Currently mdio read/write takes around ~115us as the timeout
between status check is set to 100us.
By reducing the timeout to 1us mdio read/write takes ~15us to
complete. This improves the link up event response.
Signed-off-by: Hemanth Puranik <hpuranik@codeaurora.org>
Acked-by: Timur Tabi <timur@codeaurora.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add an entry for the builtin PHYs present in the Broadcom BCM5395 switch. This
allows us to retrieve the PHY statistics among other things.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Chris Healy <cphealy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Clear Feature Endpoint Halt should reset the data toggle even if the
endpoint isn't halted. Host should manage to clear the host side data
toggle to keep in sync with the device.
Test by sending a "3 data packet URB" before and after clearing the halt.
this should create a toggle sequence with two consecutive DATA0 packets.
A successful test sequence looks like this
ClearFeature(ENDPOINT_HALT) - initial toggle clear
DATA0 (max packet sized)
DATA1 (max packet sized)
DATA0 (zero length packet)
ClearFeature(ENDPOINT_HALT) - resets toggle
DATA0 (max packet sized), if clear halt fails then toggle is DATA1
DATA1 (max packet sized)
DATA0 (zero length packet)
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
USB SS and SSP hubs provide wHubDelay values on their hub descriptor
which we should inform the USB Device about.
The USB Specification 3.0 explains, on section 9.4.11, how to
calculate the value and how to issue the request. Note that a
USB_REQ_SET_ISOCH_DELAY is valid on all device states (Default,
Address, Configured), we just *chose* to issue it from Address state
right after successfully fetching the USB Device Descriptor.
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Pull xen fixes from Juergen Gross:
"Two minor fixes for running as Xen dom0:
- when built as 32 bit kernel on large machines the Xen LAPIC
emulation should report a rather modern LAPIC in order to support
enough APIC-Ids
- The Xen LAPIC emulation is needed for dom0 only, so build it only
for kernels supporting to run as Xen dom0"
* tag 'for-linus-4.15-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
xen: XEN_ACPI_PROCESSOR is Dom0-only
x86/Xen: don't report ancient LAPIC version
The `max310x_spi_probe` function attempted to setup the GPIO bits before
the corresponding structs for each serial port were initialized. If the
DTS file specified a GPIO hog, this led to a crash because the GPIO
stack ended up calling `max310x_gpio_direction_output` which referenced
uninitialized memory:
[<c04598c0>] (max310x_gpio_direction_output) from [<c03f5a2c>] (_gpiod_direction_output_raw+0x94/0x2d4)
[<c03f5a2c>] (_gpiod_direction_output_raw) from [<c03f991c>] (gpiod_hog+0x6c/0x154)
[<c03f991c>] (gpiod_hog) from [<c03fa2d8>] (of_gpiochip_add+0x28c/0x444)
[<c03fa2d8>] (of_gpiochip_add) from [<c03f6b2c>] (gpiochip_add_data+0x4f8/0x760)
[<c03f6b2c>] (gpiochip_add_data) from [<c03f6dd4>] (devm_gpiochip_add_data+0x40/0x7c)
[<c03f6dd4>] (devm_gpiochip_add_data) from [<c0459fec>] (max310x_spi_probe+0x530/0x894)
[<c0459fec>] (max310x_spi_probe) from [<c0503294>] (spi_drv_probe+0x7c/0xac)
[<c0503294>] (spi_drv_probe) from [<c046628c>] (driver_probe_device+0x234/0x2e8)
[<c046628c>] (driver_probe_device) from [<c0464890>] (bus_for_each_drv+0x60/0x94)
[<c0464890>] (bus_for_each_drv) from [<c0465f78>] (__device_attach+0xb0/0x114)
[<c0465f78>] (__device_attach) from [<c0465548>] (bus_probe_device+0x84/0x8c)
[<c0465548>] (bus_probe_device) from [<c0463a00>] (device_add+0x3f4/0x580)
