Commit 344179fc7e ("ARM: 9106/1: traps: use get_kernel_nofault instead
of set_fs()") replaced an occurrence of __get_user() with
get_kernel_nofault(), but inverted the sense of the conditional in the
process, resulting in no values to be printed at all.
I.e., every exception stack now looks like this:
Exception stack(0xc18d1fb0 to 0xc18d1ff8)
1fa0: ???????? ???????? ???????? ????????
1fc0: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
1fe0: ???????? ???????? ???????? ???????? ???????? ????????
which is rather unhelpful.
Fixes: 344179fc7e ("ARM: 9106/1: traps: use get_kernel_nofault instead of set_fs()")
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
tglx notes:
This function [futex_detect_cmpxchg] is only needed when an
architecture has to runtime discover whether the CPU supports it or
not. ARM has unconditional support for this, so the obvious thing to
do is the below.
Fixes linkage failure from Clang randconfigs:
kernel/futex.o:(.text.fixup+0x5c): relocation truncated to fit: R_ARM_JUMP24 against `.init.text'
and boot failures for CONFIG_THUMB2_KERNEL.
Link: https://github.com/ClangBuiltLinux/linux/issues/325
Comments from Nick Desaulniers:
See-also: 03b8c7b623 ("futex: Allow architectures to skip
futex_atomic_cmpxchg_inatomic() test")
Reported-by: Arnd Bergmann <arnd@arndb.de>
Reported-by: Nathan Chancellor <nathan@kernel.org>
Suggested-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Cc: stable@vger.kernel.org # v3.14+
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
This removes the chrdev_lock from the counter subsystem. This was
intended to prevent opening the chrdev more than once. However, this
doesn't work in practice since userspace can duplicate file descriptors
and pass file descriptors to other processes. Since this protection
can't be relied on, it is best to just remove it.
Suggested-by: Greg KH <gregkh@linuxfoundation.org>
Acked-by: William Breathitt Gray <vilhelm.gray@gmail.com>
Signed-off-by: David Lechner <david@lechnology.com>
Link: https://lore.kernel.org/r/20211017185521.3468640-1-david@lechnology.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When user code execution with privilege mode, it will lead to
infinite loop in the page fault handler if ARM_LPAE enabled,
The issue could be reproduced with
"echo EXEC_USERSPACE > /sys/kernel/debug/provoke-crash/DIRECT"
As Permission fault shows in ARM spec,
IFSR format when using the Short-descriptor translation table format
Permission fault: 01101 First level 01111 Second level
IFSR format when using the Long-descriptor translation table format
Permission fault: 0011LL LL bits indicate levelb.
Add is_permission_fault() function to check permission fault and die
if permission fault occurred under instruction fault in do_page_fault().
Fixes: 1d4d37159d ("ARM: 8235/1: Support for the PXN CPU feature on ARMv7")
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Provide die_kernel_fault() helper to do the kernel fault reporting,
which with msg argument, it could report different message in different
scenes, and the later patch "ARM: mm: Fix PXN process with LPAE feature"
will use it.
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Now the show_pts() will dump the virtual (hashed) address of page
table base, it is useless, kill it.
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Now the write fault check in do_page_fault() and access_error() twice,
we can cleanup access_error(), and make the fault check and vma flags set
into do_page_fault() directly, then pass the vma flags to __do_page_fault.
No functional change.
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
The __do_page_fault() won't use task_struct argument, kill it
and also use current->mm directly in do_page_fault().
No functional change.
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Clean up the multiple goto statements and drops local variable
vm_fault_t fault, which will make the __do_page_fault() much more
readability.
No functional change.
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
iMX6 needs to write some code to OCRAM which resumes the DDR controller
after suspend. However, merely using __arm_ioremap_exec() causes the
kernel to complain of a W+X mapping. Solve this by using the newly
introduced __arm_iomem_set_ro() function to prevent inadvertent or
malicious writes to code we may later execute.
Tested-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
__arm_iomem_set_ro() marks an ioremapped area read-only. This is
intended for use with __arm_ioremap_exec() to allow the kernel to
write some code into e.g. SRAM and then write-protect it so the
kernel doesn't complain about W+X mappings.
Tested-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Add support for parsing the "linux,usable-memory-range" DT property.
This property is used to describe the usable memory reserved for the
crash dump kernel, and thus makes the memory reservation explicit.
