Bjorn Helgaas [Thu, 4 Aug 2022 16:41:57 +0000 (11:41 -0500)]
Merge branch 'pci/ctrl/mediatek-gen3'
- Fix refcount leak in mtk_pcie_init_irq_domains() (Miaoqian Lin)
- Print decoded LTSSM state when the link doesn't come up (Jianjun Wang)
* pci/ctrl/mediatek-gen3:
PCI: mediatek-gen3: Print LTSSM state when PCIe link down
PCI: mediatek-gen3: Fix refcount leak in mtk_pcie_init_irq_domains()
Bjorn Helgaas [Thu, 4 Aug 2022 16:41:55 +0000 (11:41 -0500)]
Merge branch 'pci/ctrl/imx6'
- Factor out ref clock disables to match enables (Bjorn Helgaas)
- Collect clock enables in imx6_pcie_clk_enable() (Richard Zhu)
- Propagate regulator and clock errors back to .host_init() caller (Richard
Zhu)
- Disable i.MX6QDL clock when disabling ref clocks (Richard Zhu)
- Call host init function directly in resume instead of duplicating the
code (Richard Zhu)
- Turn off regulators when suspending (Richard Zhu)
- Make link being down a non-fatal error so probe doesn't fail (Richard
Zhu)
- Start link in resume only if it was up before suspend to reduce resume
time (Richard Zhu)
- Move PHY init and power-on out of clock- and reset-related functions
(Richard Zhu)
- Rework suspend callback to be more symmetric with resume (Richard Zhu)
- Set PCIE_DBI_RO_WR_EN before writing DBI registers (Richard Zhu)
- Allow speeds faster than Gen2 (Richard Zhu)
* pci/ctrl/imx6:
PCI: imx6: Support more than Gen2 speed link mode
PCI: imx6: Set PCIE_DBI_RO_WR_EN before writing DBI registers
PCI: imx6: Reformat suspend callback to keep symmetric with resume
PCI: imx6: Move the imx6_pcie_ltssm_disable() earlier
PCI: imx6: Disable clocks in reverse order of enable
PCI: imx6: Do not hide PHY driver callbacks and refine the error handling
PCI: imx6: Reduce resume time by only starting link if it was up before suspend
PCI: imx6: Mark the link down as non-fatal error
PCI: imx6: Move regulator enable out of imx6_pcie_deassert_core_reset()
PCI: imx6: Turn off regulator when system is in suspend mode
PCI: imx6: Call host init function directly in resume
PCI: imx6: Disable i.MX6QDL clock when disabling ref clocks
PCI: imx6: Propagate .host_init() errors to caller
PCI: imx6: Collect clock enables in imx6_pcie_clk_enable()
PCI: imx6: Factor out ref clock disable to match enable
PCI: imx6: Move imx6_pcie_clk_disable() earlier
PCI: imx6: Move imx6_pcie_enable_ref_clk() earlier
PCI: imx6: Move PHY management functions together
PCI: imx6: Move imx6_pcie_grp_offset(), imx6_pcie_configure_type() earlier
PCI: imx6: Convert to NOIRQ_SYSTEM_SLEEP_PM_OPS()
- Move eDMA private data from struct dw_edma to struct dw_edma_chip (Frank
Li)
- Convert "struct dw_edma_region rg_region" to "void __iomem *reg_base"
since only the virtual address (not physical address or size) is used
(Frank Li)
- Rename "*_ch_cnt" to "ll_*_cnt" to reflect actual usage (Frank Li)
- Drop dma_slave_config.direction field usage (Serge Semin)
- Fix eDMA Rd/Wr-channels and DMA-direction semantics (Serge Semin)
- Add chip-specific DW_EDMA_CHIP_LOCAL flag to indicate that local eDMA
doesn't require generating MSIs to remote (Frank Li)
- Enable DMA tests for endpoints that support it (Frank Li)
* pci/ctrl/dwc-edma:
PCI: endpoint: Enable DMA tests for endpoints with DMA capabilities
dmaengine: dw-edma: Add support for chip-specific flags
dmaengine: dw-edma: Fix eDMA Rd/Wr-channels and DMA-direction semantics
dmaengine: dw-edma: Drop dma_slave_config.direction field usage
dmaengine: dw-edma: Rename wr(rd)_ch_cnt to ll_wr(rd)_cnt in struct dw_edma_chip
dmaengine: dw-edma: Change rg_region to reg_base in struct dw_edma_chip
dmaengine: dw-edma: Detach the private data and chip info structures
dmaengine: dw-edma: Remove unused irq field in struct dw_edma_chip
- Detect iATU region size from hardware (Serge Semin)
- Validate iATU outbound mappings against hardware constraints (Serge
Semin)
- Check for errors in iATU setup (Serge Semin)
- Allocate a 32-bit DMA-able page to be MSI target instead of using a
driver data structure that may not be addressable with 32-bit address
(Will McVicker)
- Use the bitmap API to allocate bitmaps instead of open-coding it
(Christophe JAILLET)
- Correct dw_pcie_free_msi() checking for when to remove IRQ handler and
data (Dmitry Baryshkov)
- Split MSI init to new dw_pcie_msi_host_init() function (Dmitry Baryshkov)
- Convert struct pcie_port.msi_irq to an array so we can support more than
32 MSI interrupts (Dmitry Baryshkov)
- Handle MSIs routed to multiple GIC interrupts for Qualcomm platforms with
groups of 32 MSI vectors (Dmitry Baryshkov)
- Add additional MSI interrupts to qcom DT (Dmitry Baryshkov)
* pci/ctrl/dwc:
dt-bindings: PCI: qcom: Support additional MSI vectors
PCI: dwc: Handle MSIs routed to multiple GIC interrupts
PCI: dwc: Convert struct pcie_port.msi_irq to an array
PCI: dwc: Split MSI IRQ parsing/allocation to a separate function
PCI: dwc: Correct msi_irq condition in dw_pcie_free_msi()
PCI: dwc: Use the bitmap API to allocate bitmaps
PCI: dwc: Fix MSI msi_msg DMA mapping
PCI: dwc: Check iATU in/outbound range setup status
PCI: dwc: Validate iATU outbound mappings against hardware constraints