[<c0463a00>] (device_add) from [<c0504164>] (spi_add_device+0x9c/0x134)
[<c0504164>] (spi_add_device) from [<c0504c18>] (spi_register_controller+0x484/0x910)
[<c0504c18>] (spi_register_controller) from [<c0506ee0>] (orion_spi_probe+0x2f4/0x3b4)
[<c0506ee0>] (orion_spi_probe) from [<c0467dac>] (platform_drv_probe+0x50/0xb0)
[<c0467dac>] (platform_drv_probe) from [<c046628c>] (driver_probe_device+0x234/0x2e8)
[<c046628c>] (driver_probe_device) from [<c04663f8>] (__driver_attach+0xb8/0xbc)
[<c04663f8>] (__driver_attach) from [<c04647e8>] (bus_for_each_dev+0x68/0x9c)
[<c04647e8>] (bus_for_each_dev) from [<c046574c>] (bus_add_driver+0x104/0x210)
[<c046574c>] (bus_add_driver) from [<c0466f14>] (driver_register+0x78/0xf4)
[<c0466f14>] (driver_register) from [<c0101bdc>] (do_one_initcall+0x44/0x168)
[<c0101bdc>] (do_one_initcall) from [<c0a00dc0>] (kernel_init_freeable+0x140/0x1cc)
[<c0a00dc0>] (kernel_init_freeable) from [<c078c590>] (kernel_init+0x8/0x108)
[<c078c590>] (kernel_init) from [<c0107a50>] (ret_from_fork+0x14/0x24)
This can be easily fixed by moving the corresponding code below. And
because the UARTs are already there by the time we reach this point, the
`goto` needs changing so that more stuff is freed. (I have not tested
this error path.)
Signed-off-by: Jan Kundrát <jan.kundrat@cesnet.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If uart_add_one_port() fails in mxs_auart_probe, the clks has
to be disabled.Two clks are previously enabled in mxs_get_clks().
Found by Linux Driver Verification project (linuxtesting.org).
Signed-off-by: Branislav Radocaj <branislav@radocaj.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 3840ed9548 ("tty: goldfish: Implement support for kernel
'earlycon' parameter") breaks an allmodconfig config on x86:
| LD vmlinux.o
| MODPOST vmlinux.o
|drivers/tty/serial/earlycon.o: In function `parse_options':
|drivers/tty/serial/earlycon.c:97: undefined reference to `uart_parse_earlycon'
|Makefile:1005: recipe for target 'vmlinux' failed
earlycon.c::parse_options() invokes uart_parse_earlycon() from serial_core.c
which is compiled=m because GOLDFISH_TTY itself (and most others) are =m.
To avoid that, I'm adding the _CONSOLE config option which is selected if the
GOLDFISH module itself is =y since it doesn't need the early bits for the =m
case (other drivers do the same dance).
The alternative would be to move uart_parse_earlycon() from
serial_core.c to earlycon.c (we don't have that many users of that
function).
Fixes: 3840ed9548 ("tty: goldfish: Implement support for kernel
'earlycon' parameter")
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Miodrag Dinic <miodrag.dinic@mips.com>
Acked-by: Miodrag Dinic <miodrag.dinic@mips.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If there is no clock rate for uartclk defined, disable the previously
enabled clock again.
Found by Linux Driver Verification project (linuxtesting.org).
Fixes: 23f5b3fdd0 serial: 8250_dw: only get the clock rate in one place
Signed-off-by: Stefan Potyra <Stefan.Potyra@elektrobit.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The driver may sleep under a spinlock.
The function call paths are:
isicom_activate (acquire the spinlock)
isicom_setup_board
drop_dtr_rts
WaitTillCardIsFree
msleep --> may sleep
isicom_set_termios
isicom_config_port
drop_dtr
WaitTillCardIsFree
msleep --> may sleep
isicom_tiocmset
drop_dtr
WaitTillCardIsFree
msleep --> may sleep
Though "in_atomic" is used to check atomic context,
but it is not recommended to use in driver code (see include/linux/preempt.h).