If present, Linux no longer needs to mask the program counter, and rely
on the "mem=" kernel parameter to obtain the start and size of usable
memory.
For backwards compatibility, the traditional method to derive the start
of memory is still used if "linux,usable-memory-range" is absent.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
The remove callback is only called after probe completed successfully.
In this case platform_set_drvdata() was called with a non-NULL argument
and so !sdev is never true.
The motivation for this change is to get rid of non-zero return values
for remove callbacks as their only effect is to trigger a runtime
warning. See commit e5e1c20978 ("driver core: platform: Emit a warning
if a remove callback returned non-zero") for further details.
Link: https://lore.kernel.org/linux-arm-kernel/20210721205450.2173923-1-u.kleine-koenig@pengutronix.de
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
of_amba_device_create() uses irq_of_parse_and_map() to translate
a DT interrupt specification into a Linux virtual interrupt number.
But it doesn't properly handle the case where the interrupt controller
is not yet available, eg, when pl011 interrupt is connected to MBIGEN
interrupt controller, because the mbigen initialization is too late,
which will lead to no IRQ due to no IRQ domain found, log is shown below,
"irq: no irq domain found for uart0 !"
use of_irq_get() to return -EPROBE_DEFER as above, and in the function
amba_device_try_add()/amba_device_add(), it will properly handle in such
case, also return 0 in other fail cases to be consistent as before.
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Frank Rowand <frowand.list@gmail.com>
Reported-by: Ruizhe Lin <linruizhe@huawei.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
After commit 77a7300aba ("of/irq: Get rid of NO_IRQ usage"),
no irq case has been removed, irq_of_parse_and_map() will return
0 in all cases when get error from parse and map an interrupt into
linux virq space.
amba_device_register() is only used on no-DT initialization, see
s3c64xx_pl080_init() arch/arm/mach-s3c/pl080.c
ep93xx_init_devices() arch/arm/mach-ep93xx/core.c
They won't set -1 to irq[0], so no need the warn.
This reverts commit 2eac58d502.
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
No one use the following functions, kill them.
amba_aphb_device_add()
amba_apb_device_add()
amba_apb_device_add_res()
amba_ahb_device_add()
amba_ahb_device_add_res()
Cc: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Commit 344179fc7e ("ARM: 9106/1: traps: use get_kernel_nofault instead
of set_fs()") replaced an occurrence of __get_user() with
get_kernel_nofault(), but inverted the sense of the conditional in the
process, resulting in no values to be printed at all.
I.e., every exception stack now looks like this:
Exception stack(0xc18d1fb0 to 0xc18d1ff8)
1fa0: ???????? ???????? ???????? ????????
1fc0: ???????? ???????? ???????? ???????? ???????? ???????? ???????? ????????
1fe0: ???????? ???????? ???????? ???????? ???????? ????????
which is rather unhelpful.
Fixes: 344179fc7e ("ARM: 9106/1: traps: use get_kernel_nofault instead of set_fs()")
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
On Windows systems, ASUS laptops uses the "turn display off" key
(usually fn+f6) to turn both display and keyboard backlit off. On Linux
systems, this key has no effect at all since most desktop enviroments
don't deal with KEY_DISPLAY_OFF. By mapping it to KEY_SCREENLOCK
instead, would enable desktop environments to handle this key as a
screen lock intent from the user, out of the box.
