linux.git
17 months agolib/fw_table: Remove acpi_parse_entries_array() export
Dan Williams [Thu, 2 Nov 2023 22:07:02 +0000 (15:07 -0700)]
lib/fw_table: Remove acpi_parse_entries_array() export

Stephen reports that the ACPI helper library rework,
CONFIG_FIRMWARE_TABLE, introduces a new compiler warning:

    WARNING: modpost: vmlinux: acpi_parse_entries_array: EXPORT_SYMBOL used
    for init symbol. Remove __init or EXPORT_SYMBOL.

Delete this export as it turns out it is unneeded, and future work wraps
this in another exported helper. Note that in general
EXPORT_SYMBOL_ACPI_LIB() is needed for exporting symbols that are marked
__init_or_acpilib, but in this case no export is required.

Fixes: a103f46633fd ("acpi: Move common tables helper functions to common lib")
Cc: Dave Jiang <dave.jiang@intel.com>
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Closes: http://lore.kernel.org/r/20231030160523.670a7569@canb.auug.org.au
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/169896282222.70775.940454758280866379.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/pci: Change CXL AER support check to use native AER
Terry Bowman [Thu, 2 Nov 2023 15:52:32 +0000 (10:52 -0500)]
cxl/pci: Change CXL AER support check to use native AER

Native CXL protocol errors are delivered to the OS through AER
reporting. The owner of AER owns CXL Protocol error management with
respect to _OSC negotiation.[1] CXL device errors are handled by a
separate interrupt with native control gated by _OSC control field
'CXL Memory Error Reporting Control'.

The CXL driver incorrectly checks for 'CXL Memory Error Reporting
Control' before accessing AER registers and caching RCH downport
AER registers. Replace the current check in these 2 cases with
native AER checks.

[1] CXL 3.0 - 9.17.2 CXL _OSC, Table-9-26, Interpretation of CXL
_OSC Support Fields, p.641

Fixes: f05fd10d138d ("cxl/pci: Add RCH downstream port AER register discovery")
Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Reviewed-by: Smita Koralahalli <Smita.KoralahalliChannabasappa@amd.com>
Link: https://lore.kernel.org/r/20231102155232.1421261-1-terry.bowman@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/hdm: Remove broken error path
Dan Williams [Tue, 31 Oct 2023 21:09:19 +0000 (14:09 -0700)]
cxl/hdm: Remove broken error path

Dan reports that cxl_decoder_commit() potentially leaks a hold of
cxl_dpa_rwsem. The potential error case is a "should not" happen
scenario, turn it into a "can not" happen scenario by adding the error
check to cxl_port_setup_targets() where other setting validation occurs.

Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Closes: http://lore.kernel.org/r/63295673-5d63-4919-b851-3b06d48734c0@moroto.mountain
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Fixes: 176baefb2eb5 ("cxl/hdm: Commit decoder state to hardware")
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/hdm: Fix && vs || bug
Dan Carpenter [Tue, 31 Oct 2023 09:53:52 +0000 (12:53 +0300)]
cxl/hdm: Fix && vs || bug

If "info" is NULL then this code will crash.  || was intended instead of
&&.

Fixes: 8ce520fdea24 ("cxl/hdm: Use stored Component Register mappings to map HDM decoder capability")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Robert Richter <rrichter@amd.com>
Link: https://lore.kernel.org/r/60028378-d3d5-4d6d-90fd-f915f061e731@moroto.mountain
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agoMerge branch 'for-6.7/cxl-commited' into cxl/next
Dan Williams [Tue, 31 Oct 2023 18:00:08 +0000 (11:00 -0700)]
Merge branch 'for-6.7/cxl-commited' into cxl/next

Add the committed decoder sysfs attribute for v6.7.

17 months agoMerge branch 'for-6.7/cxl' into cxl/next
Dan Williams [Tue, 31 Oct 2023 17:59:44 +0000 (10:59 -0700)]
Merge branch 'for-6.7/cxl' into cxl/next

Pickup some misc. CXL updates for v6.7.

17 months agoMerge branch 'for-6.7/cxl-qtg' into cxl/next
Dan Williams [Tue, 31 Oct 2023 17:59:26 +0000 (10:59 -0700)]
Merge branch 'for-6.7/cxl-qtg' into cxl/next

Merge some prep-work for CXL QOS class support. This cycle saw large
collisions with mm on this topic, so the bulk of this topic needs to
wait.

17 months agoMerge branch 'for-6.7/cxl-rch-eh' into cxl/next
Dan Williams [Tue, 31 Oct 2023 17:59:00 +0000 (10:59 -0700)]
Merge branch 'for-6.7/cxl-rch-eh' into cxl/next

Restricted CXL Host (RCH) Error Handling undoes the topology munging of
CXL 1.1 to enabled some AER recovery, and lands some base infrastructure
for handling Root-Complex-Event-Collectors (RCECs) with CXL. Include
this long running series finally for v6.7.

17 months agoacpi: Move common tables helper functions to common lib
Dave Jiang [Thu, 12 Oct 2023 18:53:54 +0000 (11:53 -0700)]
acpi: Move common tables helper functions to common lib

Some of the routines in ACPI driver/acpi/tables.c can be shared with
parsing CDAT. CDAT is a device-provided data structure that is formatted
similar to a platform provided ACPI table. CDAT is used by CXL and can
exist on platforms that do not use ACPI. Split out the common routine
from ACPI to accommodate platforms that do not support ACPI and move that
to /lib. The common routines can be built outside of ACPI if
FIRMWARE_TABLES is selected.

Link: https://lore.kernel.org/linux-cxl/CAJZ5v0jipbtTNnsA0-o5ozOk8ZgWnOg34m34a9pPenTyRLj=6A@mail.gmail.com/
Suggested-by: "Rafael J. Wysocki" <rafael@kernel.org>
Reviewed-by: Hanjun Guo <guohanjun@huawei.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Acked-by: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>
Link: https://lore.kernel.org/r/169713683430.2205276.17899451119920103445.stgit@djiang5-mobl3
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl: Add support for reading CXL switch CDAT table
Dave Jiang [Thu, 12 Oct 2023 18:53:48 +0000 (11:53 -0700)]
cxl: Add support for reading CXL switch CDAT table

Add read_cdat_data() call in cxl_switch_port_probe() to allow
reading of CDAT data for CXL switches. read_cdat_data() needs
to be adjusted for the retrieving of the PCIe device depending
on if the passed in port is endpoint or switch.

Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/169713682855.2205276.6418370379144967443.stgit@djiang5-mobl3
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl: Add checksum verification to CDAT from CXL
Dave Jiang [Thu, 12 Oct 2023 18:53:42 +0000 (11:53 -0700)]
cxl: Add checksum verification to CDAT from CXL

A CDAT table is available from a CXL device. The table is read by the
driver and cached in software. With the CXL subsystem needing to parse the
CDAT table, the checksum should be verified. Add checksum verification
after the CDAT table is read from device.

Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/169713682277.2205276.2687265961314933628.stgit@djiang5-mobl3
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl: Export QTG ids from CFMWS to sysfs as qos_class attribute
Dave Jiang [Thu, 12 Oct 2023 18:53:37 +0000 (11:53 -0700)]
cxl: Export QTG ids from CFMWS to sysfs as qos_class attribute

Export the QoS Throttling Group ID from the CXL Fixed Memory Window
Structure (CFMWS) under the root decoder sysfs attributes as qos_class.

CXL rev3.0 9.17.1.3 CXL Fixed Memory Window Structure (CFMWS)

cxl cli will use this id to match with the _DSM retrieved id for a
hot-plugged CXL memory device DPA memory range to make sure that the
DPA range is under the right CFMWS window.

Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/169713681699.2205276.14475306324720093079.stgit@djiang5-mobl3
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl: Add decoders_committed sysfs attribute to cxl_port
Dave Jiang [Mon, 16 Oct 2023 17:57:54 +0000 (10:57 -0700)]
cxl: Add decoders_committed sysfs attribute to cxl_port

This attribute allows cxl-cli to determine whether there are decoders
committed to a memdev.  This is only a snapshot of the state, and
doesn't offer any protection or serialization against a concurrent
disable-region operation.

Reviewed-by: Jim Harris <jim.harris@samsung.com>
Suggested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/169747907439.272156.10261062080830155662.stgit@djiang5-mobl3
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl: Add cxl_decoders_committed() helper
Dave Jiang [Mon, 16 Oct 2023 17:57:48 +0000 (10:57 -0700)]
cxl: Add cxl_decoders_committed() helper

Add a helper to retrieve the number of decoders committed for the port.
Replace all the open coding of the calculation with the helper.

Link: https://lore.kernel.org/linux-cxl/651c98472dfed_ae7e729495@dwillia2-xfh.jf.intel.com.notmuch/
Suggested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>
Link: https://lore.kernel.org/r/169747906849.272156.1729290904857372335.stgit@djiang5-mobl3
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/core/regs: Rework cxl_map_pmu_regs() to use map->dev for devm
Robert Richter [Wed, 18 Oct 2023 17:17:13 +0000 (19:17 +0200)]
cxl/core/regs: Rework cxl_map_pmu_regs() to use map->dev for devm

struct cxl_register_map carries a @dev parameter for devm operations.
Simplify the function interface to use that instead of a separate @dev
argument.

Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-21-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/core/regs: Rename phys_addr in cxl_map_component_regs()
Robert Richter [Wed, 18 Oct 2023 17:17:12 +0000 (19:17 +0200)]
cxl/core/regs: Rename phys_addr in cxl_map_component_regs()

Trivial change that renames variable phys_addr in
cxl_map_component_regs() to shorten its length to keep the 80 char
size limit for the line and also for consistency between the different
paths.

Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-20-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agoPCI/AER: Unmask RCEC internal errors to enable RCH downstream port error handling
Robert Richter [Wed, 18 Oct 2023 17:17:11 +0000 (19:17 +0200)]
PCI/AER: Unmask RCEC internal errors to enable RCH downstream port error handling

AER corrected and uncorrectable internal errors (CIE/UIE) are masked
in their corresponding mask registers per default once in power-up
state. [1][2] Enable internal errors for RCECs to receive CXL
downstream port errors of Restricted CXL Hosts (RCHs).

[1] CXL 3.0 Spec, 12.2.1.1 - RCH Downstream Port Detected Errors
[2] PCIe Base Spec r6.0, 7.8.4.3 Uncorrectable Error Mask Register,
    7.8.4.6 Correctable Error Mask Register

Co-developed-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-19-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agoPCI/AER: Forward RCH downstream port-detected errors to the CXL.mem dev handler
Robert Richter [Wed, 18 Oct 2023 17:17:10 +0000 (19:17 +0200)]
PCI/AER: Forward RCH downstream port-detected errors to the CXL.mem dev handler

In Restricted CXL Device (RCD) mode a CXL device is exposed as an
RCiEP, but CXL downstream and upstream ports are not enumerated and
not visible in the PCIe hierarchy. [1] Protocol and link errors from
these non-enumerated ports are signaled as internal AER errors, either
Uncorrectable Internal Error (UIE) or Corrected Internal Errors (CIE)
via an RCEC.

Restricted CXL host (RCH) downstream port-detected errors have the
Requester ID of the RCEC set in the RCEC's AER Error Source ID
register. A CXL handler must then inspect the error status in various
CXL registers residing in the dport's component register space (CXL
RAS capability) or the dport's RCRB (PCIe AER extended
capability). [2]

Errors showing up in the RCEC's error handler must be handled and
connected to the CXL subsystem. Implement this by forwarding the error
to all CXL devices below the RCEC. Since the entire CXL device is
controlled only using PCIe Configuration Space of device 0, function
0, only pass it there [3]. The error handling is limited to currently
supported devices with the Memory Device class code set (CXL Type 3
Device, PCI_CLASS_MEMORY_CXL, 502h), handle downstream port errors in
the device's cxl_pci driver. Support for other CXL Device Types
(e.g. a CXL.cache Device) can be added later.

To handle downstream port errors in addition to errors directed to the
CXL endpoint device, a handler must also inspect the CXL RAS and PCIe
AER capabilities of the CXL downstream port the device is connected
to.

Since CXL downstream port errors are signaled using internal errors,
the handler requires those errors to be unmasked. This is subject of a
follow-on patch.

The reason for choosing this implementation is that the AER service
driver claims the RCEC device, but does not allow it to register a
custom specific handler to support CXL. Connecting the RCEC hard-wired
with a CXL handler does not work, as the CXL subsystem might not be
present all the time. The alternative to add an implementation to the
portdrv to allow the registration of a custom RCEC error handler isn't
worth doing it as CXL would be its only user. Instead, just check for
an CXL RCEC and pass it down to the connected CXL device's error
handler. With this approach the code can entirely be implemented in
the PCIe AER driver and is independent of the CXL subsystem. The CXL
driver only provides the handler.

[1] CXL 3.0 spec: 9.11.8 CXL Devices Attached to an RCH
[2] CXL 3.0 spec, 12.2.1.1 RCH Downstream Port-detected Errors
[3] CXL 3.0 spec, 8.1.3 PCIe DVSEC for CXL Devices

Co-developed-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Cc: Oliver O'Halloran <oohall@gmail.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: linuxppc-dev@lists.ozlabs.org
Cc: linux-pci@vger.kernel.org
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-18-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/pci: Disable root port interrupts in RCH mode
Terry Bowman [Wed, 18 Oct 2023 17:17:09 +0000 (19:17 +0200)]
cxl/pci: Disable root port interrupts in RCH mode

The RCH root port contains root command AER registers that should not be
enabled.[1] Disable these to prevent root port interrupts.

