Remove an old workaround that is no longer necessary since rS343824.
There used to be a problem with FMan interrupts firing on multiple CPUS
at the same time.
This ended up being due to multicast interrupts being unsupported in the
Freescale PIC (so instead of using a selection algorithm, it would do some
unspecified action, such as interrupting multiple cpus at random.)
Reviewed by: jhibbits
Sponsored by: Tag1 Consulting, Inc.
Differential Revision: https://reviews.freebsd.org/D23829
After the network epoch was added, we lost the ability to migrate the
ithread in the middle of dispatch, as being in the network epoch will pin
the current thread (for safety reasons.)
Luckily, we don't actually have to do this workaround in the first place,
as we can just bind it to the correct cpu when we preallocate it.
Pass dev through to XX_PreallocAndBindIntr() and actually bind it to the
cpu like it was supposed to in the first place, instad of leaving it
floating and moving it to the correct cpu the first time it fires.
This fixes panics while bringing up dtsec on my X5000.
Reviewed by: jhibbits
Sponsored by: Tag1 Consulting, Inc.
Differential Revision: https://reviews.freebsd.org/D23826
switch over to opt-in instead of opt-out for epoch.
Instead of IFF_NEEDSEPOCH, provide IFF_KNOWSEPOCH. If driver marks
itself with IFF_KNOWSEPOCH, then ether_input() would not enter epoch
when processing its packets.
Now this will create recursive entrance in epoch in >90% network
drivers, but will guarantee safeness of the transition.
Mark several tested drivers as IFF_KNOWSEPOCH.
Reviewed by: hselasky, jeff, bz, gallatin
Differential Revision: https://reviews.freebsd.org/D23674
supposedly may call into ether_input() without network epoch.
They all need to be reviewed before 13.0-RELEASE. Some may need
be fixed. The flag is not planned to be used in the kernel for
a long time.
Don't hold the scheduler lock while doing context switches. Instead we
unlock after selecting the new thread and switch within a spinlock
section leaving interrupts and preemption disabled to prevent local
concurrency. This means that mi_switch() is entered with the thread
locked but returns without. This dramatically simplifies scheduler
locking because we will not hold the schedlock while spinning on
blocked lock in switch.
This change has not been made to 4BSD but in principle it would be
more straightforward.
Discussed with: markj
Reviewed by: kib
Tested by: pho
Differential Revision: https://reviews.freebsd.org/D22778
The BMan softc must exist when dtsec devices are created, else a NULL
pointer is dereferenced. QMan likely as well. Until now, we have relied on
order within the fdt parsing to attach correctly, but this obviously is not
foolproof. Mark these as BUS_PASS_SUPPORTDEV so they're probed and attached
explicitly before dtsec devices.
Change OF_getencprop_alloc semantics to be combination of malloc and
OF_getencprop and return size of the property, not number of elements
allocated.
For the use cases where number of elements is preferred introduce
OF_getencprop_alloc_multi helper function that copies semantics
of OF_getencprop_alloc prior to this change.
This is to make OF_getencprop_alloc and OF_getencprop_alloc_multi
function signatures consistent with OF_getencprop_alloc and
OF_getencprop_alloc_multi.
Functionality-wise this patch is mostly rename of OF_getencprop_alloc
to OF_getencprop_alloc_multi except two calls in ofw_bus_setup_iinfo
where 1 was used as a block size.
Official Linux dts's put the individual portals under a simple-bus, rather
than under a '*-portals' grouping. This adds a hack to permit that, which
gets us closer to using stock device trees for DPAA-based devices.
XX_VirtToPhys(), by way of pmap_kextract(), is an expensive operation.
Profiling via dtrace during a series of iperf tests I found 16111 / 30432
stack frames were located in mmu_booke_kextract(), so eliminating this
expensive call should improve performance slightly. XX_PhysToVirt() is not
as expensive, but redundant calls in this context is wasteful.
Some devices (P5040, P4080) have multiple frame managers in their DPAA
subsystems. This was prevented by use of a softc singleton in the DPAA
driver. Since if_dtsec(4) has moved to be a child of fman, it can access
the fman device data via the parent object.
Rework the dTSEC and FMan drivers to be more like a full bus relationship,
so that dtsec can use bus_alloc_resource() instead of trying to handle the
offset from the dts. This required taking some code from the sparc64 ebus
driver to allow subdividing the fman region for the dTSEC devices.
This adds some support for ARM as well as 64-bit. 64-bit on PowerPC is
currently not working, and ARM support has not been completed or tested on the
FreeBSD side.
As this was imported from a Linux tree, it includes some Linux-isms
(ioread/iowrite), so compile with the LinuxKPI for now. This may change in the
future.
All of these arguments are stored in m_ext, so there is no reason
to pass them in the argument list. Not all functions need the second
argument, some don't even need the first one. The second argument
lives in next cache line, so not dereferencing it is a performance
gain. This was discovered in sendfile(2), which will be covered by
next commits.
The second goal of this commit is to bring even more flexibility
to m_ext mbufs, allowing to create more fields in m_ext, opaque to
the generic mbuf code, and potentially set and dereferenced by
subsystems.