Signed-off-by: Vinícius Angiolucci Reis <angiolucci@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
The Microchip Icicle kit uses SiFive E51 and U54 cores, so it looks that
also Core Local Interruptor and Platform-Level Interrupt Controller are
coming from SiFive. Add proper compatibles to silence dtbs_check
warnings:
clint@2000000: compatible:0: 'sifive,clint0' is not one of ['sifive,fu540-c000-clint', 'canaan,k210-clint']
interrupt-controller@c000000: compatible:0: 'sifive,plic-1.0.0' is not one of ['sifive,fu540-c000-plic', 'canaan,k210-plic']
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20210920130412.145231-1-krzysztof.kozlowski@canonical.com
The DTSI file defines soc node and address/size cells, so there is no
point in duplicating it in DTS file.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Reviewed-by: Alexandre Ghiti <alexandre.ghiti@canonical.com>
Tested-by: Alexandre Ghiti <alexandre.ghiti@canonical.com>
Link: https://lore.kernel.org/r/20210920130248.145058-3-krzysztof.kozlowski@canonical.com
Add missing sifive,fu540 compatible to fix dtbs_check warnings:
arch/riscv/boot/dts/sifive/hifive-unleashed-a00.dt.yaml: /: compatible: 'oneOf' conditional failed, one must be fixed:
['sifive,hifive-unleashed-a00', 'sifive,fu540-c000'] is too short
'sifive,hifive-unleashed-a00' is not one of ['sifive,hifive-unmatched-a00']
'sifive,fu740-c000' was expected
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Reviewed-by: Alexandre Ghiti <alexandre.ghiti@canonical.com>
Tested-by: Alexandre Ghiti <alexandre.ghiti@canonical.com>
Link: https://lore.kernel.org/r/20210920130248.145058-2-krzysztof.kozlowski@canonical.com
Some Apple ISO keyboards have a quirk where the backtick/tilde key is
swapped with the less-than/greater-than key. Unfortunately, there is no
perfectly reliable way to detect whether a keyboard has the quirk or
not, but the quirk appears to only be present on models that support
Bluetooth, and the affected keyboards usually report country code 13 in
the HID descriptor.
Therefore, the best we can do is to change
/sys/module/hid_apple/parameters/iso_layout to a ternary:
0 = Not ISO or ISO and not quirky
1 = ISO and quirky
-1 = Guess based on product ID and country code
Table of keyboards that José, Julian and I have tested:
Product Model Shape Labels Bus Country Quirky
=========================================================
05ac:0201 M2452 ANSI Usonian USB 0 No
05ac:020b A1048 ANSI Usonian USB 0 No
05ac:020c A1048 ISO Québécois USB 13 No
05ac:0221 A1243 ISO Norwegian USB 13 No
05ac:0221 A1243 ISO Portuguese USB 13 No
05ac:0221 A1243 ISO Swedish USB 13 No
05ac:0221 A1243 ISO Swiss USB 13 No
05ac:022c A1255 ANSI Usonian BT 33 No
05ac:022d A1255 ISO Hebrew BT 13 Yes
05ac:022d A1255 ISO Québécois BT 13 Yes
05ac:022d A1255 ISO Spanish BT 13 Yes
05ac:023a A1314 ISO Russian BT 13 Yes
05ac:023a A1314 ISO Swiss BT 13 Yes
05ac:024f A1243 ANSI Usonian USB 0 No
05ac:0250 A1243 ISO British USB 13 No
05ac:0250 A1243 ISO German USB 13 No
05ac:0250 A1243 ISO Italian USB 13 No
05ac:0250 A1243 ISO Québécois USB 13 No
05ac:0251 A1243 JIS Japanese USB 15 No
05ac:0255 A1314 ANSI Usonian BT 33 No
05ac:0255 A1314 ANSI Taiwanese BT 33 No
05ac:0255 A1314 ANSI Thai BT 33 No
05ac:0256 A1314 ISO Arabic BT 13 Yes
05ac:0256 A1314 ISO French BT 13 Yes
05ac:0256 A1314 ISO German BT 13 Yes
05ac:0256 A1314 ISO Norwegian BT 13 Yes
05ac:0256 A1314 ISO Spanish BT 13 Yes
05ac:0256 A1314 ISO Swiss BT 13 Yes
05ac:0257 A1314 JIS Japanese BT 15 No
05ac:0267 A1644 ANSI Usonian USB 33 No
004c:0267 A1644 ANSI Usonian BT 0 No
05ac:0267 A1644 ISO British USB 13 Yes
004c:0267 A1644 ISO British BT 0 Yes
05ac:0267 A1644 ISO Finnish USB 13 Yes
004c:0267 A1644 ISO Finnish BT 0 Yes
05ac:0267 A1644 ISO Québécois USB 13 Yes
004c:0267 A1644 ISO Québécois BT 0 Yes
05ac:0267 A1644 ISO Spanish USB 13 Yes
004c:0267 A1644 ISO Spanish BT 0 Yes
05ac:0267 A1644 ISO Swiss USB 13 Yes
004c:0267 A1644 ISO Swiss BT 0 Yes
05ac:0267 A1644 JIS Japanese USB 15 No
004c:0267 A1644 JIS Japanese BT 0 No
05ac:029c A2450 ANSI Usonian USB 33 No
004c:029c A2450 ANSI Usonian BT 0 No
05ac:029c A2450 ISO Spanish USB 13 Yes
004c:029c A2450 ISO Spanish BT 0 Yes
05ac:029c A2450 JIS Japanese USB 15 No
004c:029c A2450 JIS Japanese BT 0 No
Reported-by: José Expósito <jose.exposito89@gmail.com>
Tested-by: José Expósito <jose.exposito89@gmail.com>
Tested-by: Julian Weigt <juw@posteo.de>
Signed-off-by: Alex Henrie <alexhenrie24@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
The ANSI, ISO, and JIS variants of this keyboard all have the same
product ID.