PCI: dwc: Add iATU regions size detection procedure
PCI: dwc: Simplify in/outbound iATU setup methods
PCI: dwc: Drop enum dw_pcie_region_type in favor of PCIE_ATU_REGION_DIR_IB/OB
PCI: dwc: Drop enum dw_pcie_as_type in favor of PCIE_ATU_TYPE_MEM/IO
PCI: dwc: Add dw_pcie_ops.host_deinit() callback
PCI: tegra194: Drop manual DW PCIe controller version setup
PCI: intel-gw: Drop manual DW PCIe controller version setup
PCI: dwc: Add macros to compare Synopsys IP core versions
PCI: dwc: Read DWC IP core version from register
PCI: dwc: Use native DWC IP core version representation
PCI: dwc: Detect iATU settings after getting "addr_space" resource
PCI: dwc: Log link speed and width if it comes up
PCI: dwc-plat: Drop dw_plat_pcie_of_match[] forward declaration
PCI: dwc-plat: Drop unused regmap pointer
PCI: dwc-plat: Simplify dw_plat_pcie_probe() return values
PCI: dwc: Rename struct pcie_port to dw_pcie_rp
PCI: dwc: Move io_cfg_atu_shared to struct pcie_port
PCI: dwc: Add start_link/stop_link inlines
PCI: dwc: Reuse local pointer to the resource data
PCI: dwc: Organize local variable usage
PCI: dwc: Convert dw_pcie_link_up() to use dw_pcie_readl_dbi()
PCI: dwc: Simplify unrolled iATU detection
PCI: dwc: Add newlines to log messages
PCI: dwc: Add braces to multi-line if-else statements
PCI: dwc: Always enable CDM check if "snps,enable-cdm-check" exists
PCI: dwc: Deallocate EPC memory on dw_pcie_ep_init() errors
PCI: dwc: Set INCREASE_REGION_SIZE flag based on limit address
PCI: dwc: Disable outbound windows only for controllers using iATU
PCI: dwc: Add unroll iATU space support to dw_pcie_disable_atu()
PCI: dwc: Stop link on host_init errors and de-initialization
- Prevent config space access when link is down (Jim Quinlan)
- Split post-link up initialization to brcm_pcie_start_link() (Jim Quinlan)
- Enable child bus device regulators described under Root Ports in DT (Jim
Quinlan)
- Disable/enable regulators in suspend/resume (Jim Quinlan)
- Rename .map_bus() functions to end with 'map_bus' as they do in other
drivers (Jim Quinlan)
* pci/ctrl/brcmstb:
PCI: brcmstb: Rename .map_bus() functions to end with 'map_bus'
PCI: brcmstb: Disable/enable regulators in suspend/resume
PCI: brcmstb: Enable child bus device regulators from DT
PCI: brcmstb: Split post-link up initialization to brcm_pcie_start_link()
PCI: brcmstb: Prevent config space access when link is down
PCI: brcmstb: Remove unnecessary forward declarations
Bjorn Helgaas [Thu, 4 Aug 2022 16:41:52 +0000 (11:41 -0500)]
Merge branch 'pci/resource'
- Replace sparc pci_mmap_page_range() wrapper. This still leaves a
sparc-specific pci_mmap_resource_range(), but it's only one interface
instead of two (Arnd Bergmann)
- Remove sparc-specific pci_mmap_resource_range() by implementing
pci_iobar_pfn(). This removes the ability to map the entire PCI I/O
space using /proc/bus/pci, but we believe that's already been broken
since v2.6.28 (Arnd Bergmann)
* pci/resource:
sparc: Use generic pci_mmap_resource_range()
PCI: Remove pci_mmap_page_range() wrapper
Bjorn Helgaas [Thu, 4 Aug 2022 16:41:52 +0000 (11:41 -0500)]
Merge branch 'pci/err'
- Recognize disconnected devices so we don't bother trying to set them to
"frozen" or "normal" state (Christoph Hellwig)
- Clear PCI Status register during enumeration in case firmware left errors
logged (Kai-Heng Feng)
- Configure ECRC for every device, including hot-added ones (Stefan Roese)
- Keep AER error reporting enabled for switches (Stefan Roese)
- Enable error reporting for all devices that support AER (Stefan Roese)
- Iterate over error counters instead of error strings to avoid printing
junk in AER sysfs counters (Mohamed Khalfella)
* pci/err:
PCI/AER: Iterate over error counters instead of error strings
PCI/AER: Enable error reporting when AER is native
PCI/portdrv: Don't disable AER reporting in get_port_device_capability()
PCI/AER: Configure ECRC for every device
PCI: Clear PCI_STATUS when setting up device
PCI/ERR: Recognize disconnected devices in report_error_detected()
Bjorn Helgaas [Thu, 4 Aug 2022 16:41:51 +0000 (11:41 -0500)]
Merge branch 'pci/enumeration'
- Split out ARI "next function" handling from the traditional one (Niklas
Schnelle)
- Move jailhouse "isolated function" (non-zero functions where function 0
doesn't exist) handling to pci_scan_slot() to avoid duplicating
multi-function scanning in pci_scan_child_bus_extend() (Niklas Schnelle)
- Extend "isolated function" probing to s390 (Niklas Schnelle).
- Allow s390 zPCI zbus without a function 0 (Niklas Schnelle)
* pci/enumeration:
s390/pci: allow zPCI zbus without a function zero
PCI: Extend isolated function probing to s390
PCI: Move jailhouse's isolated function handling to pci_scan_slot()
PCI: Split out next_ari_fn() from next_fn()
PCI: Clean up pci_scan_slot()
Richard Zhu [Wed, 18 May 2022 09:35:27 +0000 (17:35 +0800)]
PCI: imx6: Set PCIE_DBI_RO_WR_EN before writing DBI registers
The PCIE_DBI_RO_WR_EN bit should be set when write some DBI registers. To
make sure that the DBI registers are writable, set the PCIE_DBI_RO_WR_EN
properly when writing the DBI registers.
