As of commit 98fa6271cf
("tcp: refactor setting the initial congestion window") this is called
only from tcp_input.c, so it can be static.
Signed-off-by: Florian Westphal <fw@strlen.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/mlx5: Add ability to read and write ECE options
net/mlx5: Add support for RDMA TX FT headers modifying
net/mlx5: Move iseg access helper routines close to mlx5_core driver
net/mlx5: Cleanup mlx5_ifc_fte_match_set_misc2_bits
net/mlx5: Add support in forward to namespace
{IB/net}/mlx5: Simplify don't trap code
net/mlx5: Replace zero-length array with flexible-array
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Pull rdma fixes from Jason Gunthorpe:
"Nothing profound here, just a last set of long standing bug fixes:
- Incorrect error unwind in qib and pvrdma
- User triggerable NULL pointer crash in mlx5 with ODP prefetch
- syzkaller RCU race in uverbs
- Rare double free crash in ipoib"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
IB/ipoib: Fix double free of skb in case of multicast traffic in CM mode
RDMA/core: Fix double destruction of uobject
RDMA/pvrdma: Fix missing pci disable in pvrdma_pci_probe()
RDMA/mlx5: Fix NULL pointer dereference in destroy_prefetch_work
IB/qib: Call kobject_put() when kobject_init_and_add() fails
now that can be done conveniently - all non-trivial cases have
_HAVE_ARCH_COPY_AND_CSUM_FROM_USER defined, so the fallback in
net/checksum.h is used only for dummy (copy_from_user, then
csum_partial) implementation. Allowing us to get rid of all
dummy instances, both of csum_and_copy_from_user() and
csum_partial_copy_from_user().
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
The SCTP protocol allows to bind multiple address to a socket. That
feature is currently only exposed as a socket option. Add a bind_add
method struct proto that allows to bind additional addresses, and
switch the dlm code to use the method instead of going through the
socket option from kernel space.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a helper to directly set the SCTP_NODELAY sockopt from kernel space
without going through a fake uaccess.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Steffen Klassert says:
====================
pull request (net): ipsec 2020-05-29
1) Several fixes for ESP gro/gso in transport and beet mode when
IPv6 extension headers are present. From Xin Long.
2) Fix a wrong comment on XFRMA_OFFLOAD_DEV.
From Antony Antony.
3) Fix sk_destruct callback handling on ESP in TCP encapsulation.
From Sabrina Dubroca.
4) Fix a use after free in xfrm_output_gso when used with vxlan.
From Xin Long.
5) Fix secpath handling of VTI when used wiuth IPCOMP.
From Xin Long.
6) Fix an oops when deleting a x-netns xfrm interface.
From Nicolas Dichtel.
7) Fix a possible warning on policy updates. We had a case where it was
possible to add two policies with the same lookup keys.
From Xin Long.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Steffen Klassert says:
====================
pull request (net-next): ipsec-next 2020-05-29
1) Add IPv6 encapsulation support for ESP over UDP and TCP.
From Sabrina Dubroca.
2) Remove unneeded reference when initializing xfrm interfaces.
From Nicolas Dichtel.
3) Remove some indirect calls from the state_afinfo.
From Florian Westphal.
Please note that this pull request has two merge conflicts
between commit:
0c922a4850 ("xfrm: Always set XFRM_TRANSFORMED in xfrm{4,6}_output_finish")
from Linus' tree and commit:
2ab6096db2 ("xfrm: remove output_finish indirection from xfrm_state_afinfo")
from the ipsec-next tree.
and between commit:
3986912f6a ("ipv6: move SIOCADDRT and SIOCDELRT handling into ->compat_ioctl")
from the net-next tree and commit:
0146dca70b ("xfrm: add support for UDPv6 encapsulation of ESP")
from the ipsec-next tree.
Both conflicts can be resolved as done in linux-next.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Allow a ULP to ask the core to provide a completion queue based on a
least-used search on a per-device CQ pools. The device CQ pools grow in a
lazy fashion when more CQs are requested.
This feature reduces the amount of interrupts when using many QPs. Using
shared CQs allows for more effcient completion handling. It also reduces
the amount of overhead needed for CQ contexts.
