Engine->id namespace is per-tile so struct igt_live_test->reset_engine[]
needs to be two-dimensional so engine reset counts from all tiles can be
stored with no aliasing. With aliasing, if we had a real multi-tile
platform, the reset counts would be incorrect for same engine instance on
different tiles.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Fixes: 0c29efa23f ("drm/i915/selftests: Consider multi-gt instead of to_gt()")
Reported-by: Alan Previn Teres Alexis <alan.previn.teres.alexis@intel.com>
Cc: Tejas Upadhyay <tejas.upadhyay@intel.com>
Cc: Andi Shyti <andi.shyti@linux.intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231201122109.729006-1-tvrtko.ursulin@linux.intel.com
(cherry picked from commit 0647ece381)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Sifive platform dma (sf-pdma) has both in-order and out-of-order
configurations but sf-pdam driver configured to do in-order DMA
transfers, with out-of-order configuration got better throughput
in the PolarFire SoC platform.
Add a PolarFire SoC specific compatible and code to support
for out-of-order dma transfers
Reviewed-by: Emil Renner Berthing <emil.renner.berthing@canonical.com>
Signed-off-by: Shravan Chippa <shravan.chippa@microchip.com>
Link: https://lore.kernel.org/r/20231208103856.3732998-4-shravan.chippa@microchip.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Add new compatible name microchip,mpfs-pdma to support
out of order dma transfers
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Shravan Chippa <shravan.chippa@microchip.com>
Link: https://lore.kernel.org/r/20231208103856.3732998-3-shravan.chippa@microchip.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Update sf-pdma driver to adopt generic DMA device tree bindings.
It calls of_dma_controller_register() with of_dma_xlate_by_chan_id
to get the generic DMA device tree helper support and the DMA
clients can look up the sf-pdma controller using standard APIs.
Signed-off-by: Shravan Chippa <shravan.chippa@microchip.com>
Link: https://lore.kernel.org/r/20231208103856.3732998-2-shravan.chippa@microchip.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Eric Dumazet says:
====================
ipv6: more data-race annotations
Small follow up series, taking care of races around
np->mcast_oif and np->ucast_oif.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
np->ucast_oif is read locklessly in some contexts.
Make all accesses to this field lockless, adding appropriate
annotations.
This also makes setsockopt( IPV6_UNICAST_IF ) lockless.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
np->mcast_oif is read locklessly in some contexts.
Make all accesses to this field lockless, adding appropriate
annotations.
This also makes setsockopt( IPV6_MULTICAST_IF ) lockless.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This reverts commit b8dbbbc535 ("net: rtnetlink: remove local list
in __linkwatch_run_queue()"). It's evidently broken when there's a
non-urgent work that gets added back, and then the loop can never
finish.
While reverting, add a note about that.
Reported-by: Marek Szyprowski <m.szyprowski@samsung.com>
Fixes: b8dbbbc535 ("net: rtnetlink: remove local list in __linkwatch_run_queue()")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The current implementation's default Pause Forward setting is causing
unnecessary network traffic. This patch disables Pause Forward to
address this issue.
Fixes: 1121f6b02e ("octeontx2-af: Priority flow control configuration support")
Signed-off-by: Hariprasad Kelam <hkelam@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use current_top_of_stack() helper in sync_regs() and vc_switch_off_ist()
instead of open-coding the reading of the top_of_stack percpu variable
explicitly.
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: https://lore.kernel.org/r/20231204210320.114429-2-ubizjak@gmail.com
Version generation for "const_pcpu_hot" symbol failed because genksyms
doesn't know the __seg_gs keyword. Effectively revert commit 4604c052b8
("x86/percpu: Declare const_pcpu_hot as extern const variable") and
use this_cpu_read_const() instead to avoid "sparse: dereference of
noderef expression" warning when reading const_pcpu_hot.
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: https://lore.kernel.org/r/20231204210320.114429-1-ubizjak@gmail.com
Some members of struct svs_bank are not changed during runtime, so those
are not variables but constants: move all of those to a new structure
called svs_bank_pdata and refactor the code to make use of that and
reorder members by size where possible.
This effectively moves at least 50 bytes to the text segment.
While at it, also uniform the thermal zone names across the banks.
Link: https://lore.kernel.org/r/20231121125044.78642-19-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
The `freq` variable is of type unsigned long and, even though it does
currently work with u32 because no frequency is higher than U32_MAX,
it is not guaranteed that in the future we will see one.
Initialize the freq variable with ULONG_MAX instead of U32_MAX.
Link: https://lore.kernel.org/r/20231121125044.78642-18-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
The svs_init01() and svs_init02() functions are already checking if the
INIT01 and INIT02 modes are available - but that's done in for loops and
for each SVS bank.