[1] CXL 3.0 - 12.2.1.1 RCH Downstream Port-detected Errors

Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-17-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/pci: Add RCH downstream port error logging
Terry Bowman [Wed, 18 Oct 2023 17:17:08 +0000 (19:17 +0200)]
cxl/pci: Add RCH downstream port error logging

RCH downstream port error logging is missing in the current CXL driver. The
missing AER and RAS error logging is needed for communicating driver error
details to userspace. Update the driver to include PCIe AER and CXL RAS
error logging.

Add RCH downstream port error handling into the existing RCiEP handler.
The downstream port error handler is added to the RCiEP error handler
because the downstream port is implemented in a RCRB, is not PCI
enumerable, and as a result is not directly accessible to the PCI AER
root port driver. The AER root port driver calls the RCiEP handler for
handling RCD errors and RCH downstream port protocol errors.

Update existing RCiEP correctable and uncorrectable handlers to also call
the RCH handler. The RCH handler will read the RCH AER registers, check for
error severity, and if an error exists will log using an existing kernel
AER trace routine. The RCH handler will also log downstream port RAS errors
if they exist.

Co-developed-by: Robert Richter <rrichter@amd.com>
Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-16-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/pci: Map RCH downstream AER registers for logging protocol errors
Terry Bowman [Wed, 18 Oct 2023 17:17:07 +0000 (19:17 +0200)]
cxl/pci: Map RCH downstream AER registers for logging protocol errors

The restricted CXL host (RCH) error handler will log protocol errors
using AER and RAS status registers. The AER and RAS registers need to
be virtually memory mapped before enabling interrupts. Create the
initializer function devm_cxl_setup_parent_dport() for this when the
endpoint is connected with the dport. The initialization sets up the
RCH RAS and AER mappings.

Add 'struct cxl_regs' to 'struct cxl_dport' for saving a pointer to
the RCH downstream port's AER and RAS registers.

Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Co-developed-by: Robert Richter <rrichter@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-15-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/pci: Update CXL error logging to use RAS register address
Terry Bowman [Wed, 18 Oct 2023 17:17:06 +0000 (19:17 +0200)]
cxl/pci: Update CXL error logging to use RAS register address

The CXL error handler currently only logs endpoint RAS status. The CXL
topology includes several components providing RAS details to be logged
during error handling.[1] Update the current handler's RAS logging to use a
RAS register address. Also, update the error handler function names to be
consistent with correctable and uncorrectable RAS. This will allow for
adding support to log other CXL component's RAS details in the future.

[1] CXL3.0 Table 8-22 CXL_Capability_ID Assignment

Co-developed-by: Robert Richter <rrichter@amd.com>
Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-14-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agoPCI/AER: Refactor cper_print_aer() for use by CXL driver module
Terry Bowman [Wed, 18 Oct 2023 17:17:05 +0000 (19:17 +0200)]
PCI/AER: Refactor cper_print_aer() for use by CXL driver module

The CXL driver plans to use cper_print_aer() for logging restricted CXL
host (RCH) AER errors. cper_print_aer() is not currently exported and
therefore not usable by the CXL drivers built as loadable modules. Export
the cper_print_aer() function. Use the EXPORT_SYMBOL_NS_GPL() variant
to restrict the export to CXL drivers.

The CONFIG_ACPI_APEI_PCIEAER kernel config is currently used to enable
cper_print_aer(). cper_print_aer() logs the AER registers and is
useful in PCIE AER logging outside of APEI. Remove the
CONFIG_ACPI_APEI_PCIEAER dependency to enable cper_print_aer().

The cper_print_aer() function name implies CPER specific use but is useful
in non-CPER cases as well. Rename cper_print_aer() to pci_print_aer().

Also, update cxl_core to import CXL namespace imports.

Co-developed-by: Robert Richter <rrichter@amd.com>
Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Cc: Mahesh J Salgaonkar <mahesh@linux.ibm.com>
Cc: Oliver O'Halloran <oohall@gmail.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: linux-pci@vger.kernel.org
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-13-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/pci: Add RCH downstream port AER register discovery
Robert Richter [Fri, 27 Oct 2023 22:08:06 +0000 (15:08 -0700)]
cxl/pci: Add RCH downstream port AER register discovery

Restricted CXL host (RCH) downstream port AER information is not currently
logged while in the error state. One problem preventing the error logging
is the AER and RAS registers are not accessible. The CXL driver requires
changes to find RCH downstream port AER and RAS registers for purpose of
error logging.

RCH downstream ports are not enumerated during a PCI bus scan and are
instead discovered using system firmware, ACPI in this case.[1] The
downstream port is implemented as a Root Complex Register Block (RCRB).
The RCRB is a 4k memory block containing PCIe registers based on the PCIe
root port.[2] The RCRB includes AER extended capability registers used for
reporting errors. Note, the RCH's AER Capability is located in the RCRB
memory space instead of PCI configuration space, thus its register access
is different. Existing kernel PCIe AER functions can not be used to manage
the downstream port AER capabilities and RAS registers because the port was
not enumerated during PCI scan and the registers are not PCI config
accessible.

Discover RCH downstream port AER extended capability registers. Use MMIO
accesses to search for extended AER capability in RCRB register space.

[1] CXL 3.0 Spec, 9.11.2 - System Firmware View of CXL 1.1 Hierarchy
[2] CXL 3.0 Spec, 8.2.1.1 - RCH Downstream Port RCRB

Co-developed-by: Robert Richter <rrichter@amd.com>
Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-12-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/port: Remove Component Register base address from struct cxl_port
Robert Richter [Wed, 18 Oct 2023 17:17:02 +0000 (19:17 +0200)]
cxl/port: Remove Component Register base address from struct cxl_port

The Component Register base address @component_reg_phys is no longer
used after the rework of the Component Register setup which now uses
struct member @reg_map instead. Remove the base address.

Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-10-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/pci: Remove Component Register base address from struct cxl_dev_state
Robert Richter [Wed, 18 Oct 2023 17:17:01 +0000 (19:17 +0200)]
cxl/pci: Remove Component Register base address from struct cxl_dev_state

The Component Register base address @component_reg_phys is no longer
used after the rework of the Component Register setup which now uses
struct member @reg_map instead. Remove the base address.

Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-9-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/hdm: Use stored Component Register mappings to map HDM decoder capability
Robert Richter [Wed, 18 Oct 2023 17:17:00 +0000 (19:17 +0200)]
cxl/hdm: Use stored Component Register mappings to map HDM decoder capability

Now, that the Component Register mappings are stored, use them to
enable and map the HDM decoder capabilities. The Component Registers
do not need to be probed again for this, remove probing code.

The HDM capability applies to Endpoints, USPs and VH Host Bridges. The
Endpoint's component register mappings are located in the cxlds and
else in the port's structure. Duplicate the cxlds->reg_map in
port->reg_map for endpoint ports.

Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
[rework to drop cxl_port_get_comp_map()]
Link: https://lore.kernel.org/r/20231018171713.1883517-8-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/pci: Store the endpoint's Component Register mappings in struct cxl_dev_state
Robert Richter [Wed, 18 Oct 2023 17:16:59 +0000 (19:16 +0200)]
cxl/pci: Store the endpoint's Component Register mappings in struct cxl_dev_state

Same as for ports and dports, also store the endpoint's Component
Register mappings, use struct cxl_dev_state for that.

Keep the Component Register base address @component_reg_phys a bit to
not break functionality. It will be removed after the transition in a
later patch.

Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-7-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/port: Pre-initialize component register mappings
Robert Richter [Wed, 18 Oct 2023 17:16:58 +0000 (19:16 +0200)]
cxl/port: Pre-initialize component register mappings

The component registers of a component may not exist and
cxl_setup_comp_regs() will fail for that reason. In another case,
Software may not use and set those registers up. cxl_setup_comp_regs()
is then called with a base address of CXL_RESOURCE_NONE. Both are
valid cases, but the function returns without initializing the
register map.

Now, a missing component register block is not necessarily a reason to
fail (feature is optional or its existence checked later). Change
cxl_setup_comp_regs() to also use components with the component
register block missing. Thus, always initialize struct
cxl_register_map with valid values, set @dev and make @resource
CXL_RESOURCE_NONE.

The change is in preparation of follow-on patches.

Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-6-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/port: Rename @comp_map to @reg_map in struct cxl_register_map
Robert Richter [Wed, 18 Oct 2023 17:16:57 +0000 (19:16 +0200)]
cxl/port: Rename @comp_map to @reg_map in struct cxl_register_map

Name the field @reg_map, because @reg_map->host will be used for
mapping operations beyond component registers (i.e. AER registers).
This is valid for all occurrences of @comp_map. Change them all.

Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-5-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/port: Fix @host confusion in cxl_dport_setup_regs()
Dan Williams [Wed, 18 Oct 2023 17:16:56 +0000 (19:16 +0200)]
cxl/port: Fix @host confusion in cxl_dport_setup_regs()

commit 5d2ffbe4b81a ("cxl/port: Store the downstream port's Component Register mappings in struct cxl_dport")

...moved the dport component registers from a raw component_reg_phys
passed in at dport instantiation time to a 'struct cxl_register_map'
populated with both the component register data *and* the "host" device
for mapping operations.

While typical CXL switch dports are mapped by their associated 'struct
cxl_port', an RCH host bridge dport registered by cxl_acpi needs to wait
until the cxl_mem driver makes the attachment to map the registers. This
is because there are no intervening 'struct cxl_port' instances between
the root cxl_port and the endpoint port in an RCH topology.

For now just mark the host as NULL in the RCH dport case until code that
needs to map the dport registers arrives.

This patch is not flagged for -stable since nothing in the current
driver uses the dport->comp_map.

Now, I am slightly uneasy that cxl_setup_comp_regs() sets map->host to a
wrong value and then cxl_dport_setup_regs() fixes it up, but the
alternatives I came up with are more messy. For example, adding an
@logdev to 'struct cxl_register_map' that the dev_printk()s can fall
back to when @host is NULL. I settled on "post-fixup+comment" since it
is only RCH dports that have this special case where register probing is
split between a host-bridge RCRB lookup and when cxl_mem_probe() does
the association of the cxl_memdev and endpoint port.

[moved rename of @comp_map to @reg_map into next patch]

Fixes: 5d2ffbe4b81a ("cxl/port: Store the downstream port's Component Register mappings in struct cxl_dport")
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-4-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/core/regs: Rename @dev to @host in struct cxl_register_map
Robert Richter [Wed, 18 Oct 2023 17:16:55 +0000 (19:16 +0200)]
cxl/core/regs: Rename @dev to @host in struct cxl_register_map

The primary role of @dev is to host the mappings for devm operations.
@dev is too ambiguous as a name. I.e. when does @dev refer to the
'struct device *' instance that the registers belong, and when does
@dev refer to the 'struct device *' instance hosting the mapping for
devm operations?

Clarify the role of @dev in cxl_register_map by renaming it to @host.
Also, rename local variables to 'host' where map->host is used.

Signed-off-by: Terry Bowman <terry.bowman@amd.com>
Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20231018171713.1883517-3-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/port: Fix delete_endpoint() vs parent unregistration race
Dan Williams [Sat, 28 Oct 2023 03:13:23 +0000 (20:13 -0700)]
cxl/port: Fix delete_endpoint() vs parent unregistration race

The CXL subsystem, at cxl_mem ->probe() time, establishes a lineage of
ports (struct cxl_port objects) between an endpoint and the root of a
CXL topology. Each port including the endpoint port is attached to the
cxl_port driver.

Given that setup, it follows that when either any port in that lineage
goes through a cxl_port ->remove() event, or the memdev goes through a
cxl_mem ->remove() event. The hierarchy below the removed port, or the
entire hierarchy if the memdev is removed needs to come down.

The delete_endpoint() callback is careful to check whether it is being
called to tear down the hierarchy, or if it is only being called to
teardown the memdev because an ancestor port is going through
->remove().

That care needs to take the device_lock() of the endpoint's parent.
Which requires 2 bugs to be fixed:

1/ A reference on the parent is needed to prevent use-after-free
   scenarios like this signature:

    BUG: spinlock bad magic on CPU#0, kworker/u56:0/11
    Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS edk2-20230524-3.fc38 05/24/2023
    Workqueue: cxl_port detach_memdev [cxl_core]
    RIP: 0010:spin_bug+0x65/0xa0
    Call Trace:
      do_raw_spin_lock+0x69/0xa0
     __mutex_lock+0x695/0xb80
     delete_endpoint+0xad/0x150 [cxl_core]
     devres_release_all+0xb8/0x110
     device_unbind_cleanup+0xe/0x70
     device_release_driver_internal+0x1d2/0x210
     detach_memdev+0x15/0x20 [cxl_core]
     process_one_work+0x1e3/0x4c0
     worker_thread+0x1dd/0x3d0

2/ In the case of RCH topologies, the parent device that needs to be
   locked is not always @port->dev as returned by cxl_mem_find_port(), use
   endpoint->dev.parent instead.

Fixes: 8dd2bc0f8e02 ("cxl/mem: Add the cxl_mem driver")
Cc: <stable@vger.kernel.org>
Reported-by: Robert Richter <rrichter@amd.com>
Closes: http://lore.kernel.org/r/20231018171713.1883517-2-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agotools/testing/cxl: Slow down the mock firmware transfer
Vishal Verma [Thu, 26 Oct 2023 17:32:41 +0000 (11:32 -0600)]
tools/testing/cxl: Slow down the mock firmware transfer

The cxl-cli unit test for firmware update does operations like starting
an asynchronous firmware update, making sure it is in progress, and
attempting to cancel it. In some cases, such as with no or minimal
dynamic debugging turned on, the firmware update completes too quickly,
not allowing the test to have a chance to verify it was in progress.
This caused a failure of the signature:

  expected fw_update_in_progress:true
  test/cxl-update-firmware.sh: failed at line 88

Fix this by adding a delay (~1.5 - 2 ms) to each firmware transfer
request handled by the mocked interface.