Test setup:
Intel(R) Xeon(R) Platinum 8176M CPU @ 2.10GHz servers.
Running NVMeoF 4KB read IOs over ConnectX-5EX across Spectrum switch.
TX-depth = 32. The patch was applied in the nvme driver on both the target
and initiator. Four controllers are accessed from each core. In the
current test case we have exposed sixteen NVMe namespaces using four
different subsystems (four namespaces per subsystem) from one NVM port.
Each controller allocated X queues (RDMA QPs) and attached to Y CQs.
Before this series we had X == Y, i.e for four controllers we've created
total of 4X QPs and 4X CQs. In the shared case, we've created 4X QPs and
only X CQs which means that we have four controllers that share a
completion queue per core. Until fourteen cores there is no significant
change in performance and the number of interrupts per second is less than
a million in the current case.
==================================================
|Cores|Current KIOPs |Shared KIOPs |improvement|
|-----|---------------|--------------|-----------|
|14 |2332 |2723 |16.7% |
|-----|---------------|--------------|-----------|
|20 |2086 |2712 |30% |
|-----|---------------|--------------|-----------|
|28 |1971 |2669 |35.4% |
|=================================================
|Cores|Current avg lat|Shared avg lat|improvement|
|-----|---------------|--------------|-----------|
|14 |767us |657us |14.3% |
|-----|---------------|--------------|-----------|
|20 |1225us |943us |23% |
|-----|---------------|--------------|-----------|
|28 |1816us |1341us |26.1% |
========================================================
|Cores|Current interrupts|Shared interrupts|improvement|
|-----|------------------|-----------------|-----------|
|14 |1.6M/sec |0.4M/sec |72% |
|-----|------------------|-----------------|-----------|
|20 |2.8M/sec |0.6M/sec |72.4% |
|-----|------------------|-----------------|-----------|
|28 |2.9M/sec |0.8M/sec |63.4% |
====================================================================
|Cores|Current 99.99th PCTL lat|Shared 99.99th PCTL lat|improvement|
|-----|------------------------|-----------------------|-----------|
|14 |67ms |6ms |90.9% |
|-----|------------------------|-----------------------|-----------|
|20 |5ms |6ms |-10% |
|-----|------------------------|-----------------------|-----------|
|28 |8.7ms |6ms |25.9% |
|===================================================================
Performance improvement with sixteen disks (sixteen CQs per core) is
comparable.
Link: https://lore.kernel.org/r/1590568495-101621-3-git-send-email-yaminf@mellanox.com
Signed-off-by: Yamin Friedman <yaminf@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Reviewed-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Most of blk-mq drivers depend on managed IRQ's auto-affinity to setup
up queue mapping. Thomas mentioned the following point[1]:
"That was the constraint of managed interrupts from the very beginning:
The driver/subsystem has to quiesce the interrupt line and the associated
queue _before_ it gets shutdown in CPU unplug and not fiddle with it
until it's restarted by the core when the CPU is plugged in again."
However, current blk-mq implementation doesn't quiesce hw queue before
the last CPU in the hctx is shutdown. Even worse, CPUHP_BLK_MQ_DEAD is a
cpuhp state handled after the CPU is down, so there isn't any chance to
quiesce the hctx before shutting down the CPU.
Add new CPUHP_AP_BLK_MQ_ONLINE state to stop allocating from blk-mq hctxs
where the last CPU goes away, and wait for completion of in-flight
requests. This guarantees that there is no inflight I/O before shutting
down the managed IRQ.
Add a BLK_MQ_F_STACKING and set it for dm-rq and loop, so we don't need
to wait for completion of in-flight requests from these drivers to avoid
a potential dead-lock. It is safe to do this for stacking drivers as those
do not use interrupts at all and their I/O completions are triggered by
underlying devices I/O completion.
[1] https://lore.kernel.org/linux-block/alpine.DEB.2.21.1904051331270.1802@nanos.tec.linutronix.de/
[hch: different retry mechanism, merged two patches, minor cleanups]
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Daniel Wagner <dwagner@suse.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Drivers may need to bypass error injection for error recovery. Rename
__blk_mq_complete_request() to blk_mq_force_complete_rq() and export
that function so drivers may skip potential fake timeouts after they've
reclaimed lost requests.
