platform/kernel/linux-rpi.git
21 months agoMerge branch 'for-6.3/cxl-ram-region' into cxl/next
Dan Williams [Tue, 14 Feb 2023 22:58:26 +0000 (14:58 -0800)]
Merge branch 'for-6.3/cxl-ram-region' into cxl/next

Pick up some fixes from exposure of for-6.3/cxl-ram-region in
linux-next.

21 months agodax/hmem: build hmem device support as module if possible
Arnd Bergmann [Tue, 14 Feb 2023 13:18:47 +0000 (14:18 +0100)]
dax/hmem: build hmem device support as module if possible

When device_hmem.o is enabled but dax itself is a loadable module, the
dax_hmem support fails to link because Kbuild never compiles built-in
code under drivers/dax:

ERROR: modpost: "walk_hmem_resources" [drivers/dax/hmem/dax_hmem.ko] undefined!

Make sure that drivers/dax is entered for compiling built-in code
even with CONFIG_DAX=m.

Fixes: 7dab174e2e27 ("dax/hmem: Move hmem device registration to dax_hmem.ko")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Link: https://lore.kernel.org/r/20230214131913.1431969-1-arnd@kernel.org
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agodax: cxl: add CXL_REGION dependency
Arnd Bergmann [Tue, 14 Feb 2023 10:30:49 +0000 (11:30 +0100)]
dax: cxl: add CXL_REGION dependency

There is already a dependency on CXL_REGION, which depends on CXL_BUS,
but since CXL_REGION is a 'bool' symbol, it's possible to configure
DAX as built-in even though CXL itself is a loadable module:

x86_64-linux-ld: drivers/dax/cxl.o: in function `cxl_dax_region_probe':
cxl.c:(.text+0xb): undefined reference to `to_cxl_dax_region'
x86_64-linux-ld: drivers/dax/cxl.o: in function `cxl_dax_region_driver_init':
cxl.c:(.init.text+0x10): undefined reference to `__cxl_driver_register'
x86_64-linux-ld: drivers/dax/cxl.o: in function `cxl_dax_region_driver_exit':
cxl.c:(.exit.text+0x9): undefined reference to `cxl_driver_unregister'

Prevent this with another depndency on the tristate symbol.

Fixes: 09d09e04d2fc ("cxl/dax: Create dax devices for CXL RAM regions")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Link: https://lore.kernel.org/r/20230214103054.1082908-1-arnd@kernel.org
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl: avoid returning uninitialized error code
Arnd Bergmann [Mon, 13 Feb 2023 10:12:11 +0000 (11:12 +0100)]
cxl: avoid returning uninitialized error code

The new cxl_add_to_region() function returns an uninitialized
value on success:

drivers/cxl/core/region.c:2628:6: error: variable 'rc' is used uninitialized whenever 'if' condition is false [-Werror,-Wsometimes-uninitialized]
        if (IS_ERR(cxlr)) {
            ^~~~~~~~~~~~
drivers/cxl/core/region.c:2654:9: note: uninitialized use occurs here
        return rc;

Simplify the logic to have the rc variable always initialized in the
same place.

Fixes: a32320b71f08 ("cxl/region: Add region autodiscovery")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Link: https://lore.kernel.org/r/20230213101220.3821689-1-arnd@kernel.org
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/pmem: Fix nvdimm registration races
Dan Williams [Tue, 14 Feb 2023 01:01:05 +0000 (17:01 -0800)]
cxl/pmem: Fix nvdimm registration races

A loop of the form:

    while true; do modprobe cxl_pci; modprobe -r cxl_pci; done

...fails with the following crash signature:

    BUG: kernel NULL pointer dereference, address: 0000000000000040
    [..]
    RIP: 0010:cxl_internal_send_cmd+0x5/0xb0 [cxl_core]
    [..]
    Call Trace:
     <TASK>
     cxl_pmem_ctl+0x121/0x240 [cxl_pmem]
     nvdimm_get_config_data+0xd6/0x1a0 [libnvdimm]
     nd_label_data_init+0x135/0x7e0 [libnvdimm]
     nvdimm_probe+0xd6/0x1c0 [libnvdimm]
     nvdimm_bus_probe+0x7a/0x1e0 [libnvdimm]
     really_probe+0xde/0x380
     __driver_probe_device+0x78/0x170
     driver_probe_device+0x1f/0x90
     __device_attach_driver+0x85/0x110
     bus_for_each_drv+0x7d/0xc0
     __device_attach+0xb4/0x1e0
     bus_probe_device+0x9f/0xc0
     device_add+0x445/0x9c0
     nd_async_device_register+0xe/0x40 [libnvdimm]
     async_run_entry_fn+0x30/0x130

...namely that the bottom half of async nvdimm device registration runs
after the CXL has already torn down the context that cxl_pmem_ctl()
needs. Unlike the ACPI NFIT case that benefits from launching multiple
nvdimm device registrations in parallel from those listed in the table,
CXL is already marked PROBE_PREFER_ASYNCHRONOUS. So provide for a
synchronous registration path to preclude this scenario.

Fixes: 21083f51521f ("cxl/pmem: Register 'pmem' / cxl_nvdimm devices")
Cc: <stable@vger.kernel.org>
Reported-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agoMerge branch 'for-6.3/cxl-ram-region' into cxl/next
Dan Williams [Sat, 11 Feb 2023 02:11:01 +0000 (18:11 -0800)]
Merge branch 'for-6.3/cxl-ram-region' into cxl/next

Include the support for enumerating and provisioning ram regions for
v6.3. This also include a default policy change for ram / volatile
device-dax instances to assign them to the dax_kmem driver by default.

21 months agoMerge branch 'for-6.3/cxl' into cxl/next
Dan Williams [Sat, 11 Feb 2023 02:05:59 +0000 (18:05 -0800)]
Merge branch 'for-6.3/cxl' into cxl/next

Pick up some final miscellaneous updates for v6.3 including support for
communicating 'exclusive' and 'enabled' state of commands.

21 months agocxl/mem: Fix UAPI command comment
Ira Weiny [Fri, 3 Feb 2023 01:04:27 +0000 (17:04 -0800)]
cxl/mem: Fix UAPI command comment

The command comment had grammatical errors.  In an attempt to fix those
it was noted that the comment and the query command were not in sync.

Now that the query command returns excluded and device unsupported
command information.  Update the kdoc and fix the grammatical errors.

[1] https://lore.kernel.org/all/63b4ec4e37cc1_5178e2941d@dwillia2-xfh.jf.intel.com.notmuch/

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221222-cxl-misc-v4-4-62f701c1cdd1@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/uapi: Tag commands from cxl_query_cmd()
Ira Weiny [Fri, 3 Feb 2023 01:04:26 +0000 (17:04 -0800)]
cxl/uapi: Tag commands from cxl_query_cmd()

It was pointed out that commands not supported by the device or excluded
by the kernel were being returned in cxl_query_cmd().[1]

While libcxl correctly handles failing commands, it is more efficient to
not issue an invalid command in the first place.  This can't be done
without additional information being returned from cxl_query_cmd().  In
addition, information about the availability of commands can be useful
for debugging.

Add flags to struct cxl_command_info which reflect if a command is
enabled and/or exclusive to the kernel.

[1] https://lore.kernel.org/all/63b4ec4e37cc1_5178e2941d@dwillia2-xfh.jf.intel.com.notmuch/

Suggested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221222-cxl-misc-v4-3-62f701c1cdd1@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/uapi: Add warning on CXL command enum
Ira Weiny [Fri, 3 Feb 2023 01:04:25 +0000 (17:04 -0800)]
cxl/uapi: Add warning on CXL command enum

The CXL command enum is exported to user space and must maintain
backwards compatibility.

Add comment that new defines must be added to the end of the list.

Suggested-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221222-cxl-misc-v4-2-62f701c1cdd1@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/mem: Remove unused CXL_CMD_FLAG_NONE define
Ira Weiny [Fri, 3 Feb 2023 01:04:24 +0000 (17:04 -0800)]
cxl/mem: Remove unused CXL_CMD_FLAG_NONE define

CXL_CMD_FLAG_NONE is not used, remove it.

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221222-cxl-misc-v4-1-62f701c1cdd1@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/dax: Create dax devices for CXL RAM regions
Dan Williams [Fri, 10 Feb 2023 09:07:19 +0000 (01:07 -0800)]
cxl/dax: Create dax devices for CXL RAM regions

While platform firmware takes some responsibility for mapping the RAM
capacity of CXL devices present at boot, the OS is responsible for
mapping the remainder and hot-added devices. Platform firmware is also
responsible for identifying the platform general purpose memory pool,
typically DDR attached DRAM, and arranging for the remainder to be 'Soft
Reserved'. That reservation allows the CXL subsystem to route the memory
to core-mm via memory-hotplug (dax_kmem), or leave it for dedicated
access (device-dax).

The new 'struct cxl_dax_region' object allows for a CXL memory resource
(region) to be published, but also allow for udev and module policy to
act on that event. It also prevents cxl_core.ko from having a module
loading dependency on any drivers/dax/ modules.