To fix it, only using mdelay instead.
This bug is found by my static analysis tool(DSAC) and checked by my code review.
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On most Renesas ARM platforms, the SCIF serial ports can be used with
DMA, so most users will want DMA support to be enabled.
On SuperH platforms, SCI(F) serial ports cannot be used with DMA yet
(see also commit 219fb0c143 ("serial: sh-sci: Remove the platform
data dma slave rx/tx channel IDs")), so users will want it disabled to
reduce kernel size.
Hence follow the above rationale to configure the default, unless
CONFIG_EXPERT is enabled.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Renesas H8/300 and ARM platforms use DT and support earlycon, so most
users want earlycon support to be enabled.
On SuperH platforms, earlycon is not yet supported.
Hence follow the above rationale to configure the default, unless
CONFIG_EXPERT is enabled.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Most users will want to use a serial console.
Hence make that the default, unless CONFIG_EXPERT is enabled.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Auto-configure the maximum number of serial ports based on how many can
be present on the architecture:
- 3 on H8/300,
- 10 on SuperH,
- 18 on Reneas ARM.
The default can still be overridden if CONFIG_EXPERT is enabled.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Using "make kvmconfig" results in a potentially unusable linux image
on s390. The reason is that both the (default on s390) sclp consoles
as well as the 8250 console register a ttyS<x> as console. Since there
will be no 8250 on s390 let's fence 8250. This will ensure that there
is always a working sclp console.
Reported-by: Alice Frosi <alice@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Update TODO file:
- Remove TODO info about the OV564x driver, while this still needs
to be done (add a OV5642 driver or merge with OV5640 driver), it
is not relevant here.
- Update TODO about methods for retrieving CSI bus config.
- Add some TODO's about OF graph parsing restrictions.
Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Re-order some of the function prototypes in imx-media.h to
group them correctly by source file. No functional changes.
Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Although not technically necessary because imx-media has only a
maximum of 8 video devices, and once setup the video device lists
are static, in anticipation of moving control ineritance to
v4l2-core, make the vdev lists more generic by converting to
dynamic list_head's.
After doing that, 'struct imx_media_pad' is now just a list_head
of video devices reachable from a pad. Allocate an array of list_head's,
one list_head for each pad, and attach that array to sd->host_priv.
An entry in the pad lists is of type 'struct imx_media_pad_vdev', and
points to a video device from the master list.
Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Commit dfea747d2a ("drivers: base: cacheinfo: support DT overrides for
cache properties") doesn't initialise the cache type if it's present
only in DT and the architecture is not aware of it. They are unified
system level cache which are generally transparent.
This patch check if the cache type is set to NOCACHE but the DT node
indicates that it's unified cache and sets the cache type accordingly.
Fixes: dfea747d2a ("drivers: base: cacheinfo: support DT overrides for cache properties")
Reported-and-tested-by: Tan Xiaojun <tanxiaojun@huawei.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For more complex OF graphs, there will be more async subdevices
registered. Remove the static subdev[IMX_MEDIA_MAX_SUBDEVS] array,
so that imx-media places no limits on the number of async subdevs
that can be added and registered.
There were two uses for 'struct imx_media_subdev'. First was to act
as the async subdev list to be passed to v4l2_async_notifier_register().
Second was to aid in inheriting subdev controls to the capture devices,
and this is done by creating a list of capture devices that can be reached
from a subdev's source pad. So 'struct imx_media_subdev' also contained
a static array of 'struct imx_media_pad' for placing the capture device
lists at each pad.
'struct imx_media_subdev' has been completely removed. Instead, at async
completion, allocate an array of 'struct imx_media_pad' and attach it to
the subdev's host_priv pointer, in order to support subdev controls
inheritance.