Signed-off-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Daniel Wagner <dwagner@suse.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The name of pm_runtime_callbacks_present() is confusing, because
it suggests that the device has PM-runtime callbacks if 'true' is
returned by that function, but in fact that may not be the case,
so replace it with pm_runtime_has_no_callbacks() which is not
ambiguous.
No functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Hi!
This patchset is another attempt to fix the regulator coupling on
Exynos5800/5422 SoCs. Here are links to the previous attempts:
https://lore.kernel.org/linux-samsung-soc/20191008101709.qVNy8eijBi0LynOteWFMnTg4GUwKG599n6OyYoX1Abs@z/https://lore.kernel.org/lkml/20191017102758.8104-1-m.szyprowski@samsung.com/https://lore.kernel.org/linux-pm/cover.1589528491.git.viresh.kumar@linaro.org/https://lore.kernel.org/linux-pm/20200528131130.17984-1-m.szyprowski@samsung.com/
The problem is with "vdd_int" regulator coupled with "vdd_arm" on Odroid
XU3/XU4 boards family. "vdd_arm" is handled by CPUfreq. "vdd_int" is
handled by devfreq. CPUfreq initialized quite early during boot and it
starts changing OPPs and "vdd_arm" value. Sometimes CPU activity during
boot goes down and some low-frequency OPPs are selected, what in turn
causes lowering "vdd_arm". This happens before devfreq applies its
requirements on "vdd_int". Regulator balancing code reduces "vdd_arm"
voltage value, what in turn causes lowering "vdd_int" value to the lowest
possible value. This is much below the operation point of the wcore bus,
which still runs at the highest frequency.
The issue was hard to notice because in the most cases the board managed
to boot properly, even when the regulator was set to lowest value allowed
by the regulator constraints. However, it caused some random issues,
which can be observed as "Unhandled prefetch abort" or low USB stability.
Adding more and more special cases to the generic code has been rejected,
so the only way to ensure the desired behavior on Exynos5800-based SoCs
is to make a custom regulator coupler driver.
Best regards,
Marek Szyprowski
Patch summary:
Marek Szyprowski (2):
regulator: extract voltage balancing code to separate function
soc: samsung: Add simple voltage coupler for Exynos5800
arch/arm/mach-exynos/Kconfig | 1 +
drivers/regulator/core.c | 49 ++++++++-------
drivers/soc/samsung/Kconfig | 3 +
drivers/soc/samsung/Makefile | 1 +
.../soc/samsung/exynos-regulator-coupler.c | 59 +++++++++++++++++++
include/linux/regulator/coupler.h | 8 +++
6 files changed, 101 insertions(+), 20 deletions(-)
create mode 100644 drivers/soc/samsung/exynos-regulator-coupler.c
--
2.17.1
base-commit: 8f3d9f3542
Sub-devices of a real DMA device might exist on a separate segment than
the real DMA device and its IOMMU. These devices should still have a
valid device_domain_info, but the current dma alias model won't
allocate info for the subdevice.
This patch adds a segment member to struct device_domain_info and uses
the sub-device's BDF so that these sub-devices won't alias to other
devices.
Fixes: 2b0140c696 ("iommu/vt-d: Use pci_real_dma_dev() for mapping")
Cc: stable@vger.kernel.org # v5.6+
Signed-off-by: Jon Derrick <jonathan.derrick@intel.com>
Acked-by: Lu Baolu <baolu.lu@linux.intel.com>
Link: https://lore.kernel.org/r/20200527165617.297470-3-jonathan.derrick@intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Bartosz Golaszewski <bgolaszewski@baylibre.com>:
From: Bartosz Golaszewski <bgolaszewski@baylibre.com>
I noticed that oftentimes I use regmap_update_bits() for simple bit
setting or clearing. In this case the fourth argument is superfluous as
it's always 0 or equal to the mask argument.
This series proposes to add simple bit operations for setting, clearing
and testing specific bits with regmap.
