Bjorn Helgaas [Tue, 29 Aug 2023 16:03:56 +0000 (11:03 -0500)]
Merge branch 'pci/controller/resources'
- Use Use devm_platform_get_and_ioremap_resource() instead of open-coding
platform_get_resource() followed by devm_ioremap_resource() (Yang Li)
* pci/controller/resources:
PCI: imx6: Use devm_platform_get_and_ioremap_resource()
PCI: xgene-msi: Use devm_platform_get_and_ioremap_resource()
PCI: v3: Use devm_platform_get_and_ioremap_resource()
PCI: rcar-gen2: Use devm_platform_get_and_ioremap_resource()
Bjorn Helgaas [Tue, 29 Aug 2023 16:03:56 +0000 (11:03 -0500)]
Merge branch 'pci/controller/remove-void-cast'
- Add stubs for devm action functions that call clk_disable_unprepare() to
avoid casts between incompatible function types (Krzysztof Wilczyński)
* pci/controller/remove-void-cast:
PCI: microchip: Remove cast between incompatible function type
PCI: keembay: Remove cast between incompatible function type
PCI: meson: Remove cast between incompatible function type
Bjorn Helgaas [Tue, 29 Aug 2023 16:03:55 +0000 (11:03 -0500)]
Merge branch 'pci/controller/tegra194'
- Revert "PCI: tegra194: Enable support for 256 Byte payload" because Linux
doesn't know how to reduce MPS from to 256 to 128 bytes for Endpoints
below a Switch (because other devices below the Switch might already be
operating), which leads to Malformed TLP errors (Vidya Sagar)
* pci/controller/tegra194:
Revert "PCI: tegra194: Enable support for 256 Byte payload"
* pci/controller/qcom-edma:
PCI: endpoint: Add kernel-doc for pci_epc_mem_init() API
PCI: epf-mhi: Use iATU for small transfers
PCI: epf-mhi: Add support for SM8450
PCI: epf-mhi: Add eDMA support
PCI: qcom-ep: Add eDMA support
PCI: epf-mhi: Make use of the alignment restriction from EPF core
PCI: qcom-ep: Pass alignment restriction to the EPF core
- Read FPGA MSI configuration parameters from hardware instead of
hard-coding them (Daire McNamara)
- Re-partition code between probe() and init() (Daire McNamara)
* pci/controller/microchip:
PCI: microchip: Re-partition code between probe() and init()
PCI: microchip: Gather MSI information from hardware config registers
PCI: microchip: Clean up initialisation of interrupts
PCI: microchip: Enable event handlers to access bridge and control pointers
PCI: microchip: Align register, offset, and mask names with HW docs
PCI: microchip: Enable building driver as a module
PCI: microchip: Correct the DED and SEC interrupt bit offsets
Bjorn Helgaas [Tue, 29 Aug 2023 16:03:53 +0000 (11:03 -0500)]
Merge branch 'pci/controller/layerscape'
- Add support for link-down notification so the endpoint driver can process
LINK_DOWN events (Frank Li)
- Save Link Capabilities during probe so they can be restored when handling
a link-up event, since the controller loses the Link Width and Link Speed
values during reset (Xiaowei Bao)
* pci/controller/layerscape:
PCI: layerscape: Add workaround for lost link capabilities during reset
PCI: layerscape: Add support for link-down notification
Bjorn Helgaas [Tue, 29 Aug 2023 16:03:52 +0000 (11:03 -0500)]
Merge branch 'pci/controller/dwc'
- Add an imx6 .host_deinit() callback so we can clean up things like
regulators on probe failure or driver unload (Mark Brown)
- Add PCIE_PME_TO_L2_TIMEOUT_US value for controller drivers that need to
manually synchronize power removal (Frank Li)
- Add generic dwc suspend/resume APIs (dw_pcie_suspend_noirq() and
dw_pcie_resume_noirq()) to be called by controller driver suspend/resume
ops, and a controller callback to send PME_Turn_Off (Frank Li)
- Add layerscape suspend/resume support, including manual
PME_Turn_off/PME_TO_Ack handshake (Hou Zhiqiang, Frank Li)
* pci/controller/dwc:
PCI: layerscape: Add power management support for ls1028a
PCI: dwc: Implement generic suspend/resume functionality
PCI: Add PCIE_PME_TO_L2_TIMEOUT_US L2 ready timeout value
PCI: dwc: Provide deinit callback for i.MX
Bjorn Helgaas [Tue, 29 Aug 2023 16:03:49 +0000 (11:03 -0500)]
Merge branch 'pci/pcie-rmw'
- Add locking for read/modify/write PCIe Capability Register accessors for
Link Control and Root Control (Ilpo Järvinen)
- Use PCIe RMW accessors for Link Control updates in PCI core, pciehp,
amdgpu, radeon, mlx5, ath10k, ath11k, ath12k (Ilpo Järvinen)
