This file already uses sysfs_emit(). So be consistent and also use
sysfs_emit_at().
Moreover, size is always < PAGE_SIZE because scnprintf() (and now
sysfs_emit_at()) returns the number of characters written not including the
trailing '\0'. So some tests can be removed.
This slightly simplifies the code and makes it more readable.
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Helge Deller <deller@gmx.de>
Use sysfs_emit_at() instead of snprintf() + custom logic.
Using sysfs_emit_at() is much more simple.
Also, sysfs_emit() is already used in this function, so using
sysfs_emit_at() is more consistent.
Also simplify the logic:
- always add a space after an entry
- change the last space into a '\n'
Finally it is easy to see that, given the size of cabc_modes, PAGE_SIZE
can not be reached.
So better keep everything simple (and correct).
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Signed-off-by: Helge Deller <deller@gmx.de>
The devm_clk_get_enabled() helpers:
- call devm_clk_get()
- call clk_prepare_enable() and register what is needed in order to
call clk_disable_unprepare() when needed, as a managed resource.
This simplifies the code and avoids the calls to clk_disable_unprepare().
Signed-off-by: ying zuxin <yingzuxin@vivo.com>
Signed-off-by: Helge Deller <deller@gmx.de>
If an error occurs after request_mem_region(), a corresponding
release_mem_region() should be called, as already done in the remove
function.
Fixes: 1da177e4c3 ("Linux-2.6.12-rc2")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Helge Deller <deller@gmx.de>
There's only one instance of hwmon per device, and MMIO access to it is
always done through the root tile. The code has been passing around a
pointer to the root tile's primary GT, which is confusing since this
isn't really a GT-level concept. Replace that pointer with an xe_device
pointer and use xe_root_mmio_gt(xe) to get a pointer when we need to do
register MMIO. This makes things easier to follow, and also cleans up
the code in preparation for a much larger MMIO register access overhaul
that's coming soon.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240829220619.789159-6-matthew.d.roper@intel.com
There's only one instance of the pcode per tile, and for GT-related
accesses both the primary and media GT share the same register
interface. Since Xe was using per-GT locking, the pcode mutex wasn't
actually protecting everything that it should since concurrent accesses
related to a tile's primary GT and media GT were possible.
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240829220619.789159-5-matthew.d.roper@intel.com
When building with clang, there's this warning:
vdso_test_getrandom.c:145:40: warning: omitting the parameter name in a function definition is a C23 extension [-Wc23-extensions]
145 | static void *test_vdso_getrandom(void *)
| ^
vdso_test_getrandom.c:155:40: warning: omitting the parameter name in a function definition is a C23 extension [-Wc23-extensions]
155 | static void *test_libc_getrandom(void *)
| ^
vdso_test_getrandom.c:165:43: warning: omitting the parameter name in a function definition is a C23 extension [-Wc23-extensions]
165 | static void *test_syscall_getrandom(void *)
Add the named ctx parameter to quash it.
Reported-by: Mark Brown <broonie@kernel.org>
Reviewed-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
According to the documentation, when building a kernel with the C=2
parameter, all source files should be checked. But this does not happen
for the kernel/bpf/ directory.
$ touch kernel/bpf/core.o
$ make C=2 CHECK=true kernel/bpf/core.o
Outputs:
CHECK scripts/mod/empty.c
CALL scripts/checksyscalls.sh
DESCEND objtool
INSTALL libsubcmd_headers
CC kernel/bpf/core.o
As can be seen the compilation is done, but CHECK is not executed. This
happens because kernel/bpf/Makefile has defined its own rule for
compilation and forgotten the macro that does the check.
There is no need to duplicate the build code, and this rule can be
removed to use generic rules.
Acked-by: Masahiro Yamada <masahiroy@kernel.org>
Tested-by: Oleg Nesterov <oleg@redhat.com>
Tested-by: Alan Maguire <alan.maguire@oracle.com>
Signed-off-by: Alexey Gladkov <legion@kernel.org>
Link: https://lore.kernel.org/r/20240830074350.211308-1-legion@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add PCIe phy bindings for the ZCU102.
Link: https://lore.kernel.org/r/20240531161337.864994-8-sean.anderson@linux.dev
Tested-by: Thippeswamy Havalige <thippeswamy.havalige@amd.com>
Signed-off-by: Sean Anderson <sean.anderson@linux.dev>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Michal Simek <michal.simek@amd.com>
Since the shunt voltage and the current register report the same values
when the chip is calibrated, we can calculate the current directly
from the shunt voltage without relying on chip calibration.