Give those a shortcut to get out early if no SVS bank features the
desired init mode: this is especially done to avoid some locking in
the svs_init01(), but also to avoid multiple for loops to check the
same, when no bank supports a specific mode.
Link: https://lore.kernel.org/r/20231121125044.78642-17-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Cleanup the svs_probe() function: use dev_err_probe() where possible,
change some efuse read failure gotos and then remove now impossible
IS_ERR_OR_NULL() checks (as they will never return true) for nvmem
(efuse read) failures.
Also remove some unnecessary blank lines.
Link: https://lore.kernel.org/r/20231121125044.78642-16-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Drop the "pbank" pointer from struct svs_bank: this was used to simply
pass a pointer to the SVS bank that the flow was working on.
That for instance needs more locking, and it's avoidable by adding one
more parameter to functions working on specific banks, either a bank
index number, or passing the svs_bank pointer directly from the caller.
Even if the locking can now be reduced, for now, it was still left in
place for the sake of making sure to not introduce any stability and/or
reliability regression.
Link: https://lore.kernel.org/r/20231121125044.78642-12-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Remove almost all of the per-SoC .efuse_parsing() callbacks and replace
them with one common callback svs_common_parse_efuse(): to do that, also
change the function signature of the callback to add the newly required
pointer to struct svs_platform_data, containing the SVS-global fuse map.
This is done for MT8186, MT8188, MT8192, MT8195.
As for MT8183, the efuse parse function was simplified by using the new
fuse maps.
Link: https://lore.kernel.org/r/20231121125044.78642-11-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
The t-calibration-data (SVS-Thermal calibration data) shall exist for
all SoCs or SVS won't work anyway: move it to the common svs_probe()
function and remove it from all of the per-SoC efuse_parsing() probe
callbacks.
Link: https://lore.kernel.org/r/20231121125044.78642-10-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
In preparation for commonizing the efuse parsing function, add the
SVS-Thermal coefficients for all SoCs for which said function can be
commonized (MT8186, MT8188, MT8192, MT8195) and assign those to their
platform data structure.
That will be used to calculate the MTS parameter with the equation
MTS = (ts_coeff * 2) / 1000
This commit brings no functional changes.
Link: https://lore.kernel.org/r/20231121125044.78642-9-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
In preparation for adding a common efuse parsing function which will
greatly reduce code duplication, add a SoC-specific mapping that will
be used to retrieve the right SVS calibration values from the fuses.
The maps are two: one is a Global Map used for reading parameters that
are SVS-global, and one is a Bank Map for reading calibrations for
each SVS Bank.
While at it, also populate the map in the platform data for each SoC.
Being this a preparation commit, there are no functional changes.
Link: https://lore.kernel.org/r/20231121125044.78642-8-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
This driver tries to create a device link to the thermal sensor device:
change all instances of "lvts" and "thermal" to "thermal-sensor", as
that's what the devicetree node name must be.
Note for MT8183: As specified in a previous commit, this SoC never got
SVS probing, so this is not a breaking change and it does not require
fallback for older device trees.
Link: https://lore.kernel.org/r/20231121125044.78642-7-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Many 32-bit members of this struct can be size reduced to either 16-bit
or even 8-bit, for a total saving of ~61 bytes per bank. Keeping in mind
that one SoC declares at least two banks, this brings a minimum of ~122
bytes saving (depending on compiler optimization).
Link: https://lore.kernel.org/r/20231121125044.78642-6-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
In svs_bank_resource_setup() there is a "big" switch assigning different
names depending on sw_id and type and this will surely grow: for example
MT8186 has got a two-line type (high/low) SVS bank for CPU_BIG, and this
would require more switch nesting.
Simplify all of this by changing that to a devm_kasprintf() call that
will concatenate the SW_ID string (e.g. SVSB_CPU_LITTLE) with the Type
string (e.g. _LOW), resulting in the expected full bank name (e.g.
SVSB_CPU_LITTLE_LOW).
This being a dynamic allocation can be slower, but this happens only
once in the life of this driver and it's not a performance path, so it's
totally acceptable.
Link: https://lore.kernel.org/r/20231121125044.78642-5-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
The sw_id and type specifiers currently are defined as BIT(x) for
unknown reasons: nothing in this code makes any AND/OR check for
those, and that would never happen anyway because both sw_id and
type are exclusive, as in:
- There will never be a bank that is for both CPU and GPU, or
for CPU and CCI together;
- A bank cannot be contemporarily of one-line and two-line type,
as much as it cannot contemporarily have both HIGH and LOW roles
Change those definitions to enumerations and also add some kerneldoc
to better describe what they are for and what they indicate.
While at it, also change the names adding _SWID or _TYPE to increase
human readability.
Link: https://lore.kernel.org/r/20231121125044.78642-4-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
The svs_regs_v2 array of registers was offsetted by 0xc00 because the
SVS node was supposed to have the same iostart as the thermal sensors.