Likewise, remove static async_ptrs[IMX_MEDIA_MAX_SUBDEVS] array.
Instead, allocate a 'struct imx_media_async_subdev' when forming
the async list, and add it to an asd_list list_head in
imx_media_add_async_subdev(). At async completion, allocate the
asd pointer list and pull the asd's off asd_list for
v4l2_async_notifier_register().
Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Pass the subdev's fwnode_handle to imx_media_find_async_subdev() and
imx_media_add_async_subdev(), instead of a device_node.
Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
A separate string for the device name, for DEVNAME async match, was
never needed. Just assign the asd device name to the passed platform
device name pointer in imx_media_add_async_subdev().
Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Calling of_parse_subdev() recursively to a downstream path that has
already been followed is ok, it just means that call will return
immediately since the subdevice was already added to the async list.
With that there is no need to determine whether a subdevice's port
is a sink or source, so 'num_{sink|src}_pads' is no longer used and
is removed.
Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Remove the static list of media links that were formed at probe time.
These links can instead be created after all registered async subdevices
have been bound in imx_media_probe_complete().
The media links between subdevices that exist in the device tree, can
be created post-async completion by using v4l2_fwnode_parse_link() for
each endpoint node of that subdevice. Note this approach assumes
device-tree ports are equivalent to media pads (pad index equals
port id), and that device-tree endpoints are equivalent to media
links between pads.
Because links are no longer parsed by imx_media_of_parse(), its sole
function is now only to add subdevices that it encounters by walking
the OF graph to the async list, so the function has been renamed
imx_media_add_of_subdevs().
Similarly, the media links between the IPU-internal subdevice pads (the
CSI source pads, and all pads between the vdic, ic-prp, ic-prpenc, and
ic-prpvf subdevices), can be created post-async completion by looping
through the subdevice's media pads and using the const internal_subdev
table.
Because links are no longer parsed by imx_media_add_internal_subdevs(),
this function no longer needs an array of CSI subdevs to form links
from.
In summary, the following functions, which were used to form a list
of media links at probe time, are removed:
imx_media_add_pad_link()
add_internal_links()
of_add_pad_link()
replaced by these functions, called at probe time, which only populate
the async subdev list:
imx_media_add_of_subdevs()
imx_media_add_internal_subdevs()
and these functions, called at async completion, which create the
media links:
imx_media_create_of_links()
imx_media_create_csi_of_links()
imx_media_create_internal_links()
Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
Unlike the various of_* routines to fetch properties, fwnode_* routines can
have an early check against a NULL fwnode_handle reference which makes them
return -EINVAL (see fwnode_call_int_op), thus making it virtually impossible to
differentiate what type of error is going on.
Have an early check in phylink_register_sfp() so we can keep proceeding with
the initialization, there is not much we can do without a valid fwnode_handle
except return early and treat this similarly to -ENOENT.
Fixes: 8fa7b9b6af ("phylink: convert to fwnode")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
It has been reported that the dummy byte we add to avoid
ZLPs can be forwarded by the modem to the PGW/GGSN, and that
some operators will drop the connection if this happens.
In theory, QMI devices are based on CDC ECM and should as such
both support ZLPs and silently ignore the dummy byte. The latter
assumption failed. Let's test out the first.
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
There is a hunk of code that is incorrectly indented with spaces
and rather than a tab. Clean this up.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support for SFF modules, which are soldered down SFP modules.
These have a different phys_id value, and also have the present and
rate select signals omitted compared with their socketed counter-parts.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For XPB registers reads, some island IDs require special handling (e.g.
ARM island), which is already taken care of in nfp_xpb_readl(), so use
that instead of a straight CPP read.
Without this fix all "xpbm:ArmIsldXpbmMap.*" registers are reported as
0xffffffff. It has also been observed to cause a system reboot.
With this fix correct values are reported, none of which are 0xffffffff.
The values may be read using ethtool debug level 2.