With this change, the current register is no longer accessed. Its
register address is only used to indicate if reading or writing
current or shunt voltage is desired when accessing registers.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
While the chips supported by this driver do not directly support current
limits, they do support setting shunt voltage limits. The shunt voltage
divided by the shunt resistor value is the current. On top of that,
calibration values are set such that in the shunt voltage register and
the current register report the same values. That means we can report and
configure current limits based on shunt voltage limits, and we can do so
with much better accuracy than by setting shunt voltage limits.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Pass the to-be-limited register to alert functions and use it to determine
conversion from limit to register value.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Convert driver to use the with_info hardware monitoring API
to reduce its dependency on sysfs attribute functions.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
ina2xx_get_value() will be needed earlier in the next patch, so move it.
No functional change.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Alerts should only be cleared after reported, not immediately after the
alert condition has been cleared. Set the latch enable bit to keep alerts
latched until the alert register has been read from the chip.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Move all chip initialization code into a single function.
No functional change.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Module tests show various overflow problems when writing limits
and other attributes.
in0_crit: Suspected overflow: [max=82, read 0, written 2147483648]
in0_lcrit: Suspected overflow: [max=82, read 0, written 2147483648]
in1_crit: Suspected overflow: [max=40959, read 0, written 2147483647]
in1_lcrit: Suspected overflow: [max=40959, read 0, written 2147483647]
power1_crit: Suspected overflow: [max=134218750, read 0, written 2147483648]
update_interval: Suspected overflow: [max=2253, read 2, written 2147483647]
Implement missing clamping on attribute write operations to avoid those
problems.
While at it, check in the probe function if the shunt resistor value
passed from devicetree is valid, and bail out if it isn't. Also limit
mutex use to the code calling ina2xx_set_shunt() since it isn't needed
when called from the probe function.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
If it is necessary to re-initialize the chip, for example because
it has been power cycled, use regmap functions to update register
contents. This ensures that all registers, including the configuration
register and alert registers, are updated to previously configured
values without having to locally cache everything.
For this to work, volatile registers have to be marked as volatile.
Also, the cache needs to be bypassed when reading the calibration
and mask_enable registers. While the calibration register is not
volatile, it will be reset to 0 if the chip has been power cycled.
Most of the bits in the mask_enable register are configuration bits,
except for bit 4 which reports if an alert has been observed.
Both registers need to be marked as non-volatile to be updated
after a power cycle, but it is necessary to bypass the cache when
reading them to detect if the chip has been power cycled and to
read the alert status.
The chip does not support register auto-increments. It is therefore
necessary to configure regmap to use single register read/write
operations. Otherwise regmap tries to write all registers in a single
operation when synchronizing register contents with the hardware,
and the synchronization fails.
Another necessary change is to declare ina226_alert_to_reg() as u16.
So far it returned an s16 which is sign extended to a large negative
value which is then sent to regmap as unsigned int, causing an -EINVAL
error return.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
If regmap is accessed more than once in a function, declare and used
local regmap variable.
While at it, drop low value debug messages.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Recent versions of checkpatch complain that struct regmap_config
should be declared as const.
WARNING: struct regmap_config should normally be const
Doing so reveals a potential problem in the driver: If both supported
chips are present in a single system, the maximum number of registers
may race when devices are instantiated since max_registers is updated
in the probe function. Solve the problem by setting .max_registers to the
maximum register address of all supported chips. This does not make a
practical difference while fixing the potential race condition and reducing
code complexity.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Use bit operations where possible to make the code more generic and to
align it with other drivers. Also use compile time conversion from bit
to mask to reduce runtime overhead.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
There are no in-tree users of ina2xx platform data. Drop it and support
device properties instead as alternative if it should ever be needed.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Simplify driver maintenance by reordering include files to alphabetic
order.
Whule at it, drop unnecessary / unused jiffies.h.
No functional change.
Reviewed-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
pKVM relies on hypercalls to expose services such as memory sharing to
protected guests. Tentatively allocate a block of 58 hypercalls (i.e.
fill the remaining space in the first 64 function IDs) for pKVM usage,
as future extensions such as pvIOMMU support, range-based memory sharing
and validation of assigned devices will require additional services.