Reported-by: Dan Williams <dan.j.williams@intel.com>
Tested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
Link: https://lore.kernel.org/r/20231026-vv-fw_upd_test_fix-v2-1-5282fd193883@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/region: Fix x1 root-decoder granularity calculations
Jim Harris [Thu, 26 Oct 2023 17:09:06 +0000 (10:09 -0700)]
cxl/region: Fix x1 root-decoder granularity calculations

Root decoder granularity must match value from CFWMS, which may not
be the region's granularity for non-interleaved root decoders.

So when calculating granularities for host bridge decoders, use the
region's granularity instead of the root decoder's granularity to ensure
the correct granularities are set for the host bridge decoders and any
downstream switch decoders.

Test configuration is 1 host bridge * 2 switches * 2 endpoints per switch.

Region created with 2048 granularity using following command line:

cxl create-region -m -d decoder0.0 -w 4 mem0 mem2 mem1 mem3 \
  -g 2048 -s 2048M

Use "cxl list -PDE | grep granularity" to get a view of the granularity
set at each level of the topology.

Before this patch:
        "interleave_granularity":2048,
        "interleave_granularity":2048,
    "interleave_granularity":512,
        "interleave_granularity":2048,
        "interleave_granularity":2048,
    "interleave_granularity":512,
"interleave_granularity":256,

After:
        "interleave_granularity":2048,
        "interleave_granularity":2048,
    "interleave_granularity":4096,
        "interleave_granularity":2048,
        "interleave_granularity":2048,
    "interleave_granularity":4096,
"interleave_granularity":2048,

Fixes: 27b3f8d13830 ("cxl/region: Program target lists")
Cc: <stable@vger.kernel.org>
Signed-off-by: Jim Harris <jim.harris@samsung.com>
Link: https://lore.kernel.org/r/169824893473.1403938.16110924262989774582.stgit@bgt-140510-bm03.eng.stellus.in
[djbw: fixup the prebuilt cxl_test region]
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/region: Fix cxl_region_rwsem lock held when returning to user space
Li Zhijian [Wed, 25 Oct 2023 08:54:50 +0000 (16:54 +0800)]
cxl/region: Fix cxl_region_rwsem lock held when returning to user space

Fix a missed "goto out" to unlock on error to cleanup this splat:

    WARNING: lock held when returning to user space!
    6.6.0-rc3-lizhijian+ #213 Not tainted
    ------------------------------------------------
    cxl/673 is leaving the kernel with locks still held!
    1 lock held by cxl/673:
     #0: ffffffffa013b9d0 (cxl_region_rwsem){++++}-{3:3}, at: commit_store+0x7d/0x3e0 [cxl_core]

In terms of user visible impact of this bug for backports:

cxl_region_invalidate_memregion() on x86 invokes wbinvd which is a
problematic instruction for virtualized environments. So, on virtualized
x86, cxl_region_invalidate_memregion() returns an error. This failure
case got missed because CXL memory-expander device passthrough is not a
production use case, and emulation of CXL devices is typically limited
to kernel development builds with CONFIG_CXL_REGION_INVALIDATION_TEST=y,
that makes cxl_region_invalidate_memregion() succeed.

In other words, the expected exposure of this bug is limited to CXL
subsystem development environments using QEMU that neglected
CONFIG_CXL_REGION_INVALIDATION_TEST=y.

Fixes: d1257d098a5a ("cxl/region: Move cache invalidation before region teardown, and before setup")
Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20231025085450.2514906-1-lizhijian@fujitsu.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/region: Use cxl_calc_interleave_pos() for auto-discovery
Alison Schofield [Wed, 25 Oct 2023 20:01:34 +0000 (13:01 -0700)]
cxl/region: Use cxl_calc_interleave_pos() for auto-discovery

For auto-discovered regions the driver must assign each target to
a valid position in the region interleave set based on the decoder
topology.

The current implementation fails to parse valid decode topologies,
as it does not consider the child offset into a parent port. The sort
put all targets of one port ahead of another port when an interleave
was expected, causing the region assembly to fail.

Replace the existing relative sort with cxl_calc_interleave_pos() that
finds the exact position in a region interleave for an endpoint based
on a walk up the ancestral tree from endpoint to root decoder.

cxl_calc_interleave_pos() was introduced in a prior patch, so the work
here is to use it in cxl_region_sort_targets().

Remove the obsoleted helper functions from the prior sort.

Testing passes on pre-production hardware with BIOS defined regions
that natively trigger this autodiscovery path of the region driver.
Testing passes a CXL unit test using the dev_dbg() calculation test
(see cxl_region_attach()) across an expanded set of region configs:
1, 1, 1+1, 1+1+1, 2, 2+2, 2+2+2, 2+2+2+2, 4, 4+4, where each number
represents the count of endpoints per host bridge.

Fixes: a32320b71f08 ("cxl/region: Add region autodiscovery")
Reported-by: Dmytro Adamenko <dmytro.adamenko@intel.com>
Signed-off-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Link: https://lore.kernel.org/r/3946cc55ddc19678733eddc9de2c317749f43f3b.1698263080.git.alison.schofield@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/region: Calculate a target position in a region interleave
Alison Schofield [Fri, 27 Oct 2023 20:04:48 +0000 (13:04 -0700)]
cxl/region: Calculate a target position in a region interleave

Introduce a calculation to find a target's position in a region
interleave. Perform a self-test of the calculation on user-defined
regions.

The region driver uses the kernel sort() function to put region
targets in relative order. Positions are assigned based on each
target's index in that sorted list. That relative sort doesn't
consider the offset of a port into its parent port which causes
some auto-discovered regions to fail creation. In one failure case,
a 2 + 2 config (2 host bridges each with 2 endpoints), the sort
puts all the targets of one port ahead of another port when they
were expected to be interleaved.

In preparation for repairing the autodiscovery region assembly,
introduce a new method for discovering a target position in the
region interleave.

cxl_calc_interleave_pos() adds a method to find the target position by
ascending from an endpoint to a root decoder. The calculation starts
with the endpoint's local position and position in the parent port. It
traverses towards the root decoder and examines both position and ways
in order to allow the position to be refined all the way to the root
decoder.

This calculation: position = position * parent_ways + parent_pos;
applied iteratively yields the correct position.

Include a self-test that exercises this new position calculation against
every successfully configured user-defined region.

Signed-off-by: Alison Schofield <alison.schofield@intel.com>
Link: https://lore.kernel.org/r/0ac32c75cf81dd8b86bf07d70ff139d33c2300bc.1698263080.git.alison.schofield@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agoMAINTAINERS: Add tools/testing/cxl files to CXL
Vishal Verma [Thu, 26 Oct 2023 17:43:20 +0000 (11:43 -0600)]
MAINTAINERS: Add tools/testing/cxl files to CXL

tools/testing/cxl contains the unit test infrastructure for mocking CXL
hierarchies. These are under the purview of the CXL subsystem maintainers.
Add the 'F:' entry for this to MAINTAINERS so that get_maintainer.pl
works as expected for patches to this area.

Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
Link: https://lore.kernel.org/r/20231026-vv-mainteners-fix-v1-1-0a0f25634073@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/region: Prepare the decoder match range helper for reuse
Alison Schofield [Thu, 26 Oct 2023 15:46:54 +0000 (08:46 -0700)]
cxl/region: Prepare the decoder match range helper for reuse

match_decoder_by_range() and decoder_match_range() both determine
if an HPA range matches a decoder. The first does it for root
decoders and the second one operates on switch decoders.

Tidy these up with clear naming and make the switch helper more
like the root decoder helper in style and functionality. Make it
take the actual range, rather than an endpoint decoder from which
it extracts the range. Require an exact match on switch decoders,
because unlike a root decoder that maps an entire region, Linux
only supports 1:1 mapping of switch to endpoint decoders. Note that
root-decoders are a super-set of switch-decoders and the range they
cover is a super-set of a region, hence the use of range_contains() for
that case.

Aside from aesthetics and maintainability, this is in preparation
for reuse.

Signed-off-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Link: https://lore.kernel.org/r/011b1f498e1758bb8df17c5951be00bd8d489e3b.1698263080.git.alison.schofield@intel.com
[djbw: fixup root decoder vs switch decoder range checks]
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/mbox: Remove useless cast in cxl_mem_create_range_info()
Alison Schofield [Tue, 15 Aug 2023 17:20:52 +0000 (10:20 -0700)]
cxl/mbox: Remove useless cast in cxl_mem_create_range_info()

DEFINE_RES_MEM() is a wrapper around the DEFINE_RES_NAMED() macro
which already has the (struct resource) for the compound literal.
The user of the macro should not repeat the cast.

Cleans up these sparse warnings:
drivers/cxl/core/mbox.c:1184:18: warning: cast to non-scalar
drivers/cxl/core/mbox.c:1184:18: warning: cast from non-scalar

Fixes: 52c4d11f1dce ("resource: Convert DEFINE_RES_NAMED() to be compound literal")
Signed-off-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20230815172052.22514-1-alison.schofield@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/region: Do not try to cleanup after cxl_region_setup_targets() fails
Jim Harris [Wed, 11 Oct 2023 14:51:31 +0000 (14:51 +0000)]
cxl/region: Do not try to cleanup after cxl_region_setup_targets() fails

Commit 5e42bcbc3fef ("cxl/region: decrement ->nr_targets on error in
cxl_region_attach()") tried to avoid 'eiw' initialization errors when
->nr_targets exceeded 16, by just decrementing ->nr_targets when
cxl_region_setup_targets() failed.

Commit 86987c766276 ("cxl/region: Cleanup target list on attach error")
extended that cleanup to also clear cxled->pos and p->targets[pos]. The
initialization error was incidentally fixed separately by:
Commit 8d4285425714 ("cxl/region: Fix port setup uninitialized variable
warnings") which was merged a few days after 5e42bcbc3fef.

But now the original cleanup when cxl_region_setup_targets() fails
prevents endpoint and switch decoder resources from being reused:

1) the cleanup does not set the decoder's region to NULL, which results
   in future dpa_size_store() calls returning -EBUSY
2) the decoder is not properly freed, which results in future commit
   errors associated with the upstream switch

Now that the initialization errors were fixed separately, the proper
cleanup for this case is to just return immediately. Then the resources
associated with this target get cleanup up as normal when the failed
region is deleted.

The ->nr_targets decrement in the error case also helped prevent
a p->targets[] array overflow, so add a new check to prevent against
that overflow.

Tested by trying to create an invalid region for a 2 switch * 2 endpoint
topology, and then following up with creating a valid region.

Fixes: 5e42bcbc3fef ("cxl/region: decrement ->nr_targets on error in cxl_region_attach()")
Cc: <stable@vger.kernel.org>
Signed-off-by: Jim Harris <jim.harris@samsung.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Acked-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/169703589120.1202031.14696100866518083806.stgit@bgt-140510-bm03.eng.stellus.in
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agotools/testing/cxl: Add 'sanitize notifier' support
Dan Williams [Wed, 4 Oct 2023 23:50:09 +0000 (16:50 -0700)]
tools/testing/cxl: Add 'sanitize notifier' support

Allow for cxl_test regression of the sanitize notifier. Reuse the core
setup infrastructure, and trigger notifications upon any sanitize
submission with a programmable notification delay.

Cc: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agotools/testing/cxl: Make cxl_memdev_state available to other command emulation
Dan Williams [Wed, 4 Oct 2023 23:56:34 +0000 (16:56 -0700)]
tools/testing/cxl: Make cxl_memdev_state available to other command emulation

Move @mds out of the event specific 'struct mock_event_store' and into
the base 'struct cxl_mockmem_data' directly. This is in preparation for
enabling cxl_test to exercise the notifier flow for 'sanitize' operation
completion.

Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
17 months agocxl/mem: Fix shutdown order
Dan Williams [Fri, 29 Sep 2023 01:02:07 +0000 (18:02 -0700)]
cxl/mem: Fix shutdown order

Ira reports that removing cxl_mock_mem causes a crash with the following
trace:

 BUG: kernel NULL pointer dereference, address: 0000000000000044
 [..]
 RIP: 0010:cxl_region_decode_reset+0x7f/0x180 [cxl_core]
 [..]
 Call Trace:
  <TASK>
  cxl_region_detach+0xe8/0x210 [cxl_core]
  cxl_decoder_kill_region+0x27/0x40 [cxl_core]
  cxld_unregister+0x29/0x40 [cxl_core]
  devres_release_all+0xb8/0x110
  device_unbind_cleanup+0xe/0x70
  device_release_driver_internal+0x1d2/0x210
  bus_remove_device+0xd7/0x150
  device_del+0x155/0x3e0
  device_unregister+0x13/0x60
  devm_release_action+0x4d/0x90
  ? __pfx_unregister_port+0x10/0x10 [cxl_core]
  delete_endpoint+0x121/0x130 [cxl_core]
  devres_release_all+0xb8/0x110
  device_unbind_cleanup+0xe/0x70
  device_release_driver_internal+0x1d2/0x210
  bus_remove_device+0xd7/0x150
  device_del+0x155/0x3e0
  ? lock_release+0x142/0x290
  cdev_device_del+0x15/0x50
  cxl_memdev_unregister+0x54/0x70 [cxl_core]

This crash is due to the clearing out the cxl_memdev's driver context
(@cxlds) before the subsystem is done with it. This is ultimately due to
the region(s), that this memdev is a member, being torn down and expecting
to be able to de-reference @cxlds, like here:

static int cxl_region_decode_reset(struct cxl_region *cxlr, int count)
...
                if (cxlds->rcd)
                        goto endpoint_reset;
...

Fix it by keeping the driver context valid until memdev-device
unregistration, and subsequently the entire stack of related
dependencies, unwinds.

