Alesya Shapira <alesya.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
From 9000 family on, we need to get HW address from host
CSR registers.
OEM can override it by fusing the override registers - read
those first, and if those are 0 - read the OTP registers instead.
In addition - bail out if no valid mac address is present. Make
it shared for all NICs.
Signed-off-by: Sara Sharon <sara.sharon@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
We kick the allocator when we have 2 RBDs that don't have
attached RBs, and the allocator allocates 8 RBs meaning
that it needs another 6 RBDs to attach the RBs to.
The design is that allocator should always have enough RBDs
to fulfill requests, so we give in advance 6 RBDs to the
allocator so that when it is kicked, it gets additional 2 RBDs
and has enough RBDs.
These RBDs were taken from the Rx queue itself, meaning
that each Rx queue didn't have the maximal number of
RBDs, but MAX - 6.
Change initial number of RBDs in the system to include both
queue size and allocator reserves.
Note the multi-queue is always 511 instead of 512 to avoid a
full queue since we cannot detect this state easily enough in
the 9000 arch.
Signed-off-by: Sara Sharon <sara.sharon@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
When running async rx handler the framework holds the mvm->mutex
before starting the async handler, that might cause a deadlock in case
the handler calls to ops that lock the mutex as well.
Add support for running async rx handler without hold the mutex before
activating the handler.
Signed-off-by: Chaya Rachel Ivgi <chaya.rachel.ivgi@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Currently when the FW sends start/stop aux roc time event
notification with an error status, the driver returns an
error value, but does not remove the time event, and does
not notify the stack above that the time event is over.
This causes problems that the stack above assumes we are still
in the middle of a time event, and therefore can block different
events, such as scanning.
On FW failure notification, cleanup the time event parameters and
notify the stack above that the time event is over.
Signed-off-by: Matti Gottlieb <matti.gottlieb@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Our hardware de-aggregates AMSDUs but copies the mac header
as it to the de-aggregated MPDUs. We need to turn off the AMSDU
bit in the QoS control ourselves.
Signed-off-by: Sara Sharon <sara.sharon@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Add debugfs entries to get the ctdp budget average
and to stop ctdp.
Signed-off-by: Chaya Rachel Ivgi <chaya.rachel.ivgi@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Add details of the interface devm_thermal_zone_of_sensor_register()
and devm_thermal_zone_of_sensor_unregister() in the
<thermal/sysfs-api.txt>.
Signed-off-by: Laxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: Eduardo Valentin <edubezval@gmail.com>
Add resource managed version of thermal_zone_of_sensor_register() and
thermal_zone_of_sensor_unregister().
This helps in reducing the code size in error path, remove of
driver remove callbacks and making proper sequence for deallocations.
Signed-off-by: Laxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: Eduardo Valentin <edubezval@gmail.com>
Add details of the interface thermal_zone_of_sensor_register() and
thermal_zone_of_sensor_unregister() in the thermal/sysfs-api.txt.
The details describes the functionality and parameter which
are passed to these interfaces.
Signed-off-by: Laxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: Eduardo Valentin <edubezval@gmail.com>
Previously the client-core detected this condition by sheer luck!
Since we used strncpy, no NUL byte would be included on the name. The
client-core would call strlen, which would read past the end of its
buffer, but return a number large enough that the client-core would
return ENAMETOOLONG.
Signed-off-by: Martin Brandenburg <martin@omnibond.com>
Signed-off-by: Mike Marshall <hubcap@omnibond.com>
Neil recently fixed an obscure race in break_stripe_batch_list. Debug would be
quite convenient if we know the stripe state. This is what this patch does.
Signed-off-by: Shaohua Li <shli@fb.com>
Not needed any more because we need to protect the elements on the list anyway.
This reverts commit 38bf516c75b4ef0f5c716e05fa9baab7c52d6c39.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Not needed any more because we need to protect the elements on the list anyway.
This reverts commit fe237ed7efec8ac147a4572fdf81173a7f8ddda7.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Not needed any more because we need to protect the elements on the list anyway.