Richard Zhu [Thu, 14 Jul 2022 07:31:09 +0000 (15:31 +0800)]
PCI: imx6: Reformat suspend callback to keep symmetric with resume
Create imx6_pcie_stop_link() and imx6_pcie_host_exit() functions.
Encapsulate clocks, regulators disables and PHY uninitialization into
imx6_pcie_host_exit().
To keep suspend/resume symmetric as much as possible, invoke these two
new created functions in suspend callback.
To be symmetric with imx6_pcie_host_exit(), move imx6_pcie_clk_enable()
to imx6_pcie_host_init() from imx6_pcie_deassert_core_reset().
Richard Zhu [Thu, 14 Jul 2022 07:31:08 +0000 (15:31 +0800)]
PCI: imx6: Move the imx6_pcie_ltssm_disable() earlier
Move the imx6_pcie_ltssm_disable() earlier and place it just behind the
imx6_pcie_ltssm_enable(), since it might not be only used by suspend
callback directly.
To be symmetric with imx6_pcie_ltssm_enable(), add the IMX6Q and IMX8MQ
switch cases in imx6_pcie_ltssm_disable().
Richard Zhu [Thu, 14 Jul 2022 07:31:05 +0000 (15:31 +0800)]
PCI: imx6: Reduce resume time by only starting link if it was up before suspend
i.MX PCIe doesn't support hotplug. During resume, only start PCIe link
training when the link was up before system suspend to avoid the long
latency in the link training period.
Richard Zhu [Thu, 14 Jul 2022 07:31:01 +0000 (15:31 +0800)]
PCI: imx6: Call host init function directly in resume
Call imx6_pcie_host_init() instead of duplicating codes in resume. Note
that this also means we do MPLL setup again during resume, which we didn't
do before.
PCI: imx6: Factor out ref clock disable to match enable
The PCIe ref clocks are specific to different variants. The enables are
already split out into imx6_pcie_enable_ref_clk(), but the disables were
combined with the more generic bus/phy/pcie clock disables in
imx6_pcie_clk_disable().
Split out the variant-specific disables into imx6_pcie_disable_ref_clk() to
match imx6_pcie_enable_ref_clk().
Move imx6_pcie_enable_ref_clk() earlier so it's not in the middle between
imx6_pcie_assert_core_reset() and imx6_pcie_deassert_core_reset(). No
functional change intended.
Collect imx6_pcie_init_phy(), imx7d_pcie_wait_for_phy_pll_lock(), and
imx6_setup_phy_mpll() earlier with other PHY-related code. No functional
change intended.
Move imx6_pcie_grp_offset() and imx6_pcie_configure_type() earlier in the
file since they depend on nothing and are used by several other functions
that will be moved earlier. No functional change intended.
Replace SET_NOIRQ_SYSTEM_SLEEP_PM_OPS() with NOIRQ_SYSTEM_SLEEP_PM_OPS(),
which has the advantage that the compiler always sees the PM callbacks as
referenced, so they don't need to be wrapped with "#ifdef CONFIG_PM_SLEEP"
or tagged with "__maybe_unused" to avoid "defined but not used" warnings.
See 1a3c7bb08826 ("PM: core: Add new *_PM_OPS macros, deprecate old ones").
Marek Szyprowski [Tue, 28 Jun 2022 22:04:09 +0000 (00:04 +0200)]
PCI: exynos: Correct generic PHY usage
The proper initialization for generic PHYs is to call first phy_init(),
then phy_power_on().
While touching this, remove the phy_reset() call. It is just a left-over
from the obsoleted Exynos5440 support and the current exynos-pcie PHY
driver doesn't even support this function. It is also rarely used by other
drivers.
Link: https://lore.kernel.org/r/20220628220409.26545-2-m.szyprowski@samsung.com Reported-by: Bjorn Helgaas <helgaas@kernel.org> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Chanho Park <chanho61.park@samsung.com> Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
The exynos-pcie driver called phy_power_on() before phy_init() for some
historical reasons. However the generic PHY framework assumes that the
proper sequence is to call phy_init() first, then phy_power_on(). The
operations done by both functions should be considered as one action and as
such they are called by the exynos-pcie driver (without doing anything
between them). The initialization is just a sequence of register writes,
which cannot be altered without breaking the hardware operation.
To match the generic PHY framework requirement, simply move all register
writes to the phy_init()/phy_exit() and drop power_on()/power_off()
callbacks. This way the driver will also work with the old (incorrect)
PHY initialization call sequence.
Link: https://lore.kernel.org/r/20220628220409.26545-1-m.szyprowski@samsung.com Reported-by: Bjorn Helgaas <helgaas@kernel.org> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Chanho Park <chanho61.park@samsung.com> Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Acked-By: Vinod Koul <vkoul@kernel.org>
PCI: dwc: Handle MSIs routed to multiple GIC interrupts
On some Qualcomm platforms each group of 32 MSI vectors is routed to a
separate GIC interrupt. Implement support for such configurations by
parsing "msi0" ... "msiX" interrupts and attaching them to the chained
handler.
Note that if DT doesn't list an array of MSI interrupts and uses a single
"msi" IRQ, the driver will limit the number of supported MSI vectors to 32.
PCI: dwc: Convert struct pcie_port.msi_irq to an array
The Qualcomm DWC PCIe controller supports more than 32 MSI interrupts, but
they are routed to separate interrupts in groups of 32 vectors. To support
this configuration, change the msi_irq field to an array. Let the DWC core
handle all interrupts that were set in this array.