Fixes: 9cc238c7a526 ("cxl/pci: Introduce cdevm_file_operations")
Reported-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
18 months agocxl/memdev: Fix sanitize vs decoder setup locking
Dan Williams [Thu, 5 Oct 2023 01:35:01 +0000 (18:35 -0700)]
cxl/memdev: Fix sanitize vs decoder setup locking

The sanitize operation is destructive and the expectation is that the
device is unmapped while in progress. The current implementation does a
lockless check for decoders being active, but then does nothing to
prevent decoders from racing to be committed. Introduce state tracking
to resolve this race.

This incidentally cleans up unpriveleged userspace from triggering mmio
read cycles by spinning on reading the 'security/state' attribute. Which
at a minimum is a waste since the kernel state machine can cache the
completion result.

Lastly cxl_mem_sanitize() was mistakenly marked EXPORT_SYMBOL() in the
original implementation, but an export was never required.

Fixes: 0c36b6ad436a ("cxl/mbox: Add sanitization handling machinery")
Cc: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
18 months agocxl/pci: Fix sanitize notifier setup
Dan Williams [Wed, 4 Oct 2023 23:49:36 +0000 (16:49 -0700)]
cxl/pci: Fix sanitize notifier setup

Fix a race condition between the mailbox-background command interrupt
firing and the security-state sysfs attribute being removed.

The race is difficult to see due to the awkward placement of the
sanitize-notifier setup code and the multiple places the teardown calls
are made, cxl_memdev_security_init() and cxl_memdev_security_shutdown().

Unify setup in one place, cxl_sanitize_setup_notifier(). Arrange for
the paired cxl_sanitize_teardown_notifier() to safely quiet the notifier
and let the cxl_memdev + irq be unregistered later in the flow.

Note: The special wrinkle of the sanitize notifier is that it interacts
with interrupts, which are enabled early in the flow, and it interacts
with memdev sysfs which is not initialized until late in the flow. Hence
why this setup routine takes an @cxlmd argument, and not just @mds.

This fix is also needed as a preparation fix for a memdev unregistration
crash.

Reported-by: Jonathan Cameron <Jonathan.Cameron@Huawei.com>
Closes: http://lore.kernel.org/r/20230929100316.00004546@Huawei.com
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Fixes: 0c36b6ad436a ("cxl/mbox: Add sanitization handling machinery")
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
18 months agocxl/pci: Clarify devm host for memdev relative setup
Dan Williams [Wed, 4 Oct 2023 23:04:49 +0000 (16:04 -0700)]
cxl/pci: Clarify devm host for memdev relative setup

It is all too easy to get confused about @dev usage in the CXL driver
stack. Before adding a new cxl_pci_probe() setup operation that has a
devm lifetime dependent on @cxlds->dev binding, but also references
@cxlmd->dev, and prints messages, rework the devm_cxl_add_memdev() and
cxl_memdev_setup_fw_upload() function signatures to make this
distinction explicit. I.e. pass in the devm context as an @host argument
rather than infer it from other objects.

This is in preparation for adding a devm_cxl_sanitize_setup_notifier().

Note the whitespace fixup near the change of the devm_cxl_add_memdev()
signature. That uncaught typo originated in the patch that added
cxl_memdev_security_init().

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
18 months agocxl/pci: Remove inconsistent usage of dev_err_probe()
Dan Williams [Wed, 4 Oct 2023 23:24:39 +0000 (16:24 -0700)]
cxl/pci: Remove inconsistent usage of dev_err_probe()

If dev_err_probe() is to be used it should at least be used consistently
within the same function. It is also worth questioning whether
every potential -ENOMEM needs an explicit error message.

Remove the cxl_setup_fw_upload() error prints for what are rare /
hardware-independent failures.

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
18 months agocxl/pci: Remove hardirq handler for cxl_request_irq()
Dan Williams [Wed, 4 Oct 2023 03:12:12 +0000 (20:12 -0700)]
cxl/pci: Remove hardirq handler for cxl_request_irq()

Now that all callers of cxl_request_irq() are using threaded irqs, drop
the hardirq handler option.

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
18 months agoACPICA: Add defines for CDAT SSLBIS
Dave Jiang [Sun, 1 Oct 2023 14:35:29 +0000 (07:35 -0700)]
ACPICA: Add defines for CDAT SSLBIS

Add upstream port and any port definition for SSLBIS.

Link: https://github.com/acpica/acpica/pull/898
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
18 months agoLinux 6.6-rc4
Linus Torvalds [Sun, 1 Oct 2023 21:15:13 +0000 (14:15 -0700)]
Linux 6.6-rc4

18 months agoMerge tag 'kbuild-fixes-v6.6-2' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 1 Oct 2023 20:48:46 +0000 (13:48 -0700)]
Merge tag 'kbuild-fixes-v6.6-2' of git://git./linux/kernel/git/masahiroy/linux-kbuild

Pull Kbuild fixes from Masahiro Yamada:

 - Fix the module compression with xz so the in-kernel decompressor
   works

 - Document a kconfig idiom to express an optional dependency between
   modules

 - Make modpost, when W=1 is given, detect broken drivers that reference
   .exit.* sections

 - Remove unused code

* tag 'kbuild-fixes-v6.6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild:
  kbuild: remove stale code for 'source' symlink in packaging scripts
  modpost: Don't let "driver"s reference .exit.*
  vmlinux.lds.h: remove unused CPU_KEEP and CPU_DISCARD macros
  modpost: add missing else to the "of" check
  Documentation: kbuild: explain handling optional dependencies
  kbuild: Use CRC32 and a 1MiB dictionary for XZ compressed modules

18 months agoMerge tag 'mm-hotfixes-stable-2023-10-01-08-34' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Sun, 1 Oct 2023 20:33:25 +0000 (13:33 -0700)]
Merge tag 'mm-hotfixes-stable-2023-10-01-08-34' of git://git./linux/kernel/git/akpm/mm

Pull misc fixes from Andrew Morton:
 "Fourteen hotfixes, eleven of which are cc:stable. The remainder
  pertain to issues which were introduced after 6.5"

* tag 'mm-hotfixes-stable-2023-10-01-08-34' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm:
  Crash: add lock to serialize crash hotplug handling
  selftests/mm: fix awk usage in charge_reserved_hugetlb.sh and hugetlb_reparenting_test.sh that may cause error
  mm: mempolicy: keep VMA walk if both MPOL_MF_STRICT and MPOL_MF_MOVE are specified
  mm/damon/vaddr-test: fix memory leak in damon_do_test_apply_three_regions()
  mm, memcg: reconsider kmem.limit_in_bytes deprecation
  mm: zswap: fix potential memory corruption on duplicate store
  arm64: hugetlb: fix set_huge_pte_at() to work with all swap entries
  mm: hugetlb: add huge page size param to set_huge_pte_at()
  maple_tree: add MAS_UNDERFLOW and MAS_OVERFLOW states
  maple_tree: add mas_is_active() to detect in-tree walks
  nilfs2: fix potential use after free in nilfs_gccache_submit_read_data()
  mm: abstract moving to the next PFN
  mm: report success more often from filemap_map_folio_range()
  fs: binfmt_elf_efpic: fix personality for ELF-FDPIC

18 months agoMerge tag 'char-misc-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sun, 1 Oct 2023 19:50:04 +0000 (12:50 -0700)]
Merge tag 'char-misc-6.6-rc4' of git://git./linux/kernel/git/gregkh/char-misc

Pull misc driver fix from Greg KH:
 "Here is a single, much requested, fix for a set of misc drivers to
  resolve a much reported regression in the -rc series that has also
  propagated back to the stable releases. Sorry for the delay, lots of
  conference travel for a few weeks put me very far behind in patch
  wrangling.

  It has been reported by many to resolve the reported problem, and has
  been in linux-next with no reported issues"

* tag 'char-misc-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  misc: rtsx: Fix some platforms can not boot and move the l1ss judgment to probe

18 months agoMerge tag 'tty-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sun, 1 Oct 2023 19:44:45 +0000 (12:44 -0700)]
Merge tag 'tty-6.6-rc4' of git://git./linux/kernel/git/gregkh/tty

Pull tty / serial driver fixes from Greg KH:
 "Here are two tty/serial driver fixes for 6.6-rc4 that resolve some
  reported regressions:

   - revert a n_gsm change that ended up causing problems

   - 8250_port fix for irq data

  both have been in linux-next for over a week with no reported
  problems"

* tag 'tty-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  Revert "tty: n_gsm: fix UAF in gsm_cleanup_mux"
  serial: 8250_port: Check IRQ data before use

18 months agoMerge tag 'x86-urgent-2023-10-01' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 1 Oct 2023 16:50:58 +0000 (09:50 -0700)]
Merge tag 'x86-urgent-2023-10-01' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Ingo Molnar:
 "Misc fixes: a kerneldoc build warning fix, add SRSO mitigation for
  AMD-derived Hygon processors, and fix a SGX kernel crash in the page
  fault handler that can trigger when ksgxd races to reclaim the SECS
  special page, by making the SECS page unswappable"

* tag 'x86-urgent-2023-10-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/sgx: Resolves SECS reclaim vs. page fault for EAUG race
  x86/srso: Add SRSO mitigation for Hygon processors
  x86/kgdb: Fix a kerneldoc warning when build with W=1

18 months agoMerge tag 'timers-urgent-2023-10-01' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 1 Oct 2023 16:41:58 +0000 (09:41 -0700)]
Merge tag 'timers-urgent-2023-10-01' of git://git./linux/kernel/git/tip/tip

Pull timer fix from Ingo Molnar:
 "Fix a spurious kernel warning during CPU hotplug events that may
  trigger when timer/hrtimer softirqs are pending, which are otherwise
  hotplug-safe and don't merit a warning"

* tag 'timers-urgent-2023-10-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  timers: Tag (hr)timer softirq as hotplug safe

18 months agoMerge tag 'sched-urgent-2023-10-01' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 1 Oct 2023 16:38:05 +0000 (09:38 -0700)]
Merge tag 'sched-urgent-2023-10-01' of git://git./linux/kernel/git/tip/tip

Pull scheduler fix from Ingo Molnar:
 "Fix a RT tasks related lockup/live-lock during CPU offlining"

* tag 'sched-urgent-2023-10-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/rt: Fix live lock between select_fallback_rq() and RT push

18 months agoMerge tag 'perf-urgent-2023-10-01' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 1 Oct 2023 16:34:53 +0000 (09:34 -0700)]
Merge tag 'perf-urgent-2023-10-01' of git://git./linux/kernel/git/tip/tip

Pull perf event fixes from Ingo Molnar:
 "Misc fixes: work around an AMD microcode bug on certain models, and
  fix kexec kernel PMI handlers on AMD systems that get loaded on older
  kernels that have an unexpected register state"

* tag 'perf-urgent-2023-10-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/x86/amd: Do not WARN() on every IRQ
  perf/x86/amd/core: Fix overflow reset on hotplug

18 months agokbuild: remove stale code for 'source' symlink in packaging scripts
Masahiro Yamada [Sun, 1 Oct 2023 14:03:39 +0000 (23:03 +0900)]
kbuild: remove stale code for 'source' symlink in packaging scripts

Since commit d8131c2965d5 ("kbuild: remove $(MODLIB)/source symlink"),
modules_install does not create the 'source' symlink.

Remove the stale code from builddeb and kernel.spec.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
18 months agomodpost: Don't let "driver"s reference .exit.*
Uwe Kleine-König [Sat, 30 Sep 2023 16:52:04 +0000 (18:52 +0200)]
modpost: Don't let "driver"s reference .exit.*

Drivers must not reference functions marked with __exit as these likely
are not available when the code is built-in.

There are few creative offenders uncovered for example in ARCH=amd64
allmodconfig builds. So only trigger the section mismatch warning for
W=1 builds.

The dual rule that drivers must not reference .init.* is implemented
since commit 0db252452378 ("modpost: don't allow *driver to reference
.init.*") which however missed that .exit.* should be handled in the
same way.

Thanks to Masahiro Yamada and Arnd Bergmann who gave valuable hints to
find this improvement.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
18 months agovmlinux.lds.h: remove unused CPU_KEEP and CPU_DISCARD macros
Masahiro Yamada [Sat, 30 Sep 2023 07:13:35 +0000 (16:13 +0900)]
vmlinux.lds.h: remove unused CPU_KEEP and CPU_DISCARD macros

Remove the left-over of commit e24f6628811e ("modpost: remove all
traces of cpuinit/cpuexit sections").

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Acked-by: Paul Gortmaker <paul.gortmaker@windriver.com>
18 months agomodpost: add missing else to the "of" check
Mauricio Faria de Oliveira [Thu, 28 Sep 2023 20:28:07 +0000 (17:28 -0300)]
modpost: add missing else to the "of" check

Without this 'else' statement, an "usb" name goes into two handlers:
the first/previous 'if' statement _AND_ the for-loop over 'devtable',
but the latter is useless as it has no 'usb' device_id entry anyway.

Tested with allmodconfig before/after patch; no changes to *.mod.c:

    git checkout v6.6-rc3
    make -j$(nproc) allmodconfig
    make -j$(nproc) olddefconfig

    make -j$(nproc)
    find . -name '*.mod.c' | cpio -pd /tmp/before

    # apply patch

    make -j$(nproc)
    find . -name '*.mod.c' | cpio -pd /tmp/after

    diff -r /tmp/before/ /tmp/after/
    # no difference

Fixes: acbef7b76629 ("modpost: fix module autoloading for OF devices with generic compatible property")
Signed-off-by: Mauricio Faria de Oliveira <mfo@canonical.com>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
18 months agoMerge tag 'soc-fixes-6.6' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Sun, 1 Oct 2023 01:41:37 +0000 (18:41 -0700)]
Merge tag 'soc-fixes-6.6' of git://git./linux/kernel/git/soc/soc

Pull ARM SoC fixes from Arnd Bergmann:
 "These are the latest bug fixes that have come up in the soc tree. Most
  of these are fairly minor. Most notably, the majority of changes this
  time are not for dts files as usual.