This reverts commit dae6ecf9e6c9b677e577826c3ac665c6dd9c490b.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We not only need to protect the mapping tree and freed list itself,
but also the items on those list.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Off-by-one: last is inclusive, so the maximum is start + max_size - 1
Wrong unit: addr is in bytes, max_size is in pages
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
As observed on Apple iMac10,1, DCE-3.2, RV-730,
link rate of 2.7 Ghz is not selected, because
the args.v1.ucConfig flag setting for 2.7 Ghz
gets overwritten by a following assignment of
the transmitter to use.
Move link rate setup a few lines down to fix this.
In practice this didn't have any positive or
negative effect on display setup on the tested
iMac10,1 so i don't know if backporting to stable
makes sense or not.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Both are maintained by same team.
Reviewed-by: Christian König <christian.koenig@amd.com>
Acked-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Some PX laptops don't provide an ACPI method to control dGPU power. On
those systems, the driver is responsible for handling the dGPU power
state. Disable runtime PM on them until support for this is implemented.
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Some PX laptops don't provide an ACPI method to control dGPU power. On
those systems, the driver is responsible for handling the dGPU power
state. Disable runtime PM on them until support for this is implemented.
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
break_stripe_batch_list breaks up a batch and copies some flags from
the batch head to the members, preserving others.
It doesn't preserve or copy STRIPE_PREREAD_ACTIVE. This is not
normally a problem as STRIPE_PREREAD_ACTIVE is cleared when a
stripe_head is added to a batch, and is not set on stripe_heads
already in a batch.
However there is no locking to ensure one thread doesn't set the flag
after it has just been cleared in another. This does occasionally happen.
md/raid5 maintains a count of the number of stripe_heads with
STRIPE_PREREAD_ACTIVE set: conf->preread_active_stripes. When
break_stripe_batch_list clears STRIPE_PREREAD_ACTIVE inadvertently
this could becomes incorrect and will never again return to zero.
md/raid5 delays the handling of some stripe_heads until
preread_active_stripes becomes zero. So when the above mention race
happens, those stripe_heads become blocked and never progress,
resulting is write to the array handing.
So: change break_stripe_batch_list to preserve STRIPE_PREREAD_ACTIVE
in the members of a batch.
URL: https://bugzilla.kernel.org/show_bug.cgi?id=108741
URL: https://bugzilla.redhat.com/show_bug.cgi?id=1258153
URL: http://thread.gmane.org/5649C0E9.2030204@zoner.cz
Reported-by: Martin Svec <martin.svec@zoner.cz> (and others)
Tested-by: Tom Weber <linux@junkyard.4t2.com>
Fixes: 1b956f7a8f ("md/raid5: be more selective about distributing flags across batch.")
Cc: stable@vger.kernel.org (v4.1 and later)
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Shaohua Li <shli@fb.com>
Commit f37755490f ("tracepoints: Do not trace when cpu is offline") added
a check to make sure that tracepoints only get called when the cpu is
online, as it uses rcu_read_lock_sched() for protection.
Commit 3a630178fd ("tracing: generate RCU warnings even when tracepoints
are disabled") added lockdep checks (including rcu checks) for events that
are not enabled to catch possible RCU issues that would only be triggered if
a trace event was enabled. Commit f37755490f only stopped the warnings
when the trace event was enabled but did not prevent warnings if the trace
event was called when disabled.
To fix this, the cpu online check is moved to where the condition is added
to the trace event. This will place the cpu online check in all places that
it may be used now and in the future.
Cc: stable@vger.kernel.org # v3.18+
Fixes: f37755490f ("tracepoints: Do not trace when cpu is offline")
Fixes: 3a630178fd ("tracing: generate RCU warnings even when tracepoints are disabled")
Reported-by: Sudeep Holla <sudeep.holla@arm.com>
Tested-by: Sudeep Holla <sudeep.holla@arm.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
add the missing RAID Engine device node for p5040.
otherwise, the device can not be detected.
Signed-off-by: Xuelin Shi <xuelin.shi@nxp.com>
Signed-off-by: Scott Wood <oss@buserror.net>
csum_partial is often called for small fixed length packets
for which it is suboptimal to use the generic csum_partial()
function.
