- Hold power management references to all PHYs while enabling them to avoid
a race when one provides clocks to others (Siddharth Vadapalli)
* pci/controller/keystone:
PCI: keystone: Fix race condition when initializing PHYs
- Add j721e DT and driver support for 'num-lanes' for devices that support
x1, x2, or x4 Links (Matt Ranostay)
- Add j721e DT compatible strings and driver support for j784s4 (Matt Ranostay)
- Make TI J721E Kconfig depend on ARCH_K3 since the hardware is specific to
those TI SoC parts (Peter Robinson)
* pci/controller/cadence:
PCI: j721e: Make TI J721E depend on ARCH_K3
PCI: j721e: Add TI J784S4 PCIe configuration
PCI: j721e: Add PCIe 4x lane selection support
PCI: j721e: Add per platform maximum lane settings
dt-bindings: PCI: ti,j721e-pci-*: Add j784s4-pci-* compatible strings
dt-bindings: PCI: ti,j721e-pci-*: Add checks for num-lanes
- Do dma_mrpc cleanup during switchtec_pci_remove() to match its devm
ioremapping in switchtec_pci_probe(). Previously the cleanup was done in
stdev_release(), which used stale pointers if stdev->cdev happened to be
open when the PCI device was removed (Daniel Stodden)
* pci/switchtec:
PCI: switchtec: Fix stdev_release() crash after surprise hot remove
- Restructure pci_dev_for_each_resource() to avoid computing the address of
an out-of-bounds array element (the bounds check was performed later so
the element was never actually *read*, but it's nicer to avoid even
computing an out-of-bounds address) (Andy Shevchenko)
* pci/resource:
PCI: Avoid potential out-of-bounds read in pci_dev_for_each_resource()
- Log device type (Root Port, Switch Port, etc) during enumeration (Bjorn
Helgaas)
- Log resource names (BAR 0, VF BAR 0, bridge window, etc) consistently
instead of a mix of names and "reg 0x10" (Puranjay Mohan, Bjorn Helgaas)
- Log bridges before devices below the bridges (Bjorn Helgaas)
* pci/enumeration-logging:
PCI: Log bridge info when first enumerating bridge
PCI: Log bridge windows conditionally
PCI: Supply bridge device, not secondary bus, to read window details
PCI: Move pci_read_bridge_windows() below individual window accessors
PCI: Use resource names in PCI log messages
PCI: Update BAR # and window messages
PCI: Log device type during enumeration
- Convert pci-host-common.c platform .remove() callback to .remove_new()
returning 'void' since it's not useful to return error codes here (Uwe
Kleine-König)
- Log a message about updating AMD USB controller class code (so dwc3, not
xhci, claims it) only when we actually change it (Guilherme G. Piccoli)
- Use PCI_HEADER_TYPE_* instead of literals in x86, powerpc, SCSI lpfc
(Ilpo Järvinen)
- Clean up open-coded PCIBIOS return code mangling (Ilpo Järvinen)
- Fix 64GT/s effective data rate calculation to use 1b/1b encoding rather
than the 8b/10b or 128b/130b used by lower rates (Ilpo Järvinen)
* pci/enumeration:
PCI: Fix 64GT/s effective data rate calculation
x86/pci: Clean up open-coded PCIBIOS return code mangling
scsi: lpfc: Use PCI_HEADER_TYPE_MFD instead of literal
powerpc/fsl-pci: Use PCI_HEADER_TYPE_MASK instead of literal
x86/pci: Use PCI_HEADER_TYPE_* instead of literals
PCI: Only override AMD USB controller if required
PCI: host-generic: Convert to platform remove callback returning void
- Log AER errors as "Correctable" (not "Corrected") or "Uncorrectable" to
match spec terminology (Bjorn Helgaas)
- Decode Requester ID when no error info found instead of printing the raw
hex value (Bjorn Helgaas)
* pci/aer:
PCI/AER: Use explicit register sizes for struct members
PCI/AER: Decode Requester ID when no error info found
PCI/AER: Use 'Correctable' and 'Uncorrectable' spec terms for errors
The Broadcom STB/CM PCIe HW core, which is also used in RPi SOCs, must be
deliberately set by the PCIe RC HW into one of three mutually exclusive
modes:
"safe" -- No CLKREQ# expected or required, refclk is always provided. This
mode should work for all devices but is not be capable of any refclk
power savings.