   - Updates to the addresses of the broadcom and aspeed entries in the
     MAINTAINERS file.

   - Defconfig updates to address a regression on samsung and a build
     warning from an unknown Kconfig symbol

   - Build fixes for the StrongARM and Uniphier platforms

   - Code fixes for SCMI and FF-A firmware drivers, both of which had a
     simple bug that resulted in invalid data, and a lesser fix for the
     optee firmware driver

   - Multiple fixes for the recently added loongson/loongarch "guts" soc
     driver

   - Devicetree fixes for RISC-V on the startfive platform, addressing
     issues with NOR flash, usb and uart.

   - Multiple fixes for NXP i.MX8/i.MX9 dts files, fixing problems with
     clock, gpio, hdmi settings and the Makefile

   - Bug fixes for i.MX firmware code and the OCOTP soc driver

   - Multiple fixes for the TI sysc bus driver

   - Minor dts updates for TI omap dts files, to address boot time
     warnings and errors"

* tag 'soc-fixes-6.6' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (35 commits)
  MAINTAINERS: Fix Florian Fainelli's email address
  arm64: defconfig: enable syscon-poweroff driver
  ARM: locomo: fix locomolcd_power declaration
  soc: loongson: loongson2_guts: Remove unneeded semicolon
  soc: loongson: loongson2_guts: Convert to devm_platform_ioremap_resource()
  soc: loongson: loongson_pm2: Populate children syscon nodes
  dt-bindings: soc: loongson,ls2k-pmc: Allow syscon-reboot/syscon-poweroff as child
  soc: loongson: loongson_pm2: Drop useless of_device_id compatible
  dt-bindings: soc: loongson,ls2k-pmc: Use fallbacks for ls2k-pmc compatible
  soc: loongson: loongson_pm2: Add dependency for INPUT
  arm64: defconfig: remove CONFIG_COMMON_CLK_NPCM8XX=y
  ARM: uniphier: fix cache kernel-doc warnings
  MAINTAINERS: aspeed: Update Andrew's email address
  MAINTAINERS: aspeed: Update git tree URL
  firmware: arm_ffa: Don't set the memory region attributes for MEM_LEND
  arm64: dts: imx: Add imx8mm-prt8mm.dtb to build
  arm64: dts: imx8mm-evk: Fix hdmi@3d node
  soc: imx8m: Enable OCOTP clock for imx8mm before reading registers
  arm64: dts: imx8mp-beacon-kit: Fix audio_pll2 clock
  arm64: dts: imx8mp: Fix SDMA2/3 clocks
  ...

18 months agoMerge tag 'trace-v6.6-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace...
Linus Torvalds [Sun, 1 Oct 2023 01:19:02 +0000 (18:19 -0700)]
Merge tag 'trace-v6.6-rc3' of git://git./linux/kernel/git/trace/linux-trace

Pull tracing fixes from Steven Rostedt:

 - Make sure 32-bit applications using user events have aligned access
   when running on a 64-bit kernel.

 - Add cond_resched in the loop that handles converting enums in
   print_fmt string is trace events.

 - Fix premature wake ups of polling processes in the tracing ring
   buffer. When a task polls waiting for a percentage of the ring buffer
   to be filled, the writer still will wake it up at every event. Add
   the polling's percentage to the "shortest_full" list to tell the
   writer when to wake it up.

 - For eventfs dir lookups on dynamic events, an event system's only
   event could be removed, leaving its dentry with no children. This is
   totally legitimate. But in eventfs_release() it must not access the
   children array, as it is only allocated when the dentry has children.

* tag 'trace-v6.6-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  eventfs: Test for dentries array allocated in eventfs_release()
  tracing/user_events: Align set_bit() address for all archs
  tracing: relax trace_event_eval_update() execution with cond_resched()
  ring-buffer: Update "shortest_full" in polling

18 months agoeventfs: Test for dentries array allocated in eventfs_release()
Steven Rostedt (Google) [Sat, 30 Sep 2023 13:01:06 +0000 (09:01 -0400)]
eventfs: Test for dentries array allocated in eventfs_release()

The dcache_dir_open_wrapper() could be called when a dynamic event is
being deleted leaving a dentry with no children. In this case the
dlist->dentries array will never be allocated. This needs to be checked
for in eventfs_release(), otherwise it will trigger a NULL pointer
dereference.

Link: https://lore.kernel.org/linux-trace-kernel/20230930090106.1c3164e9@rorschach.local.home
Cc: Mark Rutland <mark.rutland@arm.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Fixes: ef36b4f92868 ("eventfs: Remember what dentries were created on dir open")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
18 months agotracing/user_events: Align set_bit() address for all archs
Beau Belgrave [Mon, 25 Sep 2023 23:08:28 +0000 (23:08 +0000)]
tracing/user_events: Align set_bit() address for all archs

All architectures should use a long aligned address passed to set_bit().
User processes can pass either a 32-bit or 64-bit sized value to be
updated when tracing is enabled when on a 64-bit kernel. Both cases are
ensured to be naturally aligned, however, that is not enough. The
address must be long aligned without affecting checks on the value
within the user process which require different adjustments for the bit
for little and big endian CPUs.

Add a compat flag to user_event_enabler that indicates when a 32-bit
value is being used on a 64-bit kernel. Long align addresses and correct
the bit to be used by set_bit() to account for this alignment. Ensure
compat flags are copied during forks and used during deletion clears.

Link: https://lore.kernel.org/linux-trace-kernel/20230925230829.341-2-beaub@linux.microsoft.com
Link: https://lore.kernel.org/linux-trace-kernel/20230914131102.179100-1-cleger@rivosinc.com/
Cc: stable@vger.kernel.org
Fixes: 7235759084a4 ("tracing/user_events: Use remote writes for event enablement")
Reported-by: Clément Léger <cleger@rivosinc.com>
Suggested-by: Clément Léger <cleger@rivosinc.com>
Signed-off-by: Beau Belgrave <beaub@linux.microsoft.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
18 months agotracing: relax trace_event_eval_update() execution with cond_resched()
Clément Léger [Fri, 29 Sep 2023 19:16:37 +0000 (21:16 +0200)]
tracing: relax trace_event_eval_update() execution with cond_resched()

When kernel is compiled without preemption, the eval_map_work_func()
(which calls trace_event_eval_update()) will not be preempted up to its
complete execution. This can actually cause a problem since if another
CPU call stop_machine(), the call will have to wait for the
eval_map_work_func() function to finish executing in the workqueue
before being able to be scheduled. This problem was observe on a SMP
system at boot time, when the CPU calling the initcalls executed
clocksource_done_booting() which in the end calls stop_machine(). We
observed a 1 second delay because one CPU was executing
eval_map_work_func() and was not preempted by the stop_machine() task.

Adding a call to cond_resched() in trace_event_eval_update() allows
other tasks to be executed and thus continue working asynchronously
like before without blocking any pending task at boot time.

Link: https://lore.kernel.org/linux-trace-kernel/20230929191637.416931-1-cleger@rivosinc.com
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Clément Léger <cleger@rivosinc.com>
Tested-by: Atish Patra <atishp@rivosinc.com>
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
18 months agoring-buffer: Update "shortest_full" in polling
Steven Rostedt (Google) [Fri, 29 Sep 2023 22:01:13 +0000 (18:01 -0400)]
ring-buffer: Update "shortest_full" in polling

It was discovered that the ring buffer polling was incorrectly stating
that read would not block, but that's because polling did not take into
account that reads will block if the "buffer-percent" was set. Instead,
the ring buffer polling would say reads would not block if there was any
data in the ring buffer. This was incorrect behavior from a user space
point of view. This was fixed by commit 42fb0a1e84ff by having the polling
code check if the ring buffer had more data than what the user specified
"buffer percent" had.

The problem now is that the polling code did not register itself to the
writer that it wanted to wait for a specific "full" value of the ring
buffer. The result was that the writer would wake the polling waiter
whenever there was a new event. The polling waiter would then wake up, see
that there's not enough data in the ring buffer to notify user space and
then go back to sleep. The next event would wake it up again.

Before the polling fix was added, the code would wake up around 100 times
for a hackbench 30 benchmark. After the "fix", due to the constant waking
of the writer, it would wake up over 11,0000 times! It would never leave
the kernel, so the user space behavior was still "correct", but this
definitely is not the desired effect.

To fix this, have the polling code add what it's waiting for to the
"shortest_full" variable, to tell the writer not to wake it up if the
buffer is not as full as it expects to be.

Note, after this fix, it appears that the waiter is now woken up around 2x
the times it was before (~200). This is a tremendous improvement from the
11,000 times, but I will need to spend some time to see why polling is
more aggressive in its wakeups than the read blocking code.

Link: https://lore.kernel.org/linux-trace-kernel/20230929180113.01c2cae3@rorschach.local.home
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Fixes: 42fb0a1e84ff ("tracing/ring-buffer: Have polling block on watermark")
Reported-by: Julia Lawall <julia.lawall@inria.fr>
Tested-by: Julia Lawall <julia.lawall@inria.fr>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
18 months agoMerge tag 'dma-mapping-6.6-2023-09-30' of git://git.infradead.org/users/hch/dma-mapping
Linus Torvalds [Sat, 30 Sep 2023 18:07:26 +0000 (11:07 -0700)]
Merge tag 'dma-mapping-6.6-2023-09-30' of git://git.infradead.org/users/hch/dma-mapping

Pull dma-mapping fixes from Christoph Hellwig:

 - fix the narea calculation in swiotlb initialization (Ross Lagerwall)

 - fix the check whether a device has used swiotlb (Petr Tesarik)

* tag 'dma-mapping-6.6-2023-09-30' of git://git.infradead.org/users/hch/dma-mapping:
  swiotlb: fix the check whether a device has used software IO TLB
  swiotlb: use the calculated number of areas

18 months agoMerge tag 'iomap-6.6-fixes-4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Sat, 30 Sep 2023 18:01:38 +0000 (11:01 -0700)]
Merge tag 'iomap-6.6-fixes-4' of git://git./fs/xfs/xfs-linux

Pull iomap fixes from Darrick Wong:

 - Handle a race between writing and shrinking block devices by
   returning EIO

 - Fix a typo in a comment

* tag 'iomap-6.6-fixes-4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  iomap: Spelling s/preceeding/preceding/g
  iomap: add a workaround for racy i_size updates on block devices

18 months agoMerge tag 'i2c-for-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Sat, 30 Sep 2023 17:07:33 +0000 (10:07 -0700)]
Merge tag 'i2c-for-6.6-rc4' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "Usual business: a driver fix, a DT fix, a minor core fix"

* tag 'i2c-for-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: npcm7xx: Fix callback completion ordering
  i2c: mux: Avoid potential false error message in i2c_mux_add_adapter
  dt-bindings: i2c: mxs: Pass ref and 'unevaluatedProperties: false'

18 months agoMerge tag 'acpi-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Sat, 30 Sep 2023 16:59:37 +0000 (09:59 -0700)]
Merge tag 'acpi-6.6-rc4' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI fix from Rafael Wysocki:
 "Fix a possible NULL pointer dereference in the error path of
  acpi_video_bus_add() resulting from recent changes (Dinghao Liu)"

* tag 'acpi-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI: video: Fix NULL pointer dereference in acpi_video_bus_add()

18 months agoMerge tag 'powerpc-6.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc...
Linus Torvalds [Sat, 30 Sep 2023 16:53:09 +0000 (09:53 -0700)]
Merge tag 'powerpc-6.6-3' of git://git./linux/kernel/git/powerpc/linux

Pull powerpc fixes from Michael Ellerman:

 - Fix arch_stack_walk_reliable(), used by live patching

 - Fix powerpc selftests to work with run_kselftest.sh

Thanks to Joe Lawrence and Petr Mladek.

* tag 'powerpc-6.6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  selftests/powerpc: Fix emit_tests to work with run_kselftest.sh
  powerpc/stacktrace: Fix arch_stack_walk_reliable()

18 months agoMerge tag 'nfsd-6.6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux
Linus Torvalds [Sat, 30 Sep 2023 16:44:48 +0000 (09:44 -0700)]
Merge tag 'nfsd-6.6-2' of git://git./linux/kernel/git/cel/linux

Pull nfsd fix from Chuck Lever:

 - Fix NFSv4 READ corner case

* tag 'nfsd-6.6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux:
  NFSD: Fix zero NFSv4 READ results when RQ_SPLICE_OK is not set

18 months agoMerge tag '6.6-rc3-smb3-client-fix' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sat, 30 Sep 2023 16:39:23 +0000 (09:39 -0700)]
Merge tag '6.6-rc3-smb3-client-fix' of git://git.samba.org/sfrench/cifs-2.6

Pull smb client fix from Steve French:
 "Fix for password freeing potential oops (also for stable)"

* tag '6.6-rc3-smb3-client-fix' of git://git.samba.org/sfrench/cifs-2.6:
  fs/smb/client: Reset password pointer to NULL

18 months agoCrash: add lock to serialize crash hotplug handling
Baoquan He [Tue, 26 Sep 2023 12:09:05 +0000 (20:09 +0800)]
Crash: add lock to serialize crash hotplug handling

Eric reported that handling corresponding crash hotplug event can be
failed easily when many memory hotplug event are notified in a short
period.  They failed because failing to take __kexec_lock.

=======
[   78.714569] Fallback order for Node 0: 0
[   78.714575] Built 1 zonelists, mobility grouping on.  Total pages: 1817886
[   78.717133] Policy zone: Normal
[   78.724423] crash hp: kexec_trylock() failed, elfcorehdr may be inaccurate
[   78.727207] crash hp: kexec_trylock() failed, elfcorehdr may be inaccurate
[   80.056643] PEFILE: Unsigned PE binary
=======

The memory hotplug events are notified very quickly and very many, while
the handling of crash hotplug is much slower relatively.  So the atomic
variable __kexec_lock and kexec_trylock() can't guarantee the
serialization of crash hotplug handling.