Signed-off-by: Alex Henrie <alexhenrie24@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
A new 'chassis-type' root node property has recently been approved for
the device-tree specification.
Add this property for end-user devices (such as laptops,
smartphones and tablets) based on Samsung S5Pv210 ARM SoCs.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Link: https://lore.kernel.org/r/20211017101228.19478-3-krzysztof.kozlowski@canonical.com
A new 'chassis-type' root node property has recently been approved for
the device-tree specification.
Add this property for end-user devices (such as laptops,
smartphones and tablets) based on Samsung Exynos ARM SoCs.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Link: https://lore.kernel.org/r/20211017101228.19478-2-krzysztof.kozlowski@canonical.com
A new 'chassis-type' root node property has recently been approved for
the device-tree specification.
Add this property for end-user devices (such as laptops,
smartphones and tablets) based on Samsung Exynos ARM64 SoCs.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Link: https://lore.kernel.org/r/20211017101228.19478-1-krzysztof.kozlowski@canonical.com
Add new registers to support systems with multiply cooling devices.
Modular systems support up-to four cooling devices. This capability
is detected according to the registers initial setting.
Signed-off-by: Vadim Pasternak <vadimp@nvidia.com>
Reviewed-by: Michael <michaelsh@nvidia.com>
Link: https://lore.kernel.org/r/20211002093609.3771576-1-vadimp@nvidia.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Add documentation for the new attributes for line cards:
- CPLDs versioning.
- Write protection control for 'nvram' devices.
- Line card reset reasons.
- Enabling burning of FPGA and CPLDs.
- Enabling burning of FPGA and gearbox SPI flashes,
- Enabling power of whole line card.
- Enabling power of QSFP ports equipped on line card.
- The maximum powered required for line card feeding.
- Line card configuration Id.
Signed-off-by: Vadim Pasternak <vadimp@nvidia.com>
Reviewed-by: Michael Shych <michaelsh@nvidia.com>
Link: https://lore.kernel.org/r/20211002093238.3771419-10-vadimp@nvidia.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Add documentation for the new attributes:
- "bios_active_image"; "bios_auth_fail"; "bios_upgrade_fail";
"bios_safe_mode" to represent various BIOS statuses.
- "lc{n}_enable" - for put/release the line card to/from enable state.
- "lc{n}_pwr" - for power on/off the line card.
- "lc{n}_rst_mask" - for line card reset state enforced by ASIC, when
it sets it due to some abnormal ASIC behavior.
- "psu3_on"; "psu4_on" - for connection/disconnection power supply unit
to/from the power source.
- "pm_mgmt_en" - for setting power management control ownership. When
power management control is provided by hardware, it means that
hardware will automatically power off one or more line cards in case
system power budget is under power required for feeding all powered
on line cards. It could be a case, when some of power units lost
power good state.
- "shutdown_unlock" - for unlocking system after hardware or firmware
thermal shutdown, which causes locking of the all interfaces to ASIC.
Signed-off-by: Vadim Pasternak <vadimp@nvidia.com>
Reviewed-by: Michael Shych <michaelsh@nvidia.com>
Link: https://lore.kernel.org/r/20211002093238.3771419-9-vadimp@nvidia.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Provide support for the Nvidia MSN4800-XX line cards for MSN4800
Ethernet modular switch system, providing a high performance switching
solution for Enterprise Data Centers (EDC) for building Ethernet based
clusters, High-Performance Computing (HPC) and embedded environments.
Initial version provides support for line card type MSN4800-C16. This
type of line card is equipped with:
- Lattice CPLD device, used for system and ports control.
- four Nvidia gearbox devices, used for port splitting.
- FPGA device, used for gearboxes management.
- 16x100G QSFP28 ports.