Reviewed by: gallatin, kbowling
Differential Revision: https://reviews.freebsd.org/D12615
Drop the tracking down to the pmap layer, with optimizations to only track
necessary pages. This should give a (slight) performance improvement, as well
as a stability improvement, as the tracking is already mostly handled by the
pmap layer.
Linux has a slightly different device tree definition for DPAA than originally
done in the FreeBSD driver. This changes the driver to be mostly compatible
with the Linux device tree definitions. Currently the differences are:
bman-portals: compatible = "fsl,bman-portals" (Linux is "simple-bus")
qman-portals: compatible = "fsl,qman-portals" (Linux is "simple-bus")
fman: compatible = "fsl,fman" (Linux is "simple-bus")
The Linux device tree doesn't specify anything for rgmii in the mdio. This
change still requires the device tree to specify the phy-handle, and doesn't yet
support tbi.
If the device tree doesn't contain a cpu-handle field in any bman-portal or
qman-portal, it will exit without setting up the devinfo, leaving it
uninitialized. This will lead to attempts to free random memory, and ultimately
panic.
Instead replace it with a different hack, that turns fman into a simplebus
subclass, and maps its children within its address space.
Since all PHY communication is done through dtsec0's mdio space, the FDT
contains a reference to the dtsec0 mdio handle in all nodes that need it.
Instead of using Freescale's implementation for MII access, use our own (copied
loosely from the eTSEC driver, and could possibly be merged eventually). This
lets us access the registers directly rather than needing a full dtsec interface
just to access the registers.
Future directions will include turning fman into more of a simplebus, and not
mapping the region and playing games. This will require changes to the dtsec
driver to make it a child of fman, and possibly other drivers as well.
As part of an effort to extend Book-E to the 64-bit world, make the necessary
changes to the DPAA/dTSEC driver set to be integer-pointer conversion clean.
This means no more casts to int, and use uintptr_t where needed.
Since the NCSW source is effectively obsolete, direct changes to the source tree
are safe.
This was found while reworking the device tree nodes for dtsec to match the
Linux device tree. Instead of waiting and expecting later code to call
dpaa_portal_map_registers(), do the equivalent immediately upon mapping.
Otherwise, it's possible to access the pages before that function is called, and
hang the CPU.
BMan and QMan will do this at attach time. Even though the registers are mapped
now, dpaa_portal_map_registers() will be called at BMan and QMan attach time,
updating the mappings to be private, and in the case of cache-enabled registers,
marked as coherent memory mappings.
Summary:
PowerPC Book-E SMP is currently broken for unknown reasons. Pull in
Semihalf changes made c2012 for e500mc/e5500, which enables booting SMP.
This eliminates the shared software TLB1 table, replacing it with
tlb1_read_entry() function.
This does not yet support ePAPR SMP booting, and doesn't handle resetting CPUs
already released (ePAPR boot releases APs to a spin loop waiting on a specific
address). This will be addressed in the near future by using the MPIC to reset
the AP into our own alternate boot address.
This does include a change to the dpaa/dtsec(4) driver, to mark the portals as
CPU-private.
Test Plan:
Tested on Amiga X5000/20 (P5020). Boots, prints the following
messages:
Adding CPU 0, pir=0, awake=1
Waking up CPU 1 (dev=1)
Adding CPU 1, pir=20, awake=1
SMP: AP CPU #1 launched
top(1) shows CPU1 active.
Obtained from: Semihalf
Relnotes: Yes
Differential Revision: https://reviews.freebsd.org/D5945
Freescale's QorIQ line includes a new ethernet controller, based on their
Datapath Acceleration Architecture (DPAA). This uses a combination of a Frame
manager, Buffer manager, and Queue manager to improve performance across all
interfaces by being able to pass data directly between hardware acceleration
interfaces.
As part of this import, Freescale's Netcomm Software (ncsw) driver is imported.
This was an attempt by Freescale to create an OS-agnostic sub-driver for
managing the hardware, using shims to interface to the OS-specific APIs. This
work was abandoned, and Freescale's primary work is in the Linux driver (dual
BSD/GPL license). Hence, this was imported directly to sys/contrib, rather than
going through the vendor area. Going forward, FreeBSD-specific changes may be
made to the ncsw code, diverging from the upstream in potentially incompatible
ways. An alternative could be to import the Linux driver itself, using the
linuxKPI layer, as that would maintain parity with the vendor-maintained driver.
However, the Linux driver has not been evaluated for reliability yet, and may
have issues with the import, whereas the ncsw-based driver in this commit was
completed by Semihalf 4 years ago, and is very stable.
Other SoC modules based on DPAA, which could be added in the future:
* Security and Encryption engine (SEC4.x, SEC5.x)
* RAID engine
Additional work to be done:
* Implement polling mode
* Test vlan support
* Add support for the Pattern Matching Engine, which can do regular expression
matching on packets.
This driver has been tested on the P5020 QorIQ SoC. Others listed in the
dtsec(4) manual page are expected to work as the same DPAA engine is included in
all.
Obtained from: Semihalf
Relnotes: Yes
Sponsored by: Alex Perez/Inertial Computing