Here, add a new mutex lock __crash_hotplug_lock to serialize crash hotplug
handling specifically.  This doesn't impact the usage of __kexec_lock.

Link: https://lkml.kernel.org/r/20230926120905.392903-1-bhe@redhat.com
Fixes: 247262756121 ("crash: add generic infrastructure for crash hotplug support")
Signed-off-by: Baoquan He <bhe@redhat.com>
Tested-by: Eric DeVolder <eric.devolder@oracle.com>
Reviewed-by: Eric DeVolder <eric.devolder@oracle.com>
Reviewed-by: Valentin Schneider <vschneid@redhat.com>
Cc: Sourabh Jain <sourabhjain@linux.ibm.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agoselftests/mm: fix awk usage in charge_reserved_hugetlb.sh and hugetlb_reparenting_tes...
Juntong Deng [Tue, 26 Sep 2023 18:19:44 +0000 (02:19 +0800)]
selftests/mm: fix awk usage in charge_reserved_hugetlb.sh and hugetlb_reparenting_test.sh that may cause error

According to the awk manual, the -e option does not need to be specified
in front of 'program' (unless you need to mix program-file).

The redundant -e option can cause error when users use awk tools other
than gawk (for example, mawk does not support the -e option).

Error Example:
awk: not an option: -e

Link: https://lkml.kernel.org/r/VI1P193MB075228810591AF2FDD7D42C599C3A@VI1P193MB0752.EURP193.PROD.OUTLOOK.COM
Signed-off-by: Juntong Deng <juntong.deng@outlook.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agomm: mempolicy: keep VMA walk if both MPOL_MF_STRICT and MPOL_MF_MOVE are specified
Yang Shi [Wed, 20 Sep 2023 22:32:42 +0000 (15:32 -0700)]
mm: mempolicy: keep VMA walk if both MPOL_MF_STRICT and MPOL_MF_MOVE are specified

When calling mbind() with MPOL_MF_{MOVE|MOVEALL} | MPOL_MF_STRICT, kernel
should attempt to migrate all existing pages, and return -EIO if there is
misplaced or unmovable page.  Then commit 6f4576e3687b ("mempolicy: apply
page table walker on queue_pages_range()") messed up the return value and
didn't break VMA scan early ianymore when MPOL_MF_STRICT alone.  The
return value problem was fixed by commit a7f40cfe3b7a ("mm: mempolicy:
make mbind() return -EIO when MPOL_MF_STRICT is specified"), but it broke
the VMA walk early if unmovable page is met, it may cause some pages are
not migrated as expected.

The code should conceptually do:

 if (MPOL_MF_MOVE|MOVEALL)
     scan all vmas
     try to migrate the existing pages
     return success
 else if (MPOL_MF_MOVE* | MPOL_MF_STRICT)
     scan all vmas
     try to migrate the existing pages
     return -EIO if unmovable or migration failed
 else /* MPOL_MF_STRICT alone */
     break early if meets unmovable and don't call mbind_range() at all
 else /* none of those flags */
     check the ranges in test_walk, EFAULT without mbind_range() if discontig.

Fixed the behavior.

Link: https://lkml.kernel.org/r/20230920223242.3425775-1-yang@os.amperecomputing.com
Fixes: a7f40cfe3b7a ("mm: mempolicy: make mbind() return -EIO when MPOL_MF_STRICT is specified")
Signed-off-by: Yang Shi <yang@os.amperecomputing.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Rafael Aquini <aquini@redhat.com>
Cc: Kirill A. Shutemov <kirill@shutemov.name>
Cc: David Rientjes <rientjes@google.com>
Cc: <stable@vger.kernel.org> [4.9+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agomm/damon/vaddr-test: fix memory leak in damon_do_test_apply_three_regions()
Jinjie Ruan [Mon, 25 Sep 2023 07:20:59 +0000 (15:20 +0800)]
mm/damon/vaddr-test: fix memory leak in damon_do_test_apply_three_regions()

When CONFIG_DAMON_VADDR_KUNIT_TEST=y and making CONFIG_DEBUG_KMEMLEAK=y
and CONFIG_DEBUG_KMEMLEAK_AUTO_SCAN=y, the below memory leak is detected.

Since commit 9f86d624292c ("mm/damon/vaddr-test: remove unnecessary
variables"), the damon_destroy_ctx() is removed, but still call
damon_new_target() and damon_new_region(), the damon_region which is
allocated by kmem_cache_alloc() in damon_new_region() and the damon_target
which is allocated by kmalloc in damon_new_target() are not freed.  And
the damon_region which is allocated in damon_new_region() in
damon_set_regions() is also not freed.

So use damon_destroy_target to free all the damon_regions and damon_target.

    unreferenced object 0xffff888107c9a940 (size 64):
      comm "kunit_try_catch", pid 1069, jiffies 4294670592 (age 732.761s)
      hex dump (first 32 bytes):
        00 00 00 00 00 00 00 00 06 00 00 00 6b 6b 6b 6b  ............kkkk
        60 c7 9c 07 81 88 ff ff f8 cb 9c 07 81 88 ff ff  `...............
      backtrace:
        [<ffffffff817e0167>] kmalloc_trace+0x27/0xa0
        [<ffffffff819c11cf>] damon_new_target+0x3f/0x1b0
        [<ffffffff819c7d55>] damon_do_test_apply_three_regions.constprop.0+0x95/0x3e0
        [<ffffffff819c82be>] damon_test_apply_three_regions1+0x21e/0x260
        [<ffffffff829fce6a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
        [<ffffffff81237cf6>] kthread+0x2b6/0x380
        [<ffffffff81097add>] ret_from_fork+0x2d/0x70
        [<ffffffff81003791>] ret_from_fork_asm+0x11/0x20
    unreferenced object 0xffff8881079cc740 (size 56):
      comm "kunit_try_catch", pid 1069, jiffies 4294670592 (age 732.761s)
      hex dump (first 32 bytes):
        05 00 00 00 00 00 00 00 14 00 00 00 00 00 00 00  ................
        6b 6b 6b 6b 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b  kkkkkkkk....kkkk
      backtrace:
        [<ffffffff819bc492>] damon_new_region+0x22/0x1c0
        [<ffffffff819c7d91>] damon_do_test_apply_three_regions.constprop.0+0xd1/0x3e0
        [<ffffffff819c82be>] damon_test_apply_three_regions1+0x21e/0x260
        [<ffffffff829fce6a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
        [<ffffffff81237cf6>] kthread+0x2b6/0x380
        [<ffffffff81097add>] ret_from_fork+0x2d/0x70
        [<ffffffff81003791>] ret_from_fork_asm+0x11/0x20
    unreferenced object 0xffff888107c9ac40 (size 64):
      comm "kunit_try_catch", pid 1071, jiffies 4294670595 (age 732.843s)
      hex dump (first 32 bytes):
        00 00 00 00 00 00 00 00 06 00 00 00 6b 6b 6b 6b  ............kkkk
        a0 cc 9c 07 81 88 ff ff 78 a1 76 07 81 88 ff ff  ........x.v.....
      backtrace:
        [<ffffffff817e0167>] kmalloc_trace+0x27/0xa0
        [<ffffffff819c11cf>] damon_new_target+0x3f/0x1b0
        [<ffffffff819c7d55>] damon_do_test_apply_three_regions.constprop.0+0x95/0x3e0
        [<ffffffff819c851e>] damon_test_apply_three_regions2+0x21e/0x260
        [<ffffffff829fce6a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
        [<ffffffff81237cf6>] kthread+0x2b6/0x380
        [<ffffffff81097add>] ret_from_fork+0x2d/0x70
        [<ffffffff81003791>] ret_from_fork_asm+0x11/0x20
    unreferenced object 0xffff8881079ccc80 (size 56):
      comm "kunit_try_catch", pid 1071, jiffies 4294670595 (age 732.843s)
      hex dump (first 32 bytes):
        05 00 00 00 00 00 00 00 14 00 00 00 00 00 00 00  ................
        6b 6b 6b 6b 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b  kkkkkkkk....kkkk
      backtrace:
        [<ffffffff819bc492>] damon_new_region+0x22/0x1c0
        [<ffffffff819c7d91>] damon_do_test_apply_three_regions.constprop.0+0xd1/0x3e0
        [<ffffffff819c851e>] damon_test_apply_three_regions2+0x21e/0x260
        [<ffffffff829fce6a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
        [<ffffffff81237cf6>] kthread+0x2b6/0x380
        [<ffffffff81097add>] ret_from_fork+0x2d/0x70
        [<ffffffff81003791>] ret_from_fork_asm+0x11/0x20
    unreferenced object 0xffff888107c9af40 (size 64):
      comm "kunit_try_catch", pid 1073, jiffies 4294670597 (age 733.011s)
      hex dump (first 32 bytes):
        00 00 00 00 00 00 00 00 06 00 00 00 6b 6b 6b 6b  ............kkkk
        20 a2 76 07 81 88 ff ff b8 a6 76 07 81 88 ff ff   .v.......v.....
      backtrace:
        [<ffffffff817e0167>] kmalloc_trace+0x27/0xa0
        [<ffffffff819c11cf>] damon_new_target+0x3f/0x1b0
        [<ffffffff819c7d55>] damon_do_test_apply_three_regions.constprop.0+0x95/0x3e0
        [<ffffffff819c877e>] damon_test_apply_three_regions3+0x21e/0x260
        [<ffffffff829fce6a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
        [<ffffffff81237cf6>] kthread+0x2b6/0x380
        [<ffffffff81097add>] ret_from_fork+0x2d/0x70
        [<ffffffff81003791>] ret_from_fork_asm+0x11/0x20
    unreferenced object 0xffff88810776a200 (size 56):
      comm "kunit_try_catch", pid 1073, jiffies 4294670597 (age 733.011s)
      hex dump (first 32 bytes):
        05 00 00 00 00 00 00 00 14 00 00 00 00 00 00 00  ................
        6b 6b 6b 6b 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b  kkkkkkkk....kkkk
      backtrace:
        [<ffffffff819bc492>] damon_new_region+0x22/0x1c0
        [<ffffffff819c7d91>] damon_do_test_apply_three_regions.constprop.0+0xd1/0x3e0
        [<ffffffff819c877e>] damon_test_apply_three_regions3+0x21e/0x260
        [<ffffffff829fce6a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
        [<ffffffff81237cf6>] kthread+0x2b6/0x380
        [<ffffffff81097add>] ret_from_fork+0x2d/0x70
        [<ffffffff81003791>] ret_from_fork_asm+0x11/0x20
    unreferenced object 0xffff88810776a740 (size 56):
      comm "kunit_try_catch", pid 1073, jiffies 4294670597 (age 733.025s)
      hex dump (first 32 bytes):
        3d 00 00 00 00 00 00 00 3f 00 00 00 00 00 00 00  =.......?.......
        6b 6b 6b 6b 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b  kkkkkkkk....kkkk
      backtrace:
        [<ffffffff819bc492>] damon_new_region+0x22/0x1c0
        [<ffffffff819bfcc2>] damon_set_regions+0x4c2/0x8e0
        [<ffffffff819c7dbb>] damon_do_test_apply_three_regions.constprop.0+0xfb/0x3e0
        [<ffffffff819c877e>] damon_test_apply_three_regions3+0x21e/0x260
        [<ffffffff829fce6a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
        [<ffffffff81237cf6>] kthread+0x2b6/0x380
        [<ffffffff81097add>] ret_from_fork+0x2d/0x70
        [<ffffffff81003791>] ret_from_fork_asm+0x11/0x20
    unreferenced object 0xffff888108038240 (size 64):
      comm "kunit_try_catch", pid 1075, jiffies 4294670600 (age 733.022s)
      hex dump (first 32 bytes):
        00 00 00 00 00 00 00 00 03 00 00 00 6b 6b 6b 6b  ............kkkk
        48 ad 76 07 81 88 ff ff 98 ae 76 07 81 88 ff ff  H.v.......v.....
      backtrace:
        [<ffffffff817e0167>] kmalloc_trace+0x27/0xa0
        [<ffffffff819c11cf>] damon_new_target+0x3f/0x1b0
        [<ffffffff819c7d55>] damon_do_test_apply_three_regions.constprop.0+0x95/0x3e0
        [<ffffffff819c898d>] damon_test_apply_three_regions4+0x1cd/0x210
        [<ffffffff829fce6a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
        [<ffffffff81237cf6>] kthread+0x2b6/0x380
        [<ffffffff81097add>] ret_from_fork+0x2d/0x70
        [<ffffffff81003791>] ret_from_fork_asm+0x11/0x20
    unreferenced object 0xffff88810776ad28 (size 56):
      comm "kunit_try_catch", pid 1075, jiffies 4294670600 (age 733.022s)
      hex dump (first 32 bytes):
        05 00 00 00 00 00 00 00 07 00 00 00 00 00 00 00  ................
        6b 6b 6b 6b 6b 6b 6b 6b 00 00 00 00 6b 6b 6b 6b  kkkkkkkk....kkkk
      backtrace:
        [<ffffffff819bc492>] damon_new_region+0x22/0x1c0
        [<ffffffff819bfcc2>] damon_set_regions+0x4c2/0x8e0
        [<ffffffff819c7dbb>] damon_do_test_apply_three_regions.constprop.0+0xfb/0x3e0
        [<ffffffff819c898d>] damon_test_apply_three_regions4+0x1cd/0x210
        [<ffffffff829fce6a>] kunit_generic_run_threadfn_adapter+0x4a/0x90
        [<ffffffff81237cf6>] kthread+0x2b6/0x380
        [<ffffffff81097add>] ret_from_fork+0x2d/0x70
        [<ffffffff81003791>] ret_from_fork_asm+0x11/0x20

Link: https://lkml.kernel.org/r/20230925072100.3725620-1-ruanjinjie@huawei.com
Fixes: 9f86d624292c ("mm/damon/vaddr-test: remove unnecessary variables")
Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
Reviewed-by: SeongJae Park <sj@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agomm, memcg: reconsider kmem.limit_in_bytes deprecation
Michal Hocko [Thu, 21 Sep 2023 07:38:29 +0000 (09:38 +0200)]
mm, memcg: reconsider kmem.limit_in_bytes deprecation

This reverts commits 86327e8eb94c ("memcg: drop kmem.limit_in_bytes") and
partially reverts 58056f77502f ("memcg, kmem: further deprecate
kmem.limit_in_bytes") which have incrementally removed support for the
kernel memory accounting hard limit.  Unfortunately it has turned out that
there is still userspace depending on the existence of
memory.kmem.limit_in_bytes [1].  The underlying functionality is not
really required but the non-existent file just confuses the userspace
which fails in the result.  The patch to fix this on the userspace side
has been submitted but it is hard to predict how it will propagate through
the maze of 3rd party consumers of the software.

