This change adds a flag to indicate that a UART is has an external means
of synchronising its FIFO, without needing CTSRTS or XON/XOFF.

This allows us to use the throttle/unthrottle callbacks, without having
to claim other methods of flow control.
Signed-off-by: NJeremy Kerr <jk@ozlabs.org>
Tested-by: NEddie James <eajames@linux.vnet.ibm.com>
Tested-by: NJoel Stanley <joel@jms.id.au>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Remove dependencies on HAS_DMA where a Kconfig symbol depends on another
symbol that implies HAS_DMA, and, optionally, on "|| COMPILE_TEST".
In most cases this other symbol is an architecture or platform specific
symbol, or PCI.

Generic symbols and drivers without platform dependencies keep their
dependencies on HAS_DMA, to prevent compiling subsystems or drivers that
cannot work anyway.

This simplifies the dependencies, and allows to improve compile-testing.
Signed-off-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: NMark Brown <broonie@kernel.org>
Acked-by: NRobin Murphy <robin.murphy@arm.com>
Acked-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

On DT based platforms when current-speed property is present baudrate
is setup. Also port->uartclk is initialized to bogus BASE_BAUD * 16
value. Drivers like uartps/ns16550 contain logic when baudrate and
uartclk is used for baudrate calculation.

The patch is reading optional clock-frequency property to replace bogus
BASE_BAUD * 16 calculation to have proper baudrate calculation.
Signed-off-by: NMichal Simek <michal.simek@xilinx.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

On some systems IRQ lines between multiple UARTs might be shared. If so, the
irqflags have to be configured accordingly. The reason is: The 8250 port startup
code performs IRQ tests *before* the IRQ handler for that particular port is
registered. This is performed in serial8250_do_startup(). This function checks
whether IRQF_SHARED is configured and only then disables the IRQ line while
testing.

This test is performed upon each open() of the UART device. Imagine two UARTs
share the same IRQ line: On is already opened and the IRQ is active. When the
second UART is opened, the IRQ line has to be disabled while performing IRQ
tests. Otherwise an IRQ might handler might be invoked, but the the IRQ itself
cannot be handled, because the corresponding handler isn't registered,
yet. That's because the 8250 code uses a chain-handler and invokes the
corresponding port's IRQ handling rountines himself.

Unfortunately this IRQF_SHARED flag isn't configured for UARTs probed via device
tree even if the IRQs are shared. This way, the actual and shared IRQ line isn't
disabled while performing tests and the kernel correctly detects a spurious
IRQ. So, adding this flag to the DT probe solves the issue.

Note: The UPF_SHARE_IRQ flag is configured unconditionally. Therefore, the
IRQF_SHARED flag can be set unconditionally as well.

Example stacktrace by performing echo 1 > /dev/ttyS2 on a non-patched system:

|irq 85: nobody cared (try booting with the "irqpoll" option)
| [...]
|handlers:
|[<ffff0000080fc628>] irq_default_primary_handler threaded [<ffff00000855fbb8>] serial8250_interrupt
|Disabling IRQ #85
Signed-off-by: NKurt Kanzenbach <kurt@linutronix.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Enable/Clear module level UART wakeup in UART_OMAP_WER register based on
return value of device_may_wakeup() in .suspend(). This allows
userspace to use sysfs to control the ability of UART to wakeup the
system from deep sleep state. Register is restored back in .startup()
call that happens as part of resume sequence.

With this patch, userspace can control UART wakeup capability via sysfs:
To enable wakeup capability:
echo enabled >  /sys/class/tty/ttyXX/device/power/wakeup
For disabling wakeup capability:
echo disabled > /sys/class/tty/ttyXX/device/power/wakeup

Note that the UART wakeup events configured in the 8250 hardware only
work for idle modes that do not cut off power for the UART. For deeper
idle states, dedicated padconf wakeirqs must be used. Or in some cases
the UART RX pin can be remuxed to GPIO input if the GPIO block stays
powered.
Signed-off-by: NVignesh R <vigneshr@ti.com>
Tested-by: NTony Lindgren <tony@atomide.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The previous implementation has had a detrimental effect on devices using
high bitrates (bluetooth), as the fifo being non-empty for a single check
would result in a 10 µs delay.