# ethtool -W <netdev> 2
# ethtool -w <netdev> data dump.dat
Fixes: 0e6c4955e1 ("nfp: dump CPP, XPB and direct ME CSRs")
Signed-off-by: Carl Heymann <carl.heymann@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In TLV-based ethtool debug dumps, don't do a CPP read for absolute
rtsyms, use the addr field in the symbol table directly as the value.
Without this fix rtsym gro_release_ring_0 is 4 bytes of zeros.
With this fix the correct value, 0x0000004a 0x00000000 is reported.
The values may be read using ethtool debug level 2.
# ethtool -W <netdev> 2
# ethtool -w <netdev> data dump.dat
Fixes: e1e798e3fd ("nfp: dump rtsyms")
Signed-off-by: Carl Heymann <carl.heymann@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a suffix to distinguish kernel mainline version and aquantia releases
Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
On very first start we should read out current HW counter values
to make diff based calculations later.
This also should be done each time NIC gets down/up or wakes up
after sleep state. We reset link state explicitly to prevent diffs
from being summed this first time.
Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Reduce timeout from 2 secs to 1 sec. If link is down,
reduce it to 500msec. This speeds up link detection.
Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This metric comes from HW and is also diff-calculated, like other counters
Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Originally they were filled from ring sw counters.
These sometimes incorrectly calculate byte and packet amounts
when using LRO/LSO and jumboframes. Filling ndev counters from
hardware makes them precise.
Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Device hardware provides only 32bit counters. Using these directly
causes byte counters to overflow soon. A separate nic level structure
with 64 bit counters is now used to collect incrementally all the stats
and report these counters to ethtool stats and ndev stats.
Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Systems with large MRRS on device (2K, 4K) with high data rates and/or
large MTU, atlantic observes DMA packet buffer overflow. On some systems
that causes PCIe transaction errors, hardware NMIs or datapath freeze.
This patch
1) Limits MRRS from device side to 2K (thats maximum our hardware supports)
2) Limit maximum size of outstanding TX DMA data read requests. This makes
hardware buffers running fine.
Signed-off-by: Pavel Belous <pavel.belous@aquantia.com>
Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Different hardware device Ids correspond to different maximum speed
available. Extra checks were added for devices D108 and D109 to
remove unsupported speeds from these device capabilities list.
Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Firmware flashing takes around 60s (specified to not take more than
70s). Prevent hogging the RTNL lock in this time and make use of the
longer timeout for the NSP command. The timeout is set to 2.5 * 70
seconds.
We only allow flashing the firmware from reprs or PF netdevs. VFs do not
have an app reference.
Signed-off-by: Dirk van der Merwe <dirk.vandermerwe@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The firmware flashing NSP operation takes longer to execute than the
current default timeout. We need a mechanism to set a longer timeout for
some commands. This patch adds the infrastructure to this.
The default timeout is still 30 seconds.
Signed-off-by: Dirk van der Merwe <dirk.vandermerwe@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The imx-media driver currently supports a device tree graph of
limited complexity. This patch is a first step in allowing imx-media
to work with more general OF graphs.
The CSI subdevice assumes the originating upstream subdevice (the
"sensor") is connected directly to either the CSI mux or the MIPI
CSI-2 receiver. But for more complex graphs, the sensor can be distant,
with possible bridge entities in between. Thus the sensor's bus type
could be quite different from what is entering the CSI. For example
a distant sensor could have a parallel interface, but the stream
entering the i.MX is MIPI CSI-2.
To remove this assumption, get the entering bus config from the entity
that is directly upstream from either the CSI mux, or the CSI-2 receiver.
If the CSI-2 receiver is not in the enabled pipeline, the bus type to the
CSI is parallel, otherwise the CSI is receiving MIPI CSI-2.
Note that we can't use the direct upstream source connected to CSI
(which is either the CSI mux or the CSI-2 receiver) to determine
bus type. The bus entering the CSI from the CSI-2 receiver is a 32-bit
parallel bus containing the demultiplexed MIPI CSI-2 virtual channels.