Tested-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/167602003896.1924368.10335442077318970468.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agodax: Assign RAM regions to memory-hotplug by default
Dan Williams [Fri, 10 Feb 2023 09:07:13 +0000 (01:07 -0800)]
dax: Assign RAM regions to memory-hotplug by default

The default mode for device-dax instances is backwards for RAM-regions
as evidenced by the fact that it tends to catch end users by surprise.
"Where is my memory?". Recall that platforms are increasingly shipping
with performance-differentiated memory pools beyond typical DRAM and
NUMA effects. This includes HBM (high-bandwidth-memory) and CXL (dynamic
interleave, varied media types, and future fabric attached
possibilities).

For this reason the EFI_MEMORY_SP (EFI Special Purpose Memory => Linux
'Soft Reserved') attribute is expected to be applied to all memory-pools
that are not the general purpose pool. This designation gives an
Operating System a chance to defer usage of a memory pool until later in
the boot process where its performance properties can be interrogated
and administrator policy can be applied.

'Soft Reserved' memory can be anything from too limited and precious to
be part of the general purpose pool (HBM), too slow to host hot kernel
data structures (some PMEM media), or anything in between. However, in
the absence of an explicit policy, the memory should at least be made
usable by default. The current device-dax default hides all
non-general-purpose memory behind a device interface.

The expectation is that the distribution of users that want the memory
online by default vs device-dedicated-access by default follows the
Pareto principle. A small number of enlightened users may want to do
userspace memory management through a device, but general users just
want the kernel to make the memory available with an option to get more
advanced later.

Arrange for all device-dax instances not backed by PMEM to default to
attaching to the dax_kmem driver. From there the baseline memory hotplug
policy (CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE / memhp_default_state=)
gates whether the memory comes online or stays offline. Where, if it
stays offline, it can be reliably converted back to device-mode where it
can be partitioned, or fronted by a userspace allocator.

So, if someone wants device-dax instances for their 'Soft Reserved'
memory:

1/ Build a kernel with CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE=n or boot
   with memhp_default_state=offline, or roll the dice and hope that the
   kernel has not pinned a page in that memory before step 2.

2/ Write a udev rule to convert the target dax device(s) from
   'system-ram' mode to 'devdax' mode:

   daxctl reconfigure-device $dax -m devdax -f

Cc: Michal Hocko <mhocko@suse.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Gregory Price <gregory.price@memverge.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/167602003336.1924368.6809503401422267885.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agodax/hmem: Move hmem device registration to dax_hmem.ko
Dan Williams [Fri, 10 Feb 2023 09:07:07 +0000 (01:07 -0800)]
dax/hmem: Move hmem device registration to dax_hmem.ko

In preparation for the CXL region driver to take over the responsibility
of registering device-dax instances for CXL regions, move the
registration of "hmem" devices to dax_hmem.ko.

Previously the builtin component of this enabling
(drivers/dax/hmem/device.o) would register platform devices for each
address range and trigger the dax_hmem.ko module to load and attach
device-dax instances to those devices. Now, the ranges are collected
from the HMAT and EFI memory map walking, but the device creation is
deferred. A new "hmem_platform" device is created which triggers
dax_hmem.ko to load and register the platform devices.

Tested-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/167602002771.1924368.5653558226424530127.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agodax/hmem: Convey the dax range via memregion_info()
Dan Williams [Fri, 10 Feb 2023 09:07:02 +0000 (01:07 -0800)]
dax/hmem: Convey the dax range via memregion_info()

In preparation for hmem platform devices to be unregistered, stop using
platform_device_add_resources() to convey the address range. The
platform_device_add_resources() API causes an existing "Soft Reserved"
iomem resource to be re-parented under an inserted platform device
resource. When that platform device is deleted it removes the platform
device resource and all children.

Instead, it is sufficient to convey just the address range and let
request_mem_region() insert resources to indicate the devices active in
the range. This allows the "Soft Reserved" resource to be re-enumerated
upon the next probe event.

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/167602002217.1924368.7036275892522551624.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agodax/hmem: Drop unnecessary dax_hmem_remove()
Dan Williams [Fri, 10 Feb 2023 09:06:56 +0000 (01:06 -0800)]
dax/hmem: Drop unnecessary dax_hmem_remove()

Empty driver remove callbacks can just be elided.

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Gregory Price <gregory.price@memverge.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/167602001664.1924368.9102029637928071240.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agodax/hmem: Move HMAT and Soft reservation probe initcall level
Dan Williams [Fri, 10 Feb 2023 09:06:51 +0000 (01:06 -0800)]
dax/hmem: Move HMAT and Soft reservation probe initcall level

In preparation for moving more filtering of "hmem" ranges into the
dax_hmem.ko module, update the initcall levels. HMAT range registration
moves to subsys_initcall() to be done before Soft Reservation probing,
and Soft Reservation probing is moved to device_initcall() to be done
before dax_hmem.ko initialization if it is built-in.

Tested-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/167602001107.1924368.11562316181038595611.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agotools/testing/cxl: Define a fixed volatile configuration to parse
Dan Williams [Fri, 10 Feb 2023 09:06:45 +0000 (01:06 -0800)]
tools/testing/cxl: Define a fixed volatile configuration to parse

Take two endpoints attached to the first switch on the first host-bridge
in the cxl_test topology and define a pre-initialized region. This is a
x2 interleave underneath a x1 CXL Window.

$ modprobe cxl_test
$ # cxl list -Ru
{
  "region":"region3",
  "resource":"0xf010000000",
  "size":"512.00 MiB (536.87 MB)",
  "interleave_ways":2,
  "interleave_granularity":4096,
  "decode_state":"commit"
}

Tested-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/167602000547.1924368.11613151863880268868.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Add region autodiscovery
Dan Williams [Sat, 11 Feb 2023 01:31:17 +0000 (17:31 -0800)]
cxl/region: Add region autodiscovery

Region autodiscovery is an asynchronous state machine advanced by
cxl_port_probe(). After the decoders on an endpoint port are enumerated
they are scanned for actively enabled instances. Each active decoder is
flagged for auto-assembly CXL_DECODER_F_AUTO and attached to a region.
If a region does not already exist for the address range setting of the
decoder one is created. That creation process may race with other
decoders of the same region being discovered since cxl_port_probe() is
asynchronous. A new 'struct cxl_root_decoder' lock, @range_lock, is
introduced to mitigate that race.

Once all decoders have arrived, "p->nr_targets == p->interleave_ways",
they are sorted by their relative decode position. The sort algorithm
involves finding the point in the cxl_port topology where one leg of the
decode leads to deviceA and the other deviceB. At that point in the
topology the target order in the 'struct cxl_switch_decoder' indicates
the relative position of those endpoint decoders in the region.

>From that point the region goes through the same setup and validation
steps as user-created regions, but instead of programming the decoders
it validates that driver would have written the same values to the
decoders as were already present.

Tested-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/167601999958.1924368.9366954455835735048.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/port: Split endpoint and switch port probe
Dan Williams [Fri, 10 Feb 2023 09:06:33 +0000 (01:06 -0800)]
cxl/port: Split endpoint and switch port probe

Jonathan points out that the shared code between the switch and endpoint
case is small. Before adding another is_cxl_endpoint() conditional,
just split the two cases.

Rather than duplicate the "Couldn't enumerate decoders" error message
take the opportunity to improve the error messages in
devm_cxl_enumerate_decoders().

Reported-by: Jonathan Cameron <Jonathan.Cameron@Huawei.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/167601999378.1924368.15071142145866277623.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Enable CONFIG_CXL_REGION to be toggled
Dan Williams [Fri, 10 Feb 2023 09:06:27 +0000 (01:06 -0800)]
cxl/region: Enable CONFIG_CXL_REGION to be toggled

Add help text and a label so the CXL_REGION config option can be
toggled. This is mainly to enable compile testing without region
support.

Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Gregory Price <gregory.price@memverge.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Link: https://lore.kernel.org/r/167601998765.1924368.258370414771847699.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agokernel/range: Uplevel the cxl subsystem's range_contains() helper
Dan Williams [Fri, 10 Feb 2023 09:06:21 +0000 (01:06 -0800)]
kernel/range: Uplevel the cxl subsystem's range_contains() helper

In support of the CXL subsystem's use of 'struct range' to track decode
address ranges, add a common range_contains() implementation with
identical semantics as resource_contains();

The existing 'range_contains()' in lib/stackinit_kunit.c is namespaced
with a 'stackinit_' prefix.

Cc: Kees Cook <keescook@chromium.org>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Gregory Price <gregory.price@memverge.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Link: https://lore.kernel.org/r/167601998163.1924368.6067392174077323935.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Move region-position validation to a helper
Dan Williams [Fri, 10 Feb 2023 09:06:15 +0000 (01:06 -0800)]
cxl/region: Move region-position validation to a helper

In preparation for region autodiscovery, that needs all devices
discovered before their relative position in the region can be
determined, consolidate all position dependent validation in a helper.