The second patch uses all three in a driver that got recently picked into
the net-next tree.
The patches obviously target different trees so - if you're ok with
the change itself - I propose you pick the first one into your regmap
tree for v5.8 and then I'll resend the second patch to add the first
user for these macros for v5.9.
v1 -> v2:
- convert the new macros to static inline functions
v2 -> v3:
- drop unneeded ternary operator
Bartosz Golaszewski (2):
regmap: provide helpers for simple bit operations
net: ethernet: mtk-star-emac: use regmap bitops
drivers/base/regmap/regmap.c | 22 +++++
drivers/net/ethernet/mediatek/mtk_star_emac.c | 80 ++++++++-----------
include/linux/regmap.h | 36 +++++++++
3 files changed, 93 insertions(+), 45 deletions(-)
base-commit: 8f3d9f3542
--
2.26.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.orghttp://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Looking at the Felix and Ocelot drivers, Maxim asked if it would be
possible to use them as a base for a new driver for the switch inside
NXP T1040. Turns out, it is! The result is a driver eerily similar to
Felix.
The biggest challenge seems to be getting register read/write API
generic enough to cover such wild bitfield variations between hardware
generations. There is a patch on the regmap core which I would like to
get in through the networking subsystem, if possible (and if Mark is
ok), since it's a trivial addition.
Maxim Kochetkov (4):
soc/mscc: ocelot: add MII registers description
net: mscc: ocelot: convert SYS_PAUSE_CFG register access to regfield
net: mscc: ocelot: extend watermark encoding function
net: dsa: ocelot: introduce driver for Seville VSC9953 switch
Vladimir Oltean (7):
regmap: add helper for per-port regfield initialization
net: mscc: ocelot: unexport ocelot_probe_port
net: mscc: ocelot: convert port registers to regmap
net: mscc: ocelot: convert QSYS_SWITCH_PORT_MODE and SYS_PORT_MODE to
regfields
net: dsa: ocelot: create a template for the DSA tags on xmit
net: mscc: ocelot: split writes to pause frame enable bit and to
thresholds
net: mscc: ocelot: disable flow control on NPI interface
drivers/net/dsa/ocelot/Kconfig | 12 +
drivers/net/dsa/ocelot/Makefile | 6 +
drivers/net/dsa/ocelot/felix.c | 49 +-
drivers/net/dsa/ocelot/felix_vsc9959.c | 72 +-
drivers/net/dsa/ocelot/seville.c | 742 +++++++++++++++
drivers/net/dsa/ocelot/seville.h | 50 +
drivers/net/dsa/ocelot/seville_vsc9953.c | 1064 ++++++++++++++++++++++
drivers/net/ethernet/mscc/ocelot.c | 87 +-
drivers/net/ethernet/mscc/ocelot.h | 9 +-
drivers/net/ethernet/mscc/ocelot_board.c | 21 +-
drivers/net/ethernet/mscc/ocelot_io.c | 18 +-
drivers/net/ethernet/mscc/ocelot_regs.c | 57 ++
include/linux/regmap.h | 8 +
include/soc/mscc/ocelot.h | 68 +-
include/soc/mscc/ocelot_dev.h | 78 --
include/soc/mscc/ocelot_qsys.h | 13 -
include/soc/mscc/ocelot_sys.h | 23 -
net/dsa/tag_ocelot.c | 21 +-
18 files changed, 2196 insertions(+), 202 deletions(-)
create mode 100644 drivers/net/dsa/ocelot/seville.c
create mode 100644 drivers/net/dsa/ocelot/seville.h
create mode 100644 drivers/net/dsa/ocelot/seville_vsc9953.c
base-commit: 8f3d9f3542
--
2.25.1
In many instances regmap_update_bits() is used for simple bit setting
and clearing. In these cases the last argument is redundant and we can
hide it with a static inline function.
This adds three new helpers for simple bit operations: set_bits,
clear_bits and test_bits (the last one defined as a regular function).
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Link: https://lore.kernel.org/r/20200528154503.26304-2-brgl@bgdev.pl
Signed-off-by: Mark Brown <broonie@kernel.org>
After binding a device to an mm, device drivers currently need to
register a mm_exit handler. This function is called when the mm exits,
to gracefully stop DMA targeting the address space and flush page faults
to the IOMMU.