But the CSI and its IDMAC channels must be configured based on whether
it is receiving data from the CSI-2 receiver or from the CSI mux's
parallel interface pins.
The function csi_get_upstream_endpoint() is used to find this
endpoint. It makes use of a new utility function
imx_media_find_upstream_pad(), that if given a grp_id of 0, will
return the closest upstream pad from start_entity.
With these changes, imx_media_find_sensor() is no longer used and
is removed. As a result there is also no longer a need to identify
any sensor or set the sensor subdev's group id as a method to search
for it. So IMX_MEDIA_GRP_ID_SENSOR is removed. Also the video-mux group
id IMX_MEDIA_GRP_ID_VIDMUX was never used so that is removed as well.
The remaining IMX_MEDIA_GRP_ID_* definitions are entities internal
to the i.MX.
Another use of imx_media_find_sensor() in the CSI was to call the
sensor's g_skip_frames op to determine if a delay was needed before
enabling the CSI at stream on. If necessary this will have to be
re-addressed at a later time.
Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
A typical code fragment was copied across many dvb-frontend drivers and
causes large stack frames when built with with CONFIG_KASAN on gcc-5/6/7:
drivers/media/dvb-frontends/cxd2841er.c:3225:1: error: the frame size of 3992 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
drivers/media/dvb-frontends/cxd2841er.c:3404:1: error: the frame size of 3136 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
drivers/media/dvb-frontends/stv0367.c:3143:1: error: the frame size of 4016 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
drivers/media/dvb-frontends/stv090x.c:3430:1: error: the frame size of 5312 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
drivers/media/dvb-frontends/stv090x.c:4248:1: error: the frame size of 4872 bytes is larger than 3072 bytes [-Werror=frame-larger-than=]
gcc-8 now solves this by consolidating the stack slots for the argument
variables, but on older compilers we can get the same behavior by taking
the pointer of a local variable rather than the inline function argument.
Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81715
Cc: stable@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
With CONFIG_KASAN, we get an overly long stack frame due to inlining
the register access functions:
drivers/media/tuners/r820t.c: In function 'generic_set_freq.isra.7':
drivers/media/tuners/r820t.c:1334:1: error: the frame size of 2880 bytes is larger than 2048 bytes [-Werror=frame-larger-than=]
This is caused by a gcc bug that has now been fixed in gcc-8.
To work around the problem, we can pass the register data
through a local variable that older gcc versions can optimize
out as well.
Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81715
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@s-opensource.com>
IPvlan currently scrubs packets at every location where packets may be
crossing namespace boundary. Though this is desirable, currently IPvlan
does it more than necessary. e.g. packets that are going to take
dev_forward_skb() path will get scrubbed so no point in scrubbing them
before forwarding. Another side-effect of scrubbing is that pkt-type gets
set to PACKET_HOST which overrides what was already been set by the
earlier path making erroneous delivery of the packets.
Also scrubbing packets just before calling dev_queue_xmit() has detrimental
effects since packets lose skb->sk and because of that miss prio updates,
incorrect socket back-pressure and would even break TSQ.
Fixes: b93dd49c1a ('ipvlan: Scrub skb before crossing the namespace boundary')
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This reverts commit 92ff426450.
Even though the check added is not that taxing, it's not really needed.
First of all this will be per packet cost and second thing is that the
eth_type_trans() already does this correctly. The excessive scrubbing
in IPvlan was changing the pkt-type skb metadata of the packet which
made it necessary to re-check the mac. The subsequent patch in this
series removes the faulty packet-scrub.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Any modification to the takeover IP-ranges requires that we re-evaluate
which IP addresses are takeover-eligible. Otherwise we might do takeover
for some addresses when we no longer should, or vice-versa.
Signed-off-by: Julian Wiedmann <jwi@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