Recall that in the on-demand region creation flow the end-user picks the
position of a given endpoint decoder in a region. In the autodiscovery
case the position of an endpoint decoder can only be determined after
all other endpoint decoders that claim to decode the region's address
range have been enumerated and attached. So, in the autodiscovery case
endpoint decoders may be attached before their relative position is
known. Once all decoders arrive, then positions can be determined and
validated with cxl_region_validate_position() the same as user initiated
on-demand creation.

Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/167601997584.1924368.4615769326126138969.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Cleanup target list on attach error
Dan Williams [Fri, 10 Feb 2023 09:06:09 +0000 (01:06 -0800)]
cxl/region: Cleanup target list on attach error

Jonathan noticed that the target list setup is not unwound completely
upon error. Undo all the setup in the 'err_decrement:' exit path.

Fixes: 27b3f8d13830 ("cxl/region: Program target lists")
Reported-by: Jonathan Cameron <Jonathan.Cameron@Huawei.com>
Link: http://lore.kernel.org/r/20230208123031.00006990@Huawei.com
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/167601996980.1924368.390423634911157277.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Refactor attach_target() for autodiscovery
Dan Williams [Fri, 10 Feb 2023 09:06:04 +0000 (01:06 -0800)]
cxl/region: Refactor attach_target() for autodiscovery

Region autodiscovery is the process of kernel creating 'struct
cxl_region' object to represent active CXL memory ranges it finds
already active in hardware when the driver loads. Typically this happens
when platform firmware establishes CXL memory regions and then publishes
them in the memory map. However, this can also happen in the case of
kexec-reboot after the kernel has created regions.

In the autodiscovery case the region creation process starts with a
known endpoint decoder. Refactor attach_target() into a helper that is
suitable to be called from either sysfs, for runtime region creation, or
from cxl_port_probe() after it has enumerated all endpoint decoders.

The cxl_port_probe() context is an async device-core probing context, so
it is not appropriate to allow SIGTERM to interrupt the assembly
process. Refactor attach_target() to take @cxled and @state as arguments
where @state indicates whether waiting from the region rwsem is
interruptible or not.

No behavior change is intended.

Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Link: https://lore.kernel.org/r/167601996393.1924368.2202255054618600069.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Add volatile region creation support
Dan Williams [Fri, 10 Feb 2023 09:05:57 +0000 (01:05 -0800)]
cxl/region: Add volatile region creation support

Expand the region creation infrastructure to enable 'ram'
(volatile-memory) regions. The internals of create_pmem_region_store()
and create_pmem_region_show() are factored out into helpers
__create_region() and __create_region_show() for the 'ram' case to
reuse.

Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Gregory Price <gregory.price@memverge.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Link: https://lore.kernel.org/r/167601995775.1924368.352616146815830591.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Validate region mode vs decoder mode
Dan Williams [Fri, 10 Feb 2023 09:05:51 +0000 (01:05 -0800)]
cxl/region: Validate region mode vs decoder mode

In preparation for a new region mode, do not, for example, allow
'ram' decoders to be assigned to 'pmem' regions and vice versa.

Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Gregory Price <gregory.price@memverge.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Link: https://lore.kernel.org/r/167601995111.1924368.7459128614177994602.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Support empty uuids for non-pmem regions
Dan Williams [Fri, 10 Feb 2023 09:05:45 +0000 (01:05 -0800)]
cxl/region: Support empty uuids for non-pmem regions

Shipping versions of the cxl-cli utility expect all regions to have a
'uuid' attribute. In preparation for 'ram' regions, update the 'uuid'
attribute to return an empty string which satisfies the current
expectations of 'cxl list -R'. Otherwise, 'cxl list -R' fails in the
presence of regions with the 'uuid' attribute missing. Force the
attribute to be read-only as there is no facility or expectation for a
'ram' region to recall its uuid from one boot to the next.

Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/167601994558.1924368.12612811533724694444.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Add a mode attribute for regions
Dan Williams [Fri, 10 Feb 2023 09:05:39 +0000 (01:05 -0800)]
cxl/region: Add a mode attribute for regions

In preparation for a new region type, "ram" regions, add a mode
attribute to clarify the mode of the decoders that can be added to a
region. Share the internals of mode_show() (for decoders) with the
region case.

Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Gregory Price <gregory.price@memverge.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Link: https://lore.kernel.org/r/167601993930.1924368.4305018565539515665.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/Documentation: Update references to attributes added in v6.0
Dan Williams [Fri, 10 Feb 2023 09:05:33 +0000 (01:05 -0800)]
cxl/Documentation: Update references to attributes added in v6.0

Prior to Linus deciding that the kernel that following v5.19 would be
v6.0, the CXL ABI documentation already referenced v5.20. In preparation
for updating these entries update the kernel version to v6.0.

Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Gregory Price <gregory.price@memverge.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Tested-by: Fan Ni <fan.ni@samsung.com>
Link: https://lore.kernel.org/r/167601993360.1924368.14122892663883462813.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/memdev: Fix endpoint port removal
Dan Williams [Sat, 11 Feb 2023 01:29:09 +0000 (17:29 -0800)]
cxl/memdev: Fix endpoint port removal

Testing of ram region support [1], stimulates a long standing bug in
cxl_detach_ep() where some cxl_ep_remove() cleanup is skipped due to
inability to walk ports after dports have been unregistered. That
results in a failure to re-register a memdev after the port is
re-enabled leading to a crash like the following:

    cxl_port_setup_targets: cxl region4: cxl_host_bridge.0:port4 iw: 1 ig: 256
    general protection fault, ...
    [..]
    RIP: 0010:cxl_region_setup_targets+0x897/0x9e0 [cxl_core]
    dev_name at include/linux/device.h:700
    (inlined by) cxl_port_setup_targets at drivers/cxl/core/region.c:1155
    (inlined by) cxl_region_setup_targets at drivers/cxl/core/region.c:1249
    [..]
    Call Trace:
     <TASK>
     attach_target+0x39a/0x760 [cxl_core]
     ? __mutex_unlock_slowpath+0x3a/0x290
     cxl_add_to_region+0xb8/0x340 [cxl_core]
     ? lockdep_hardirqs_on+0x7d/0x100
     discover_region+0x4b/0x80 [cxl_port]
     ? __pfx_discover_region+0x10/0x10 [cxl_port]
     device_for_each_child+0x58/0x90
     cxl_port_probe+0x10e/0x130 [cxl_port]
     cxl_bus_probe+0x17/0x50 [cxl_core]

Change the port ancestry walk to be by depth rather than by dport. This
ensures that even if a port has unregistered its dports a deferred
memdev cleanup will still be able to cleanup the memdev's interest in
that port.

The parent_port->dev.driver check is only needed for determining if the
bottom up removal beat the top-down removal, but cxl_ep_remove() can
always proceed given the port is pinned. That is, the two sources of
cxl_ep_remove() are in cxl_detach_ep() and cxl_port_release(), and
cxl_port_release() can not run if cxl_detach_ep() holds a reference.

Fixes: 2703c16c75ae ("cxl/core/port: Add switch port enumeration")
Link: http://lore.kernel.org/r/167564534874.847146.5222419648551436750.stgit@dwillia2-xfh.jf.intel.com
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Link: https://lore.kernel.org/r/167601992789.1924368.8083994227892600608.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/mem: Correct full ID range allocation
Davidlohr Bueso [Wed, 8 Feb 2023 18:19:44 +0000 (10:19 -0800)]
cxl/mem: Correct full ID range allocation

For ID allocations we want 0-(max-1), ie: smatch complains:

 error: Calling ida_alloc_range() with a 'max' argument which is a power of 2. -1 missing?

Correct this and also replace the call to use the max() flavor instead.

Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20230208181944.240261-1-dave@stgolabs.net
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agoMerge branch 'for-6.3/cxl-events' into cxl/next
Dan Williams [Tue, 7 Feb 2023 19:14:06 +0000 (11:14 -0800)]
Merge branch 'for-6.3/cxl-events' into cxl/next

Add the CXL event and interrupt support for the v6.3 update.

21 months agoMerge branch 'for-6.3/cxl' into cxl/next
Dan Williams [Tue, 7 Feb 2023 19:12:24 +0000 (11:12 -0800)]
Merge branch 'for-6.3/cxl' into cxl/next

Merge the general CXL updates with fixes targeting v6.2-rc for v6.3.
Resolve a conflict with the fix and move of cxl_report_and_clear() from
pci.c to core/pci.c.

21 months agocxl/region: Fix passthrough-decoder detection
Dan Williams [Tue, 7 Feb 2023 19:04:30 +0000 (11:04 -0800)]
cxl/region: Fix passthrough-decoder detection

A passthrough decoder is a decoder that maps only 1 target. It is a
special case because it does not impose any constraints on the
interleave-math as compared to a decoder with multiple targets. Extend
the passthrough case to multi-target-capable decoders that only have one
target selected. I.e. the current code was only considering passthrough
*ports* which are only a subset of the potential passthrough decoder
scenarios.