"no-l1ss" -- CLKREQ# is expected to be driven by the downstream device for
CPM and ASPM L0s and L1. Provides Clock Power Management, L0s, and L1,
but cannot provide L1 substate (L1SS) power savings. If the downstream
device connected to the RC is L1SS capable AND the OS enables L1SS, all
PCIe traffic may abruptly halt, potentially hanging the system.
"default" -- Bidirectional CLKREQ# between the RC and downstream device.
Provides ASPM L0s, L1, and L1SS, but not compliant to provide Clock
Power Management; specifically, may not be able to meet the T_CLRon max
timing of 400ns as specified in "Dynamic Clock Control", section
3.2.5.2.2 of the PCIe Express Mini CEM 2.1 specification. This
situation is atypical and should happen only with older devices.
Previously, this driver always set the mode to "no-l1ss", as almost all
STB/CM boards operate in this mode. But now there is interest in
activating L1SS power savings from STB/CM customers, which requires "aspm"
mode. In addition, a bug was filed for RPi4 CM platform because most
devices did not work in "no-l1ss" mode.
Note that the mode is specified by the DT property "brcm,clkreq-mode". If
this property is omitted, then "default" mode is chosen.
Note: Since L1 substates are now possible, a modification was made
regarding an internal bus timeout: During long periods of the PCIe RC HW
being in an L1SS sleep state, there may be a timeout on an internal bus
access, even though there may not be any PCIe access involved. Such a
timeout will cause a subsequent CPU abort.
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217276
Link: https://lore.kernel.org/linux-pci/20231113185607.1756-3-james.quinlan@broadcom.com
Tested-by: Cyril Brulebois <cyril@debamax.com>
Tested-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: Jim Quinlan <james.quinlan@broadcom.com>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
The PCI driver invokes the PHY APIs using the ks_pcie_enable_phy()
function. The PHY in this case is the Serdes. It is possible that the
PCI instance is configured for two lane operation across two different
Serdes instances, using one lane of each Serdes.
In such a configuration, if the reference clock for one Serdes is
provided by the other Serdes, it results in a race condition. After the
Serdes providing the reference clock is initialized by the PCI driver by
invoking its PHY APIs, it is not guaranteed that this Serdes remains
powered on long enough for the PHY APIs based initialization of the
dependent Serdes. In such cases, the PLL of the dependent Serdes fails
to lock due to the absence of the reference clock from the former Serdes
which has been powered off by the PM Core.
Fix this by obtaining reference to the PHYs before invoking the PHY
initialization APIs and releasing reference after the initialization is
complete.
Link: https://lore.kernel.org/linux-pci/20230927041845.1222080-1-s-vadapalli@ti.com
Fixes: 49229238ab ("PCI: keystone: Cleanup PHY handling")
Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Acked-by: Ravi Gunasekaran <r-gunasekaran@ti.com>
When building with clang 18 with the -Werror compiler option enabled,
the following error will be reported:
drivers/pci/controller/dwc/pcie-rcar-gen4.c:439:15: error: cast to smaller integer type 'enum dw_pcie_device_mode' from 'const void *' [-Werror,-Wvoid-pointer-to-enum-cast]
439 | rcar->mode = (enum dw_pcie_device_mode)of_device_get_match_data(&rcar->pdev->dev);
To fix this issue, cast the data the of_device_get_match_data() helper
returns to uintptr_t rather than the dw_pcie_device_mode enum.
[kwilczynski: commit log]
Link: https://lore.kernel.org/linux-pci/20231220053829.1921187-7-yoshihiro.shimoda.uh@renesas.com
Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
aer_irq() reads the AER Root Error Status and Error Source Identification
(PCI_ERR_ROOT_STATUS and PCI_ERR_ROOT_ERR_SRC) registers directly into
struct aer_err_source. Both registers are 32 bits, so declare the members
explicitly as "u32" instead of "unsigned int".
Similarly, aer_get_device_error_info() reads the AER Header Log
(PCI_ERR_HEADER_LOG) registers, which are also 32 bits, into struct
aer_header_log_regs. Declare those members as "u32" as well.