Suggested-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/86a5h5yg5y.wl-maz@kernel.org
Acked-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20240830130150.8568-8-will@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
Hook up pKVM's MMIO_GUARD hypercall so that ioremap() and friends will
register the target physical address as MMIO with the hypervisor,
allowing guest exits to that page to be emulated by the host with full
syndrome information.
Acked-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20240830130150.8568-7-will@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
Confidential Computing environments such as pKVM and Arm's CCA
distinguish between shared (i.e. emulated) and private (i.e. assigned)
MMIO regions.
Introduce a hook into our implementation of ioremap_prot() so that MMIO
regions can be shared if necessary.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Acked-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20240830130150.8568-6-will@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
If we detect the presence of pKVM's SHARE and UNSHARE hypercalls, then
register a backend implementation of the mem_encrypt API so that things
like DMA buffers can be shared appropriately with the host.
Acked-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20240830130150.8568-5-will@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
Implementing the internal mem_encrypt API for arm64 depends entirely on
the Confidential Computing environment in which the kernel is running.
Introduce a simple dispatcher so that backend hooks can be registered
depending upon the environment in which the kernel finds itself.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Acked-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20240830130150.8568-4-will@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
Implement a pKVM protected guest driver to probe the presence of pKVM
and determine the memory protection granule using the HYP_MEMINFO
hypercall.
Acked-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20240830130150.8568-3-will@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
arm64 will soon require its own callback to initialise services
that are only available on this architecture. Introduce a hook
that can be overloaded by the architecture.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20240830130150.8568-2-will@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
A "%s" is missing in ksft_exit_fail_msg(); instead, use the newly
introduced ksft_exit_fail_perror().
Signed-off-by: Dev Jain <dev.jain@arm.com>
Reviewed-by: Shuah Khan <skhan@linuxfoundation.org>
Reviewed-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20240830052911.4040970-1-dev.jain@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
At the moment, the SMMUv3 unit tests assume ATS is always enabled,
although this is sufficient to test hitless/non-hitless transitions,
but exercising other features is useful to check ste/cd population
logic (for example the .get_used logic).
Add an enum where bits define features per-master, at the moment there
is only ATS and STALLs which are mutually exclusive, but this would
make it easier to extend with other features in the future.
Also, Add 2 more tests for s1 <-> s2 transitions with stalls enabled.
Signed-off-by: Mostafa Saleh <smostafa@google.com>
Link: https://lore.kernel.org/r/20240830110349.797399-3-smostafa@google.com
Signed-off-by: Will Deacon <will@kernel.org>
According to the spec (ARM IHI 0070 F.b), in
"5.5 Fault configuration (A, R, S bits)":
A STE with stage 2 translation enabled and STE.S2S == 0 is
considered ILLEGAL if SMMU_IDR0.STALL_MODEL == 0b10.
Also described in the pseudocode “SteIllegal()”
if STE.Config == '11x' then
[..]
if eff_idr0_stall_model == '10' && STE.S2S == '0' then
// stall_model forcing stall, but S2S == 0
return TRUE;
Which means, S2S must be set when stall model is
"ARM_SMMU_FEAT_STALL_FORCE", but currently the driver ignores that.
Although, the driver can do the minimum and only set S2S for
“ARM_SMMU_FEAT_STALL_FORCE”, it is more consistent to match S1
behaviour, which also sets it for “ARM_SMMU_FEAT_STALL” if the
master has requested stalls.
Also, since S2 stalls are enabled now, report them to the IOMMU layer
and for VFIO devices it will fail anyway as VFIO doesn’t register an
iopf handler.
Signed-off-by: Mostafa Saleh <smostafa@google.com>
Link: https://lore.kernel.org/r/20240830110349.797399-2-smostafa@google.com
Signed-off-by: Will Deacon <will@kernel.org>
Add a test case for commit bb4485562f ("ASoC: cs-amp-lib: Ignore
empty UEFI calibration entries").
Any entries in the calibration blob that have calTime==0 are empty
entries. So they must not be returned by a lookup.
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>
Link: https://patch.msgid.link/20240830144819.118362-1-rf@opensource.cirrus.com
Signed-off-by: Mark Brown <broonie@kernel.org>
There is a clk == NULL check after the switch to check for
unsupported clk types. Since clk is re-assigned in a loop,
this check is useless right now for anything but the first
round. Let's fix this up by assigning clk = NULL in the
loop before the switch statement.