Fixes: e4f6dfa9ef75 ("cxl/region: Fix 'distance' calculation with passthrough ports")
Cc: <stable@vger.kernel.org>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/167564540422.847146.13816934143225777888.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Fix null pointer dereference for resetting decoder
Fan Ni [Thu, 15 Dec 2022 17:09:14 +0000 (17:09 +0000)]
cxl/region: Fix null pointer dereference for resetting decoder

Not all decoders have a reset callback.

The CXL specification allows a host bridge with a single root port to
have no explicit HDM decoders. Currently the region driver assumes there
are none.  As such the CXL core creates a special pass through decoder
instance without a commit/reset callback.

Prior to this patch, the ->reset() callback was called unconditionally when
calling cxl_region_decode_reset. Thus a configuration with 1 Host Bridge,
1 Root Port, and one directly attached CXL type 3 device or multiple CXL
type 3 devices attached to downstream ports of a switch can cause a null
pointer dereference.

Before the fix, a kernel crash was observed when we destroy the region, and
a pass through decoder is reset.

The issue can be reproduced as below,
    1) create a region with a CXL setup which includes a HB with a
    single root port under which a memdev is attached directly.
    2) destroy the region with cxl destroy-region regionX -f.

Fixes: 176baefb2eb5 ("cxl/hdm: Commit decoder state to hardware")
Cc: <stable@vger.kernel.org>
Signed-off-by: Fan Ni <fan.ni@samsung.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Tested-by: Gregory Price <gregory.price@memverge.com>
Reviewed-by: Gregory Price <gregory.price@memverge.com>
Link: https://lore.kernel.org/r/20221215170909.2650271-1-fan.ni@samsung.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/pci: Fix irq oneshot expectations
Dan Williams [Mon, 30 Jan 2023 23:39:26 +0000 (15:39 -0800)]
cxl/pci: Fix irq oneshot expectations

The IRQ core expects that users of the default hardirq handler specify
IRQF_ONESHOT to keep interrupts disabled until the threaded handler
runs. That meets the CXL driver's expectations since it is an edge
triggered MSI and this flag would have been passed by default using
pci_request_irq() instead of devm_request_threaded_irq().

Fixes: a49aa8141b65 ("cxl/mem: Wire up event interrupts")
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Julia Lawall <julia.lawall@lip6.fr>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/pci: Set the device timestamp
Jonathan Cameron [Mon, 30 Jan 2023 15:13:27 +0000 (15:13 +0000)]
cxl/pci: Set the device timestamp

CXL r3.0 section 8.2.9.4.2 "Set Timestamp" recommends that the host sets
the timestamp after every Conventional or CXL Reset to ensure accurate
timestamps. This should include on initial boot up. The time base that
is being set is used by a device for the poison list overflow timestamp
and all event timestamps.  Note that the command is optional and if
not supported and the device cannot return accurate timestamps it will
fill the fields in with an appropriate marker (see the specification
description of each timestamp).

Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230130151327.32415-1-Jonathan.Cameron@huawei.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/mbox: Add missing parameter to docs.
Jonathan Cameron [Mon, 30 Jan 2023 15:34:37 +0000 (15:34 +0000)]
cxl/mbox: Add missing parameter to docs.

Kernel-doc should be complete, so add documentation for the status
parameter.

Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/20230130153437.3153-1-Jonathan.Cameron@huawei.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agoLinux 6.2-rc6
Linus Torvalds [Sun, 29 Jan 2023 21:59:43 +0000 (13:59 -0800)]
Linux 6.2-rc6

21 months agoMerge tag 'irq_urgent_for_v6.2_rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 29 Jan 2023 19:26:49 +0000 (11:26 -0800)]
Merge tag 'irq_urgent_for_v6.2_rc6' of git://git./linux/kernel/git/tip/tip

Pull irq fix from Borislav Petkov:

 - Cleanup the firmware node for the new IRQ MSI domain properly, to
   avoid leaking memory

* tag 'irq_urgent_for_v6.2_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  genirq/msi: Free the fwnode created by msi_create_device_irq_domain()

21 months agoMerge tag 'x86_urgent_for_v6.2_rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 29 Jan 2023 19:17:34 +0000 (11:17 -0800)]
Merge tag 'x86_urgent_for_v6.2_rc6' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Borislav Petkov:

 - Start checking for -mindirect-branch-cs-prefix clang support too now
   that LLVM 16 will support it

 - Fix a NULL ptr deref when suspending with Xen PV

 - Have a SEV-SNP guest check explicitly for features enabled by the
   hypervisor and fail gracefully if some are unsupported by the guest
   instead of failing in a non-obvious and hard-to-debug way

 - Fix a MSI descriptor leakage under Xen

 - Mark Xen's MSI domain as supporting MSI-X

 - Prevent legacy PIC interrupts from being resent in software by
   marking them level triggered, as they should be, which lead to a NULL
   ptr deref

* tag 'x86_urgent_for_v6.2_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/build: Move '-mindirect-branch-cs-prefix' out of GCC-only block
  acpi: Fix suspend with Xen PV
  x86/sev: Add SEV-SNP guest feature negotiation support
  x86/pci/xen: Fixup fallout from the PCI/MSI overhaul
  x86/pci/xen: Set MSI_FLAG_PCI_MSIX support in Xen MSI domain
  x86/i8259: Mark legacy PIC interrupts with IRQ_LEVEL

21 months agoMerge tag 'input-for-v6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor...
Linus Torvalds [Sun, 29 Jan 2023 19:06:47 +0000 (11:06 -0800)]
Merge tag 'input-for-v6.2-rc5' of git://git./linux/kernel/git/dtor/input

Pull input fixes from Dmitry Torokhov:

 - touchpads on HP 15-* laptops switched back to PS/2 emulation mode

 - a quirk for Clevo PCX0DX/TUXEDO XP1511 to make sure keyboard is
   responding after resume

* tag 'input-for-v6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: i8042 - add Clevo PCX0DX to i8042 quirk table
  Revert "Input: synaptics - switch touchpad on HP Laptop 15-da3001TU to RMI mode"

21 months agoMerge tag 'cxl-fixes-for-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 29 Jan 2023 18:47:22 +0000 (10:47 -0800)]
Merge tag 'cxl-fixes-for-6.2-rc6' of git://git./linux/kernel/git/cxl/cxl

Pull cxl fixes from Dan Williams:
 "A couple of fixes for bugs introduced during the merge window. One is
  a regression, the other was a bug in the CXL AER handler:

   - Fix a crash regression due to module load order of cxl_pmem.ko

   - Fix wrong register offset read in CXL AER handling path"

* tag 'cxl-fixes-for-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/cxl/cxl:
  cxl/pmem: Fix nvdimm unregistration when cxl_pmem driver is absent
  cxl: fix cxl_report_and_clear() RAS UE addr mis-assignment

21 months agoRevert "mm/compaction: fix set skip in fast_find_migrateblock"
Vlastimil Babka [Fri, 13 Jan 2023 17:33:45 +0000 (18:33 +0100)]
Revert "mm/compaction: fix set skip in fast_find_migrateblock"

This reverts commit 7efc3b7261030da79001c00d92bc3392fd6c664c.

We have got openSUSE reports (Link 1) for 6.1 kernel with khugepaged
stalling CPU for long periods of time.  Investigation of tracepoint data
shows that compaction is stuck in repeating fast_find_migrateblock()
based migrate page isolation, and then fails to migrate all isolated
pages.

Commit 7efc3b726103 ("mm/compaction: fix set skip in fast_find_migrateblock")
was suspected as it was merged in 6.1 and in theory can indeed remove a
termination condition for fast_find_migrateblock() under certain
conditions, as it removes a place that always marks a scanned pageblock
from being re-scanned.  There are other such places, but those can be
skipped under certain conditions, which seems to match the tracepoint
data.

Testing of revert also appears to have resolved the issue, thus revert
the commit until a more robust solution for the original problem is
developed.

It's also likely this will fix qemu stalls with 6.1 kernel reported in
Link 2, but that is not yet confirmed.

Link: https://bugzilla.suse.com/show_bug.cgi?id=1206848
Link: https://lore.kernel.org/kvm/b8017e09-f336-3035-8344-c549086c2340@kernel.org/
Link: https://lore.kernel.org/lkml/20230125134434.18017-1-mgorman@techsingularity.net/
Fixes: 7efc3b726103 ("mm/compaction: fix set skip in fast_find_migrateblock")
Cc: <stable@vger.kernel.org>
Tested-by: Pedro Falcato <pedro.falcato@gmail.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
21 months agoFix up more non-executable files marked executable
Linus Torvalds [Sat, 28 Jan 2023 19:17:57 +0000 (11:17 -0800)]
Fix up more non-executable files marked executable

Joe found another DT file that shouldn't be executable, and that
frustrated me enough that I went hunting with this script:

    git ls-files -s |
        grep '^100755' |
        cut -f2 |
        xargs grep -L '^#!'

and that found another file that shouldn't have been marked executable
either, despite being in the scripts directory.