Limit the change to devices with the new "marvell,armada-38x-uart"
compatible string. Also update the code to allow the first 1000 retries
to not perform a delay.

The maximum duration of retries has been increased to cover a worst-case
seen on the Armada 385 SoC. "dmesg ; resize", will fill the buffer with
text to output before doing a resize. At 9600 baud this took up to 13 ms
to flush all characters and avoid some getting lost.
Signed-off-by: NJoshua Scott <joshua.scott@alliedtelesis.co.nz>
Reviewed-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

HSCIF has facilities that allow moving the RX sampling point by between
-8 and 7 sampling cycles (one sampling cycles equals 1/15 of a bit
by default) to improve the error margin in case of slightly mismatched
bit rates between sender and receiver.

This patch tries to determine if shifting the sampling point can improve
the error margin and will enable it if so.
Signed-off-by: NUlrich Hecht <ulrich.hecht+renesas@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

On DT platforms, the sh-sci driver requires the presence of "serialN"
aliases in DT, from which instance IDs are derived.  If a DT alias is
missing, the drivers fails to probe the corresponding serial port.

This becomes cumbersome when considering DT overlays, as currently
there is no upstream support for dynamically updating the /aliases node
in DT.  Furthermore, even in the presence of such support, hardcoded
instance IDs in independent overlays are prone to conflicts.

Hence add support for dynamic instance IDs, to be used in the absence of
a DT alias.  This makes serial ports behave similar to I2C and SPI
buses, which already support dynamic instances.

Ports in use are tracked using a simple bitmask of type unsigned long,
which is sufficient to handle all current hardware (max. 18 ports).
The maximum number of serial ports is still fixed, and configurable
through Kconfig.  Range validation is done through both Kconfig and a
compile-time check.

Due to the fixed maximum number of serial ports, dynamic and static
instances share the same ID space.  Static instances added later are
rejected when conflicting with dynamic instances registered earlier.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Add suspend and resume hooks to save/restore the registers content
during S2RAM operation.

Also save/restore the oversampling rate register (OSAMP) as earlier
stages already tuned that register to get a precise UART clock.
Suggested-by: NAllen Yan <yanwei@marvell.com>
Signed-off-by: NMiquel Raynal <miquel.raynal@bootlin.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Without this, tx stops working after resume. By adding
these calls, everything seems to work fine.
Signed-off-by: NPascal Huerst <pascal.huerst@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

We should get drvdata from struct device directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
Acked-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: NMichal Simek <michal.simek@xilinx.com>
Acked-by: NPatrice Chotard <patrice.chotard@st.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

RFC 4122 asks for letters a-f in UUID to be lowercase. Follow this
recommendation.

Suggested by Paul Dagnelie at:
https://savannah.nongnu.org/bugs/index.php?53569Signed-off-by: NJean Delvare <jdelvare@suse.de>

OEM strings are defined by each OEM and they contain customized and
useful OEM information. Supporting it provides more flexible uses of
the dmi_matches function.
Signed-off-by: NAlex Hung <alex.hung@canonical.com>
Signed-off-by: NJean Delvare <jdelvare@suse.de>

The test which ensures that the DMI type 1 structure is long enough
to hold the UUID is off by one. It would fail if the structure is
exactly 24 bytes long, while that's sufficient to hold the UUID.

I don't expect this bug to cause problem in practice because all
implementations I have seen had length 8, 25 or 27 bytes, in line
with the SMBIOS specifications. But let's fix it still.
Signed-off-by: NJean Delvare <jdelvare@suse.de>
Fixes: a814c359 ("firmware: dmi_scan: Check DMI structure length")
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>

This seems to cause flickering and lock-ups for a wide range of users.
Revert until we've found a proper fix for the flickering and lock-ups.

This reverts commit 36cc549d.