Now, reverting alone 86327e8eb94c is not an option because there is
another set of userspace which cannot cope with ENOTSUPP returned when
writing to the file.  Therefore we have to go and revisit 58056f77502f as
well.  There are two ways to go ahead.  Either we give up on the
deprecation and fully revert 58056f77502f as well or we can keep
kmem.limit_in_bytes but make the write a noop and warn about the fact.
This should work for both known breaking workloads which depend on the
existence but do not depend on the hard limit enforcement.

Note to backporters to stable trees.  a8c49af3be5f ("memcg: add per-memcg
total kernel memory stat") introduced in 4.18 has added memcg_account_kmem
so the accounting is not done by obj_cgroup_charge_pages directly for v1
anymore.  Prior kernels need to add it explicitly (thanks to Johannes for
pointing this out).

[akpm@linux-foundation.org: fix build - remove unused local]
Link: http://lkml.kernel.org/r/20230920081101.GA12096@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net
Link: https://lkml.kernel.org/r/ZRE5VJozPZt9bRPy@dhcp22.suse.cz
Fixes: 86327e8eb94c ("memcg: drop kmem.limit_in_bytes")
Fixes: 58056f77502f ("memcg, kmem: further deprecate kmem.limit_in_bytes")
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Shakeel Butt <shakeelb@google.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jeremi Piotrowski <jpiotrowski@linux.microsoft.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Tejun heo <tj@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agomm: zswap: fix potential memory corruption on duplicate store
Domenico Cerasuolo [Fri, 22 Sep 2023 17:22:11 +0000 (19:22 +0200)]
mm: zswap: fix potential memory corruption on duplicate store

While stress-testing zswap a memory corruption was happening when writing
back pages.  __frontswap_store used to check for duplicate entries before
attempting to store a page in zswap, this was because if the store fails
the old entry isn't removed from the tree.  This change removes duplicate
entries in zswap_store before the actual attempt.

[cerasuolodomenico@gmail.com: add a warning and a comment, per Johannes]
Link: https://lkml.kernel.org/r/20230925130002.1929369-1-cerasuolodomenico@gmail.com
Link: https://lkml.kernel.org/r/20230922172211.1704917-1-cerasuolodomenico@gmail.com
Fixes: 42c06a0e8ebe ("mm: kill frontswap")
Signed-off-by: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Nhat Pham <nphamcs@gmail.com>
Cc: Dan Streetman <ddstreet@ieee.org>
Cc: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: Seth Jennings <sjenning@redhat.com>
Cc: Vitaly Wool <vitaly.wool@konsulko.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agoarm64: hugetlb: fix set_huge_pte_at() to work with all swap entries
Ryan Roberts [Fri, 22 Sep 2023 11:58:04 +0000 (12:58 +0100)]
arm64: hugetlb: fix set_huge_pte_at() to work with all swap entries

When called with a swap entry that does not embed a PFN (e.g.
PTE_MARKER_POISONED or PTE_MARKER_UFFD_WP), the previous implementation of
set_huge_pte_at() would either cause a BUG() to fire (if CONFIG_DEBUG_VM
is enabled) or cause a dereference of an invalid address and subsequent
panic.

arm64's huge pte implementation supports multiple huge page sizes, some of
which are implemented in the page table with multiple contiguous entries.
So set_huge_pte_at() needs to work out how big the logical pte is, so that
it can also work out how many physical ptes (or pmds) need to be written.
It previously did this by grabbing the folio out of the pte and querying
its size.

However, there are cases when the pte being set is actually a swap entry.
But this also used to work fine, because for huge ptes, we only ever saw
migration entries and hwpoison entries.  And both of these types of swap
entries have a PFN embedded, so the code would grab that and everything
still worked out.

But over time, more calls to set_huge_pte_at() have been added that set
swap entry types that do not embed a PFN.  And this causes the code to go
bang.  The triggering case is for the uffd poison test, commit
99aa77215ad0 ("selftests/mm: add uffd unit test for UFFDIO_POISON"), which
causes a PTE_MARKER_POISONED swap entry to be set, coutesey of commit
8a13897fb0da ("mm: userfaultfd: support UFFDIO_POISON for hugetlbfs") -
added in v6.5-rc7.  Although review shows that there are other call sites
that set PTE_MARKER_UFFD_WP (which also has no PFN), these don't trigger
on arm64 because arm64 doesn't support UFFD WP.

Arguably, the root cause is really due to commit 18f3962953e4 ("mm:
hugetlb: kill set_huge_swap_pte_at()"), which aimed to simplify the
interface to the core code by removing set_huge_swap_pte_at() (which took
a page size parameter) and replacing it with calls to set_huge_pte_at()
where the size was inferred from the folio, as descibed above.  While that
commit didn't break anything at the time, it did break the interface
because it couldn't handle swap entries without PFNs.  And since then new
callers have come along which rely on this working.  But given the
brokeness is only observable after commit 8a13897fb0da ("mm: userfaultfd:
support UFFDIO_POISON for hugetlbfs"), that one gets the Fixes tag.

Now that we have modified the set_huge_pte_at() interface to pass the huge
page size in the previous patch, we can trivially fix this issue.

Link: https://lkml.kernel.org/r/20230922115804.2043771-3-ryan.roberts@arm.com
Fixes: 8a13897fb0da ("mm: userfaultfd: support UFFDIO_POISON for hugetlbfs")
Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
Reviewed-by: Axel Rasmussen <axelrasmussen@google.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Gordeev <agordeev@linux.ibm.com>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Christian Borntraeger <borntraeger@linux.ibm.com>
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Helge Deller <deller@gmx.de>
Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Qi Zheng <zhengqi.arch@bytedance.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: Sven Schnelle <svens@linux.ibm.com>
Cc: Uladzislau Rezki (Sony) <urezki@gmail.com>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Will Deacon <will@kernel.org>
Cc: <stable@vger.kernel.org> [6.5+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agomm: hugetlb: add huge page size param to set_huge_pte_at()
Ryan Roberts [Fri, 22 Sep 2023 11:58:03 +0000 (12:58 +0100)]
mm: hugetlb: add huge page size param to set_huge_pte_at()

Patch series "Fix set_huge_pte_at() panic on arm64", v2.

This series fixes a bug in arm64's implementation of set_huge_pte_at(),
which can result in an unprivileged user causing a kernel panic.  The
problem was triggered when running the new uffd poison mm selftest for
HUGETLB memory.  This test (and the uffd poison feature) was merged for
v6.5-rc7.

Ideally, I'd like to get this fix in for v6.6 and I've cc'ed stable
(correctly this time) to get it backported to v6.5, where the issue first
showed up.

Description of Bug
==================

arm64's huge pte implementation supports multiple huge page sizes, some of
which are implemented in the page table with multiple contiguous entries.
So set_huge_pte_at() needs to work out how big the logical pte is, so that
it can also work out how many physical ptes (or pmds) need to be written.
It previously did this by grabbing the folio out of the pte and querying
its size.

However, there are cases when the pte being set is actually a swap entry.
But this also used to work fine, because for huge ptes, we only ever saw
migration entries and hwpoison entries.  And both of these types of swap
entries have a PFN embedded, so the code would grab that and everything
still worked out.

But over time, more calls to set_huge_pte_at() have been added that set
swap entry types that do not embed a PFN.  And this causes the code to go
bang.  The triggering case is for the uffd poison test, commit
99aa77215ad0 ("selftests/mm: add uffd unit test for UFFDIO_POISON"), which
causes a PTE_MARKER_POISONED swap entry to be set, coutesey of commit
8a13897fb0da ("mm: userfaultfd: support UFFDIO_POISON for hugetlbfs") -
added in v6.5-rc7.  Although review shows that there are other call sites
that set PTE_MARKER_UFFD_WP (which also has no PFN), these don't trigger
on arm64 because arm64 doesn't support UFFD WP.

If CONFIG_DEBUG_VM is enabled, we do at least get a BUG(), but otherwise,
it will dereference a bad pointer in page_folio():

    static inline struct folio *hugetlb_swap_entry_to_folio(swp_entry_t entry)
    {
        VM_BUG_ON(!is_migration_entry(entry) && !is_hwpoison_entry(entry));

        return page_folio(pfn_to_page(swp_offset_pfn(entry)));
    }

Fix
===

The simplest fix would have been to revert the dodgy cleanup commit
18f3962953e4 ("mm: hugetlb: kill set_huge_swap_pte_at()"), but since
things have moved on, this would have required an audit of all the new
set_huge_pte_at() call sites to see if they should be converted to
set_huge_swap_pte_at().  As per the original intent of the change, it
would also leave us open to future bugs when people invariably get it
wrong and call the wrong helper.

So instead, I've added a huge page size parameter to set_huge_pte_at().
This means that the arm64 code has the size in all cases.  It's a bigger
change, due to needing to touch the arches that implement the function,
but it is entirely mechanical, so in my view, low risk.

I've compile-tested all touched arches; arm64, parisc, powerpc, riscv,
s390, sparc (and additionally x86_64).  I've additionally booted and run
mm selftests against arm64, where I observe the uffd poison test is fixed,
and there are no other regressions.

This patch (of 2):

In order to fix a bug, arm64 needs to be told the size of the huge page
for which the pte is being set in set_huge_pte_at().  Provide for this by
adding an `unsigned long sz` parameter to the function.  This follows the
same pattern as huge_pte_clear().

This commit makes the required interface modifications to the core mm as
well as all arches that implement this function (arm64, parisc, powerpc,
riscv, s390, sparc).  The actual arm64 bug will be fixed in a separate
commit.

No behavioral changes intended.

Link: https://lkml.kernel.org/r/20230922115804.2043771-1-ryan.roberts@arm.com
Link: https://lkml.kernel.org/r/20230922115804.2043771-2-ryan.roberts@arm.com
Fixes: 8a13897fb0da ("mm: userfaultfd: support UFFDIO_POISON for hugetlbfs")
Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu> [powerpc 8xx]
Reviewed-by: Lorenzo Stoakes <lstoakes@gmail.com> [vmalloc change]
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Gordeev <agordeev@linux.ibm.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Axel Rasmussen <axelrasmussen@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Christian Borntraeger <borntraeger@linux.ibm.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Helge Deller <deller@gmx.de>
Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Qi Zheng <zhengqi.arch@bytedance.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: Sven Schnelle <svens@linux.ibm.com>
Cc: Uladzislau Rezki (Sony) <urezki@gmail.com>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Will Deacon <will@kernel.org>
Cc: <stable@vger.kernel.org> [6.5+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agomaple_tree: add MAS_UNDERFLOW and MAS_OVERFLOW states
Liam R. Howlett [Thu, 21 Sep 2023 18:12:36 +0000 (14:12 -0400)]
maple_tree: add MAS_UNDERFLOW and MAS_OVERFLOW states

When updating the maple tree iterator to avoid rewalks, an issue was
introduced when shifting beyond the limits.  This can be seen by trying to
go to the previous address of 0, which would set the maple node to
MAS_NONE and keep the range as the last entry.

Subsequent calls to mas_find() would then search upwards from mas->last
and skip the value at mas->index/mas->last.  This showed up as a bug in
mprotect which skips the actual VMA at the current range after attempting
to go to the previous VMA from 0.

Since MAS_NONE may already be set when searching for a value that isn't
contained within a node, changing the handling of MAS_NONE in mas_find()
would make the code more complicated and error prone.  Furthermore, there
was no way to tell which limit was hit, and thus which action to take
(next or the entry at the current range).

This solution is to add two states to track what happened with the
previous iterator action.  This allows for the expected behaviour of the
next command to return the correct item (either the item at the range
requested, or the next/previous).

Tests are also added and updated accordingly.

Link: https://lkml.kernel.org/r/20230921181236.509072-3-Liam.Howlett@oracle.com
Link: https://gist.github.com/heatd/85d2971fae1501b55b6ea401fbbe485b
Link: https://lore.kernel.org/linux-mm/20230921181236.509072-1-Liam.Howlett@oracle.com/
Fixes: 39193685d585 ("maple_tree: try harder to keep active node with mas_prev()")
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Reported-by: Pedro Falcato <pedro.falcato@gmail.com>
Closes: https://gist.github.com/heatd/85d2971fae1501b55b6ea401fbbe485b
Closes: https://bugs.archlinux.org/task/79656
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agomaple_tree: add mas_is_active() to detect in-tree walks
Liam R. Howlett [Thu, 21 Sep 2023 18:12:35 +0000 (14:12 -0400)]
maple_tree: add mas_is_active() to detect in-tree walks

Patch series "maple_tree: Fix mas_prev() state regression".