- hotpswap controllers, voltage regulators, analog-to-digital
convertors, nvram devices.
- status LED.
During initialization driver creates:
- line card's I2C tree through "i2c-mux-mlxcpd" driver.
- line card's LED objects through "leds-mlxreg" driver.
- line card's CPLD register space input / output "hwmon" attributes for
line control and monitoring through "mlxreg-io" driver. These
attributes provide CPLD and FPAG versioning, control for upgradable
components burning, NVRAM devices write protection, line card
revision, line card power consuming, line card reset cause
indication, etcetera.
Lattice CPLD device and nvram devices are feeding from auxiliary power
domain and accessible, when line card is powered off. These devices
are connected by line card driver probing routine, invoked after line
card security verification is done by hardware and event lc#n_verified
is received for line card located in slot #n.
Gearboxes, FPGA, hotpswap controllers, voltage regulators,
analog-to-digital convertors are feeding from main power domain. These
devices are connected after power good event "lc#n_powered" is received
for line card located in slot #n.
The driver 'mlxreg-lc' is driven by 'mlxreg-hotplug' driver following
relevant "hotplug" events.
Signed-off-by: Vadim Pasternak <vadimp@nvidia.com>
Reviewed-by: Michael Shych <michaelsh@nvidia.com>
Link: https://lore.kernel.org/r/20211002093238.3771419-8-vadimp@nvidia.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Extend structure 'mlxreg_core_data' with the field "secured". The
purpose of this field is to restrict access to some attributes, if
kernel is configured with security options, like:
LOCK_DOWN_KERNEL_FORCE_CONFIDENTIALITY.
Access to some attributes, which for example, allow burning of some
hardware components, like FPGA, CPLD, SPI, etcetera can break the
system. In case user does not want to allow such access, it can disable
it by setting security options.
Signed-off-by: Vadim Pasternak <vadimp@nvidia.com>
Reviewed-by: Michael Shych <michaelsh@nvidia.com>
Link: https://lore.kernel.org/r/20211002093238.3771419-7-vadimp@nvidia.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Extend maximum number of the attributes, exposed to 'sysfs'.
It is requires in order to support modular systems, which
provide more attributes for system control, statuses and info.
Signed-off-by: Vadim Pasternak <vadimp@nvidia.com>
Reviewed-by: Michael Shych <michaelsh@nvidia.com>
Link: https://lore.kernel.org/r/20211002093238.3771419-6-vadimp@nvidia.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Add event notifier callbacks for modular system line cards. These
callbacks are to be passed to "mlxreg-hotplug" driver by line card
driver during probing. Then, when any line card related hotplug event
is received (insertion ,power, synch, ready), hotplug driver will
invoke callback for the relevant line card.
Signed-off-by: Vadim Pasternak <vadimp@nvidia.com>
Reviewed-by: Michael Shych <michaelsh@nvidia.com>
Link: https://lore.kernel.org/r/20211002093238.3771419-5-vadimp@nvidia.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Extend the structure 'mlxreg_hotplug_device" with platform device field
to allow transition of the register map and system interrupt line number
to underlying hotplug devices, sharing the same register map and
same interrupt line with 'mlxreg-hotplug' driver.
Extend logic for hotplug devices creation and removing according to
the action associated with the hotplug device description. Previously
hotplug driver was capable to attach / de-attach upon hotplug events
only I2C devices handled by simple I2C drivers. Now it should be able
to attach also devices handled by the platform drivers.
The motivation is to allow transition of platform data like:
- system interrupt line number, sharing with 'mlxreg-hotplug' to
underlying hotplug devices.
- shared register map of programmable devices on main board to
underlying hotplug devices.
Additioanlly the number of 'sysfs' attributes is increased, since
modular system defines more 'sysfs' attributes.
Signed-off-by: Vadim Pasternak <vadimp@nvidia.com>
Reviewed-by: Michael Shych <michaelsh@nvidia.com>
Link: https://lore.kernel.org/r/20211002093238.3771419-4-vadimp@nvidia.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Add initial chassis management support for Nvidia modular Ethernet
switch systems MSN4800, providing a high performance switching solution
for Enterprise Data Centers (EDC) for building Ethernet based clusters,
High-Performance Computing (HPC) and embedded environments.
