Move the source files out of staging into their final locations:
-mc.h include file in drivers/staging/fsl-mc/include go to include/linux/fsl
-source files in drivers/staging/fsl-mc/bus go to drivers/bus/fsl-mc
-overview.rst, providing an overview of DPAA2, goes to
Documentation/networking/dpaa2/overview.rst
Update or delete other remaining staging files -- Makefile, Kconfig, TODO.
Update dpaa2_eth and dpio staging drivers.
Add integration bits for the documentation build system.
Signed-off-by: Stuart Yoder <stuyoder@gmail.com>
[rebased, add dpaa2_eth and dpio #include updates]
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
[rebased, split irqchip to separate patch]
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use the newly added DPIO service API to map cpu-affine DPIO services
to channels.
The DPAA2 Ethernet driver already had mappings of frame queues and
channels to cpus, but had no control over the DPIOs used. We can
now ensure full affinity of hotpath hardware resources to cores,
which improves performance and almost eliminates some resource
contentions (e.g. enqueue/dequeue busy counters should be close to
zero from now on).
Making the pull channel operation core affine brings the most
significant benefits. This ensures the same DPIO service will be
used for all dequeue commands issued for a certain frame queue,
which is in line with the way hardware is optimized.
Additionally, we also use affine DPIOs for the frame enqueue and
buffer release operations in order to avoid resource contention.
dpaa2_io_service_register() and dpaa2_io_service_rearm()
functions receive an affine DPIO as argument mostly for uniformity,
but this doesn't change the previous functionality.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The previous patch ensures Tx flow affinity for forwarded frames,
but for termination traffic the initial flow affinity is determined
based on the skb hash, which is expected to hit only a few Tx queues
when there is a small number of flows.
Instead, use XPS (transmit packet steering) to set netdevice queue
affinity to the sending core.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The driver xmit function chooses an egress FQ based on the current
core id. The network stack itself sets a mapping field in the skb
based on many things - the default one being a hash on packet fields,
which the current driver ignores.
This patch saves the ingress frame flow affinity information in the
skb. In case of forwarded frames, this info will then be used for Tx
and Tx confirmation hardware queue selection, ensuring all processing
of the given frame is done on a single core.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Aligning the Tx buffers at 64B is a performance optimization
recommendation, not a hard requirement.
Make optional the alignment of Tx FD buffers, without enforcing
a reallocation in case there is not enough headroom for it.
On Rx, we keep allocating buffers with enough headroom to allow
Tx alignment of forwarded frames.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For non-linear skbs we build scatter-gather frames and allocate
a new buffer for the S/G table in which we reserve the required
headroom, so the actual skb headroom size doesn't matter.
Rather than use a one-size-fits-all approach, decide when to
enforce headroom requirements on a frame by frame basis.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For Tx confirmed frames that have an error indication in the frame
descriptor, we look at the Frame Annotation Status field (in the
buffer headroom) for details on the root cause and then print
a debug message with that information.
While useful in initial development stages, it doesn't bring
enough added value to justify reserving 64B of headroom for all
Tx frames (FAS is only 8B long, but we must reserve chunks of 64B
from the hardware annotation area).
If we remove the need for FAS field from egress frames, we can
renounce hardware annotation completely, since FAS is the only
HWA field we currently use.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add a counter for the number of egress frames that need to be
realloc'ed due to insufficient headroom space.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 4b2d9fe879 ("staging: fsl-dpaa2/eth: Extra headroom in RX
buffers") tried to avoid the performance penalty of doing skb
reallocations in the network stack for IP forwarded frames between
two DPAA2 Ethernet interfaces. This led to a (too) complicated
formula that relies on the stack's internal implementation.
Instead, it's safer and easier to just not request any guarantee
from the stack. We already double check in the driver the required
headroom size of egress frames and realloc the skb if needed, so
we don't need to add any extra code.
On forwarding between two of our own interfaces, there is no
functional change; for traffic forwarded from a different device or
generated on the core, skb realloc operations are moved from the stack
to our driver, with no visible impact on performance.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 4b2d9fe879 ("staging: fsl-dpaa2/eth: Extra headroom in RX
buffers") removes the software annotation (SWA) area from the RX
buffer layout, as it's not used by anyone, but fails to update the
macros for accessing hardware annotation (HWA) fields, which is
right after the SWA in the buffer headroom.
This may lead to some frame annotation status fields (e.g. indication
if L3/L4 checksum is valid) to be read incorrectly.
Turn the accessor macros into inline functions and add a bool param
to specify if SWA is present or not.
Fixes: 4b2d9fe879 ("staging: fsl-dpaa2/eth: Extra headroom in RX buffers")
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The needed headroom that we ask the stack to reserve for us in TX
skbs is larger than the headroom available in RX frames, which
leads to skb reallocations in forwarding scenarios involving two
DPNI interfaces.
Configure the hardware to reserve some extra space in the RX
frame headroom to avoid this situation. The value is chosen based
on the Tx frame data offset, the Rx buffer alignment value and the
netdevice required headroom.
The network stack will take care to reserve space for HH_DATA_MOD when
building the skb, so there's no need to account for it in the netdevice
needed headroom.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The WRIOP hardware block v1.0.0 (found on LS2080A board)
requires data in RX buffers to be aligned to 256B, but
newer revisions (e.g. on LS2088A, LS1088A) only require
64B alignment.
Check WRIOP version and decide at runtime which alignment
requirement to configure for ingress buffers.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When configuring the Tx buffer layout, the software annotation size is
mentioned, and MC accounts for it when configuring the frame
tx_data_offset. No need to handle it in the driver as well.