That's wrong for two reasons:
1. Two different devices cannot have the same iostart in devicetree,
as those would technically be the same device otherwise; and
2. SVS and Thermal Sensor (be it LVTS or AUXADC thermal) are not the
same IP, and those two do obviously have a different iospace.
Even though there already are users of this register array, the only
one that declares a devicetree node for SVS is MT8183 - but it never
actually worked because the "tzts1" thermal zone missed thermal trips,
hence this driver's probe always failed on that SoC.
Knowing this - it is safe to say that keeping compatibility with older
device trees is pointless, hence simply subtract the 0xc00 offset from
the register offset array.
Link: https://lore.kernel.org/r/20231121125044.78642-3-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Add MT8188 reset bit map for VDOSYS0 and VDOSYS1.
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hsiao Chien Sung <shawn.sung@mediatek.com>
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
- Reset ID must starts from 0 and be consecutive, but
the reset bits in our hardware design is not continuous,
some bits are left unused, we need a map to solve the problem
- Use old style 1-to-1 mapping if .rst_tb is not defined
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hsiao Chien Sung <shawn.sung@mediatek.com>
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
- Add register definitions for MT8188
- Add VDOSYS1 routing table
- Update MUTEX definitions accordingly
- Set VSYNC length from 0x40 (default) to 1 since ETHDR is bypassed
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hsiao Chien Sung <shawn.sung@mediatek.com>
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Add MT8188 VPPSYS0 and VPPSYS1 driver data.
Signed-off-by: yu-chang.lee <yu-chang.lee@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
MT8186 svs has a number of banks which used as optimization of opp
voltage table for corresponding dvfs drivers.
MT8186 svs big core uses 2-line high bank and low bank to optimize the
voltage of opp table for higher and lower frequency respectively.
Signed-off-by: Mark Tseng <chun-jen.tseng@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
MT8195 svs gpu uses 2-line high bank and low bank to optimize the
voltage of opp table for higher and lower frequency respectively.
Signed-off-by: Mark Tseng <chun-jen.tseng@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Due to electrical and mechanical constraints in certain platform designs
there may be likely interference of relatively high-powered harmonics of
the (G-)DDR memory clocks with local radio module frequency bands used
by Wifi 6/6e/7.
To mitigate this, AMD has introduced a mechanism that devices can use to
notify active use of particular frequencies so that other devices can make
relative internal adjustments as necessary to avoid this resonance.
Co-developed-by: Evan Quan <quanliangl@hotmail.com>
Signed-off-by: Evan Quan <quanliangl@hotmail.com>
Signed-off-by: Ma Jun <Jun.Ma2@amd.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Add documentation about AMD's Wifi band RFI mitigation (WBRF) mechanism
explaining the theory and how it is used.
Signed-off-by: Ma Jun <Jun.Ma2@amd.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Add the MediaTek SVS node: this will lower the voltage of various
components of the SoC based on chip quality (read from fuses) in
order to save power and generate less heat.
Also, reduce the LVTS_AP iospace to 0xc00, because that's exactly
where SVS starts.
- LVTS_AP start: 0x1100b000 length: 0xc00
- SVS start: 0x1100bc00 length: 0x400
Link: https://lore.kernel.org/r/20231121125044.78642-21-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
The SVS iospace starts at 0x1100bc00 and not at 0x1100b000 as the
latter is the thermal sensor iospace instead.
Change the iospaces for both as following:
- Thermal: 0x1100b000, length 0xc00
- SVS: 0x1100bc00, length 0x400
Please note that while this would be a breaking change for SVS (but
not for thermal sensors), it doesn't matter because the svs driver
never worked anyway because of the missing trips in tzts2, causing
that thermal zone to never actually register, hence the SVS driver
to fail probing anyway.
Link: https://lore.kernel.org/r/20231121125044.78642-2-angelogioacchino.delregno@collabora.com
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
The design of the WMI chardev interface is broken:
- it assumes that WMI drivers are not instantiated twice
- it offers next to no abstractions, the WMI driver gets
a raw byte buffer
- it is only used by a single driver, something which is
unlikely to change
Since the only user (dell-smbios-wmi) has been migrated
to his own ioctl interface, remove it.
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Link: https://lore.kernel.org/r/20231210202443.646427-6-W_Armin@gmx.de
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
The WMI chardev API will be removed in the near future.
Reimplement the necessary bits used by this driver so
that userspace software depending on it does no break.
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Link: https://lore.kernel.org/r/20231210202443.646427-5-W_Armin@gmx.de
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Use devres version of __get_free_pages() to simplify the
error handling code.
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Link: https://lore.kernel.org/r/20231210202443.646427-4-W_Armin@gmx.de
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Users can already listen to ACPI WMI events through
the ACPI netlink interface. The old wmi_notify_debug()
interface also uses the deprecated GUID-based interface.