Maybe these two are the last ones at least for now.  But I'm sure we'll
be back in a few years, fixing things up again.

Fixes: 8c6789f4e2d4 ("ASoC: dt-bindings: Add Everest ES8326 audio CODEC")
Fixes: 4d8e5cd233db ("locking/atomics: Fix scripts/atomic/ script permissions")
Reported-by: Joe Perches <joe@perches.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
21 months agoMerge tag '6.2-rc5-ksmbd-server-fixes' of git://git.samba.org/ksmbd
Linus Torvalds [Sat, 28 Jan 2023 18:52:51 +0000 (10:52 -0800)]
Merge tag '6.2-rc5-ksmbd-server-fixes' of git://git.samba.org/ksmbd

Pull ksmbd server fixes from Steve French:
 "Four smb3 server fixes, all also for stable:

   - fix for signing bug

   - fix to more strictly check packet length

   - add a max connections parm to limit simultaneous connections

   - fix error message flood that can occur with newer Samba xattr
     format"

* tag '6.2-rc5-ksmbd-server-fixes' of git://git.samba.org/ksmbd:
  ksmbd: downgrade ndr version error message to debug
  ksmbd: limit pdu length size according to connection status
  ksmbd: do not sign response to session request for guest login
  ksmbd: add max connections parameter

21 months agoMerge tag '6.2-rc5-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sat, 28 Jan 2023 01:41:47 +0000 (17:41 -0800)]
Merge tag '6.2-rc5-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull cifs fix from Steve French:
 "Fix for reconnect oops in smbdirect (RDMA), also is marked for stable"

* tag '6.2-rc5-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: Fix oops due to uncleared server->smbd_conn in reconnect

21 months agoMerge tag 'block-6.2-2023-01-27' of git://git.kernel.dk/linux
Linus Torvalds [Sat, 28 Jan 2023 00:16:57 +0000 (16:16 -0800)]
Merge tag 'block-6.2-2023-01-27' of git://git.kernel.dk/linux

Pull block fixes from Jens Axboe:
 "Minor tweaks for this release:

   - NVMe pull request via Christoph:
        - Flush initial scan_work for async probe (Keith Busch)
        - Fix passthrough csi check (Keith Busch)
        - Fix nvme-fc initialization order (Ross Lagerwall)

   - Fix for tearing down non-started device in ublk (Ming)"

* tag 'block-6.2-2023-01-27' of git://git.kernel.dk/linux:
  block: ublk: move ublk_chr_class destroying after devices are removed
  nvme: fix passthrough csi check
  nvme-pci: flush initial scan_work for async probe
  nvme-fc: fix initialization order

21 months agoMerge tag 'io_uring-6.2-2023-01-27' of git://git.kernel.dk/linux
Linus Torvalds [Sat, 28 Jan 2023 00:15:06 +0000 (16:15 -0800)]
Merge tag 'io_uring-6.2-2023-01-27' of git://git.kernel.dk/linux

Pull io_uring fixes from Jens Axboe:
 "Two small fixes for this release:

   - Sanitize how async prep is done for drain requests, so we ensure
     that it always gets done (Dylan)

   - A ring provided buffer recycling fix for multishot receive (me)"

* tag 'io_uring-6.2-2023-01-27' of git://git.kernel.dk/linux:
  io_uring: always prep_async for drain requests
  io_uring/net: cache provided buffer group value for multishot receives

21 months agoMerge tag 'hardening-v6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/kees...
Linus Torvalds [Sat, 28 Jan 2023 00:09:12 +0000 (16:09 -0800)]
Merge tag 'hardening-v6.2-rc6' of git://git./linux/kernel/git/kees/linux

Pull hardening fixes from Kees Cook:

 - Split slow memcpy tests into MEMCPY_SLOW_KUNIT_TEST

 - Reorganize gcc-plugin includes for GCC 13

 - Silence bcache memcpy run-time false positive warnings

* tag 'hardening-v6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  bcache: Silence memcpy() run-time false positive warnings
  gcc-plugins: Reorganize gimple includes for GCC 13
  kunit: memcpy: Split slow memcpy tests into MEMCPY_SLOW_KUNIT_TEST

21 months agoMerge tag 'trace-v6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/trace...
Linus Torvalds [Sat, 28 Jan 2023 00:03:32 +0000 (16:03 -0800)]
Merge tag 'trace-v6.2-rc5' of git://git./linux/kernel/git/trace/linux-trace

Pull tracing fixes from Steven Rostedt:

 - Fix filter memory leak by calling ftrace_free_filter()

 - Initialize trace_printk() earlier so that ftrace_dump_on_oops shows
   data on early crashes.

 - Update the outdated instructions in scripts/tracing/ftrace-bisect.sh

 - Add lockdep_is_held() to fix lockdep warning

 - Add allocation failure check in create_hist_field()

 - Don't initialize pointer that gets set right away in enabled_monitors_write()

 - Update MAINTAINER entries

 - Fix help messages in Kconfigs

 - Fix kernel-doc header for update_preds()

* tag 'trace-v6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  bootconfig: Update MAINTAINERS file to add tree and mailing list
  rv: remove redundant initialization of pointer ptr
  ftrace: Maintain samples/ftrace
  tracing/filter: fix kernel-doc warnings
  lib: Kconfig: fix spellos
  trace_events_hist: add check for return value of 'create_hist_field'
  tracing/osnoise: Use built-in RCU list checking
  tracing: Kconfig: Fix spelling/grammar/punctuation
  ftrace/scripts: Update the instructions for ftrace-bisect.sh
  tracing: Make sure trace_printk() can output as soon as it can be used
  ftrace: Export ftrace_free_filter() to modules

21 months agocxl/mbox: Fix Payload Length check for Get Log command
Robert Richter [Thu, 19 Jan 2023 09:49:34 +0000 (10:49 +0100)]
cxl/mbox: Fix Payload Length check for Get Log command

Commit 2aeaf663b85e introduced strict checking for variable length
payload size validation. The payload length of received data must
match the size of the requested data by the caller except for the case
where the min_out value is set.

The Get Log command does not have a header with a length field set.
The Log size is determined by the Get Supported Logs command (CXL 3.0,
8.2.9.5.1). However, the actual size can be smaller and the number of
valid bytes in the payload output must be determined reading the
Payload Length field (CXL 3.0, Table 8-36, Note 2).

Two issues arise: The command can successfully complete with a payload
length of zero. And, the valid payload length must then also be
consumed by the caller.

Change cxl_xfer_log() to pass the number of payload bytes back to the
caller to determine the number of log entries. Implement the payload
handling as a special case where mbox_cmd->size_out is consulted when
cxl_internal_send_cmd() returns -EIO. A WARN_ONCE() is added to check
that -EIO is only returned in case of an unexpected output size.

Logs can be bigger than the maximum payload length and multiple Get
Log commands can be issued. If the received payload size is smaller
than the maximum payload size we can assume all valid bytes have been
fetched. Stop sending further Get Log commands then.

On that occasion, change debug messages to also report the opcodes of
supported commands.

The variable payload commands GET_LSA and SET_LSA are not affected by
this strict check: SET_LSA cannot be broken because SET_LSA does not
return an output payload, and GET_LSA never expects short reads.

Fixes: 2aeaf663b85e ("cxl/mbox: Add variable output size validation for internal commands")
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/20230119094934.86067-1-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agoMerge tag 'i2c-for-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Fri, 27 Jan 2023 21:52:38 +0000 (13:52 -0800)]
Merge tag 'i2c-for-6.2-rc6' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "A bunch of driver fixes with a tiny bit of new IDs"

* tag 'i2c-for-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: rk3x: fix a bunch of kernel-doc warnings
  i2c: axxia: use 'struct' for kernel-doc notation
  dt-bindings: i2c: renesas,rzv2m: Fix SoC specific string
  i2c: mxs: suppress probe-deferral error message
  i2c: designware-pci: Add new PCI IDs for AMD NAVI GPU
  i2c: designware: Fix unbalanced suspended flag
  i2c: designware: use casting of u64 in clock multiplication to avoid overflow

21 months agoMerge tag 'gpio-fixes-for-v6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 27 Jan 2023 21:47:40 +0000 (13:47 -0800)]
Merge tag 'gpio-fixes-for-v6.2-rc6' of git://git./linux/kernel/git/brgl/linux

Pull gpio fixes from Bartosz Golaszewski:

 - fix the -c option in the gpio-event-mode user-space example program

 - fix the irq number translation in gpio-ep93xx and make its irqchip
   immutable

 - add a missing spin_unlock in error path in gpio-mxc

 - fix a suspend breakage on System76 and Lenovo Gen2a introduced in
   GPIO ACPI

* tag 'gpio-fixes-for-v6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux:
  tools: gpio: fix -c option of gpio-event-mon
  gpio: ep93xx: remove unused variable
  gpio: ep93xx: Make irqchip immutable
  gpio: ep93xx: Fix port F hwirq numbers in handler
  gpio: mxc: Unlock on error path in mxc_flip_edge()
  gpiolib-acpi: Don't set GPIOs for wakeup in S3 mode