This is deemed too complex for the MMU release() notifier, which may be
triggered by any mmput() invocation, from about 120 callsites [1]. The
upcoming SVA module has an example of such complexity: the I/O Page
Fault handler would need to call mmput_async() instead of mmput() after
handling an IOPF, to avoid triggering the release() notifier which would
in turn drain the IOPF queue and lock up.
Another concern is the DMA stop function taking too long, up to several
minutes [2]. For some mmput() callers this may disturb other users. For
example, if the OOM killer picks the mm bound to a device as the victim
and that mm's memory is locked, if the release() takes too long, it
might choose additional innocent victims to kill.
To simplify the MMU release notifier, don't forward the notification to
device drivers. Since they don't stop DMA on mm exit anymore, the PASID
lifetime is extended:
(1) The device driver calls bind(). A PASID is allocated.
Here any DMA fault is handled by mm, and on error we don't print
anything to dmesg. Userspace can easily trigger errors by issuing DMA
on unmapped buffers.
(2) exit_mmap(), for example the process took a SIGKILL. This step
doesn't happen during normal operations. Remove the pgd from the
PASID table, since the page tables are about to be freed. Invalidate
the IOTLBs.
Here the device may still perform DMA on the address space. Incoming
transactions are aborted but faults aren't printed out. ATS
Translation Requests return Successful Translation Completions with
R=W=0. PRI Page Requests return with Invalid Request.
(3) The device driver stops DMA, possibly following release of a fd, and
calls unbind(). PASID table is cleared, IOTLB invalidated if
necessary. The page fault queues are drained, and the PASID is
freed.
If DMA for that PASID is still running here, something went seriously
wrong and errors should be reported.
For now remove iommu_sva_ops entirely. We might need to re-introduce
them at some point, for example to notify device drivers of unhandled
IOPF.
[1] https://lore.kernel.org/linux-iommu/20200306174239.GM31668@ziepe.ca/
[2] https://lore.kernel.org/linux-iommu/4d68da96-0ad5-b412-5987-2f7a6aa796c3@amd.com/
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Acked-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
Acked-by: Lu Baolu <baolu.lu@linux.intel.com>
Link: https://lore.kernel.org/r/20200423125329.782066-3-jean-philippe@linaro.org
Signed-off-by: Joerg Roedel <jroedel@suse.de>
In simple cases we can instantiate SPD EEPROMs on the SMBus
automatically. Start with just DDR2, DDR3 and DDR4 on x86 for now,
and only for systems with no more than 4 memory slots. These
limitations may be lifted later.
Signed-off-by: Jean Delvare <jdelvare@suse.de>
[wsa: minor change for new API]
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Commit e8759ad17d ("serial: uapi: Add support for bus termination")
introduced the ability to enable rs485 bus termination from user space.
So far the feature is only used by a single driver, 8250_exar.c, using a
hardcoded GPIO pin specific to Siemens IOT2040 products.
Provide for a more generic solution by allowing specification of an
rs485 bus termination GPIO pin in the device tree: Amend the serial
core to retrieve the GPIO from the device tree (or ACPI table) and amend
the default ->rs485_config() callback for 8250 drivers to change the
GPIO on request from user space.
Perhaps 8250_exar.c can be converted to the generic approach in a
follow-up patch.
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Jan Kiszka <jan.kiszka@siemens.com>
Link: https://lore.kernel.org/r/94c6c800d1ca9fa04766dd1d43a8272c5ad4bedd.1589811297.git.lukas@wunner.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
All macro names for SDIO device IDs are prefixed by vendor name to which
device ID belongs. So for consistency add Broadcom string vendor prefix to
all Cypress macro names as they belong to SDIO Broadcom vendor ID.
Change also Cypress 43012 value from decimal do hexadecimal notation to be
consistent with all other values.
Signed-off-by: Pali Rohár <pali@kernel.org>
Link: https://lore.kernel.org/r/20200522144412.19712-11-pali@kernel.org
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
