| Age | Commit message (Collapse) | Author |
|---|
|
This reverts commit 842002f3ef06e6ca88d90968733878660994b5b8.
|
|
On the LX2160A, there are lots (about 160) of IOMMU messages produced
during boot; this is excessive. Reduce the severity of these messages
to debug level.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Add QSFP cage network interfaces.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Add experimental QSFP+ support for the SolidRun Clearfog-CX
platform.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Use a Kconfig symbol to control the build of sfp-bus.c
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
There is a erratum on lx2160a which is: "SATA link is
going down sometime during sata initialization"
The workaround for it is to reset the lane. This patch
implements this workaround.
This erratum only exists on lx2160 Rev1, will be addressed
on Rev2 and later.
Signed-off-by: Peng Ma <peng.ma@nxp.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
When modifying the objects attached to a DPRC, we may end up reading
the list of objects from the firmware while another thread is changing
changing the list. Since we read the objects via:
- Read the number of DPRC objects
- Iterate over this number of objects retrieving their details
and objects can be added in the middle of the list, this causes the
last few objects to unexpectedly disappear. The side effect of this
is if network interfaces are added after boot, they come and go. This
can result in already configured interfaces unexpectedly disappearing.
This has been easy to provoke with the restool interface added, and a
script which adds network interfaces one after each other; the kernel
rescanning runs asynchronously to restool.
NXP's approach to fixing this was to introduce a sysfs "attribute" in
their vendor tree, /sys/bus/fsl-mc/rescan, which userspace poked at to
request the kernel to rescan the DPRC object tree each time the
"restool" command completed (whether or not the tool changed anything.)
This has the effect of making the kernel's rescan synchronous with a
scripted restool, but still fails if we have multiple restools running
concurrently.
This patch takes a different approach:
- Read the number of DPRC objects
- Iterate over this number of objects retrieving their details
- Re-read the number of DPRC objects
- If the number of DPRC objects has changed while reading, repeat.
This solves the issue where network interfaces unexpectedly disappear
while adding others via ls-addni, because they've fallen off the end
of the object list.
This does *not* solve the issue that if an object is deleted while
another is added while we are reading the objects - that requires
firmware modification, or a more elaborate solution on the Linux side
(e.g., CRCing the object details and reading all objects at least
twice to check the CRC is stable.)
However, without firmware modification, this is probably the best way
to ensure that we read all the objects.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
The MC firmware (which runs the networking subsystem) is loaded into
RAM by U-Boot, and this region is omitted from the memory passed to
the kernel via DT. Prior to booting the kernel, the MC processing is
halted to allow IOMMU setup.
When booting the kernel with IOMMU support enabled and without using
both bypass and passthrough mode, the MC firmware crashes as soon as
it is released, as the MC is unable to access the RAM that has been
assigned to it for both the firmware image and other purposes, and
also the DCFG to retrieve the SoC version.
In order to avoid this, we need to setup identity mappings in the MC
domain. For the MC RAM region, we read the firmware base address
registers which tell us where the firmware is located. According to
the MC design document, the firmware is loaded within the upper 512M
of the MC RAM region, aligned to 512M, and the RAM region is also
aligned to 512M. The lower 8 bits of the firmware base address low
register tells us how large the RAM region is. Use this to calculate
its size and location in order to create an indentity mapping.
We also search DT for the DCFG node to retrieve its address, and create
a read-only identity mapping to allow the MC firmware to read the SoC
version. If we are unable to find the DCFG node, we use a default
address for this. [XXX This needs to be improved XXX]
This allows "arm-smmu.disable_bypass=1" to be dropped from the kernel
command line for LX2160A platforms.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Enable SMMU management for the MC firmware by adding the required
iommus property in the device tree node.