This system could be equipped with the different types of replaceable
line cards and management board. The first system flavor will support
the line card type MSN4800-C16 equipped with Lattice CPLD devices aimed
for system and ASIC control, one Nvidia FPGA for gearboxes (PHYs)
management, and four Nvidia gearboxes for the port control and with
16x100GbE QSFP28 ports and also with various devices for electrical
control.
The system is equipped with eight slots for line cards, four slots for
power supplies and six slots for fans. It could be configured as fully
populated or with even only one line card. The line cards are
hot-pluggable.
In the future when more line card flavors are to be available (for
example line cards with 8x200Gb Eth port, with 4x400 Eth ports, or with
some kind of smart cards for offloading purpose), any type of line card
could be inserted at any slot.
The system is based on Nvidia Spectrum-3 ASIC. The switch height is
4U and it fits standard rack size.
System could be configured as fully populated or with even only one
line card. The line cards are hot-pluggable.
Line cards are connected to the chassis through I2C interface for the
chassis management operations and through PCIe for the networking
operations. Future line cards could be connected to the chassis through
InfiniBand fabric, instead of PCIe.
The first type of line card supports 16x100GbE QSFP28 Ethernet ports.
Those line cards equipped with the programmable devices aimed for
system control of Nvidia Ethernet switch ASIC control, Nvidia FPGA,
Nvidia gearboxes (PHYs).
The next coming card generations are supposed to support:
- Line cards with 8x200Gbe QSFP28 Ethernet ports.
- Line cards with 4x400Gbe QSFP-DD Ethernet ports.
- Smart cards equipped with Nvidia ARM CPU for offloading and for fast
access to the storage (EBoF).
- Fabric cards for inter-connection.
The basic system initialization flow with input signals from the
programmable device to kernel hotplug driver and with OS response
to some of these signals is depicted below.
lc#n_prsnt *-> Input: line card presence in/out events.
Informational event. Required action - 'udev' event
generation for logging.
lc#n_verified *-> Input: line card verification status events coming
after line card security signature validation by
hardware. Required action - connect line card
driver and initialized line card devices feeding
from system auxiliary power domain.
lc#n_pwr <-* Output: line card power on / off from OS. Action
should be performed by platform power management
driver.
lc#n_powered *-> Input: line card power on/off events coming after
line card "power good" on/off events, mean that
line card power up sequence has been successfully
completed or line card "power good" status has been
dropped. Required action - connect line card
devices feeding from system main power domain.
lc#n_synced *-> Input: line card synchronization events, coming
after hardware-firmware synchronization handshake.
Required action - to enable line card, in case
lc#n_ready has been received before.
lc#n_ready *-> Input: line card ready events, indicating line card
PHYs ready / unready states. Required action -
enable line card, in case lc#n_synced has been
received before.
lc#n_enable <-* Output: line card enable from OS - release FPGA and
PHYs line card devices from reset state. Action
should be performed by platform power management
driver.
lc#n_active *-> Input: when line card "active event" is received
for particular line card, its network, hardware
monitoring and thermal interfaces should be
configured according to the configuration obtained
from the firmware. When opposite "inactive event"
is received all the above interfaces should be
teared down. Required action - connect / disconnect
the above line card interfaces through ASIC I2C
chassis management driver.
For initial support:
- Define new system type 'VMOD0011' to support new modular system.
- Provide initial platform configuration for new system type.
- Extend the registers definitions.
- Add support for modular system registers related to line card
specific events - insertion/removal, power on/off, verification
and activation.
- Add hotplug configuration for the above events.
- Add configurations for hotplug actions for the modular system.
Signed-off-by: Vadim Pasternak <vadimp@nvidia.com>
Reviewed-by: Michael Shych <michaelsh@nvidia.com>
Link: https://lore.kernel.org/r/20211002093238.3771419-3-vadimp@nvidia.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Add new types for the Nvidia modular systems MSN4800 which could
be equipped with the different types of replaceable line cards.
Add new type to specify the kind of hotplug events for the line cards.
The line card events are generated by the programmable device located
on the main board. This device implements interrupt controller logic.
Line card interrupts are associated with different line cards states
during its initialization: insertion, security signature validation,
power good state, security validation, hardware-firmware
synchronization state, line card PHYs readiness state, firmware
availability for line card ports. Also under some circumstances
hardware can generate thermal shutdown for particular line card.