- Convert PCIBIOS error values in mlx5 to generic errnos (Ilpo Järvinen)
- Simplify pcie_capability_clear_and_set_word() control flow (Bjorn
Helgaas)
* pci/pcie-rmw:
PCI: Simplify pcie_capability_clear_and_set_word() control flow
net/mlx5: Convert PCI error values to generic errnos
PCI: Document the Capability accessor RMW improvements
wifi: ath10k: Use RMW accessors for changing LNKCTL
wifi: ath12k: Use RMW accessors for changing LNKCTL
wifi: ath11k: Use RMW accessors for changing LNKCTL
net/mlx5: Use RMW accessors for changing LNKCTL
drm/radeon: Use RMW accessors for changing LNKCTL
drm/amdgpu: Use RMW accessors for changing LNKCTL
PCI/ASPM: Use RMW accessors for changing LNKCTL
PCI: pciehp: Use RMW accessors for changing LNKCTL
PCI: Make link retraining use RMW accessors for changing LNKCTL
PCI: Add locking to RMW PCI Express Capability Register accessors
Bjorn Helgaas [Tue, 29 Aug 2023 16:03:49 +0000 (11:03 -0500)]
Merge branch 'pci/ioport'
- Make I/O resources depend on CONFIG_HAS_IOPORT so inw() and friends can
be completely omitted on architectures without I/O port support (Niklas
Schnelle)
* pci/ioport:
PCI/sysfs: Make I/O resource depend on HAS_IOPORT
PCI: Make quirk using inw() depend on HAS_IOPORT
Add support for voting interconnect (ICC) bandwidth based
on the link speed and width.
This commit is inspired from the basic interconnect support added
to pcie-qcom driver in commit c4860af88d0c ("PCI: qcom: Add basic
interconnect support").
The interconnect support is kept optional to be backward compatible
with legacy device trees.
Some platforms may not boot if a device driver doesn't
initialize the interconnect path. Mostly it is handled
by the bootloader but we have starting to see cases
where bootloader simply ignores them.
Add the "pcie-mem" & "cpu-pcie" interconnect path as a required
property to the bindings.
PCI: qcom-ep: Treat unknown IRQ events as an error
Sometimes, the Qcom PCIe EP controller can receive some interrupts
unknown to the driver, like safety interrupts in newer SoCs. In those
cases, if the driver doesn't clear the interrupts, it will end up in an
interrupt storm. However, the users will not know about it because the
log is treated as a debug message.
So let's treat the unknown event log as an error so that it at least
makes the user aware, thereby getting fixed eventually.
Commit c0aba9f32801 ("dt-bindings: PCI: qcom: Add SDX65 SoC") adding
SDX65 was never tested and is clearly bogus. The qcom,sdx65-pcie-ep
compatible is followed by a fallback in DTS, and there is no driver
matched by this compatible. Driver matches by its fallback
qcom,sdx55-pcie-ep. This also fixes dtbs_check warnings like:
qcom-sdx65-mtp.dtb: pcie-ep@1c00000: compatible: ['qcom,sdx65-pcie-ep', 'qcom,sdx55-pcie-ep'] is too long
For transfers below 4K, let's use iATU since using eDMA for such small
transfers is inefficient.
This is mainly because setting up an eDMA transfer and waiting for
completion adds some latency. This latency is negligible for large
transfers but not for the smaller ones.
With using iATU, there is an increase in ~50Mbps throughput on both MHI
UL (Uplink) and DL (Downlink) channels.
Add support for Qualcomm Snapdragon SM8450 SoC to the EPF driver. SM8450
has the dedicated PID (0x0306) and supports eDMA. Currently, it has no
fixed PCI class, so it is being advertised as "PCI_CLASS_OTHERS".
Add support for Embedded DMA (eDMA) available in the DesignWare PCIe IP
to transfer the MHI buffers between the host and the endpoint. The eDMA
use helps achieve greater throughput as the transfers are offloaded from
CPUs.
For differentiating the iATU and eDMA APIs, the pci_epf_mhi_{read/write}
APIs are renamed to pci_epf_mhi_iatu_{read/write} and separate eDMA
specific APIs pci_epf_mhi_edma_{read/write} are introduced.
Platforms that require eDMA support can pass the MHI_EPF_USE_DMA flag
through pci_epf_mhi_ep_info.
PCI: epf-mhi: Make use of the alignment restriction from EPF core
Instead of hardcoding the alignment restriction in the EPF_MHI driver, make
use of the info available from the EPF core that reflects the alignment
restriction of the endpoint controller.
For this purpose, let's introduce the get_align_offset() static function.