21 months agoMerge tag 'regulator-fix-v6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 27 Jan 2023 21:43:46 +0000 (13:43 -0800)]
Merge tag 'regulator-fix-v6.2-rc5' of git://git./linux/kernel/git/broonie/regulator

Pull regulator fix from Mark Brown:
 "A fix for the DT binding documentation which dropped a property when
  being converted to YAML format causing spurious errors validating
  device trees for platforms using the device"

* tag 'regulator-fix-v6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
  regulator: dt-bindings: samsung,s2mps14: add lost samsung,ext-control-gpios

21 months agoMerge tag 'ovl-fixes-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/mszere...
Linus Torvalds [Fri, 27 Jan 2023 21:39:30 +0000 (13:39 -0800)]
Merge tag 'ovl-fixes-6.2-rc6' of git://git./linux/kernel/git/mszeredi/vfs

Pull overlayfs fixes from Miklos Szeredi:
 "Fix two bugs, a recent one introduced in the last cycle, and an older
  one from v5.11"

* tag 'ovl-fixes-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs:
  ovl: fail on invalid uid/gid mapping at copy up
  ovl: fix tmpfile leak

21 months agoMerge tag 'drm-fixes-2023-01-27' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 27 Jan 2023 21:18:14 +0000 (13:18 -0800)]
Merge tag 'drm-fixes-2023-01-27' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "Fairly small this week as well, i915 has a memory leak fix and some
  minor changes, and amdgpu has some MST fixes, and some other minor
  ones:

  drm:
   - DP MST kref fix
   - fb_helper: check return value

  i915:
   - Fix BSC default context for Meteor Lake
   - Fix selftest-scheduler's modify_type
   - memory leak fix

  amdgpu:
   - GC11.x fixes
   - SMU13.0.0 fix
   - Freesync video fix
   - DP MST fixes
   - build fix"

* tag 'drm-fixes-2023-01-27' of git://anongit.freedesktop.org/drm/drm:
  amdgpu: fix build on non-DCN platforms.
  drm/amd/display: Fix timing not changning when freesync video is enabled
  drm/display/dp_mst: Correct the kref of port.
  drm/amdgpu/display/mst: update mst_mgr relevant variable when long HPD
  drm/amdgpu/display/mst: limit payload to be updated one by one
  drm/amdgpu/display/mst: Fix mst_state->pbn_div and slot count assignments
  drm/amdgpu: declare firmware for new MES 11.0.4
  drm/amdgpu: enable imu firmware for GC 11.0.4
  drm/amd/pm: add missing AllowIHInterrupt message mapping for SMU13.0.0
  drm/amdgpu: remove unconditional trap enable on add gfx11 queues
  drm/fb-helper: Use a per-driver FB deferred I/O handler
  drm/fb-helper: Check fb_deferred_io_init() return value
  drm/i915/selftest: fix intel_selftest_modify_policy argument types
  drm/i915/mtl: Fix bcs default context
  drm/i915: Fix a memory leak with reused mmap_offset
  drm/drm_vma_manager: Add drm_vma_node_allow_once()

21 months agoMerge tag 'acpi-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Fri, 27 Jan 2023 21:11:19 +0000 (13:11 -0800)]
Merge tag 'acpi-6.2-rc6' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI fixes from Rafael Wysocki:
 "Add ACPI backlight handling quirks for 3 machines (Hans de Goede)"

* tag 'acpi-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI: video: Add backlight=native DMI quirk for Asus U46E
  ACPI: video: Add backlight=native DMI quirk for HP EliteBook 8460p
  ACPI: video: Add backlight=native DMI quirk for HP Pavilion g6-1d80nr

21 months agoMerge tag 'thermal-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Fri, 27 Jan 2023 21:01:36 +0000 (13:01 -0800)]
Merge tag 'thermal-6.2-rc6' of git://git./linux/kernel/git/rafael/linux-pm

Pull thermal control fixes from Rafael Wysocki:
 "Add locking to the Intel int340x thermal control driver to prevent its
  thermal zone callbacks from racing with firmware-induced thermal trip
  point updates (Srinivas Pandruvada, Rafael Wysocki)"

* tag 'thermal-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  thermal: intel: int340x: Add locking to int340x_thermal_get_trip_type()
  thermal: intel: int340x: Protect trip temperature from concurrent updates

21 months agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Fri, 27 Jan 2023 20:56:45 +0000 (12:56 -0800)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fix from Will Deacon:

 - Fix event counting regression in Arm CMN PMU driver due to broken
   optimisation

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  Partially revert "perf/arm-cmn: Optimise DTC counter accesses"

21 months agoMerge tag 'riscv-for-linus-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 27 Jan 2023 20:52:45 +0000 (12:52 -0800)]
Merge tag 'riscv-for-linus-6.2-rc6' of git://git./linux/kernel/git/riscv/linux

Pull RISC-V fixes from Palmer Dabbelt:

 - A few DT bindings fixes to more closely align the ISA string
   requirements between the bindings and the ISA manual.

 - A handful of build error/warning fixes.

 - A fix to move init_cpu_topology() later in the boot flow, so it can
   allocate memory.

 - The IRC channel is now in the MAINTAINERS file, so it's easier to
   find.

* tag 'riscv-for-linus-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux:
  riscv: Move call to init_cpu_topology() to later initialization stage
  riscv/kprobe: Fix instruction simulation of JALR
  riscv: fix -Wundef warning for CONFIG_RISCV_BOOT_SPINWAIT
  MAINTAINERS: add an IRC entry for RISC-V
  RISC-V: fix compile error from deduplicated __ALTERNATIVE_CFG_2
  dt-bindings: riscv: fix single letter canonical order
  dt-bindings: riscv: fix underscore requirement for multi-letter extensions

21 months agoMerge tag 'for-linus' of git://git.armlinux.org.uk/~rmk/linux-arm
Linus Torvalds [Fri, 27 Jan 2023 20:49:00 +0000 (12:49 -0800)]
Merge tag 'for-linus' of git://git.armlinux.org.uk/~rmk/linux-arm

Pull ARM fixes from Russell King:

 - fix nommu assignment build warning

 - fix -Wundef preprocessor warning

 - reduce __thumb2__ definitions for crypto files that require it

* tag 'for-linus' of git://git.armlinux.org.uk/~rmk/linux-arm:
  ARM: 9287/1: Reduce __thumb2__ definition to crypto files that require it
  ARM: 9284/1: include <asm/pgtable.h> from proc-macros.S to fix -Wundef warnings
  ARM: 9280/1: mm: fix warning on phys_addr_t to void pointer assignment

21 months agoMerge tag 'linux-kselftest-fixes-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kerne...
Linus Torvalds [Fri, 27 Jan 2023 20:41:09 +0000 (12:41 -0800)]
Merge tag 'linux-kselftest-fixes-6.2-rc6' of git://git./linux/kernel/git/shuah/linux-kselftest

Pull Kselftest fixes from Shuah Khan:
 "A single fix to a amd-pstate test Makefile bug that deletes source
  files during make clean run"

* tag 'linux-kselftest-fixes-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
  selftests: amd-pstate: Don't delete source files via Makefile

21 months agoovl: fail on invalid uid/gid mapping at copy up
Miklos Szeredi [Tue, 24 Jan 2023 15:41:18 +0000 (16:41 +0100)]
ovl: fail on invalid uid/gid mapping at copy up

If st_uid/st_gid doesn't have a mapping in the mounter's user_ns, then
copy-up should fail, just like it would fail if the mounter task was doing
the copy using "cp -a".

There's a corner case where the "cp -a" would succeed but copy up fail: if
there's a mapping of the invalid uid/gid (65534 by default) in the user
namespace.  This is because stat(2) will return this value if the mapping
doesn't exist in the current user_ns and "cp -a" will in turn be able to
create a file with this uid/gid.

This behavior would be inconsistent with POSIX ACL's, which return -1 for
invalid uid/gid which result in a failed copy.

For consistency and simplicity fail the copy of the st_uid/st_gid are
invalid.

Fixes: 459c7c565ac3 ("ovl: unprivieged mounts")
Cc: <stable@vger.kernel.org> # v5.11
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Reviewed-by: Christian Brauner <brauner@kernel.org>
Reviewed-by: Seth Forshee <sforshee@kernel.org>
21 months agoovl: fix tmpfile leak
Miklos Szeredi [Tue, 24 Jan 2023 15:41:18 +0000 (16:41 +0100)]
ovl: fix tmpfile leak

Missed an error cleanup.

Reported-by: syzbot+fd749a7ea127a84e0ffd@syzkaller.appspotmail.com
Fixes: 2b1a77461f16 ("ovl: use vfs_tmpfile_open() helper")
Cc: <stable@vger.kernel.org> # v6.1
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
21 months agoio_uring: always prep_async for drain requests
Dylan Yudaken [Fri, 27 Jan 2023 10:59:11 +0000 (02:59 -0800)]
io_uring: always prep_async for drain requests

Drain requests all go through io_drain_req, which has a quick exit in case
there is nothing pending (ie the drain is not useful). In that case it can
run the issue the request immediately.