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
|
|
|
|
While the Mellanox driver is binding, the following kernel panic
occurred:
SError Interrupt on CPU1, code 0xbf000002 -- SError
CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.3.0+ #392
Hardware name: SolidRun LX2160A COM express type 7 module (DT)
pstate: 60400085 (nZCv daIf +PAN -UAO)
pc : pci_generic_config_read+0xb0/0xc0
lr : pci_generic_config_read+0x1c/0xc0
sp : ffffff8010f9baf0
x29: ffffff8010f9baf0 x28: ffffff8010d620a0
x27: ffffff8010d79000 x26: ffffff8010d62000
x25: ffffff8010cb06d4 x24: 0000000000000000
x23: ffffff8010e499b8 x22: ffffff8010f9bbaf
x21: 0000000000000000 x20: ffffffe2eda11800
x19: ffffff8010f62158 x18: ffffff8010bdede0
x17: ffffff8010bdede8 x16: ffffff8010b96970
x15: ffffffffffffffff x14: ffffffffff000000
x13: ffffffffffffffff x12: 0000000000000030
x11: 0101010101010101 x10: 7f7f7f7f7f7f7f7f
x9 : 2dff716475687163 x8 : ffffffffffffffff
x7 : fefefefefefefefe x6 : 0000000000000000
x5 : 0000000000000000 x4 : ffffff8010f9bb6c
x3 : 0000000000000001 x2 : 0000000000000003
x1 : 0000000000000000 x0 : 0000000000000000
Kernel panic - not syncing: Asynchronous SError Interrupt
CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.3.0+ #392
Hardware name: SolidRun LX2160A COM express type 7 module (DT)
Call trace:
dump_backtrace+0x0/0x120
show_stack+0x14/0x1c
dump_stack+0x9c/0xc0
panic+0x148/0x34c
print_tainted+0x0/0xa8
arm64_serror_panic+0x74/0x80
do_serror+0x8c/0x13c
el1_error+0xbc/0x160
pci_generic_config_read+0xb0/0xc0
pci_bus_read_config_byte+0x64/0x90
pci_read_config_byte+0x40/0x48
pci_assign_irq+0x34/0xc8
pci_device_probe+0x28/0x148
really_probe+0x1c4/0x2d0
driver_probe_device+0x58/0xfc
device_driver_attach+0x68/0x70
__driver_attach+0x94/0xdc
bus_for_each_dev+0x50/0xa0
driver_attach+0x20/0x28
bus_add_driver+0x14c/0x200
driver_register+0x6c/0x124
__pci_register_driver+0x48/0x50
mlx4_init+0x154/0x180
do_one_initcall+0x30/0x250
kernel_init_freeable+0x23c/0x32c
kernel_init+0x10/0xfc
ret_from_fork+0x10/0x18
SMP: stopping secondary CPUs
Kernel Offset: disabled
CPU features: 0x0002,21006008
Memory Limit: none
which appears to be due to:
pci_read_config_byte(dev, PCI_INTERRUPT_PIN, &pin);
in pci_assign_irq(). Avoiding that access just moves the SError later
(e.g. while accessing the command register in pci_enable_device()
instead.)
This patch resolves the SError problem by preventing configuration
accesses triggering a SError interrupt.
Reported-by: Russell King <rmk+kernel@armlinux.org.uk>
Tested-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
[description modified -- rmk]
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
When LX2 PCIe controller is sending multiple split completions and
ACK latency expires indicating that ACK should be send at priority.
But because of large number of split completions and FC update DLLP,
the controller does not give priority to ACK transmission. This
results into ACK latency timer timeout error at the link partner and
the pending TLPs are replayed by the link partner again.
Workaround:
1. Reduce the ACK latency timeout value to a very small value.
2. Restrict the number of completions from the LX2 PCIe controller
to 1, by changing the Max Read Request Size (MRRS) of link partner
to the same value as Max Packet size (MPS).
This patch implemented part 1, the part 2 can be set by kernel parameter
'pci=pcie_bus_perf'
This ERRATA is only for LX2160A Rev1.0, and it will be fixed
in Rev2.0.
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
[fixed up for mainline -- rmk]
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
PCIe configuration access to non-existent function triggered
SERROR interrupt exception.
Workaround:
Disable error reporting on AXI bus during the Vendor ID read
transactions in enumeration.
This ERRATA is only for LX2160A Rev1.0, and it will be fixed
in Rev2.0.
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
This reverts commit 42d7a8dc195f99e2e99d8f38a683e0852a29f6af.
|
|
Fix errors in the mobiveil version that was merged in 5.7 kernels:
- the type of "root_bus_nr" was changed from "u8" to "char", but
it is compared to values that are typed as "unsigned char".
Depending whether a platform has "char" as signed or unsigned,
this may not do what is intended.
- ls_g4_pcie_reinit_hw() now returns a success/failure value, and
follows the Linux style of return 0 on success and -ve errno on
failure. However, the testing in ls_pcie_g4_reset() expects 0
on failure, so we won't call ls_ig4_pcie_enable_interrupt() except
if ls_g4_pcie_reinit_hw() has failed - which is likely not what
was intended.
Fixes: d29ad70a813b ("PCI: mobiveil: Add PCIe Gen4 RC driver for Layerscape SoCs")
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Add the PHY interface mode to the LPI methods and validate that LPI
is being entered for a supported mode.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Add the PHY interface mode to the LPI methods and validate that LPI
is being entered for a supported mode.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Ass the PHY interface mode into mac_disable_tx_lpi() and
mac_enable_tx_lpi() methods.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Handle -EOPNOTSUPP from mac_enable_tx_lpi() to print a sensible error.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Select the initial SFP PHY interface mode from the PHY supported
interface bitmaps.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Add a supported_interfaces member to phylib so we know which
interfaces a PHY supports. Currently, set any unconverted driver
to indicate all interfaces are supported.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Display SFP module information verbosely, splitting the generic parts
into a separate file.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
|
|
Add a compatible for SFP+ cages. SFP+ cages are backwards compatible,
but the ethernet device behind them may not support the slower speeds
of SFP modules.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Cooled SFP+ transceivers need a longer initialisation (startup) time.