[bhelgaas: reorder, drop "irq" temporary to make patch cleaner] Link: https://lore.kernel.org/r/20220707134733.2436629-3-dmitry.baryshkov@linaro.org Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Rob Herring <robh@kernel.org> Reviewed-by: Johan Hovold <johan+linaro@kernel.org> Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
PCI: dwc: Correct msi_irq condition in dw_pcie_free_msi()
The dwc-based drivers set pp->msi_irq to -ENODEV if they do not want the
dwc core to do anything with pp->msi_irq.
dw_pcie_host_init() sets the handler and data when "pp->msi_irq > 0", so
use the same condition when removing the handler and data in
dw_pcie_free_msi().
Will McVicker [Mon, 13 Jun 2022 22:26:11 +0000 (17:26 -0500)]
PCI: dwc: Fix MSI msi_msg DMA mapping
As of 07940c369a6b ("PCI: dwc: Fix MSI page leakage in suspend/resume"),
the PCIe designware host driver has been using the driver data allocation
for the msi_msg DMA mapping which can result in a DMA_MAPPING_ERROR due to
the DMA overflow check in dma_direct_map_page() when the address is greater
than 32 bits (reported in [1]). The commit was trying to address a memory
leak on suspend/resume by moving the MSI mapping to dw_pcie_host_init(),
but subsequently dropped the page allocation thinking it wasn't needed.
To fix the DMA mapping issue as well as make msi_msg DMA'able, switch back
to allocating a 32-bit page for the msi_msg. To avoid the suspend/resume
leak, allocate the page in dw_pcie_host_init() since that shouldn't be
called during suspend/resume.
Serge Semin [Fri, 24 Jun 2022 14:39:46 +0000 (17:39 +0300)]
PCI: dwc: Check iATU in/outbound range setup status
Make the DWC PCIe RC/EP safer and more verbose for invalid or failed
inbound and outbound iATU window setups. Silently ignoring iATU regions
setup errors may cause unpredictable errors. For instance if a cfg or IO
window fails to be activated, then any CFG/IO requested won't reach target
PCIe devices and the corresponding accessors will return platform-specific
random values.
Serge Semin [Fri, 24 Jun 2022 14:39:45 +0000 (17:39 +0300)]
PCI: dwc: Validate iATU outbound mappings against hardware constraints
Make __dw_pcie_prog_outbound_atu() check the requested region base and size
against what the hardware can support. Return error if the region is not
correctly aligned or of a supported size.
Serge Semin [Fri, 24 Jun 2022 14:39:44 +0000 (17:39 +0300)]
PCI: dwc: Add iATU regions size detection procedure
The DWC PCIe RC/EP/DM IP core configuration parameters determine the number
of inbound and outbound iATU windows, alignment requirements (which is also
the minimum window size), minimum and maximum sizes. If internal ATU is
enabled, the former settings are determined by CX_ATU_MIN_REGION_SIZE; the
latter are determined by CX_ATU_MAX_REGION_SIZE.
Determine the required alignment and maximum size supported by the
controller and log it to help verify whether the requested inbound or
outbound memory mappings can be fully created.
Note 1. The extended iATU regions have been supported since DWC PCIe
v4.60a. There is no need in testing the upper limit register availability
for the older cores.
Note 2. The regions alignment is determined with using the fls() method
since the lower four bits of the ATU Limit register can be occupied with
the Circular Buffer Increment setting, which can be initialized with zeros.
Unify those by pushing the unroll address computation and viewport
selection down into dw_pcie_writel_atu() so we can use the same
dw_pcie_writel_atu_ob() accessor for both paths:
In the non-unroll case, this does involve more MMIO writes to
PCIE_ATU_VIEWPORT, but it's mainly in initialization paths and the code
simplification is significant.
Serge Semin [Fri, 24 Jun 2022 14:39:42 +0000 (17:39 +0300)]
PCI: dwc: Drop enum dw_pcie_region_type in favor of PCIE_ATU_REGION_DIR_IB/OB
Previously callers of dw_pcie_disable_atu() supplied enum
dw_pcie_region_type (DW_PCIE_REGION_INBOUND, DW_PCIE_REGION_OUTBOUND),
which dw_pcie_disable_atu() converted to the PCIE_ATU_REGION_DIR_IB or
PCIE_ATU_REGION_DIR_OB values needed to program the ATU registers.
Simplify the code by dropping the dw_pcie_region_type enum and passing
PCIE_ATU_REGION_DIR_IB or PCIE_ATU_REGION_DIR_OB directly.
Reorder dw_pcie_disable_atu() arguments to (dir, index) since "index"
indicates an ATU window in the regions of the corresponding direction.
Serge Semin [Fri, 24 Jun 2022 14:39:41 +0000 (17:39 +0300)]
PCI: dwc: Drop enum dw_pcie_as_type in favor of PCIE_ATU_TYPE_MEM/IO
Previously dw_pcie_ep_set_bar() converted the BAR PCI_BASE_ADDRESS_SPACE
bit to the internal dw_pcie_as_type enum (DW_PCIE_AS_MEM, DW_PCIE_AS_IO)
and passed it down to dw_pcie_prog_inbound_atu(), which converted the enum
to the PCIE_ATU_TYPE_MEM/PCIE_ATU_TYPE_IO values needed to program the ATU
registers.
Simplify the code by dropping the dw_pcie_as_type enum and passing
PCIE_ATU_TYPE_MEM or PCIE_ATU_TYPE_IO directly.
Reorder inbound ATU function arguments to match the outbound functions,
with address-related parameters at the end.
Serge Semin [Fri, 24 Jun 2022 14:39:39 +0000 (17:39 +0300)]
PCI: tegra194: Drop manual DW PCIe controller version setup
Since the DW PCIe common code (dw_pcie_version_detect()) now reads the IP
core version directly from the hardware, there is no point manually setting
the version for controllers newer than v4.70a.
Tegra194 only supports v4.90a, so remove the now-superfluous code that sets
struct dw_pcie.version.