Feiyang Chen [Thu, 24 Aug 2023 01:37:38 +0000 (09:37 +0800)]
PCI/PM: Only read PCI_PM_CTRL register when available
For a device with no Power Management Capability, pci_power_up() previously
returned 0 (success) if the platform was able to put the device in D0,
which led to pci_set_full_power_state() trying to read PCI_PM_CTRL, even
though it doesn't exist.
Since dev->pm_cap == 0 in this case, pci_set_full_power_state() actually
read the wrong register, interpreted it as PCI_PM_CTRL, and corrupted
dev->current_state. This led to messages like this in some cases:
pci 0000:01:00.0: Refused to change power state from D3hot to D0
To prevent this, make pci_power_up() always return a negative failure code
if the device lacks a Power Management Capability, even if non-PCI platform
power management has been able to put the device in D0. The failure will
prevent pci_set_full_power_state() from trying to access PCI_PM_CTRL.
PCI: qcom-ep: Pass alignment restriction to the EPF core
Qcom PCIe EP controllers have 4K alignment restriction for the outbound
window address. Hence, pass this info to the EPF core so that the EPF
drivers can make use of this info.
Jim Quinlan [Fri, 23 Jun 2023 14:40:57 +0000 (10:40 -0400)]
PCI: brcmstb: Assert PERST# on BCM2711
The current PCIe driver assumes PERST# is asserted when probe() is invoked.
Some older versions of the 2711/RPi bootloader left PERST# unasserted, as
the Raspian OS does assert PERST# on probe(). For this reason, we assert
PERST# for BCM2711 SOCs (i.e. RPi).
Xiaowei Bao [Thu, 20 Jul 2023 13:58:34 +0000 (09:58 -0400)]
PCI: layerscape: Add workaround for lost link capabilities during reset
The endpoint controller loses the Maximum Link Width and Supported Link Speed
value from the Link Capabilities Register - initially configured by the Reset
Configuration Word (RCW) - during a link-down or hot reset event.
Sui Jingfeng [Tue, 8 Aug 2023 22:34:12 +0000 (06:34 +0800)]
PCI/VGA: Replace full MIT license text with SPDX identifier
Per Documentation/process/license-rules.rst, the SPDX MIT identifier is
equivalent to including the entire MIT license text from
LICENSES/preferred/MIT.
Replace the MIT license text with the equivalent SPDX identifier.
Sui Jingfeng [Tue, 8 Aug 2023 22:34:08 +0000 (06:34 +0800)]
PCI/VGA: Simplify vga_client_register()
Reorganize vga_client_register() to avoid the goto and the need to save the
return value. Update the kernel-doc to reflect -ENODEV on failure. No
functional change intended.
Sui Jingfeng [Tue, 8 Aug 2023 22:34:06 +0000 (06:34 +0800)]
PCI/VGA: Simplify vga_arbiter_notify_clients()
In vga_arbiter_notify_clients(), "new_state" was computed during every loop
iteration even though it doesn't depend on anything that changes during the
loop. Move the computation outside the loop.
Sui Jingfeng [Tue, 8 Aug 2023 22:34:05 +0000 (06:34 +0800)]
PCI/VGA: Correct vga_update_device_decodes() parameter type
Previously vga_update_device_decodes() took "int new_decodes", but the
callers pass "unsigned int new_decodes". Correct the
vga_update_device_decodes() parameter type to "unsigned int" to match.
In vga_arbiter_notify_clients(), the return from vgadev->set_decode() is
"unsigned int" but was stored as "uint32_t new_decodes". Correct the
new_decodes type to "unsigned int".
Sui Jingfeng [Tue, 8 Aug 2023 22:34:02 +0000 (06:34 +0800)]
PCI/VGA: Correct vga_str_to_iostate() io_state parameter type
Previously vga_str_to_iostate() took "int *io_state", but vga_arb_write()
is the only caller and it passes "unsigned int *". Make the
vga_str_to_iostate() parameter type "unsigned int *" to match.
Yong-Xuan Wang [Mon, 7 Aug 2023 05:56:21 +0000 (05:56 +0000)]
PCI: fu740: Set the number of MSI vectors
The iMSI-RX module of the DW PCIe controller provides multiple sets of
MSI_CTRL_INT_i_* registers, and each set is capable of handling 32 MSI
interrupts. However, the fu740 PCIe controller driver only enabled one set
of MSI_CTRL_INT_i_* registers, as the total number of supported interrupts
was not specified.
Set the supported number of MSI vectors to enable all the MSI_CTRL_INT_i_*
registers on the fu740 PCIe core, allowing the system to fully utilize the
available MSI interrupts.