For instance, in my configuration, I got:
* One place calling it with constant len 4
* Seven places calling it with constant len 8
* Three places calling it with constant len 14
* One place calling it with constant len 20
* One place calling it with constant len 24
* One place calling it with constant len 32
This patch renames csum_partial() to __csum_partial() and
implements csum_partial() as a wrapper inline function which
* uses csum_add() for small 16bits multiple constant length
* uses ip_fast_csum() for other 32bits multiple constant
* uses __csum_partial() in all other cases
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
Modify platform driver suspend/resume to syscore
suspend/resume. This is because p1022ds needs to use
localbus when entering the PCIE resume.
Signed-off-by: Raghav Dogra <raghav.dogra@nxp.com>
[scottwood: dropped makefile churn]
Signed-off-by: Scott Wood <oss@buserror.net>
When CONFIG_DEBUG_PAGEALLOC is activated, the initial TLB mapping gets
flushed to track accesses to wrong areas. Therefore, kernel addresses
will also generate ITLB misses.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
The MPC885 reference manual says that SDCR shall have value 0x40, but
most exemples set SDCR to 0x1
With 0x1 in SDCR, we observe TX underruns on SCC when using it in
QMC mode.
According the NXP technical support, this is a copy/paste error from
MPC860 reference manual, 0x40 being the only value supported
by the MPC885 HW.
Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Scott Wood <oss@buserror.net>
This patch disables deprecated IDE subsystem in mpc8610_hpcd_defconfig
(no IDE host drivers are selected in this config so there is no valid
reason to enable IDE subsystem itself).
Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Signed-off-by: Scott Wood <oss@buserror.net>
This patch disables deprecated IDE subsystem in stx_gp3_defconfig
(no IDE host drivers are selected in this config so there is no valid
reason to enable IDE subsystem itself).
Cc: Scott Wood <oss@buserror.net>
Cc: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Signed-off-by: Scott Wood <oss@buserror.net>
This patch disables deprecated IDE subsystem in ksi8560_defconfig
(no IDE host drivers are selected in this config so there is no valid
reason to enable IDE subsystem itself).
Cc: Scott Wood <oss@buserror.net>
Cc: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Signed-off-by: Scott Wood <oss@buserror.net>
This patch disables deprecated IDE subsystem in mpc834x_itx_defconfig
(no IDE host drivers are selected in this config so there is no valid
reason to enable IDE subsystem itself).
Cc: Scott Wood <oss@buserror.net>
Cc: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Signed-off-by: Scott Wood <oss@buserror.net>
cpm_muram_alloc_common is called twice and both the times
spin_lock_irqsave is held.
Using GFP_KERNEL can sleep in spin_lock_irqsave context and cause
deadlock
Signed-off-by: Saurabh Sengar <saurabh.truth@gmail.com>
Signed-off-by: Scott Wood <oss@buserror.net>
as cpm_muram_alloc_common is used only in this file,
making it static
Signed-off-by: Saurabh Sengar <saurabh.truth@gmail.com>
Signed-off-by: Scott Wood <oss@buserror.net>
127 is the theoretical up boundary of QEIC number,
in fact there only be 44 qe_ic_info now.
add check to overflow for qe_ic_info
Signed-off-by: Zhao Qiang <qiang.zhao@nxp.com>
Acked-by: Li Yang <leoyang.li@nxp.com>
Signed-off-by: Scott Wood <oss@buserror.net>
Commit 66b3923a1a ("arm64: hugetlb: add support for PTE contiguous bit")
introduced support for huge pages using the contiguous bit in the PTE
as opposed to block mappings, which may be slightly unwieldy (512M) in
64k page configurations.
Unfortunately, this support has resulted in some late regressions when
running the libhugetlbfs test suite with 64k pages and CONFIG_DEBUG_VM
as a result of a BUG:
| readback (2M: 64): ------------[ cut here ]------------
| kernel BUG at fs/hugetlbfs/inode.c:446!
| Internal error: Oops - BUG: 0 [#1] SMP
| Modules linked in:
| CPU: 7 PID: 1448 Comm: readback Not tainted 4.5.0-rc7 #148
| Hardware name: linux,dummy-virt (DT)
| task: fffffe0040964b00 ti: fffffe00c2668000 task.ti: fffffe00c2668000
| PC is at remove_inode_hugepages+0x44c/0x480
| LR is at remove_inode_hugepages+0x264/0x480
Rather than revert the entire patch, simply avoid advertising the
contiguous huge page sizes for now while people are actively working on
a fix. This patch can then be reverted once things have been sorted out.