Serge Semin [Fri, 24 Jun 2022 14:39:38 +0000 (17:39 +0300)]
PCI: intel-gw: Drop manual DW PCIe controller version setup
Since the DW PCIe common code (dw_pcie_version_detect()) now reads the IP
core version directly from the hardware, there is no point manually setting
the version for controllers newer than v4.70a.
Remove the now-superfluous intel-gw code that sets struct dw_pcie.version.
Serge Semin [Fri, 24 Jun 2022 14:39:36 +0000 (17:39 +0300)]
PCI: dwc: Read DWC IP core version from register
Since DWC PCIe v4.70a, the controller version and version type can be read
from the PORT_LOGIC.PCIE_VERSION_OFF and PORT_LOGIC.PCIE_VERSION_TYPE_OFF
registers respectively.
Read the version from those registers and warn if if's different from the
version we got from the device tree.
We can only read the version after platform-specific drivers have done any
DBI-related initialization, such as reference clock activation.
Serge Semin [Fri, 24 Jun 2022 14:39:35 +0000 (17:39 +0300)]
PCI: dwc: Use native DWC IP core version representation
Save the DWC IP core version in the same format as the
PORT_LOGIC.PCIE_VERSION_OFF register, similar to what other drivers for DWC
IP do (dw_spi_hw_init(), dwc3_core_is_valid(), stmmac_hwif_init()).
Serge Semin [Fri, 24 Jun 2022 14:39:34 +0000 (17:39 +0300)]
PCI: dwc: Detect iATU settings after getting "addr_space" resource
Previously, dw_pcie_ep_init() did:
dw_pcie_iatu_detect(pci);
res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "addr_space");
if (!res)
return -EINVAL;
The platform_get_resource_byname() can fail, and dw_pcie_iatu_detect()
doesn't depend on the "addr_space" resource, so delay it until afterwards,
i.e.,
Serge Semin [Fri, 24 Jun 2022 14:39:33 +0000 (17:39 +0300)]
PCI: dwc: Log link speed and width if it comes up
Printing just "link up" isn't very informative for PCI Express. Even if the
link is up, bus performance can degrade to slower speeds or to narrower
width than both Root Port and its partner is capable of. In that case it
would be handy to know the link specifications as early as possible.
If the link comes up, log the link speed (PCIe generation) and width.
Allow ASPM L1 and its substates. By default this is disabled in the qcom
specific hardware. Enable it explicitly only for controllers belonging to
2_7_0.
This does not affect any link capability registers; it will allow the link
transitions to L1 and its substates only if they are already supported.
The main feature of the sparc-specific implementation of
pci_mmap_resource_range() is that it allows mapping the entire PCI I/O
space for a PCI host bridge using the /proc/bus/pci interface on a bridge
device.
The generic implementation cannot do this, but it also appears that this
got broken for sparc by commit 9eff02e2042f ("PCI: check mmap range of
/proc/bus/pci files too"), which enforces that each address is part of a
BAR for kernels after 2.6.28.
Remove it all, assuming that the corresponding user space code has already
been changed to access /dev/ioport instead a long time ago. Add
pci_iobar_pfn() to make it possible to map I/O resources. This is adapted
from the powerpc version.
The ARCH_GENERIC_PCI_MMAP_RESOURCE symbol came up in a recent discussion,
and I noticed that this was left behind by an unfinished cleanup from 2017.
The only architecture that still relies on providing its own
pci_mmap_page_range() helper instead of using the generic
pci_mmap_resource_range() is sparc. Presumably the reasons for this have
not changed, but at least this can be simplified by converting sparc to use
the same interface as the others.
The only difference between the two is the device-specific offset that gets
added to or subtracted from vma->vm_pgoff.
Change the only caller of pci_mmap_page_range() in common code to subtract
this offset and call the modern interface, while adding it back in the
sparc implementation to preserve the existing behavior.
This removes the complexities of the dual interfaces from the common code,
and keeps it all specific to the sparc architecture code. According to
David Miller, the sparc code lets user space poke into the VGA I/O port
registers by mmapping the I/O space of the parent bridge device, which is
something that the generic pci_mmap_resource_range() code apparently does
not.
Jim Quinlan [Tue, 26 Jul 2022 17:39:11 +0000 (12:39 -0500)]
PCI: brcmstb: Rename .map_bus() functions to end with 'map_bus'
Rename the .map_bus() functions to end with 'map_bus' so they're easy to
find with, e.g., 'git grep "^static.*_map_bus" drivers/pci/'.
[bhelgaas: rename brcm_pcie_map_bus32() to brcm7425_pcie_map_bus() for
better cscope-ability (".*_map_bus" is not the same as ".*_map_bus.*")] Link: https://lore.kernel.org/r/20220725151258.42574-8-jim2101024@gmail.com Signed-off-by: Jim Quinlan <jim2101024@gmail.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Jim Quinlan [Mon, 25 Jul 2022 15:12:53 +0000 (11:12 -0400)]
PCI: brcmstb: Enable child bus device regulators from DT
Some platforms have power regulators for slots or devices below Root Ports.
On platforms like Raspberry Pi 4, these regulators are described in the
Root Port device tree node, since they logically belong to the Root Port,
not to the host bridge itself.
Add an .add_bus() hook (called when pci_alloc_child_bus() allocates the
secondary ("child") bus for a bridge), and look for such regulators. If we
find some, enable them before bringing up the link and enumerating devices
on the child bus.
Similarly, when pci_remove_bus() calls the ops->remove_bus() hook, disable
the regulators.
The regulators that may be described in a Root Port DT device are:
vpcie3v3
vpcie3v3aux
vpcie12v
These control power to the device downstream from the Root Port.