Fixes: a245fecbb8 ("clk: rockchip: add basic infrastructure for clock branches")
Cc: stable@vger.kernel.org
Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com>
[added fixes + stable-cc]
Link: https://lore.kernel.org/r/20240325193609.237182-6-sebastian.reichel@collabora.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
All clocks are registered early using CLK_OF_DECLARE(), which marks
the DT node as processed. For the processed DT node the probe routine
is never called. Thus this whole code is never executed. This could
be "fixed" by using CLK_OF_DECLARE_DRIVER, which avoids marking the
DT node as processed. But then the probe routine would re-register
all the clocks by calling rk3588_clk_init() again.
Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com>
Link: https://lore.kernel.org/r/20240325193609.237182-2-sebastian.reichel@collabora.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
On an NVMe namespace that does not support metadata, it is possible to
send an IO command with metadata through io-passthru. This allows issues
like [1] to trigger in the completion code path.
nvme_map_user_request() doesn't check if the namespace supports metadata
before sending it forward. It also allows admin commands with metadata to
be processed as it ignores metadata when bdev == NULL and may report
success.
Reject an IO command with metadata when the NVMe namespace doesn't
support it and reject an admin command if it has metadata.
[1] https://lore.kernel.org/all/mb61pcylvnym8.fsf@amazon.com/
Suggested-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Puranjay Mohan <pjy@amazon.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Anuj Gupta <anuj20.g@samsung.com>
Signed-off-by: Keith Busch <kbusch@kernel.org>
It is not safe to dereference fl->c.flc_owner without first confirming
fl->fl_lmops is the expected manager. nfsd4_deleg_getattr_conflict()
tests fl_lmops but largely ignores the result and assumes that flc_owner
is an nfs4_delegation anyway. This is wrong.
With this patch we restore the "!= &nfsd_lease_mng_ops" case to behave
as it did before the change mentioned below. This is the same as the
current code, but without any reference to a possible delegation.
Fixes: c5967721e1 ("NFSD: handle GETATTR conflict with write delegation")
Signed-off-by: NeilBrown <neilb@suse.de>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
MULTI GBT AN Control Register is IEEE Standard Register 7.32 (not a mask).
The right place should be in igc_reg.h file. In accordance with the
registers naming convention added IGC_' prefix.
Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>
Tested-by: Avigail Dahan <avigailx.dahan@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
When adding devm_regulator_bulk_get_const() I missed adding a stub for
when CONFIG_REGULATOR is not enabled. Under certain conditions (like
randconfig testing) this can cause the compiler to reports errors
like:
error: implicit declaration of function 'devm_regulator_bulk_get_const';
did you mean 'devm_regulator_bulk_get_enable'?
Add the stub.
Fixes: 1de452a0ed ("regulator: core: Allow drivers to define their init data as const")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202408301813.TesFuSbh-lkp@intel.com/
Cc: Neil Armstrong <neil.armstrong@linaro.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://patch.msgid.link/20240830073511.1.Ib733229a8a19fad8179213c05e1af01b51e42328@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
According to the IEEE standard report the EEE ability (registers 7.60 and
7.62) and the EEE Link Partner ability (registers 7.61 and 7.63). Use the
kernel's 'ethtool_keee' structure and report EEE link modes.
Example:
ethtool --show-eee <device>
Before:
Advertised EEE link modes: Not reported
Link partner advertised EEE link modes: Not reported
After:
Advertised EEE link modes: 100baseT/Full
1000baseT/Full
2500baseT/Full
Link partner advertised EEE link modes: 100baseT/Full
1000baseT/Full
2500baseT/Full
Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Tested-by: Avigail Dahan <avigailx.dahan@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
When running the igc with XDP/ZC in busy polling mode with deferral of hard
interrupts, interrupts still happen from time to time. That is caused by
the igc task watchdog which triggers Rx interrupts periodically.
That mechanism has been introduced to overcome skb/memory allocation
failures [1]. So the Rx clean functions stop processing the Rx ring in case
of such failure. The task watchdog triggers Rx interrupts periodically in
the hope that memory became available in the mean time.
The current behavior is undesirable for real time applications, because the
driver induced Rx interrupts trigger also the softirq processing. However,
all real time packets should be processed by the application which uses the
busy polling method.
Therefore, only trigger the Rx interrupts in case of real allocation
failures. Introduce a new flag for signaling that condition.