Cc: Shirish S <shirish.s@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Reviewed-by: NMichel Dänzer <michel.daenzer@amd.com>
Signed-off-by: NHarry Wentland <harry.wentland@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This reverts commit cd2d6c92.

Cc: Shirish S <shirish.s@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Reviewed-by: NMichel Dänzer <michel.daenzer@amd.com>
Signed-off-by: NHarry Wentland <harry.wentland@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Hardware understands the regamma LUT as a piecewise linear function,
with points spaced exponentially along the range. We previously
programmed the LUT for range [2^-10, 2^0). This causes (normalized)
color values of 1 (=2^0) to miss the programmed LUT, and fall onto the
end region.

For DCE, the end region is extrapolated using a single (base, slope)
pair, using the max y-value from the last point in the curve as base.
This presents a problem, since this value affects all three color
channels. Scaling down the intensity of say - the blue regamma curve -
will not affect it's end region. This is especially noticiable when
using RedShift. It scales down the blue and green channels, but leaves
full-intensity colors unshifted.

Therefore, extend the range to cover [2^-10, 2^1) by programming another
hardware segment, containing only one point. That way, we won't be
hitting the end region.

Note that things are a bit different for DCN, since the end region can
be set per-channel.
Signed-off-by: NLeo (Sunpeng) Li <sunpeng.li@amd.com>
Reviewed-by: NKrunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: NHarry Wentland <harry.wentland@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Signed-off-by: NRoman Li <roman.li@amd.com>
Reviewed-by: NCharlene Liu <Charlene.Liu@amd.com>
Acked-by: NHarry Wentland <harry.wentland@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Signed-off-by: NHarry Wentland <harry.wentland@amd.com>
Reviewed-by: NTony Cheng <Tony.Cheng@amd.com>
Acked-by: NHarry Wentland <harry.wentland@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

The nvmf_check_if_ready() checks that were added are very simplistic.
As such, the routine allows a lot of cases to fail ios during windows
of reset or re-connection. In cases where there are not multi-path
options present, the error goes back to the callee - the filesystem
or application. Not good.

The common routine was rewritten and calling syntax slightly expanded
so that per-transport is_ready routines don't need to be present.
The transports now call the routine directly. The routine is now a
fabrics routine rather than an inline function.

The routine now looks at controller state to decide the action to
take. Some states mandate io failure. Others define the condition where
a command can be accepted.  When the decision is unclear, a generic
queue-or-reject check is made to look for failfast or multipath ios and
only fails the io if it is so marked. Otherwise, the io will be queued
and wait for the controller state to resolve.

Admin commands issued via ioctl share a live admin queue with commands
from the transport for controller init. The ioctls could be intermixed
with the initialization commands. It's possible for the ioctl cmd to
be issued prior to the controller being enabled. To block this, the
ioctl admin commands need to be distinguished from admin commands used
for controller init. Added a USERCMD nvme_req(req)->rq_flags bit to
reflect this division and set it on ioctls requests.  As the
nvmf_check_if_ready() routine is called prior to nvme_setup_cmd(),
ensure that commands allocated by the ioctl path (actually anything
in core.c) preps the nvme_req(req) before starting the io. This will
preserve the USERCMD flag during execution and/or retry.
Signed-off-by: NJames Smart <james.smart@broadcom.com>
Reviewed-by: NSagi Grimberg <sagi@grimberg.e>
Reviewed-by: NJohannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Commit 42de82a8 previously attempted to fix this, and it did
correctly pad the MN and FR fields with spaces, but the SN field still
contains 0 bytes.  The current code fills out the first 16 bytes with
hex2bin, leaving the last 4 bytes zeroed.  Rather than adding a lot of
error-prone math to avoid overwriting SN twice, just set the whole thing
to spaces up front (it's only 20 bytes).