Dexuan Cui [Wed, 16 Aug 2023 17:59:39 +0000 (10:59 -0700)]
PCI: hv: Fix a crash in hv_pci_restore_msi_msg() during hibernation
When a Linux VM with an assigned PCI device runs on Hyper-V, if the PCI
device driver is not loaded yet (i.e. MSI-X/MSI is not enabled on the
device yet), doing a VM hibernation triggers a panic in
hv_pci_restore_msi_msg() -> msi_lock_descs(&pdev->dev), because
pdev->dev.msi.data is still NULL.
Avoid the panic by checking if MSI-X/MSI is enabled.
Nirmal Patel [Thu, 10 Aug 2023 21:50:29 +0000 (17:50 -0400)]
PCI: vmd: Disable bridge window for domain reset
During domain reset process vmd_domain_reset() clears PCI
configuration space of VMD root ports. But certain platform
has observed following errors and failed to boot.
...
DMAR: VT-d detected Invalidation Queue Error: Reason f
DMAR: VT-d detected Invalidation Time-out Error: SID ffff
DMAR: VT-d detected Invalidation Completion Error: SID ffff
DMAR: QI HEAD: UNKNOWN qw0 = 0x0, qw1 = 0x0
DMAR: QI PRIOR: UNKNOWN qw0 = 0x0, qw1 = 0x0
DMAR: Invalidation Time-out Error (ITE) cleared
The root cause is that memset_io() clears prefetchable memory base/limit
registers and prefetchable base/limit 32 bits registers sequentially.
This seems to be enabling prefetchable memory if the device disabled
prefetchable memory originally.
So, prefetchable memory is ffffffff00000000-575000fffff, which is
disabled. When memset_io() clears prefetchable base 32 bits register,
the prefetchable memory becomes 0000000000000000-575000fffff, which is
enabled and incorrect.
Here is the quote from section 7.5.1.3.9 of PCI Express Base 6.0 spec:
The Prefetchable Memory Limit register must be programmed to a smaller
value than the Prefetchable Memory Base register if there is no
prefetchable memory on the secondary side of the bridge.
This is believed to be the reason for the failure and in addition the
sequence of operation in vmd_domain_reset() is not following the PCIe
specs.
Disable the bridge window by executing a sequence of operations
borrowed from pci_disable_bridge_window() and pci_setup_bridge_io(),
that comply with the PCI specifications.
Ilpo Järvinen [Mon, 14 Aug 2023 13:27:20 +0000 (16:27 +0300)]
net/mlx5: Convert PCI error values to generic errnos
mlx5_pci_link_toggle() returns a mix of PCI-specific error codes and
generic errnos.
Convert the PCI-specific error values to generic errno using
pcibios_err_to_errno() before returning them.
Fixes: eabe8e5e88f5 ("net/mlx5: Handle sync reset now event") Fixes: 212b4d7251c1 ("net/mlx5: Wait for firmware to enable CRS before pci_restore_state") Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Ilpo Järvinen <[email protected]>
[bhelgaas: rebase to pci/pcie-rmw, also convert in mlx5_check_dev_ids()] Signed-off-by: Bjorn Helgaas <[email protected]>
Ilpo Järvinen [Mon, 17 Jul 2023 12:05:03 +0000 (15:05 +0300)]
PCI: Document the Capability accessor RMW improvements
Documentation claims port service drivers should play nice with respect to
PCIe Capability changes, but the concurrency control is now provided in the
Capability accessors as long as the correct ones are used.
Update the documention to match the RMW accessor behavior.
Ilpo Järvinen [Mon, 17 Jul 2023 12:05:02 +0000 (15:05 +0300)]
wifi: ath10k: Use RMW accessors for changing LNKCTL
Don't assume that only the driver would be accessing LNKCTL. ASPM policy
changes can trigger write to LNKCTL outside of driver's control.
Use RMW capability accessors which does proper locking to avoid losing
concurrent updates to the register value. On restore, clear the ASPMC field
properly.
Ilpo Järvinen [Mon, 17 Jul 2023 12:05:01 +0000 (15:05 +0300)]
wifi: ath12k: Use RMW accessors for changing LNKCTL
Don't assume that only the driver would be accessing LNKCTL. ASPM policy
changes can trigger write to LNKCTL outside of driver's control.
Use RMW capability accessors which do proper locking to avoid losing
concurrent updates to the register value. On restore, clear the ASPMC field
properly.
Ilpo Järvinen [Mon, 17 Jul 2023 12:05:00 +0000 (15:05 +0300)]
wifi: ath11k: Use RMW accessors for changing LNKCTL
Don't assume that only the driver would be accessing LNKCTL. ASPM policy
changes can trigger write to LNKCTL outside of driver's control.
Use RMW capability accessors which do proper locking to avoid losing
concurrent updates to the register value. On restore, clear the ASPMC field
properly.