[bhelgaas: commit log, name hooks brcm_pcie_add_bus(), etc, since we only
support one set of subregulator info, save info in struct brcm_pcie instead
of dev->driver_data, move brcm_pcie_start_link() from probe to .add_bus()
(from subsequent patch)] Link: https://lore.kernel.org/r/20220725151258.42574-5-jim2101024@gmail.com Signed-off-by: Jim Quinlan <jim2101024@gmail.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Jim Quinlan [Mon, 25 Jul 2022 15:12:52 +0000 (11:12 -0400)]
PCI: brcmstb: Prevent config space access when link is down
When the link is down, config accesses to downstream devices cause CPU
aborts. Allow config accesses only when the link is up.
As the following scenario shows, this check is racy and cannot completely
avoid CPU aborts, but it makes them less likely:
pci_generic_config_read
addr = brcm_pcie_map_conf # bus->ops->map_bus()
brcm_pcie_link_up # returns "true"; link is up
<link goes down>
*val = readb(addr) # link is now down
<CPU abort>
Note that config space accesses to the Root Port are not affected by link
status.
[bhelgaas: commit log, use PCIE_ECAM_REG() instead of magic 0xfff masks;
note that pci_generic_config_read32() masks low two bits already] Link: https://lore.kernel.org/r/20220725151258.42574-4-jim2101024@gmail.com Signed-off-by: Jim Quinlan <jim2101024@gmail.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Remove forward function declarations in this driver. Also move some
constant structure definitions lower in the file. There are no changes to
the code that has been moved.
Niklas Schnelle [Tue, 28 Jun 2022 14:31:00 +0000 (16:31 +0200)]
s390/pci: allow zPCI zbus without a function zero
Currently the zPCI code blocks PCI bus creation and probing of a zPCI zbus
unless there is a PCI function with devfn 0. This is always the case for
the PCI functions with hidden RID, but may keep PCI functions from a
multi-function PCI device with RID information invisible until the function
0 becomes visible. Worse, as a PCI bus is necessary to even present a PCI
hotplug slot, even that remains invisible.
With the probing of these so-called isolated PCI functions enabled for s390
in common code, this restriction is no longer necessary. On network cards
with multiple ports and a PF per port this also allows using each port on
its own while still providing the physical PCI topology information in the
devfn needed to associate VFs with their parent PF.
Niklas Schnelle [Tue, 28 Jun 2022 14:30:59 +0000 (16:30 +0200)]
PCI: Extend isolated function probing to s390
Like the jailhouse hypervisor, s390's PCI architecture allows passing
isolated PCI functions to a guest OS instance. As of now this is was not
utilized even with multi-function support as the s390 PCI code makes sure
that only virtual PCI busses including a function with devfn 0 are
presented to the PCI subsystem. A subsequent change will remove this
restriction.
Allow probing such functions by replacing the existing check for
jailhouse_paravirt() with a new hypervisor_isolated_pci_functions() helper.
Niklas Schnelle [Tue, 28 Jun 2022 14:30:58 +0000 (16:30 +0200)]
PCI: Move jailhouse's isolated function handling to pci_scan_slot()
The special case of the jailhouse hypervisor passing through individual PCI
functions handles scanning for PCI functions even if function 0 does not
exist. Previously this was done with an extra loop duplicating the one in
pci_scan_slot(). By incorporating the check for jailhouse_paravirt() into
pci_scan_slot() we can instead do this as part of the normal slot scan.
Note that with the assignment of dev->multifunction gated by fn > 0 we set
dev->multifunction unconditionally for all functions if function 0 is
missing just as in the existing jailhouse loop.
The only functional change is that we now call pcie_aspm_init_link_state()
for these functions, but this already happened if function 0 was passed
through and should not be a problem.
Niklas Schnelle [Tue, 28 Jun 2022 14:30:57 +0000 (16:30 +0200)]
PCI: Split out next_ari_fn() from next_fn()
In commit b1bd58e448f2 ("PCI: Consolidate "next-function" functions") the
next_fn() function subsumed the traditional and ARI-based next function
determination. This got rid of some needlessly complex function pointer
handling but also reduced the separation between these very different
methods of finding the next function. With the next_fn() cleaned up a bit
we can re-introduce this separation by moving out the ARI handling while
sticking with direct function calls.
Niklas Schnelle [Tue, 28 Jun 2022 14:30:56 +0000 (16:30 +0200)]
PCI: Clean up pci_scan_slot()
While determining the next PCI function is factored out of pci_scan_slot()
into next_fn(), the former still handles the first function as a special
case, which duplicates the code from the scan loop.
Furthermore the non-ARI branch of next_fn() is generally hard to understand
and especially the check for multifunction devices is hidden in the
handling of NULL devices for non-contiguous multifunction. It also signals
that no further functions need to be scanned by returning 0 via wraparound
and this is a valid function number.
Improve upon this by transforming the conditions in next_fn() to be easier
to understand.
By changing next_fn() to return -ENODEV instead of 0 when there is no next
function we can then handle the initial function inside the loop and
deduplicate the shared handling. This also makes it more explicit that only
function 0 must exist.
PCI: loongson: Work around LS7A incorrect Interrupt Pin registers
Several devices integrated into LS7A report 1 (which means they use
INTA) in their Interrupt Pin registers, but they actually use a different
interrupt.
Add a quirk to override the incorrect Interrupt Pin values.
This is only needed by ACPI-based systems. For DT-based systems,
pci_assign_irq() ignores the Interrupt Pin register except to learn that
the device uses INTx and the host bridge .map_irq() function
(loongson_map_irq()) learns the IRQ mapping via DT and of_irq_parse_pci().
[bhelgaas: drop PCIE_PORT_x, OHCI, GPU since they are function 0 and don't
need the quirk, squash in updates from
https://lore.kernel.org/r/CAAhV-H40_o+9KS1t67O98GusM38pDaiB4bssxd3KQZpAByfnLg@mail.gmail.com] Link: https://lore.kernel.org/r/20220714124216.1489304-8-chenhuacai@loongson.cn Signed-off-by: Jianmin Lv <lvjianmin@loongson.cn> Signed-off-by: Huacai Chen <chenhuacai@loongson.cn> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
fu740 uses no syscon or regman interfaces, so it doesn't need to include
mfs/syscon.h. It uses no regulator interfaces, so it doesn't need to
include regulator/consumer.h either.