Fixes: 42de82a8 ("nvmet: don't report 0-bytes in serial number")
Signed-off-by: NDaniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: NMartin Wilck <mwilck@suse.com>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Also add error flow in case srcu initialization function fails.
Signed-off-by: NMax Gurtovoy <maxg@mellanox.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

We have to increment the number of logical blocks to a 1's based value
in the native format prior to converting to 512b units.
Signed-off-by: NRodrigo R. Galvao <rosattig@linux.vnet.ibm.com>
[changelog]
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

The admin and first IO queues shared the first irq vector, which has an
affinity mask including cpu0. If a system allows cpu0 to be offlined,
the admin queue may not be usable if no other CPUs in the affinity mask
are online. This is a problem since unlike IO queues, there is only
one admin queue that always needs to be usable.

To fix, this patch allocates one pre_vector for the admin queue that
is assigned all CPUs, so will always be accessible. The IO queues are
assigned the remaining managed vectors.

In case a controller has only one interrupt vector available, the admin
and IO queues will share the pre_vector with all CPUs assigned.

Cc: Jianchao Wang <jianchao.w.wang@oracle.com>
Cc: Ming Lei <ming.lei@redhat.com>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NMing Lei <ming.lei@redhat.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

All the queue memory is allocated up front. We don't take the node
into consideration when creating queues anymore, so removing the unused
parameter.
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NMing Lei <ming.lei@redhat.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

User reported controller always retains CSTS.RDY to 1, which fails
controller disabling when resetting the controller. This is also before
the admin queue is allocated, and trying to disable an unallocated queue
results in a NULL dereference.
Reported-by: NAlex Gagniuc <Alex_Gagniuc@Dellteam.com>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

nvmet_execute_get_disc_log_page() passes a fixed-length string into
nvmet_format_discovery_entry(), which then does a longer memcpy() on
it, as pointed out by gcc-8:

In function 'nvmet_format_discovery_entry',
    inlined from 'nvmet_execute_get_disc_log_page' at drivers/nvme/target/discovery.c:126:4:
drivers/nvme/target/discovery.c:62:2: error: 'memcpy' forming offset [38, 223] is out of the bounds [0, 37] [-Werror=array-bounds]
  memcpy(e->subnqn, subsys_nqn, NVMF_NQN_SIZE);

Using strncpy() will make this well-defined, filling the rest of the
buffer with zeroes, under the assumption that the input is either
a NUL-terminated string, or a byte sequence containing no zeroes.
If the input is a string that is longer than NVMF_NQN_SIZE, we
continue to have no NUL-termination in the output.

Fixes: a07b4970 ("nvmet: add a generic NVMe target")
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

NVMe over Fabrics 1.0 Section 5.2 "Discovery Controller Properties and
Command Support" Figure 31 "Discovery Controller – Admin Commands"
explicitly listst all commands but "Get Log Page" and "Identify" as
reserved, but NetApp report the Linux host is sending Keep Alive
commands to the discovery controller, which is a violation of the
Spec.

We're already checking for discovery controllers when configuring the
keep alive timeout but when creating a discovery controller we're not
hard wiring the keep alive timeout to 0 and thus remain on
NVME_DEFAULT_KATO for the discovery controller.

This can be easily remproduced when issuing a direct connect to the
discovery susbsystem using:
'nvme connect [...] --nqn=nqn.2014-08.org.nvmexpress.discovery'
Signed-off-by: NJohannes Thumshirn <jthumshirn@suse.de>
Fixes: 07bfcd09 ("nvme-fabrics: add a generic NVMe over Fabrics library")
Reported-by: NMartin George <marting@netapp.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

nvme_start_keep_alive() isn't used outside core.c so unexport it and
make it static.
Signed-off-by: NJohannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

When nvmet_req_init() fails, __nvmet_req_complete() is called
to handle the target request via .queue_response(), so
nvme_loop_queue_response() shouldn't be called again for
handling the failure.

This patch fixes this case by the following way:

- move blk_mq_start_request() before nvmet_req_init(), so
nvme_loop_queue_response() may work well to complete this
host request

- don't call nvme_cleanup_cmd() which is done in nvme_loop_complete_rq()

- don't call nvme_loop_queue_response() which is done via
.queue_response()
Signed-off-by: NMing Lei <ming.lei@redhat.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
[trimmed changelog]
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Compiling on 32 bits system produces a warning for the shift width
when shifting 32 bit integer with 64bit integer.