Ilpo Järvinen [Mon, 17 Jul 2023 12:04:59 +0000 (15:04 +0300)]
net/mlx5: Use RMW accessors for changing LNKCTL
Don't assume that only the driver would be accessing LNKCTL of the upstream
bridge. ASPM policy changes can trigger write to LNKCTL outside of driver's
control.
Use RMW capability accessors which do proper locking to avoid losing
concurrent updates to the register value.
Ilpo Järvinen [Mon, 17 Jul 2023 12:04:58 +0000 (15:04 +0300)]
drm/radeon: Use RMW accessors for changing LNKCTL
Don't assume that only the driver would be accessing LNKCTL. ASPM policy
changes can trigger write to LNKCTL outside of driver's control. And in
the case of upstream bridge, the driver does not even own the device it's
changing the registers for.
Use RMW capability accessors which do proper locking to avoid losing
concurrent updates to the register value.
Ilpo Järvinen [Mon, 17 Jul 2023 12:04:57 +0000 (15:04 +0300)]
drm/amdgpu: Use RMW accessors for changing LNKCTL
Don't assume that only the driver would be accessing LNKCTL. ASPM policy
changes can trigger write to LNKCTL outside of driver's control. And in
the case of upstream bridge, the driver does not even own the device it's
changing the registers for.
Use RMW capability accessors which do proper locking to avoid losing
concurrent updates to the register value.
Alex Williamson [Thu, 3 Aug 2023 17:12:33 +0000 (11:12 -0600)]
PCI: Fix runtime PM race with PME polling
Testing that a device is not currently in a low power state provides no
guarantees that the device is not imminently transitioning to such a state.
Increment the PM usage counter before accessing the device. Since we don't
wish to wake the device for PME polling, do so only if the device is
already active by using pm_runtime_get_if_active().
Alex Williamson [Thu, 3 Aug 2023 17:12:32 +0000 (11:12 -0600)]
PCI/VPD: Add runtime power management to sysfs interface
Unlike default access to config space through sysfs, the VPD read and write
functions don't actively manage the runtime power management state of the
device during access. Since commit 7ab5e10eda02 ("vfio/pci: Move the
unused device into low power state with runtime PM"), the vfio-pci driver
will use runtime power management and release unused devices to make use of
low power states. Attempting to access VPD information in D3cold can
result in incorrect information or kernel crashes depending on the system
behavior.
Wrap the VPD read/write bin attribute handlers in runtime PM and take into
account the potential quirk to select the correct device to wake.
Ilpo Järvinen [Mon, 17 Jul 2023 12:04:56 +0000 (15:04 +0300)]
PCI/ASPM: Use RMW accessors for changing LNKCTL
Don't assume that the device is fully under the control of ASPM and use RMW
capability accessors which do proper locking to avoid losing concurrent
updates to the register values.
If configuration fails in pcie_aspm_configure_common_clock(), the
function attempts to restore the old PCI_EXP_LNKCTL_CCC settings. Store
only the old PCI_EXP_LNKCTL_CCC bit for the relevant devices rather
than the content of the whole LNKCTL registers. It aligns better with
how pcie_lnkctl_clear_and_set() expects its parameter and makes the
code more obvious to understand.
Ilpo Järvinen [Mon, 17 Jul 2023 12:04:54 +0000 (15:04 +0300)]
PCI: Make link retraining use RMW accessors for changing LNKCTL
Don't assume that the device is fully under the control of PCI core. Use
RMW capability accessors in link retraining which do proper locking to
avoid losing concurrent updates to the register values.
Ilpo Järvinen [Mon, 17 Jul 2023 12:04:53 +0000 (15:04 +0300)]
PCI: Add locking to RMW PCI Express Capability Register accessors
Many places in the kernel write the Link Control and Root Control PCI
Express Capability Registers without proper concurrency control and this
could result in losing the changes one of the writers intended to make.
Add pcie_cap_lock spinlock into the struct pci_dev and use it to protect
bit changes made in the RMW capability accessors. Protect only a selected
set of registers by differentiating the RMW accessor internally to
locked/unlocked variants using a wrapper which has the same signature as
pcie_capability_clear_and_set_word(). As the Capability Register (pos)
given to the wrapper is always a constant, the compiler should be able to
simplify all the dead-code away.
So far only the Link Control Register (ASPM, hotplug, link retraining,
various drivers) and the Root Control Register (AER & PME) seem to
require RMW locking.