Baruch Siach [Tue, 21 Jun 2022 08:54:53 +0000 (11:54 +0300)]
PCI: qcom: Define slot capabilities using PCI_EXP_SLTCAP_*
The PCIE_CAP_LINK1_VAL macro actually defines slot capabilities. Use
PCI_EXP_SLTCAP_* macros to spell its value, and rename it to better
describe its meaning.
Manual reparenting of pipe_clk_src is being replaced with the parking of
the clock with clk_disable()/clk_enable() in the PHY driver. Drop
redundant code switching of the pipe clock between the PHY clock source
and the safe bi_tcxo.
Link: https://lore.kernel.org/r/20220608105238.2973600-6-dmitry.baryshkov@linaro.org Tested-by: Johan Hovold <johan+linaro@kernel.org> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org> Reviewed-by: Johan Hovold <johan+linaro@kernel.org> Reviewed-by: Stephen Boyd <sboyd@kernel.org> Acked-by: Stanimir Varbanov <svarbanov@mm-sol.com>
On LS2K/LS7A, config reads to some non-existent devices don't return
0xffffffff (they are hidden devices for debug, and accessing the config
space may cause machine hang). This is a hardware flaw but we can only
avoid it by software now.
Loongson PCH (LS7A chipset) will be used by both MIPS-based and LoongArch-
based Loongson processors. MIPS-based Loongson uses FDT, while LoongArch-
based Loongson uses ACPI.
Add ACPI init support for the driver in pci-loongson.c because it is
currently FDT-only.
LoongArch is a new RISC ISA, mainline support will come soon, and
documentations are here (in translation):
PCI: loongson: Use generic 8/16/32-bit config ops on LS2K/LS7A
LS2K/LS7A support 8/16/32-bits PCI config access operations via CFG1, so
we can disable CFG0 for them and safely use pci_generic_config_read()/
pci_generic_config_write() instead of pci_generic_config_read32()/pci_
generic_config_write32().
Dmitry Baryshkov [Thu, 23 Jun 2022 20:03:19 +0000 (15:03 -0500)]
PCI: qcom: Remove unnecessary pipe_clk handling
PCIe PHY drivers (both QMP and PCIe2) already do clk_prepare_enable() /
clk_prepare_disable() pipe_clk. Remove extra calls to enable/disable
this clock from the PCIe driver, so that the PHY driver can manage the
clock on its own.
[bhelgaas: rebase on Robert Marko's DBI cleanup:
https://lore.kernel.org/r/20220623155004.688090-2-robimarko@gmail.com] Link: https://lore.kernel.org/r/20220608105238.2973600-5-dmitry.baryshkov@linaro.org Tested-by: Johan Hovold <johan+linaro@kernel.org> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org> Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> Reviewed-by: Johan Hovold <johan+linaro@kernel.org> Reviewed-by: Stephen Boyd <sboyd@kernel.org> Acked-by: Stanimir Varbanov <svarbanov@mm-sol.com>
Robert Marko [Thu, 23 Jun 2022 15:50:04 +0000 (17:50 +0200)]
PCI: qcom: Power on PHY before DBI register accesses
IPQ8074 requires the PHY to be powered on before accessing DBI registers.
It's not clear whether other variants have the same dependency, but there
seems to be no reason for them to be different, so move all the DBI
accesses from .init() to .post_init() so they are all after phy_power_on().
Robert Marko [Thu, 23 Jun 2022 15:50:03 +0000 (17:50 +0200)]
PCI: qcom: Power on PHY before IPQ8074 DBI register accesses
Currently the Gen2 port in IPQ8074 will cause the system to hang as it
accesses DBI registers in qcom_pcie_init_2_3_3(), and those are only
accesible after phy_power_on().
Move the DBI read/writes to a new qcom_pcie_post_init_2_3_3(), which is
executed after phy_power_on().
Link: https://lore.kernel.org/r/20220623155004.688090-1-robimarko@gmail.com Fixes: a0fd361db8e5 ("PCI: dwc: Move "dbi", "dbi2", and "addr_space" resource setup into common code") Signed-off-by: Robert Marko <robimarko@gmail.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Cc: stable@vger.kernel.org # v5.11+
PCI: qcom: Set up rev 2.1.0 PARF_PHY before enabling clocks
We currently enable clocks BEFORE we write to PARF_PHY_CTRL reg to enable
clocks and resets. This causes the driver to never set to a ready state
with the error 'Phy link never came up'.
This is caused by the PHY clock getting enabled before setting the required
bits in the PARF regs.
A workaround for this was set but with this new discovery we can drop
the workaround and use a proper solution to the problem by just enabling
the clock only AFTER the PARF_PHY_CTRL bit is set.
This correctly sets up the PCIe link and makes it usable even when a
bootloader leaves the PCIe link in an undefined state.
PCI/AER: Iterate over error counters instead of error strings
Previously we iterated over AER stat *names*, e.g.,
aer_correctable_error_string[32], but the actual stat *counters* may not be
that large, e.g., pdev->aer_stats->dev_cor_errs[16], which means that we
printed junk in the sysfs stats files.
Iterate over the stat counter arrays instead of the names to avoid this
junk.
Also, added a build time check to make sure all
counters have entries in strings array.
Note that these bits are all in the Device Control register and are not
AER-specific.
This affects all devices with an AER capability, including hot-added
devices.
Please note that this change is quite invasive, as error reporting now will
be enabled for all available PCIe Endpoints, which was previously not the
case.
When "pci=noaer" is selected, error reporting stays disabled of course.