Pedro Falcato retported an mprotect regression [1] which was bisected back
to the iterator changes for maple tree.  Root cause analysis showed the
mas_prev() running off the end of the VMA space (previous from 0) followed
by mas_find(), would skip the first value.

This patchset introduces maple state underflow/overflow so the sequence of
calls on the maple state will return what the user expects.

Users who encounter this bug may see mprotect(), userfaultfd_register(),
and mlock() fail on VMAs mapped with address 0.

This patch (of 2):

Instead of constantly checking each possibility of the maple state,
create a fast path that will skip over checking unlikely states.

Link: https://lkml.kernel.org/r/20230921181236.509072-1-Liam.Howlett@oracle.com
Link: https://lkml.kernel.org/r/20230921181236.509072-2-Liam.Howlett@oracle.com
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Pedro Falcato <pedro.falcato@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agonilfs2: fix potential use after free in nilfs_gccache_submit_read_data()
Pan Bian [Thu, 21 Sep 2023 14:17:31 +0000 (23:17 +0900)]
nilfs2: fix potential use after free in nilfs_gccache_submit_read_data()

In nilfs_gccache_submit_read_data(), brelse(bh) is called to drop the
reference count of bh when the call to nilfs_dat_translate() fails.  If
the reference count hits 0 and its owner page gets unlocked, bh may be
freed.  However, bh->b_page is dereferenced to put the page after that,
which may result in a use-after-free bug.  This patch moves the release
operation after unlocking and putting the page.

NOTE: The function in question is only called in GC, and in combination
with current userland tools, address translation using DAT does not occur
in that function, so the code path that causes this issue will not be
executed.  However, it is possible to run that code path by intentionally
modifying the userland GC library or by calling the GC ioctl directly.

[konishi.ryusuke@gmail.com: NOTE added to the commit log]
Link: https://lkml.kernel.org/r/1543201709-53191-1-git-send-email-bianpan2016@163.com
Link: https://lkml.kernel.org/r/20230921141731.10073-1-konishi.ryusuke@gmail.com
Fixes: a3d93f709e89 ("nilfs2: block cache for garbage collection")
Signed-off-by: Pan Bian <bianpan2016@163.com>
Reported-by: Ferry Meng <mengferry@linux.alibaba.com>
Closes: https://lkml.kernel.org/r/20230818092022.111054-1-mengferry@linux.alibaba.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Tested-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agomm: abstract moving to the next PFN
Matthew Wilcox (Oracle) [Wed, 20 Sep 2023 04:09:58 +0000 (05:09 +0100)]
mm: abstract moving to the next PFN

In order to fix the L1TF vulnerability, x86 can invert the PTE bits for
PROT_NONE VMAs, which means we cannot move from one PTE to the next by
adding 1 to the PFN field of the PTE.  This results in the BUG reported at
[1].

Abstract advancing the PTE to the next PFN through a pte_next_pfn()
function/macro.

Link: https://lkml.kernel.org/r/20230920040958.866520-1-willy@infradead.org
Fixes: bcc6cc832573 ("mm: add default definition of set_ptes()")
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reported-by: syzbot+55cc72f8cc3a549119df@syzkaller.appspotmail.com
Closes: https://lkml.kernel.org/r/000000000000d099fa0604f03351@google.com [1]
Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agomm: report success more often from filemap_map_folio_range()
Matthew Wilcox (Oracle) [Wed, 20 Sep 2023 03:53:35 +0000 (04:53 +0100)]
mm: report success more often from filemap_map_folio_range()

Even though we had successfully mapped the relevant page, we would rarely
return success from filemap_map_folio_range().  That leads to falling back
from the VMA lock path to the mmap_lock path, which is a speed &
scalability issue.  Found by inspection.

Link: https://lkml.kernel.org/r/20230920035336.854212-1-willy@infradead.org
Fixes: 617c28ecab22 ("filemap: batch PTE mappings")
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agofs: binfmt_elf_efpic: fix personality for ELF-FDPIC
Greg Ungerer [Thu, 7 Sep 2023 01:18:08 +0000 (11:18 +1000)]
fs: binfmt_elf_efpic: fix personality for ELF-FDPIC

The elf-fdpic loader hard sets the process personality to either
PER_LINUX_FDPIC for true elf-fdpic binaries or to PER_LINUX for normal ELF
binaries (in this case they would be constant displacement compiled with
-pie for example).  The problem with that is that it will lose any other
bits that may be in the ELF header personality (such as the "bug
emulation" bits).

On the ARM architecture the ADDR_LIMIT_32BIT flag is used to signify a
normal 32bit binary - as opposed to a legacy 26bit address binary.  This
matters since start_thread() will set the ARM CPSR register as required
based on this flag.  If the elf-fdpic loader loses this bit the process
will be mis-configured and crash out pretty quickly.

Modify elf-fdpic loader personality setting so that it preserves the upper
three bytes by using the SET_PERSONALITY macro to set it.  This macro in
the generic case sets PER_LINUX and preserves the upper bytes.
Architectures can override this for their specific use case, and ARM does
exactly this.

The problem shows up quite easily running under qemu using the ARM
architecture, but not necessarily on all types of real ARM hardware.  If
the underlying ARM processor does not support the legacy 26-bit addressing
mode then everything will work as expected.

Link: https://lkml.kernel.org/r/20230907011808.2985083-1-gerg@kernel.org
Fixes: 1bde925d23547 ("fs/binfmt_elf_fdpic.c: provide NOMMU loader for regular ELF binaries")
Signed-off-by: Greg Ungerer <gerg@kernel.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: Greg Ungerer <gerg@kernel.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
18 months agoMerge tag '6.6-rc3-ksmbd-server-fixes' of git://git.samba.org/ksmbd
Linus Torvalds [Fri, 29 Sep 2023 23:51:38 +0000 (16:51 -0700)]
Merge tag '6.6-rc3-ksmbd-server-fixes' of git://git.samba.org/ksmbd

Pull smb server fixes from Steve French:
 "Two SMB3 server fixes for null pointer dereferences:

   - invalid SMB3 request case (fixes issue found in testing the read
     compound patch)

   - iovec error case in response processing"

* tag '6.6-rc3-ksmbd-server-fixes' of git://git.samba.org/ksmbd:
  ksmbd: check iov vector index in ksmbd_conn_write()
  ksmbd: return invalid parameter error response if smb2 request is invalid

18 months agoMerge tag 'ceph-for-6.6-rc4' of https://github.com/ceph/ceph-client
Linus Torvalds [Fri, 29 Sep 2023 23:46:24 +0000 (16:46 -0700)]
Merge tag 'ceph-for-6.6-rc4' of https://github.com/ceph/ceph-client

Pull ceph fixes from Ilya Dryomov:
 "A series that fixes an involved 'double watch error' deadlock in RBD
  marked for stable and two cleanups"

* tag 'ceph-for-6.6-rc4' of https://github.com/ceph/ceph-client:
  rbd: take header_rwsem in rbd_dev_refresh() only when updating
  rbd: decouple parent info read-in from updating rbd_dev
  rbd: decouple header read-in from updating rbd_dev->header
  rbd: move rbd_dev_refresh() definition
  Revert "ceph: make members in struct ceph_mds_request_args_ext a union"
  ceph: remove unnecessary check for NULL in parse_longname()

18 months agoMerge tag 'xfs-6.6-fixes-2' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Fri, 29 Sep 2023 23:41:25 +0000 (16:41 -0700)]
Merge tag 'xfs-6.6-fixes-2' of git://git./fs/xfs/xfs-linux

Pull xfs fix from Chandan Babu:

 - fix for commit 68b957f64fca ("xfs: load uncached unlinked inodes into
   memory on demand") which address review comments provided by Dave
   Chinner

* tag 'xfs-6.6-fixes-2' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  xfs: fix reloading entire unlinked bucket lists

18 months agocxl/pci: Cleanup 'sanitize' to always poll
Dan Williams [Fri, 29 Sep 2023 19:44:20 +0000 (12:44 -0700)]
cxl/pci: Cleanup 'sanitize' to always poll

In preparation for fixing the init/teardown of the 'sanitize' workqueue
and sysfs notification mechanism, arrange for cxl_mbox_sanitize_work()
to be the single location where the sysfs attribute is notified. With
that change there is no distinction between polled mode and interrupt
mode. All the interrupt does is accelerate the polling interval.

The change to check for "mds->security.sanitize_node" under the lock is
there to ensure that the interrupt, the work routine and the
setup/teardown code can all have a consistent view of the registered
notifier and the workqueue state. I.e. the expectation is that the
interrupt is live past the point that the sanitize sysfs attribute is
published, and it may race teardown, so it must be consulted under a
lock. Given that new locking requirement, cxl_pci_mbox_irq() is moved
from hard to thread irq context.

Lastly, some opportunistic replacements of
"queue_delayed_work(system_wq, ...)", which is just open coded
schedule_delayed_work(), are included.

Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
18 months agoMAINTAINERS: Fix Florian Fainelli's email address
Uwe Kleine-König [Thu, 28 Sep 2023 07:06:52 +0000 (09:06 +0200)]
MAINTAINERS: Fix Florian Fainelli's email address

Commit 31345a0f5901 ("MAINTAINERS: Replace my email address") added 13
instances of ...@broadcom.com and one of only ...@broadcom. I didn't
double check if Broadcom really owns that TLD, but git send-email
doesn't accept it, so add ".com" to that one bogous(?) instance.

Fixes: 31345a0f5901 ("MAINTAINERS: Replace my email address")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
18 months agocxl/pci: Remove unnecessary device reference management in sanitize work
Dan Williams [Fri, 29 Sep 2023 21:44:46 +0000 (14:44 -0700)]
cxl/pci: Remove unnecessary device reference management in sanitize work

Given that any particular put_device() could be the final put of the
device, the fact that there are usages of cxlds->dev after
put_device(cxlds->dev) is a red flag. Drop the reference counting since
the device is pinned by being registered and will not be unregistered
without triggering the driver + workqueue to shutdown.

Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
18 months agoMerge tag 'ata-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal...
Linus Torvalds [Fri, 29 Sep 2023 20:38:34 +0000 (13:38 -0700)]
Merge tag 'ata-6.6-rc4' of git://git./linux/kernel/git/dlemoal/libata

Pull ATA fixes from Damien Le Moal:
 "A larger than usual set of fixes for 6.6-rc4 due to the unexpected
  number of fixes needed to address ATA disks suspend/resume issues.

  In more detail:

   - Add missing additionalProperties on child nodes to the pata-common
     DT bindings (Rob)

   - Fix handling of the REPORT SUPPORTED OPERATION CODES command to
     ignore reserved bits (Niklas)

   - Increase port multiplier soft reset timeout to accomodate slow
     devices and avoid issues on wakeup (Matthias)

   - A couple of minor code fixes to avoid compilation warnings in
     libata-core and libata-eh (me)

   - Many patches from me to address suspend/resume issues, and in
     particular a potential deadlock on resume due to the SCSI disk
     driver resume operation not being synchronized with libata EH port
     resume handling.

     This is addressed by changing the scsi disk driver disk start/stop
     control to allow libata to execute disk suspend (spin down) and
     resume (spin up) on its own during system suspend/resume. Runtime
     suspend/resume control remains with the SCSI disk driver.

     Other fixes include:
      - Fix libata power management request issuing to avoid races
      - Establish a link between ATA ports and SCSI devices to order PM
        operations
      - Fix device removal to avoid issues with driver rmmod removal
      - Fix synchronization of libata device rescan and SCSI disk resume
        operation
      - Remove libsas PM operations as suspend/resume is handled
        directly by the sas controller resume
      - Fix the SCSI disk driver to not issue commands to suspended
        disks, thus avoiding potential system lock-up on resume"

* tag 'ata-6.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal/libata:
  ata: libata-eh: Fix compilation warning in ata_eh_link_report()
  ata: libata-core: Fix compilation warning in ata_dev_config_ncq()
  scsi: sd: Do not issue commands to suspended disks on shutdown
  ata: libata-core: Do not register PM operations for SAS ports
  ata: libata-scsi: Fix delayed scsi_rescan_device() execution
  scsi: Do not attempt to rescan suspended devices
  ata: libata-scsi: Disable scsi device manage_system_start_stop
  scsi: sd: Differentiate system and runtime start/stop management
  ata: libata-scsi: link ata port and scsi device
  ata: libata-core: Fix port and device removal
  ata: libata-core: Fix ata_port_request_pm() locking
  ata: libata-sata: increase PMP SRST timeout to 10s
  ata: libata-scsi: ignore reserved bits for REPORT SUPPORTED OPERATION CODES
  dt-bindings: ata: pata-common: Add missing additionalProperties on child nodes

18 months agoMerge tag 'block-6.6-2023-09-28' of git://git.kernel.dk/linux
Linus Torvalds [Fri, 29 Sep 2023 20:28:49 +0000 (13:28 -0700)]
Merge tag 'block-6.6-2023-09-28' of git://git.kernel.dk/linux

Pull block fixes from Jens Axboe:
 "Just two minor comment / documentation fixes for the block side"

* tag 'block-6.6-2023-09-28' of git://git.kernel.dk/linux:
  block: fix kernel-doc for disk_force_media_change()
  block: correct stale comment in rq_qos_wait

18 months agoMerge tag 'io_uring-6.6-2023-09-28' of git://git.kernel.dk/linux
Linus Torvalds [Fri, 29 Sep 2023 19:56:34 +0000 (12:56 -0700)]
Merge tag 'io_uring-6.6-2023-09-28' of git://git.kernel.dk/linux

Pull io_uring fix from Jens Axboe:
 "A single fix going to stable for the IORING_OP_LINKAT flag handling"

* tag 'io_uring-6.6-2023-09-28' of git://git.kernel.dk/linux:
  io_uring/fs: remove sqe->rw_flags checking from LINKAT