Suggested-by: Lukas Wunner <[email protected]> Fixes: c7f486567c1d ("PCI PM: PCIe PME root port service driver") Fixes: f12eb72a268b ("PCI/ASPM: Use PCI Express Capability accessors") Fixes: 7d715a6c1ae5 ("PCI: add PCI Express ASPM support") Fixes: affa48de8417 ("staging/rdma/hfi1: Add support for enabling/disabling PCIe ASPM") Fixes: 849a9366cba9 ("misc: rtsx: Add support new chip rts5228 mmc: rtsx: Add support MMC_CAP2_NO_MMC") Fixes: 3d1e7aa80d1c ("misc: rtsx: Use pcie_capability_clear_and_set_word() for PCI_EXP_LNKCTL") Fixes: c0e5f4e73a71 ("misc: rtsx: Add support for RTS5261") Fixes: 3df4fce739e2 ("misc: rtsx: separate aspm mode into MODE_REG and MODE_CFG") Fixes: 121e9c6b5c4c ("misc: rtsx: modify and fix init_hw function") Fixes: 19f3bd548f27 ("mfd: rtsx: Remove LCTLR defination") Fixes: 773ccdfd9cc6 ("mfd: rtsx: Read vendor setting from config space") Fixes: 8275b77a1513 ("mfd: rts5249: Add support for RTS5250S power saving") Fixes: 5da4e04ae480 ("misc: rtsx: Add support for RTS5260") Fixes: 0f49bfbd0f2e ("tg3: Use PCI Express Capability accessors") Fixes: 5e7dfd0fb94a ("tg3: Prevent corruption at 10 / 100Mbps w CLKREQ") Fixes: b726e493e8dc ("r8169: sync existing 8168 device hardware start sequences with vendor driver") Fixes: e6de30d63eb1 ("r8169: more 8168dp support.") Fixes: 8a06127602de ("Bluetooth: hci_bcm4377: Add new driver for BCM4377 PCIe boards") Fixes: 6f461f6c7c96 ("e1000e: enable/disable ASPM L0s and L1 and ERT according to hardware errata") Fixes: 1eae4eb2a1c7 ("e1000e: Disable L1 ASPM power savings for 82573 mobile variants") Fixes: 8060e169e02f ("ath9k: Enable extended synch for AR9485 to fix L0s recovery issue") Fixes: 69ce674bfa69 ("ath9k: do btcoex ASPM disabling at initialization time") Fixes: f37f05503575 ("mt76: mt76x2e: disable pcie_aspm by default") Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Ilpo Järvinen <[email protected]> Signed-off-by: Bjorn Helgaas <[email protected]> Reviewed-by: "Rafael J. Wysocki" <[email protected]>
NVIDIA T4 GPUs do not work with SBR. This problem is found when the T4 card
is direct attached to a Root Port only. Avoid bus reset by marking T4 GPUs
PCI_DEV_FLAGS_NO_BUS_RESET.
Mark Brown [Mon, 31 Jul 2023 11:55:01 +0000 (12:55 +0100)]
PCI: dwc: Provide deinit callback for i.MX
The i.MX integration for the DesignWare PCI controller has a _host_exit()
operation which undoes everything that the _host_init() operation does but
does not wire this up as the host_deinit callback for the core, or call it
in any path other than suspend. This means that if we ever unwind the
initial probe of the device, for example because it fails, the regulator
core complains that the regulators for the device were left enabled:
imx6q-pcie 33800000.pcie: iATU: unroll T, 4 ob, 4 ib, align 64K, limit 16G
imx6q-pcie 33800000.pcie: Phy link never came up
imx6q-pcie 33800000.pcie: Phy link never came up
imx6q-pcie: probe of 33800000.pcie failed with error -110
------------[ cut here ]------------
WARNING: CPU: 2 PID: 46 at drivers/regulator/core.c:2396 _regulator_put+0x110/0x128
Wire up the callback so that the core can clean up after itself.
PCI: microchip: Re-partition code between probe() and init()
Continuing to use pci_host_common_probe() for the PCIe Root Complex on
PolarFire SoC is leading to an extremely large _init() function and some
unnatural code flow. Re-partition the code so that some tasks are done
in a _probe() routine, which calls pci_host_common_probe() and then use
a much smaller _init() function, mainly to enable interrupts after
address translation tables are set up.
PCI: microchip: Gather MSI information from hardware config registers
The PCIe Root Complex on PolarFire SoC is configured at bitstream creation
time using Libero. Key MSI-related parameters include the number of
MSIs (1/2/4/8/16/32) and the MSI address. In the device driver, extract
this information from hardware registers at init time, and use it to configure
MSI system, including configuring MSI capability structure correctly in
configuration space.
PCI: microchip: Clean up initialisation of interrupts
Refactor interrupt handling in _init() function into
disable_interrupts(), init_interrupts(), clear_sec_errors() and clear
ded_errors() because current code is unwieldy and prone to bugs.
Disable interrupts as soon as possible and only enable interrupts after
address translation is setup to prevent spurious axi2pcie and pcie2axi
translation errors being reported.
PCI: microchip: Enable event handlers to access bridge and control pointers
Minor re-organisation so that event handlers can access both a pointer
to the bridge area of the PCIe Root Port and the control area of the PCIe
Root Port.