[bhelgaas: commit log, note error reporting is not AER-specific] Link: https://lore.kernel.org/r/20220125071820.2247260-4-sr@denx.de Signed-off-by: Stefan Roese <sr@denx.de> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Pali Rohár <pali@kernel.org> Cc: Bharat Kumar Gogada <bharat.kumar.gogada@xilinx.com> Cc: Michal Simek <michal.simek@xilinx.com> Cc: Yao Hongbo <yaohongbo@linux.alibaba.com> Cc: Naveen Naidu <naveennaidu479@gmail.com>
Stefan Roese [Tue, 25 Jan 2022 07:18:19 +0000 (08:18 +0100)]
PCI/portdrv: Don't disable AER reporting in get_port_device_capability()
AER reporting is currently disabled in the DevCtl registers of all non Root
Port PCIe devices on systems using pcie_ports_native || host->native_aer,
disabling AER completely in such systems. This is because 2bd50dd800b5
("PCI: PCIe: Disable PCIe port services during port initialization"), added
a call to pci_disable_pcie_error_reporting() *after* the AER setup was
completed for the PCIe device tree.
Here a longer analysis about the current status of AER enabling /
disabling upon bootup provided by Bjorn:
pcie_portdrv_probe
pcie_port_device_register
get_port_device_capability
pci_disable_pcie_error_reporting
clear CERE NFERE FERE URRE # <-- disable for RP USP DSP
pcie_device_init
device_register # new AER service device
aer_probe
aer_enable_rootport # RP only
set_downstream_devices_error_reporting
set_device_error_reporting # self (RP)
if (RP || USP || DSP)
pci_enable_pcie_error_reporting
set CERE NFERE FERE URRE # <-- enable for RP
pci_walk_bus
set_device_error_reporting
if (RP || USP || DSP)
pci_enable_pcie_error_reporting
set CERE NFERE FERE URRE # <-- enable for USP DSP
In a typical Root Port -> Endpoint hierarchy, the above:
- Disables Error Reporting for the Root Port,
- Enables Error Reporting for the Root Port,
- Does NOT enable Error Reporting for the Endpoint because it is not a
Root Port or Switch Port.
In a deeper Root Port -> Upstream Switch Port -> Downstream Switch
Port -> Endpoint hierarchy:
- Disables Error Reporting for the Root Port,
- Enables Error Reporting for the Root Port,
- Enables Error Reporting for both Switch Ports,
- Does NOT enable Error Reporting for the Endpoint because it is not a
Root Port or Switch Port,
- Disables Error Reporting for the Switch Ports when pcie_portdrv_probe()
claims them. AER does not re-enable it because these are not Root
Ports.
Remove this call to pci_disable_pcie_error_reporting() from
get_port_device_capability(), leaving the already enabled AER configuration
intact. With this change, AER is enabled in the Root Port and the PCIe
switch upstream and downstream ports. Only the PCIe Endpoints don't have
AER enabled yet. A follow-up patch will take care of this Endpoint
enabling.
Fixes: 2bd50dd800b5 ("PCI: PCIe: Disable PCIe port services during port initialization") Link: https://lore.kernel.org/r/20220125071820.2247260-3-sr@denx.de Signed-off-by: Stefan Roese <sr@denx.de> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Pali Rohár <pali@kernel.org> Cc: Rafael J. Wysocki <rjw@rjwysocki.net> Cc: Bharat Kumar Gogada <bharat.kumar.gogada@xilinx.com> Cc: Michal Simek <michal.simek@xilinx.com> Cc: Yao Hongbo <yaohongbo@linux.alibaba.com> Cc: Naveen Naidu <naveennaidu479@gmail.com>
Bjorn Helgaas [Fri, 24 Jun 2022 21:27:19 +0000 (16:27 -0500)]
PCI/ASPM: Unexport pcie_aspm_support_enabled()
pcie_aspm_support_enabled() is used only by the acpi/pci_root.c driver,
which cannot be built as a module, so it does not need to be exported.
Unexport it.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Frank Li [Tue, 24 May 2022 15:21:59 +0000 (10:21 -0500)]
PCI: endpoint: Enable DMA tests for endpoints with DMA capabilities
Some PCI Endpoint controllers integrate an eDMA (embedded DMA). eDMA can
bypass the outbound memory address translation unit to access all RC memory
space.
Add eDMA support for pci-epf-test.
Depending on HW availability, the EPF test can use either eDMA or general
system DMA controllers to perform DMA. The test tries to use eDMA first and
falls back to general system DMA controllers if there's no eDMA
Separate dma_chan to dma_chan_tx and dma_chan_rx. Search for an eDMA
channel first, then search for a memory-to-memory DMA channel. If general
memory to memory channels are used, dma_chan_rx = dma_chan_tx.
Add dma_addr_t dma_remote in pci_epf_test_data_transfer() because eDMA uses
remote RC physical address directly.
Add enum dma_transfer_direction dir in pci_epf_test_data_transfer() because
eDMA chooses the correct RX/TX channel by dir.
The overall steps are:
1. Execute dma_request_channel() and filter function to find correct eDMA
RX and TX Channel. If a channel does not exist, fallback to try to
allocate general memory to memory DMA channel.
2. Execute dmaengine_slave_config() to configure remote side physical
address.
3. Execute dmaengine_prep_slave_single() to create transfer descriptor.
4. Execute tx_submit().
5. Execute dma_async_issue_pending()
[bhelgaas: squash in fix from Dan Carpenter <dan.carpenter@oracle.com>:
https://lore.kernel.org/r/Ys2GSTnZhuLzzQG5@kili, also previously posted by
Peng Wu <wupeng58@huawei.com>:
https://lore.kernel.org/all/CANXvt5rK98-cEMgpzopY9POOK8a5=VDib8uKPLgJakOG=hRfwQ@mail.gmail.com/] Link: https://lore.kernel.org/r/20220524152159.2370739-9-Frank.Li@nxp.com Signed-off-by: Frank Li <Frank.Li@nxp.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> Acked-by: Kishon Vijay Abraham I <kishon@ti.com> Acked-By: Vinod Koul <vkoul@kernel.org>
projects / kernel.git / log