No functional changes intended.
Link: https://lore.kernel.org/r/20231206224231.732765-4-helgaas@kernel.org
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
When a device with AER detects an error, it logs error information in its
own AER Error Status registers. It may send an Error Message to the Root
Port (RCEC in the case of an RCiEP), which logs the fact that an Error
Message was received (Root Error Status) and the Requester ID of the
message source (Error Source Identification).
aer_print_port_info() prints the Requester ID from the Root Port Error
Source in the usual Linux "bb:dd.f" format, but when find_source_device()
finds no error details in the hierarchy below the Root Port, it printed the
raw Requester ID without decoding it.
Decode the Requester ID in the usual Linux format so it matches other
messages.
Sample message changes:
- pcieport 0000:00:1c.5: AER: Correctable error received: 0000:00:1c.5
- pcieport 0000:00:1c.5: AER: can't find device of ID00e5
+ pcieport 0000:00:1c.5: AER: Correctable error message received from 0000:00:1c.5
+ pcieport 0000:00:1c.5: AER: found no error details for 0000:00:1c.5
Link: https://lore.kernel.org/r/20231206224231.732765-3-helgaas@kernel.org
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com>
The PCIe spec classifies errors as either "Correctable" or "Uncorrectable".
Previously we printed these as "Corrected" or "Uncorrected". To avoid
confusion, use the same terms as the spec.
One confusing situation is when one agent detects an error, but another
agent is responsible for recovery, e.g., by re-attempting the operation.
The first agent may log a "correctable" error but it has not yet been
corrected. The recovery agent must report an uncorrectable error if it is
unable to recover. If we print the first agent's error as "Corrected", it
gives the false impression that it has already been resolved.
Sample message change:
- pcieport 0000:00:1c.5: AER: Corrected error received: 0000:00:1c.5
+ pcieport 0000:00:1c.5: AER: Correctable error received: 0000:00:1c.5
Link: https://lore.kernel.org/r/20231206224231.732765-2-helgaas@kernel.org
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com>
Commit 6f5e193bfb ("PCI: dwc: Fix dw_pcie_ep_raise_msix_irq() to get
correct MSI-X table address") modified dw_pcie_ep_raise_msix_irq() to
support iATUs which require a specific alignment.
However, this support cannot have been properly tested.
The whole point is for the iATU to map an address that is aligned,
using dw_pcie_ep_map_addr(), and then let the writel() write to
ep->msi_mem + aligned_offset.
Thus, modify the address that is mapped such that it is aligned.
With this change, dw_pcie_ep_raise_msix_irq() matches the logic in
dw_pcie_ep_raise_msi_irq().
Link: https://lore.kernel.org/linux-pci/20231128132231.2221614-1-nks@flawful.org
Fixes: 6f5e193bfb ("PCI: dwc: Fix dw_pcie_ep_raise_msix_irq() to get correct MSI-X table address")
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Cc: stable@vger.kernel.org # 5.7
Cc: Kishon Vijay Abraham I <kishon@kernel.org>
Log bridge secondary/subordinate bus and window information at the same
time we log the bridge BARs, just after discovering the bridge and before
scanning the bridge's secondary bus. This logs the bridge and downstream
devices in a more logical order:
- pci 0000:00:01.0: [8086:1901] type 01 class 0x060400
- pci 0000:01:00.0: [10de:13b6] type 00 class 0x030200
- pci 0000:01:00.0: reg 0x10: [mem 0xec000000-0xecffffff]
- pci 0000:00:01.0: PCI bridge to [bus 01]
- pci 0000:00:01.0: bridge window [io 0xe000-0xefff]
+ pci 0000:00:01.0: [8086:1901] type 01 class 0x060400
+ pci 0000:00:01.0: PCI bridge to [bus 01]
+ pci 0000:00:01.0: bridge window [io 0xe000-0xefff]
+ pci 0000:01:00.0: [10de:13b6] type 00 class 0x030200
+ pci 0000:01:00.0: reg 0x10: [mem 0xec000000-0xecffffff]
Note that we read the windows into a temporary struct resource that is
thrown away, not into the resources in the struct pci_bus.