However for safety it queues it through task work.
The problem is that in this case the request is run asynchronously, but
the async work has not been prepared through io_req_prep_async.

This has not been a problem up to now, as the task work always would run
before returning to userspace, and so the user would not have a chance to
race with it.

However - with IORING_SETUP_DEFER_TASKRUN - this is no longer the case and
the work might be defered, giving userspace a chance to change data being
referred to in the request.

Instead _always_ prep_async for drain requests, which is simpler anyway
and removes this issue.

Cc: stable@vger.kernel.org
Fixes: c0e0d6ba25f1 ("io_uring: add IORING_SETUP_DEFER_TASKRUN")
Signed-off-by: Dylan Yudaken <dylany@meta.com>
Link: https://lore.kernel.org/r/20230127105911.2420061-1-dylany@meta.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
21 months agotools: gpio: fix -c option of gpio-event-mon
Ivo Borisov Shopov [Thu, 26 Jan 2023 13:10:33 +0000 (15:10 +0200)]
tools: gpio: fix -c option of gpio-event-mon

Following line should listen for a rising edge and exit after the first
one since '-c 1' is provided.

    # gpio-event-mon -n gpiochip1 -o 0 -r -c 1

It works with kernel 4.19 but it doesn't work with 5.10. In 5.10 the
above command doesn't exit after the first rising edge it keep listening
for an event forever. The '-c 1' is not taken into an account.
The problem is in commit 62757c32d5db ("tools: gpio: add multi-line
monitoring to gpio-event-mon").
Before this commit the iterator 'i' in monitor_device() is used for
counting of the events (loops). In the case of the above command (-c 1)
we should start from 0 and increment 'i' only ones and hit the 'break'
statement and exit the process. But after the above commit counting
doesn't start from 0, it start from 1 when we listen on one line.
It is because 'i' is used from one more purpose, counting of lines
(num_lines) and it isn't restore to 0 after following code

    for (i = 0; i < num_lines; i++)
        gpiotools_set_bit(&values.mask, i);

Restore the initial value of the iterator to 0 in order to allow counting
of loops to work for any cases.

Fixes: 62757c32d5db ("tools: gpio: add multi-line monitoring to gpio-event-mon")
Signed-off-by: Ivo Borisov Shopov <ivoshopov@gmail.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
[Bartosz: tweak the commit message]
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
21 months agogpio: ep93xx: remove unused variable
Arnd Bergmann [Fri, 27 Jan 2023 09:35:05 +0000 (10:35 +0100)]
gpio: ep93xx: remove unused variable

This one was left behind by a previous cleanup patch:

drivers/gpio/gpio-ep93xx.c: In function 'ep93xx_gpio_add_bank':
drivers/gpio/gpio-ep93xx.c:366:34: error: unused variable 'ic' [-Werror=unused-variable]

Fixes: 216f37366e86 ("gpio: ep93xx: Make irqchip immutable")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
21 months agoMerge tag 'drm-misc-fixes-2023-01-26' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Fri, 27 Jan 2023 02:31:02 +0000 (12:31 +1000)]
Merge tag 'drm-misc-fixes-2023-01-26' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

A fix and a preliminary patch to fix a memory leak in i915, and a use
after free fix for fbdev deferred io

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20230126104018.cbrcjxl5wefdbb2f@houat
21 months agoamdgpu: fix build on non-DCN platforms.
Dave Airlie [Fri, 27 Jan 2023 02:15:13 +0000 (12:15 +1000)]
amdgpu: fix build on non-DCN platforms.

This fixes the build here locally on my 32-bit arm build.

Signed-off-by: Dave Airlie <airlied@redhat.com>
21 months agoMerge tag 'amd-drm-fixes-6.2-2023-01-25' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Fri, 27 Jan 2023 01:50:08 +0000 (11:50 +1000)]
Merge tag 'amd-drm-fixes-6.2-2023-01-25' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.2-2023-01-25:

amdgpu:
- GC11.x fixes
- SMU13.0.0 fix
- Freesync video fix
- DP MST fixes

drm:
- DP MST kref fix

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230125220153.320248-1-alexander.deucher@amd.com
21 months agoMerge tag 'drm-intel-fixes-2023-01-26' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Fri, 27 Jan 2023 01:39:55 +0000 (11:39 +1000)]
Merge tag 'drm-intel-fixes-2023-01-26' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

- Fix BSC default context for Meteor Lake (Lucas)
- Fix selftest-scheduler's modify_type (Andi)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/Y9LKD2J5bmICTyIP@intel.com
21 months agocxl/test: Simulate event log overflow
Ira Weiny [Wed, 18 Jan 2023 05:53:43 +0000 (21:53 -0800)]
cxl/test: Simulate event log overflow

Log overflow is marked by a separate trace message.

Simulate a log with lots of messages and flag overflow until space is
cleared.

Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221216-cxl-ev-log-v7-8-2316a5c8f7d8@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/test: Add specific events
Ira Weiny [Wed, 18 Jan 2023 05:53:42 +0000 (21:53 -0800)]
cxl/test: Add specific events

Each type of event has different trace point outputs.

Add mock General Media Event, DRAM event, and Memory Module Event
records to the mock list of events returned.

Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221216-cxl-ev-log-v7-7-2316a5c8f7d8@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/test: Add generic mock events
Ira Weiny [Wed, 18 Jan 2023 05:53:41 +0000 (21:53 -0800)]
cxl/test: Add generic mock events

Facilitate testing basic Get/Clear Event functionality by creating
multiple logs and generic events with made up UUID's.

Data is completely made up with data patterns which should be easy to
spot in trace output.

A single sysfs entry resets the event data and triggers collecting the
events for testing.

Test traces are easy to obtain with a small script such as this:

#!/bin/bash -x

devices=`find /sys/devices/platform -name cxl_mem*`

# Turn on tracing
echo "" > /sys/kernel/tracing/trace
echo 1 > /sys/kernel/tracing/events/cxl/enable
echo 1 > /sys/kernel/tracing/tracing_on

# Generate fake interrupt
for device in $devices; do
        echo 1 > $device/event_trigger
done

# Turn off tracing and report events
echo 0 > /sys/kernel/tracing/tracing_on
cat /sys/kernel/tracing/trace

Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221216-cxl-ev-log-v7-6-2316a5c8f7d8@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/mem: Trace Memory Module Event Record
Ira Weiny [Wed, 18 Jan 2023 05:53:40 +0000 (21:53 -0800)]
cxl/mem: Trace Memory Module Event Record

CXL rev 3.0 section 8.2.9.2.1.3 defines the Memory Module Event Record.

Determine if the event read is memory module record and if so trace the
record.

Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221216-cxl-ev-log-v7-5-2316a5c8f7d8@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/mem: Trace DRAM Event Record
Ira Weiny [Wed, 18 Jan 2023 05:53:39 +0000 (21:53 -0800)]
cxl/mem: Trace DRAM Event Record

CXL rev 3.0 section 8.2.9.2.1.2 defines the DRAM Event Record.

Determine if the event read is a DRAM event record and if so trace the
record.

Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221216-cxl-ev-log-v7-4-2316a5c8f7d8@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/mem: Trace General Media Event Record
Ira Weiny [Wed, 18 Jan 2023 05:53:38 +0000 (21:53 -0800)]
cxl/mem: Trace General Media Event Record

CXL rev 3.0 section 8.2.9.2.1.1 defines the General Media Event Record.

Determine if the event read is a general media record and if so trace
the record as a General Media Event Record.

Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221216-cxl-ev-log-v7-3-2316a5c8f7d8@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/mem: Wire up event interrupts
Davidlohr Bueso [Wed, 18 Jan 2023 05:53:37 +0000 (21:53 -0800)]
cxl/mem: Wire up event interrupts

Currently the only CXL features targeted for irq support require their
message numbers to be within the first 16 entries.  The device may
however support less than 16 entries depending on the support it
provides.

Attempt to allocate these 16 irq vectors.  If the device supports less
then the PCI infrastructure will allocate that number.  Upon successful
allocation, users can plug in their respective isr at any point
thereafter.

CXL device events are signaled via interrupts.  Each event log may have
a different interrupt message number.  These message numbers are
reported in the Get Event Interrupt Policy mailbox command.

Add interrupt support for event logs.  Interrupts are allocated as
shared interrupts.  Therefore, all or some event logs can share the same
message number.

In addition all logs are queried on any interrupt in order of the most
to least severe based on the status register.

Finally place all event configuration logic into cxl_event_config().
Previously the logic was a simple 'read all' on start up.  But
interrupts must be configured prior to any reads to ensure no events are
missed.  A single event configuration function results in a cleaner over
all implementation.