[1] - https://git.kernel.org/pub/scm/linux/kernel/git/tglx/history.git/commit/?id=3be507547e6177e5c808544bd6a2efa2c7f1d436
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Acked-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Kurt Kanzenbach <kurt@linutronix.de>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Mor Bar-Gabay <morx.bar.gabay@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add support for offloading MQPRIO. The hardware has four priorities as well
as four queues. Each queue must be a assigned with a unique priority.
However, the priorities are only considered in TSN Tx mode. There are two
TSN Tx modes. In case of MQPRIO the Qbv capability is not required.
Therefore, use the legacy TSN Tx mode, which performs strict priority
arbitration.
Example for mqprio with hardware offload:
|tc qdisc replace dev ${INTERFACE} handle 100 parent root mqprio num_tc 4 \
| map 0 0 0 0 0 1 2 3 0 0 0 0 0 0 0 0 \
| queues 1@0 1@1 1@2 1@3 \
| hw 1
The mqprio Qdisc also allows to configure the `preemptible_tcs'. However,
frame preemption is not supported yet.
Tested on Intel i225 and implemented by following data sheet section 7.5.2,
Transmit Scheduling.
Signed-off-by: Kurt Kanzenbach <kurt@linutronix.de>
Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com>
Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Mor Bar-Gabay <morx.bar.gabay@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
When VCMDQs are assigned to a VINTF owned by a guest (HYP_OWN bit unset),
only TLB and ATC invalidation commands are supported by the VCMDQ HW. So,
implement the new cmdq->supports_cmd op to scan the input cmd in order to
make sure that it is supported by the selected queue.
Note that the guest VM shouldn't have HYP_OWN bit being set regardless of
guest kernel driver writing it or not, i.e. the hypervisor running in the
host OS should wire this bit to zero when trapping a write access to this
VINTF_CONFIG register from a guest kernel.
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Link: https://lore.kernel.org/r/8160292337059b91271045800e5c62f7295e2c24.1724970714.git.nicolinc@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
The VCMDQ in the tegra241-cmdqv driver has a guest mode that supports only
a few invalidation commands. A batch is initialized with a cmdq, so it has
to confirm whether a new command is supported or not.
Add a supports_cmd function pointer to the cmdq structure, where the vcmdq
driver should hook a command scan function. Add an inline helper too so it
can be used by both sides.
If a new command is not supported, simply issue the existing batch and re-
init it as a new batch.
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Link: https://lore.kernel.org/r/aafb24b881504f18c5d0c7c15f2134e40ad2c486.1724970714.git.nicolinc@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
NVIDIA's Tegra241 Soc has a CMDQ-Virtualization (CMDQV) hardware, extending
the standard ARM SMMU v3 IP to support multiple VCMDQs with virtualization
capabilities. In terms of command queue, they are very like a standard SMMU
CMDQ (or ECMDQs), but only support CS_NONE in the CS field of CMD_SYNC.
Add a new tegra241-cmdqv driver, and insert its structure pointer into the
existing arm_smmu_device, and then add related function calls in the SMMUv3
driver to interact with the CMDQV driver.
In the CMDQV driver, add a minimal part for the in-kernel support: reserve
VINTF0 for in-kernel use, and assign some of the VCMDQs to the VINTF0, and
select one VCMDQ based on the current CPU ID to execute supported commands.
This multi-queue design for in-kernel use gives some limited improvements:
up to 20% reduction of invalidation time was measured by a multi-threaded
DMA unmap benchmark, compared to a single queue.
The other part of the CMDQV driver will be user-space support that gives a
hypervisor running on the host OS to talk to the driver for virtualization
use cases, allowing VMs to use VCMDQs without trappings, i.e. no VM Exits.
This is designed based on IOMMUFD, and its RFC series is also under review.
It will provide a guest OS a bigger improvement: 70% to 90% reductions of
TLB invalidation time were measured by DMA unmap tests running in a guest,
compared to nested SMMU CMDQ (with trappings).
As the initial version, the CMDQV driver only supports ACPI configurations.
Signed-off-by: Nate Watterson <nwatterson@nvidia.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Co-developed-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Link: https://lore.kernel.org/r/dce50490b2c10b7254fb36aa73ed7ffd812b283a.1724970714.git.nicolinc@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
For model-specific implementation, repurpose the acpi_smmu_get_options()
to a wider acpi_smmu_acpi_probe_model(). A new model can add to the list
in this new function.
Suggested-by: Will Deacon <will@kernel.org>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Link: https://lore.kernel.org/r/79716299829aeab2e55b8c7932f2634b209bb4d5.1724970714.git.nicolinc@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