The windows may be adjusted after we know what downstream devices require,
and those adjustments are logged as they are made.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Previously pci_read_bridge_io(), pci_read_bridge_mmio(), and
pci_read_bridge_mmio_pref() unconditionally logged the bridge window
resource. A future change will call these functions earlier and more
often. Add a "log" parameter so callers can control whether to generate
the log message. No functional change intended.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Previously we logged information about devices *below* the bridge before
logging information about the bridge itself, e.g.,
pci 0000:00:01.0: [8086:1901] type 01 class 0x060400
pci 0000:01:00.0: [10de:13b6] type 00 class 0x030200
pci 0000:01:00.0: reg 0x10: [mem 0xec000000-0xecffffff]
pci 0000:00:01.0: PCI bridge to [bus 01]
pci 0000:00:01.0: bridge window [io 0xe000-0xefff]
This is partly because the bridge windows are read in this path:
pci_scan_child_bus_extend
for (devfn = 0; devfn < 256; devfn += 8)
pci_scan_slot(bus, devfn) # scan below bridge
pcibios_fixup_bus(bus)
pci_read_bridge_bases(bus) # read bridge windows
pci_read_bridge_io(bus)
Remove the assumption that the secondary (child) pci_bus already exists by
passing in the bridge device (instead of the pci_bus) and a resource
pointer when reading bridge windows. A future change can use this to log
the bridge details before we enumerate the devices below the bridge.
No functional change intended.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Move pci_read_bridge_windows() below the functions that read the I/O,
memory, and prefetchable memory windows, so pci_read_bridge_windows() can
use them in the future. No functional change intended.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
The PCI log messages print the register offsets at some places and BAR
numbers at other places. There is no uniformity in this logging mechanism.
It would be better to print names than register offsets.
Add a helper function that aids in printing more meaningful information
about the BAR numbers like "VF BAR", "ROM", "bridge window", etc. This
function can be called while printing PCI log messages.
[bhelgaas: fold in Lukas' static array suggestion from
https://lore.kernel.org/all/20211106115831.GA7452@wunner.de/]
Link: https://lore.kernel.org/r/20211106112606.192563-2-puranjay12@gmail.com
Signed-off-by: Puranjay Mohan <puranjay12@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Log the device type when enumeration a device. Sample output changes:
- pci 0000:00:00.0: [8086:1237] type 00 class 0x060000
+ pci 0000:00:00.0: [8086:1237] type 00 class 0x060000 conventional PCI endpoint
- pci 0000:00:1c.0: [8086:a110] type 01 class 0x060400
+ pci 0000:00:1c.0: [8086:a110] type 01 class 0x060400 PCIe Root Port
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
In pci_mmconfig_insert(), there's no reference to "addr" between locking
pci_mmcfg_lock and testing "addr", so it *looks* like we should move the
test before the lock.
But 07f9b61c39 ("x86/PCI: MMCONFIG: Check earlier for MMCONFIG region at
address zero") did that, which broke things by returning -EINVAL when
"addr" is zero instead of -EEXIST.
So 07f9b61c39 was reverted by 67d470e0e1 ("Revert "x86/PCI: MMCONFIG:
Check earlier for MMCONFIG region at address zero"").
Add a comment about this issue to prevent it from happening again.
Link: https://lore.kernel.org/r/20231121183643.249006-8-helgaas@kernel.org
Tested-by: Tomasz Pala <gotar@polanet.pl>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
"pci_mmcfg_check_reserved()" doesn't give a hint about what the boolean
return value means. Rename it to pci_mmcfg_reserved() so testing
"if (pci_mmcfg_reserved())" makes sense.
Update callers to treat the return value as boolean instead of comparing
with 0.
Link: https://lore.kernel.org/r/20231121183643.249006-7-helgaas@kernel.org
Tested-by: Tomasz Pala <gotar@polanet.pl>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
"acpi_mcfg_check_entry()" doesn't give a hint about what the return value
means. Rename it to "acpi_mcfg_valid_entry()", convert the return value to
bool, and update the return values and callers to match so testing
"if (acpi_mcfg_valid_entry())" makes sense.