Add new type specifying the action, which should be performed when
particular hotplug event is received. This action defines in which way
hotplug event should be handled by hotplug driver. There are the next
actions types:
- Connect I2C device with empty 'platform_data' field according to the
platform topology, if device is configured (for example, power unit
micro-controller driver, when power unit is connected to power source
(this is what is currently supported).
- Connect device with 'platform_data' field set according to the
platform topology. The purpose is to pass 'platform_data' through
hotplug driver to underlying device (for example line card driver).
- No device is associated with hotplug event - just send "udev" event
(this is what is currently supported).
Extend structure 'mlxreg_hotplug_device' with hotplug action field.
Extend structure 'mlxreg_core_data' with:
- Registers for line card power and enabling control.
- Slot number field, to indicate at which physical slot replaceable
line card device is located.
Signed-off-by: Vadim Pasternak <vadimp@nvidia.com>
Reviewed-by: Michael Shych <michaelsh@nvidia.com>
Link: https://lore.kernel.org/r/20211002093238.3771419-2-vadimp@nvidia.com
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
struct gtt_range represents a GEM object and should not be used for GTT
setup. Change psb_gtt_insert() and psb_gtt_remove() to receive all
necessary parameters from their caller. This also eliminates possible
failure from psb_gtt_insert().
There's one exception in psb_gtt_restore(), which requires an upcast
from struct resource to struct gtt_range when restoring the GTT after
hibernation. A possible solution would track the GEM objects that need
restoration separately from the GTT resource.
Rename the functions to psb_gtt_insert_pages() and psb_gtt_remove_pages()
to reflect their similarity to MMU interfaces.
v3:
* restore the comments about locking rules (Patrik)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211015084053.13708-10-tzimmermann@suse.de
psb_gtt_alloc_range() allocates struct gtt_range, create the GTT resource
and performs some half-baked initialization. Inline the function into its
only caller psb_gem_create(). For creating the GTT resource, introduce a
new helper, psb_gtt_alloc_resource() that hides the details of the GTT.
For psb_gtt_free_range(), inline the function into its only caller
psb_gem_free_object(). While at it, remove the explicit invocation of
drm_gem_free_mmap_offset(). The mmap offset is already released by
drm_gem_object_release().
v3:
* replace offset[static 1] with pointer notation (Patrik)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211015084053.13708-8-tzimmermann@suse.de
psb_gtt_attach_pages() are not GTT functions but deal with the GEM
object's SHMEM pages. The only callers of psb_gtt_attach_pages() and
psb_gtt_detach_pages() are the GEM pin helpers. Inline the calls and
cleanup the resulting code.
v2:
* unlock gtt_mutex in pin-error handling (Patrik)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211015084053.13708-7-tzimmermann@suse.de
Support private objects for stolen memory in psb_gem_create() and
convert users to psb_gem_create(). For stolen memory, psb_gem_create()
now initializes the GEM object via drm_gem_private_object_init().
In the fbdev setup, replace the open-coded initialization of struct
gtt_range with a call to psb_gem_create(). Use drm_gem_object_put()
for release.
In the cursor setup, use psb_gem_create() and get a real GEM object.
Previously the allocated instance of struct gtt_range was only partially
initialized. Release the cursor GEM object in gma_crtc_destroy(). The
release was missing from the original code.
With the conversion of all callers to psb_gem_create(), the extern
declarations of psb_gtt_alloc_range, psb_gtt_free_range and
psb_gem_object_func are not required any longer. Declare them as
static.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211015084053.13708-5-tzimmermann@suse.de
Implement psb_gem_create() for general use. Create the GEM handle in
psb_gem_create_dumb(). Allows to use psb_gem_create() for creating all
of the GEM objects.
While at it, clean-up drm_gem_dumb_create() to make it more readable.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211015084053.13708-4-tzimmermann@suse.de
Allocation and pinning helpers for struct gtt_range are GEM functions,
so move them to gem.c. No functional changes.
v2:
* keep docs for psb_gtt_{attach,detach}_pages() (Patrik)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211015084053.13708-2-tzimmermann@suse.de
Reuse the 8b/10b link training delay helpers. Functionally this skips
the check for invalid values for DPCD 1.4 and later at clock recovery
delay (as it's a fixed delay and bypasses the rd_interval) but the same
value will be checked and invalid values reported at channel
equalization.
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211014150059.28957-2-jani.nikula@intel.com