Cc: David Woods <dwoods@ezchip.com>
Reported-by: Steve Capper <steve.capper@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Some models reduce the number of available isochronous streams for higher
sampling transfer frequency. Such models bring an issue about how to add
PCM substreams. When at lower sampling transfer frequency, the
models reports whole available streams, thus this driver can add enough
number of PCM substreams at probing time. On the other hand, at higher
sampling transfer frequency, this driver can just add reduced number of
PCM substreams. After probed, even if the sampling transfer frequency is
changed to lower rate, fewer PCM substreams are actually available. This
is inconvenience.
For the reason, this commit adds a list so that this driver assume models
on the list to have two pairs of PCM substreams. This list keeps the name
of model in which the number of available streams differs depending on
sampling transfer frequency.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
In former commits, ALSA dice driver can handle available isochronous
streams. This commit adds support for several PCM substreams on the
streams.
The additional PCM substreams are available via another ALSA PCM character
devices so that one ALSA PCM application can handle them without cumbersome
operations. For example, two PCM substreams are available on each stream,
two ALSA character devices are added for them. In configuration space of
alsa-lib, it's represented with 'hw:0,0' and 'hw:0,1'.
The PCM substreams are constraint to parameters of the corresponding
streams. If the PCM substreams are unavailable for some reasons,
open(2) to ALSA PCM character device returns error and reports ENXIO.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
This commit enables ALSA dice driver to handle whole available streams.
In Dice, certain registers represent the number of available streams at
current sampling transfer frequency for both directions. The parameters
of each stream are represented in a block of register. This block is
aligned sequentially. These streams start simultaneously by writing
enable bit to a register.
This commit operates these registers when starting/stopping streams.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Currently ALSA dice driver handles a pair of isochronous resources for
IEC 61883-1/6 packet streaming. While, according to some documents about
ASICs named as 'Dice', several isochronous streams are available.
Here, I start to describe ASICs produced under 'Dice' name.
* Dice II (designed by wavefront semiconductor, including TCAT's IP)
* STD (with limited functionality of DTCP)
* CP (with full functionality of DTCP)
* TCD2210/2210-E (so-called 'Dice Mini')
* TCD2220/2220-E (so-called 'Dice Jr.')
* TCD3070-CH (so-called 'Dice III')
Some documents are public and we can see hardware design of them. We can
find some articles about hardware internal register definitions
(not registers exported to IEEE 1394 bus).
* DICE II User Guide
* http://www.tctechnologies.tc/archive/downloads/dice_ii_user_guide.pdf
* 6.1 AVS Audio Receivers
* Table 6.1: AVS Audio Receiver Memory Map
* ARX1-ARX4
* 6.2 AVS Audio Transmitters
* Table 6.2: AVS Audio Transmitter Memory Map
* ATX1, ATX2
* TCD22xx User Guide
* http://www.tctechnologies.tc/downloads/tcd22xx_user_guide.pdf
* 6.1 AVS Audio Receivers
* Table 66: AVS Audio Receiver Memory Map
* ARX1, ARX2
* 6/2 AVS Audio Transmitters
* Table 67: AVS Audio Transmitter Memory Map
* ATX1, ATX2
* DICE III
* http://www.tctechnologies.tc/downloads/TCD3070-CH.pdf
* Dual stream 63 channel transmitter/receiver
For Dice II and TCD22xx series, maximum 16 data channels are transferred in
an AMDTP packet, while for Dice III, maximum 32 data channels are
transferred.
According to the design of the series of these ASICs, this commit allows
this driver to handle additional set of isochronous resources. For
practical reason, two pair of isochronous resources are added. As of this
commit, this driver still use a pair of the first isochronous resources.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
I screwed up while merging the immutable branch for TPS65912,
so fixing it unbroken again.
Cc: Lee Jones <lee.jones@linaro.org>
Cc: Andrew F. Davis <afd@ti.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