Select the initialisation time depending on the cooled option bit.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
phy_error() is called from phy_interrupt() or phy_state_machine(), and
uses WARN_ON() to print a backtrace. The backtrace is not useful when
reporting a PHY error.
However, a system may contain multiple ethernet PHYs, and phy_error()
gives no clue which one caused the problem.
Replace WARN_ON() with a call to phydev_err() so that we can see which
PHY had an error, and also inform the user that we are halting the PHY.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Allow the PHY to probe when there is no firmware, but do not allow the
link to come up by forcing the PHY state to PHY_HALTED in a similar way
to phy_error().
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Provide phy driver start/stop hooks so that the PHY driver knows when
the network driver is starting or stopping. This will be used for the
Marvell 10G driver so that we can sanely refuse to start if the PHYs
firmware is not present, and also so that we can sanely support SFPs
behind the PHY.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Support reporting the hardware resolved pause enablement states via
phylib, overriding our software implementation.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Allow phylib drivers to pass the hardware-resolved pause state to MAC
drivers, rather than using the software-based pause resolution code.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
If MDIO_USXGMII_LINK is not set, it means that the PHYs media side
link is down. Indicate back to phylink that the link as a whole is
down.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Configure the Macchiatobin 10G PHY LED modes to correct their polarity.
We keep the existing LED behaviours, but switch their polarity to
reflect how they are connected. Tweak the LED modes as well to be:
left: off = no link
solid green = RJ45 link up (not SFP+ cage)
flash green = traffic
right: off = no link
solid green = 10G
solid yellow = 1G
flash green = 100M
flash yellow = 10M
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Add support for configuring the LEDs. Macchiatobin has an oddity in
that the left LED goes out when the cable is connected, and flashes
when there's link activity. This is because the reset default for
the LED outputs assume that the LED is connected to supply, not to
ground. Add support for configuring the LED modes and polarities.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
Add a DT bindings document for the Marvell 10G driver, which will
augment the generic ethernet PHY binding by having LED mode
configuration.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
The modalias string provided in the uevent sysfs file does not conform
to the format used in PHY driver modules. One of the reasons is that
udev loading of PHY driver modules has not been an expected use case.
This patch changes the MODALIAS entry for only PHY devices from:
MODALIAS=of:Nethernet-phyT(null)
to:
MODALIAS=mdio:00000000001000100001010100010011
Other MDIO devices (such as DSA) remain as before.
However, having udev automatically load the module has the advantage
of making use of existing functionality to have the module loaded
before the device is bound to the driver, thus taking advantage of
multithreaded boot systems, potentially decreasing the boot time.
However, this patch will not solve any issues with the driver module
not being loaded prior to the network device needing to use the PHY.
This is something that is completely out of control of any patch to
change the uevent mechanism.
Reported-by: Yinbo Zhu <zhuyinbo@loongson.cn>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
The hardware timestamps for packets contain a truncated seconds field,
only containing two bits of seconds. In order to provide the full
number of seconds, we need to keep track of the full hardware clock by
reading it every two seconds.
However, if we fail to read the clock, we silently ignore the error.
Print a warning indicating that the PP2 TAI clock timestamps have
become unreliable.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
|
|
No driver now uses the DSA switch phylink members, so we can now remove
the shim functions and method pointers. Arrange to print an error
message and fail registration if a DSA driver does not provide the
phylink MAC operations structure.
Signed-off-by: Russell King (oracle) <rmk+kernel@armlinux.org.uk>
|
|
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Add phylink EEE control methods to allow EEE to be configured. When
LPI is to be disabled, we force the port to have EEE disabled, but
when enabling EEE, if the port is under the control of the PPU, we
stop forcing it, otherwise we force-enable EEE.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Add support for EEE forcing using the MAC control register. Replace
the 88e6393x errata 4.5 EEE disable code with a call to the new EEE
forcing code.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Add a port_set_eee() method to allow the EEE settings for a port to be
configured.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Older XPCS IP requires SGMII_LINK and PHY_SIDE_SGMII to be set when
operating in 1000BASE-X mode even though the XPCS is not configured for
SGMII. An example of a device with older XPCS IP is KSZ9477.
We already don't clear these bits if we switch from SGMII to 1000BASE-X
on TXGBE - which would result in 1000BASE-X with the PHY_SIDE_SGMII bit
left set.
It is currently believed to be safe to set both bits on newer IP
without side-effects.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
|
Older revisions of the XPCS IP do not support the MAC_AUTO_SW flag and
need the BMCR register updated with the speed information from the PHY.
Split the DW_XPCS_SGMII_MODE_MAC mode into _AUTO and _MANUAL variants,
where _AUTO mode means the update happens in hardware autonomously,
whereas the _MANUAL mode means that we need to update the BMCR register
when the link comes up.
This will be required for the older XPCS IP found in KSZ9477.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
|