Make sure that offset always is 64bit, and use macros for retrieving
lower and upper bits of the offset.
Signed-off-by: NMatias Bjørling <mb@lightnvm.io>
Signed-off-by: NKeith Busch <keith.busch@intel.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

The API docs describe i2c_transfer() as taking a pointer to an array
of i2c_msg containing at least 1 entry, but leaves it to the individual
drivers to sanity check the msgs and num parameters. Let's do this in
core code instead.
Signed-off-by: NArd Biesheuvel <ard.biesheuvel@linaro.org>
[wsa: changed '<= 0' to '< 1']
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

On some systems, the BIOS expects certain SMBus register values to
match the hardware defaults. Restore these configuration registers at
shutdown time to avoid confusing the BIOS. This avoids hard-locking
such systems upon reboot.
Signed-off-by: NJean Delvare <jdelvare@suse.de>
Tested-by: NJason Andryuk <jandryuk@gmail.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Cc: stable@vger.kernel.org

Saving the original value of register SMBSLVCMD in
i801_enable_host_notify() doesn't work, because this function is
called not only at probe time but also at resume time. Do it in
i801_probe() instead, so that the saved value is not overwritten at
resume time.
Signed-off-by: NJean Delvare <jdelvare@suse.de>
Fixes: 22e94bd6 ("i2c: i801: store and restore the SLVCMD register at load and unload")
Reviewed-by: NBenjamin Tissoires <benjamin.tissoires@redhat.com>
Tested-by: NJason Andryuk <jandryuk@gmail.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Cc: stable@vger.kernel.org	# v4.10+

I added dumping of link information about tun devices over netlink in
commit 1ec010e7 ("tun: export flags, uid, gid, queue information
over netlink"), but didn't add the missing netlink notifications when
the device's exported properties change.

This patch adds notifications when owner/group or flags are modified,
when queues are attached/detached, and when a tun fd is closed.
Reported-by: NThomas Haller <thaller@redhat.com>
Fixes: 1ec010e7 ("tun: export flags, uid, gid, queue information over netlink")
Signed-off-by: NSabrina Dubroca <sd@queasysnail.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Otherwise, register_netdevice advertises the creation of the device with
the default flags, instead of what the user requested.
Reported-by: NThomas Haller <thaller@redhat.com>
Fixes: 1ec010e7 ("tun: export flags, uid, gid, queue information over netlink")
Signed-off-by: NSabrina Dubroca <sd@queasysnail.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit 92571a1a ("lan78xx: Connect phy early") moves the PHY
initialisation into lan78xx_probe, but lan78xx_open subsequently calls
lan78xx_reset. As well as forcing a second round of link negotiation,
this reset frequently prevents the phy interrupt from being generated
(even though the link is up), rendering the interface unusable.

Fix this issue by removing the lan78xx_reset call from lan78xx_open.

Fixes: 92571a1a ("lan78xx: Connect phy early")
Signed-off-by: NPhil Elwell <phil@raspberrypi.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When open fails during ethtool -L ring change, for example, the driver
may crash at bnxt_free_irq() because bp->bnapi is NULL.

If we fail to allocate all the new rings, bnxt_open_nic() will free
all the memory including bp->bnapi.  Subsequent call to bnxt_close_nic()
will try to dereference bp->bnapi in bnxt_free_irq().

Fix it by checking for !bp->bnapi in bnxt_free_irq().

Fixes: e5811b8c ("bnxt_en: Add IRQ remapping logic.")
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

With recent changes to reserve both L2 and RDMA rings, we need to include
the RDMA rings in bnxt_check_rings().  Otherwise we will under-estimate
the rings we need during ethtool -L and may lead to failure.

Fixes: fbcfc8e4 ("bnxt_en: Reserve completion rings and MSIX for bnxt_re RDMA driver.")
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>