PCI: microchip: Align register, offset, and mask names with HW docs
Minor code re-organisation so that macros representing registers ascend in
numerical order and use the same names as their hardware documentation.
Removed registers not used by the driver.
PCI: microchip: Enable building driver as a module
Enable building driver as a module. The expected use case is the
driver is built as a module, is installed when needed, and cannot be
removed once installed since it is not possible to clean-up
the irq_chip data structures on removal, as described in:
The driver has .suppress_bind_attrs set to true for the same
reasons (ie prevent unbinding, that would leave the kernel
with stale IRQ configuration that cannot be cleaned up).
PCI: microchip: Correct the DED and SEC interrupt bit offsets
The SEC and DED interrupt bits are laid out the wrong way round so the SEC
interrupt handler attempts to mask, unmask, and clear the DED interrupt
and vice versa. Correct the bit offsets so that each interrupt handler
operates properly.
Vidya Sagar [Mon, 19 Jun 2023 10:26:04 +0000 (15:56 +0530)]
Revert "PCI: tegra194: Enable support for 256 Byte payload"
After commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256 Byte
payload"), we initialize MPS=256 for tegra194 Root Ports before enumerating
the hierarchy.
Consider an Endpoint that supports only MPS=128. In the default situation
(CONFIG_PCIE_BUS_DEFAULT set and no "pci=pcie_bus_*" parameter), Linux
tries to configure the MPS of every device to match the upstream bridge.
If the Endpoint is directly below the Root Port, Linux can reduce the Root
Port MPS to 128 to match the Endpoint. But if there's a switch in the
middle, Linux doesn't reduce the Root Port MPS because other devices below
the switch may already be configured with MPS larger than 128.
This scenario results in uncorrectable Malformed TLP errors if the Root
Port sends TLPs with payloads larger than 128 bytes. These errors can
be avoided by using the "pci=pcie_bus_safe" parameter, but it doesn't
seem to be a good idea to always have this parameter even for basic
functionality to work.
Revert commit 4fb8e46c1bc4 ("PCI: tegra194: Enable support for 256 Byte
payload") so the Root Ports default to MPS=128, which all devices
support.
If peer-to-peer DMA is not required, one can use "pci=pcie_bus_perf" to
get the benefit of larger MPS settings.
If legacy I/O spaces are not supported simply return an error when
trying to access them via pci_resource_io(). This allows inb() and
friends to become undefined when they are known at compile time to be
non-functional in a later patch.
PCI: rockchip: Use 64-bit mask on MSI 64-bit PCI address
A 32-bit mask was used on the 64-bit PCI address used for mapping MSIs.
This would result in the upper 32 bits being unintentionally zeroed and
MSIs getting mapped to incorrect PCI addresses if the address had any
of the upper bits set.
PCI: microchip: Remove cast between incompatible function type
Rather than casting void(*)(struct clk *) to void (*)(void *), that
forces conversion to an incompatible function type, replace the cast
with a small local stub function with a signature that matches what
the devm_add_action_or_reset() function expects.
The sub function takes a void *, then passes it directly to
clk_disable_unprepare(), which handles the more specific type.
Reported by clang when building with warnings enabled:
drivers/pci/controller/pcie-microchip-host.c:866:32: warning: cast from 'void (*)(struct clk *)' to 'void (*)(void *)' converts to incompatible function type [-Wcast-function-type-strict]
devm_add_action_or_reset(dev, (void (*) (void *))clk_disable_unprepare,
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
No functional changes are intended.
PCI: keembay: Remove cast between incompatible function type
Rather than casting void(*)(struct clk *) to void (*)(void *), that
forces conversion to an incompatible function type, replace the cast
with a small local stub function with a signature that matches what
the devm_add_action_or_reset() function expects.
The sub function takes a void *, then passes it directly to
clk_disable_unprepare(), which handles the more specific type.
Reported by clang when building with warnings enabled:
drivers/pci/controller/dwc/pcie-keembay.c:172:12: warning: cast from 'void (*)(struct clk *)' to 'void (*)(void *)' converts to incompatible function type [-Wcast-function-type-strict]
(void(*)(void *))clk_disable_unprepare,
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
No functional changes are intended.
PCI: meson: Remove cast between incompatible function type
Rather than casting void(*)(struct clk *) to void (*)(void *), that
forces conversion to an incompatible function type, replace the cast
with a small local stub function with a signature that matches what
the devm_add_action_or_reset() function expects.
The sub function takes a void *, then passes it directly to
clk_disable_unprepare(), which handles the more specific type.