Link: https://lore.kernel.org/r/20231121183643.249006-6-helgaas@kernel.org
Tested-by: Tomasz Pala <gotar@polanet.pl>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
fd3a8cff4d ("x86/pci: Treat EfiMemoryMappedIO as reservation of ECAM
space") added the concept of using the EFI memory map to help decide
whether ECAM space mentioned in the MCFG table is valid.
Unfortunately it described that EfiMemoryMappedIO space as "reserved", but
it is actually not *reserved* by the EFI memory map. EfiMemoryMappedIO
only means the firmware requested that the OS map this space for use by
firmware runtime services.
Change the dmesg logging to describe it as simply "EfiMemoryMappedIO", not
as "reserved as EfiMemoryMappedIO". A previous commit actually *does*
reserve the space if ACPI PNP0C01/02 devices haven't done so:
- PCI: ECAM at [mem 0xe0000000-0xefffffff] reserved as EfiMemoryMappedIO
+ PCI: ECAM at [mem 0xe0000000-0xefffffff] is EfiMemoryMappedIO; assuming valid
PCI: ECAM [mem 0xe0000000-0xefffffff] reserved to work around lack of ACPI motherboard _CRS
Link: https://lore.kernel.org/r/20231121183643.249006-3-helgaas@kernel.org
Tested-by: Tomasz Pala <gotar@polanet.pl>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Tomasz, Sebastian, and some Proxmox users reported problems initializing
ixgbe NICs.
I think the problem is that ECAM space described in the ACPI MCFG table is
not reserved via a PNP0C02 _CRS method as required by the PCI Firmware spec
(r3.3, sec 4.1.2), but it *is* included in the PNP0A03 host bridge _CRS as
part of the MMIO aperture.
If we allocate space for a PCI BAR, we're likely to allocate it from that
ECAM space, which obviously cannot work.
This could happen for any device, but in the ixgbe case it happens because
it's an SR-IOV device and the BIOS didn't allocate space for the VF BARs,
so Linux reallocated the bridge window leading to ixgbe and put it on top
of the ECAM space. From Tomasz' system:
PCI: MMCONFIG for domain 0000 [bus 00-ff] at [mem 0x80000000-0x8fffffff] (base 0x80000000)
PCI: MMCONFIG at [mem 0x80000000-0x8fffffff] not reserved in ACPI motherboard resources
pci_bus 0000:00: root bus resource [mem 0x80000000-0xfbffffff window]
pci 0000:00:01.1: PCI bridge to [bus 02-03]
pci 0000:00:01.1: bridge window [mem 0xfb900000-0xfbbfffff]
pci 0000:02:00.0: [8086:10fb] type 00 class 0x020000 # ixgbe
pci 0000:02:00.0: reg 0x10: [mem 0xfba80000-0xfbafffff 64bit]
pci 0000:02:00.0: VF(n) BAR0 space: [mem 0x00000000-0x000fffff 64bit] (contains BAR0 for 64 VFs)
pci 0000:02:00.0: BAR 7: no space for [mem size 0x00100000 64bit] # VF BAR 0
pci_bus 0000:00: No. 2 try to assign unassigned res
pci 0000:00:01.1: resource 14 [mem 0xfb900000-0xfbbfffff] released
pci 0000:00:01.1: BAR 14: assigned [mem 0x80000000-0x806fffff]
pci 0000:02:00.0: BAR 0: assigned [mem 0x80000000-0x8007ffff 64bit]
pci 0000:02:00.0: BAR 7: assigned [mem 0x80204000-0x80303fff 64bit] # VF BAR 0
Fixes: 07eab0901e ("efi/x86: Remove EfiMemoryMappedIO from E820 map")
Fixes: fd3a8cff4d ("x86/pci: Treat EfiMemoryMappedIO as reservation of ECAM space")
Reported-by: Tomasz Pala <gotar@polanet.pl>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=218050
Reported-by: Sebastian Manciulea <manciuleas@protonmail.com>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=218107
Link: https://forum.proxmox.com/threads/proxmox-8-kernel-6-2-16-4-pve-ixgbe-driver-fails-to-load-due-to-pci-device-probing-failure.131203/
Link: https://lore.kernel.org/r/20231121183643.249006-2-helgaas@kernel.org
Tested-by: Tomasz Pala <gotar@polanet.pl>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: stable@vger.kernel.org # v6.2+