This results in 64B less memory allocated per frame.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since setup_dpni() became a bit too long, move the buffer layout
configuration to a separate function.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Clean up goto labels in a couple of functions, by
removing/renaming redundant ones.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Include firmware version in the driver information exported
through ethtool.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Early during probe the netdevice name is not initialized yet,
so use dev_err instead of netdev_err when printing error
messages.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
dpni_get_irq_status() also looks at the input value of its
status parameter, and if not null it automatically clears
from pending state the bits that are set there.
Use this feature to avoid a separate MC command for clearing
the interrupt event bits after reading the status.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In case we fail to allocate a skb for a fragmented ingress
frame, the cleanup function will attempt to unmap again the
first frame fragment, which had already been unmapped during
early Rx processing.
Avoid this by freeing the first buffer immediately in case
we hit an error, leaving the cleanup function to free only
the subsequent buffers.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Take into consideration the return value of napi_complete_done(),
since there might be an indication that it's not suitable to
enable driver interrupts yet.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In case dpaa2_io_service_rearm() fails with an error other then
EBUSY, it will do so silently; add a check for this and a warning
message, as a failure here means we're unable to receive any
more traffic on the current cpu.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For scatter-gather ingress frames, we expect to receive
a list of fragments from the hardware, last of which is
marked with a "final" bit.
Add a check to make sure the Rx frame has this bit set
correctly; there's not much we can do in case of a
malformed frame, but at least issue a warning.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On Rx path, if we fail to build an skb from the incoming FD,
we still need to update the channel buffer count accordingly,
otherwise we risk depleting the pool while the software counter
still sees available buffers.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We incorrectly assumed that dpaa2_io_release() can only
return -EBUSY as an error code, when in fact it can also
fail in case some of its arguments don't have valid values.
Make sure we only retry the operation while the portal is
busy and abort for all other error cases, otherwise we risk
entering an endless loop.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix the values of DPAA2_FD_CTRL_FSE and DPAA2_FD_CTRL_FAERR,
which are shifted off by one bit.
Fixes: 39163c0ce0 ("staging: fsl-dpaa2/eth: Errors checking update")
Cc: <stable@vger.kernel.org>
Signed-off-by: Horia Geantă <horia.geanta@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fixed alignment so that it matched open parenthesis
Signed-off-by: Srishti Sharma <srishtishar@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Once a Tx frame descriptor is enqueued, an interrupt might be triggered
to process the Tx confirmation and free the skb, hitting a memory use
after free when updating the tx_bytes statistic based on skb->len.
Use the frame descriptor length instead.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add an alternate dependency on COMPILE_TEST, thus leaving
this driver compile on other architectures.
Also, other drivers depending on the bus are updated to depend
on ARCH_LAYERSCAPE until they'll also be made multi-arch.
This was compiled tested on:
- booke powerpc (corenet{32,64}_smp_defconfig)
- x86 (i386_defconfig, x86_64_defconfig, needs CONFIG_OF)
- arm64 (defconfig)
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
All possible values of the switch statement are explicitly
handled, so there's no need to have a default branch.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Acked-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
mc-cmd.h contains some low level functions used to encode and decode
commands to the MC. They are used by the drivers so move them to the
public headers and get rid of the mc-cmd.h header.
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
mc-sys.h contains the API to send commands to the MC and is used
by drivers. Move it to the public headers and get rid of the mc-sys.h
header.
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This reverts commit 7d6e221d73.
The whole series is broken, so back it all out.
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Cc: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This reverts commit 9b1aa45539.
The whole series is broken, so back it all out.
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Cc: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
mc-cmd.h contains some low level functions used to encode and decode
commands to the MC. They are used by the drivers so move them to the
public headers and get rid of the mc-cmd.h header.
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
mc-sys.h contains the API to send commands to the MC and is used
by drivers. Move it to the public headers and get rid of the mc-sys.h
header.
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
dpni.c is using byte order macros and error codes but does
not explicitly include the required kernel header, so add it.
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Acked-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently, the netdevice is allocated with a default number of Rx/Tx
queues equal to CONFIG_NR_CPUS, meaning the maximum number of cores
supported by the current kernel. The actual number of queues is
reflected by the DPNI object attribute, so update the netdevice
configuration based on that.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The driver logic for allocating a MAC address to a net device
is complicated enough to deserve a function of its own. While
here, cleanup a bit the code comments.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On the egress path, frame errors are reported using both a FD control
field and the frame annotation status. The current code only handles
FAS errors. Update to look at both fields when accounting Tx errors.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@nxp.com>
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We only need to know the buffer pool id, so save exactly
that in the device's private structure, instead of the
entire DPBP attributes struct.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Remove the dpni_id and buffer_layout fields from device's
private structure. They're only used at probe so we don't
need to store them for further use.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Introduce a helper macro for accessing the frame annotation
status field in a frame buffer.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add a label to the ethtool statistics counters, to differentiate
between hardware counters and driver specific ones.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the Ethernet driver doesn't find any DPIO devices during probe,
it may be either because there's none available or because they
haven't been probed yet. Request deferred probing in case it's
the latter.
Signed-off-by: Bharat Bhushan <Bharat.Bhushan@nxp.com>
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reset the buffer pool object before using it, like we do
for the other DPAA2 objects.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The function itself checks whether GRO support is enabled
and acts accordingly, so we don't need to verify it in the
driver as well.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Don't use GFP_DMA when allocating memory for the hash key,
as we don't actually need to allocate from the lowest zone.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We already have a variable for the DMA mapping device,
so use that directly.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>