Reported by clang when building with warnings enabled:
drivers/pci/controller/dwc/pci-meson.c:191:6: warning: cast from 'void (*)(struct clk *)' to 'void (*)(void *)' converts to incompatible function type [-Wcast-function-type-strict]
(void (*) (void *))clk_disable_unprepare,
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
No functional changes are intended.
Yang Li [Thu, 23 Mar 2023 07:45:53 +0000 (15:45 +0800)]
PCI: imx6: Use devm_platform_get_and_ioremap_resource()
Convert calls to platform_get_resource() and devm_ioremap_resource()
into a single call to devm_platform_get_and_ioremap_resource(), to
simplify the code.
This new helper added in commit 890cc39a8799 ("drivers: provide
devm_platform_get_and_ioremap_resource()") does what both of the
other separate functions do.
Yang Li [Thu, 23 Mar 2023 09:00:11 +0000 (17:00 +0800)]
PCI: xgene-msi: Use devm_platform_get_and_ioremap_resource()
Convert calls to platform_get_resource() and devm_ioremap_resource()
into a single call to devm_platform_get_and_ioremap_resource(), to
simplify the code.
This new helper added in commit 890cc39a8799 ("drivers: provide
devm_platform_get_and_ioremap_resource()") does what both of the
other separate functions do.
Yang Li [Thu, 23 Mar 2023 09:04:31 +0000 (17:04 +0800)]
PCI: v3: Use devm_platform_get_and_ioremap_resource()
Convert calls to platform_get_resource() and devm_ioremap_resource()
into a single call to devm_platform_get_and_ioremap_resource(), to
simplify the code.
This new helper added in commit 890cc39a8799 ("drivers: provide
devm_platform_get_and_ioremap_resource()") does what both of the
other separate functions do.
Yang Li [Thu, 23 Mar 2023 09:16:44 +0000 (17:16 +0800)]
PCI: rcar-gen2: Use devm_platform_get_and_ioremap_resource()
Convert calls to platform_get_resource() and devm_ioremap_resource()
into a single call to devm_platform_get_and_ioremap_resource(), to
simplify the code.
This new helper added in commit 890cc39a8799 ("drivers: provide
devm_platform_get_and_ioremap_resource()") does what both of the
other separate functions do.
PCI: qcom-ep: Switch MHI bus master clock off during L1SS
Currently, as part of the qcom_pcie_perst_deassert() function, instead
of writing the updated value to clear PARF_MSTR_AXI_CLK_EN, the variable
"val" is re-read.
This must be fixed to ensure that the master clock supplied to the MHI
bus is correctly gated during L1.1/L1.2 to save power.
Thus, replace the line that re-reads "val" with a line that writes the
updated value to the register to clear PARF_MSTR_AXI_CLK_EN.
Rob Herring [Fri, 9 Jun 2023 19:33:25 +0000 (13:33 -0600)]
PCI: iproc: Use of_property_read_bool() for boolean properties
Replace low-level property access functions like of_get_property() and
of_find_property() with their typed equivalents (i.e., of_property_read_<type>
functions), as it's preferred to use typed property access functions for
reading properties.
Sven Peter [Sat, 11 Mar 2023 13:34:53 +0000 (14:34 +0100)]
PCI: apple: Initialize pcie->nvecs before use
The apple_pcie_setup_port() function computes ilog2(pcie->nvecs) to set
up the number of MSIs available for each port. However, it's called
before apple_msi_init(), which initializes pcie->nvecs.
Luckily, pcie->nvecs is part of kzalloc()-ed structure and, as such,
initialized as zero. ilog2(0) happens to be 0xffffffff which then simply
configures more MSIs in hardware than we have. This doesn't break
anything because we never hand out those vectors.
Thus, swap the order of the two calls so that the correctly initialized
value is then used.
We just sorted the entries and fields last release, so just out of a
perverse sense of curiosity, I decided to see if we can keep things
ordered for even just one release.
The answer is "No. No we cannot".
I suggest that all kernel developers will need weekly training sessions,
involving a lot of Big Bird and Sesame Street. And at the yearly
maintainer summit, we will all sing the alphabet song together.
I doubt I will keep doing this. At some point "perverse sense of
curiosity" turns into just a cold dark place filled with sadness and
despair.
Repeats: 80e62bc8487b ("MAINTAINERS: re-sort all entries and fields") Signed-off-by: Linus Torvalds <[email protected]>
Merge tag 'dma-mapping-6.5-2023-07-09' of git://git.infradead.org/users/hch/dma-mapping
Pull dma-mapping fixes from Christoph Hellwig:
- swiotlb area sizing fixes (Petr Tesarik)
* tag 'dma-mapping-6.5-2023-07-09' of git://git.infradead.org/users/hch/dma-mapping:
swiotlb: reduce the number of areas to match actual memory pool size
swiotlb: always set the number of areas before allocating the pool
projects / linux.git / log