Remove it to make the event handling code more readable.
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Link: https://lore.kernel.org/r/20231210202443.646427-3-W_Armin@gmx.de
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
The functionality of dumping WDG entries is better provided by
userspace tools like "fwts wmi", which also does not suffer from
garbled printk output caused by pr_cont().
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Link: https://lore.kernel.org/r/20231210202443.646427-2-W_Armin@gmx.de
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Highlights:
- asus-wmi: Solve i8042 filter resource handling, input, and
suspend issues
- wmi: Skip zero instance WMI blocks to avoid issues with
some laptops
- mlxbf-bootctl: Differentiate dev/production keys
- platform/surface: Correct serdev related return value to avoid
leaking errno into userspace
- Error checking fixes
The following is an automated shortlog grouped by driver:
asus-wmi:
- Change q500a_i8042_filter() into a generic i8042-filter
- disable USB0 hub on ROG Ally before suspend
- Filter Volume key presses if also reported via atkbd
- Move i8042 filter install to shared asus-wmi code
mellanox:
- Add null pointer checks for devm_kasprintf()
- Check devm_hwmon_device_register_with_groups() return value
mlxbf-bootctl:
- correctly identify secure boot with development keys
surface: aggregator:
- fix recv_buf() return value
wmi:
- Skip blocks with zero instances
-----BEGIN PGP SIGNATURE-----
iHUEABYIAB0WIQSCSUwRdwTNL2MhaBlZrE9hU+XOMQUCZXC60QAKCRBZrE9hU+XO
McbwAPwIlGYIdYzUXGVGCrCBq7wtBR8ADfFqIa42hUlkDyZ4tAEA8LgrZutAFwhq
35kVk6FSwrUGuIhX5TkxTxnp814nzwE=
=iiB1
-----END PGP SIGNATURE-----
Merge tag 'platform-drivers-x86-v6.7-3' into pdx86/for-next
Back merge pdx86 fixes into pdx86/for-next for further WMI work
depending on some of the fixes.
platform-drivers-x86 for v6.7-3
Highlights:
- asus-wmi: Solve i8042 filter resource handling, input, and
suspend issues
- wmi: Skip zero instance WMI blocks to avoid issues with
some laptops
- mlxbf-bootctl: Differentiate dev/production keys
- platform/surface: Correct serdev related return value to avoid
leaking errno into userspace
- Error checking fixes
The following is an automated shortlog grouped by driver:
asus-wmi:
- Change q500a_i8042_filter() into a generic i8042-filter
- disable USB0 hub on ROG Ally before suspend
- Filter Volume key presses if also reported via atkbd
- Move i8042 filter install to shared asus-wmi code
mellanox:
- Add null pointer checks for devm_kasprintf()
- Check devm_hwmon_device_register_with_groups() return value
mlxbf-bootctl:
- correctly identify secure boot with development keys
surface: aggregator:
- fix recv_buf() return value
wmi:
- Skip blocks with zero instances
efivar operation is updated when the tee_stmm_efi module is probed.
tee_stmm_efi module supports SetVariable runtime service, but user needs
to manually remount the efivarfs as RW to enable the write access if the
previous efivar operation does not support SetVariable and efivarfs is
mounted as read-only.
This commit notifies the update of efivar operation to efivarfs
subsystem, then drops SB_RDONLY flag if the efivar operation supports
SetVariable.
Signed-off-by: Masahisa Kojima <masahisa.kojima@linaro.org>
[ardb: use per-superblock instance of the notifier block]
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
When the flash is not owned by the non-secure world, accessing the EFI
variables is straight-forward and done via EFI Runtime Variable
Services. In this case, critical variables for system integrity and
security are normally stored in the dedicated secure storage and can
only be manipulated directly from the secure world.
Usually, small embedded devices don't have the special dedicated secure
storage. The eMMC device with an RPMB partition is becoming more common,
and we can use this RPMB partition to store the EFI Variables.
The eMMC device is typically owned by the non-secure world (Linux in our
case). There is an existing solution utilizing eMMC RPMB partition for
EFI Variables, it is implemented by interacting with TEE (OP-TEE in this
case), StandaloneMM (as EFI Variable Service Pseudo TA), eMMC driver and
tee-supplicant. The last piece is the tee-based variable access driver
to interact with TEE and StandaloneMM.
So let's add the kernel functions needed.
This feature is implemented as a kernel module. StMM PTA has
TA_FLAG_DEVICE_ENUM_SUPP flag when registered to OP-TEE so that this
tee_stmm_efi module is probed after tee-supplicant starts, since
"SetVariable" EFI Runtime Variable Service requires to interact with
tee-supplicant.
Acked-by: Sumit Garg <sumit.garg@linaro.org>
Co-developed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: Masahisa Kojima <masahisa.kojima@linaro.org>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>