Cc: Bjorn Helgaas <helgaas@kernel.org>
Cc: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Co-developed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221216-cxl-ev-log-v7-2-2316a5c8f7d8@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl: fix spelling mistakes
Randy Dunlap [Wed, 25 Jan 2023 03:22:21 +0000 (19:22 -0800)]
cxl: fix spelling mistakes

Correct spelling mistakes (reported by codespell).

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Alison Schofield <alison.schofield@intel.com>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Ira Weiny <ira.weiny@intel.com>
Cc: Ben Widawsky <bwidawsk@kernel.org>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: linux-cxl@vger.kernel.org
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>
Link: https://lore.kernel.org/r/20230125032221.21277-1-rdunlap@infradead.org
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agotools/testing/cxl: Remove cxl_test module math loading message
Alison Schofield [Thu, 26 Jan 2023 17:05:55 +0000 (09:05 -0800)]
tools/testing/cxl: Remove cxl_test module math loading message

Commit "tools/testing/cxl: Add XOR Math support to cxl_test" added
a module parameter to cxl_test for the interleave_arithmetic option.

In doing so, it also added this dev_dbg() message describing which
option cxl_test used during load:
"[  111.743246] (NULL device *): cxl_test loading modulo math option"
That "(NULL device *)" has raised needless user concern.

Remove the dev_dbg() message and make the module_param readable via
sysfs for users that need to know which math option is active.

Suggested-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Link: https://lore.kernel.org/r/20230126170555.701240-1-alison.schofield@intel.com
Signed-off-by: Alison Schofield <alison.schofield@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/mbox: Add debug messages for enabled mailbox commands
Robert Richter [Wed, 25 Jan 2023 08:57:28 +0000 (09:57 +0100)]
cxl/mbox: Add debug messages for enabled mailbox commands

Only unsupported mailbox commands are reported in debug messages. A
list of enabled commands is useful too. Change debug messages to also
report the opcodes of enabled commands. Esp. if card initialization
fails there is no way to get this information from userland.

On that occasion also add missing trailing newlines.

Signed-off-by: Robert Richter <rrichter@amd.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230125085728.234697-1-rrichter@amd.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/mem: Read, trace, and clear events on driver load
Ira Weiny [Wed, 18 Jan 2023 05:53:36 +0000 (21:53 -0800)]
cxl/mem: Read, trace, and clear events on driver load

CXL devices have multiple event logs which can be queried for CXL event
records.  Devices are required to support the storage of at least one
event record in each event log type.

Devices track event log overflow by incrementing a counter and tracking
the time of the first and last overflow event seen.

Software queries events via the Get Event Record mailbox command; CXL
rev 3.0 section 8.2.9.2.2 and clears events via CXL rev 3.0 section
8.2.9.2.3 Clear Event Records mailbox command.

If the result of negotiating CXL Error Reporting Control is OS control,
read and clear all event logs on driver load.

Ensure a clean slate of events by reading and clearing the events on
driver load.

The status register is not used because a device may continue to trigger
events and the only requirement is to empty the log at least once.  This
allows for the required transition from empty to non-empty for interrupt
generation.  Handling of interrupts is in a follow on patch.

The device can return up to 1MB worth of event records per query.
Allocate a shared large buffer to handle the max number of records based
on the mailbox payload size.

This patch traces a raw event record and leaves specific event record
type tracing to subsequent patches.  Macros are created to aid in
tracing the common CXL Event header fields.

Each record is cleared explicitly.  A clear all bit is specified but is
only valid when the log overflows.

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221216-cxl-ev-log-v7-1-2316a5c8f7d8@intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agoMerge tag 'nvme-6.2-2023-01-26' of git://git.infradead.org/nvme into block-6.2
Jens Axboe [Thu, 26 Jan 2023 18:43:33 +0000 (11:43 -0700)]
Merge tag 'nvme-6.2-2023-01-26' of git://git.infradead.org/nvme into block-6.2

Pull NVMe fixes from Christoph:

"nvme fixes for Linux 6.2

 - flush initial scan_work for async probe (Keith Busch)
 - fix passthrough csi check (Keith Busch)
 - fix nvme-fc initialization order (Ross Lagerwall)"

* tag 'nvme-6.2-2023-01-26' of git://git.infradead.org/nvme:
  nvme: fix passthrough csi check
  nvme-pci: flush initial scan_work for async probe
  nvme-fc: fix initialization order

21 months agoMerge tag 'platform-drivers-x86-v6.2-3' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 26 Jan 2023 18:29:49 +0000 (10:29 -0800)]
Merge tag 'platform-drivers-x86-v6.2-3' of git://git./linux/kernel/git/pdx86/platform-drivers-x86

Pull x86 platform driver fixes from Hans de Goede:

 -  Fix false positive apple_gmux backlight detection on older iGPU only
    MacBook models

 -  Various other small fixes and hardware-id additions

* tag 'platform-drivers-x86-v6.2-3' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86:
  platform/x86: thinkpad_acpi: Fix profile modes on Intel platforms
  ACPI: video: Fix apple gmux detection
  platform/x86: apple-gmux: Add apple_gmux_detect() helper
  platform/x86: apple-gmux: Move port defines to apple-gmux.h
  platform/x86: hp-wmi: Fix cast to smaller integer type warning
  platform/x86/amd: pmc: Add a module parameter to disable workarounds
  platform/x86/amd: pmc: Disable IRQ1 wakeup for RN/CZN
  platform/x86: asus-wmi: Fix kbd_dock_devid tablet-switch reporting
  platform/x86: gigabyte-wmi: add support for B450M DS3H WIFI-CF
  platform/x86: hp-wmi: Handle Omen Key event
  platform/x86: dell-wmi: Add a keymap for KEY_MUTE in type 0x0010 table

21 months agoMerge tag 'net-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 26 Jan 2023 18:20:12 +0000 (10:20 -0800)]
Merge tag 'net-6.2-rc6' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Paolo Abeni:
 "Including fixes from netfilter.

  Current release - regressions:

   - sched: sch_taprio: do not schedule in taprio_reset()

  Previous releases - regressions:

   - core: fix UaF in netns ops registration error path

   - ipv4: prevent potential spectre v1 gadgets

   - ipv6: fix reachability confirmation with proxy_ndp

   - netfilter: fix for the set rbtree

   - eth: fec: use page_pool_put_full_page when freeing rx buffers

   - eth: iavf: fix temporary deadlock and failure to set MAC address

  Previous releases - always broken:

   - netlink: prevent potential spectre v1 gadgets

   - netfilter: fixes for SCTP connection tracking

   - mctp: struct sock lifetime fixes

   - eth: ravb: fix possible hang if RIS2_QFF1 happen

   - eth: tg3: resolve deadlock in tg3_reset_task() during EEH

  Misc:

   - Mat stepped out as MPTCP co-maintainer"

* tag 'net-6.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (40 commits)
  net: mdio-mux-meson-g12a: force internal PHY off on mux switch
  docs: networking: Fix bridge documentation URL
  tsnep: Fix TX queue stop/wake for multiple queues
  net/tg3: resolve deadlock in tg3_reset_task() during EEH
  net: mctp: mark socks as dead on unhash, prevent re-add
  net: mctp: hold key reference when looking up a general key
  net: mctp: move expiry timer delete to unhash
  net: mctp: add an explicit reference from a mctp_sk_key to sock
  net: ravb: Fix possible hang if RIS2_QFF1 happen
  net: ravb: Fix lack of register setting after system resumed for Gen3
  net/x25: Fix to not accept on connected socket
  ice: move devlink port creation/deletion
  sctp: fail if no bound addresses can be used for a given scope
  net/sched: sch_taprio: do not schedule in taprio_reset()
  Revert "Merge branch 'ethtool-mac-merge'"
  netrom: Fix use-after-free of a listening socket.
  netfilter: conntrack: unify established states for SCTP paths
  Revert "netfilter: conntrack: add sctp DATA_SENT state"
  netfilter: conntrack: fix bug in for_each_sctp_chunk
  netfilter: conntrack: fix vtag checks for ABORT/SHUTDOWN_COMPLETE
  ...

21 months agotreewide: fix up files incorrectly marked executable
Linus Torvalds [Thu, 26 Jan 2023 18:05:39 +0000 (10:05 -0800)]
treewide: fix up files incorrectly marked executable

I'm not exactly clear on what strange workflow causes people to do it,
but clearly occasionally some files end up being committed as executable
even though they clearly aren't.

This is a reprise of commit 90fda63fa115 ("treewide: fix up files
incorrectly marked executable"), just with a different set of files (but
with the same trivial shell scripting).

So apparently we need to re-do this every five years or so, and Joe
needs to just keep reminding me to do so ;)

Reported-by: Joe Perches <joe@perches.com>
Fixes: 523375c943e5 ("drm/vmwgfx: Port vmwgfx to arm64")
Fixes: 5c439937775d ("ASoC: codecs: add support for ES8326")
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
21 months agoblock: ublk: move ublk_chr_class destroying after devices are removed
Ming Lei [Thu, 26 Jan 2023 11:53:46 +0000 (19:53 +0800)]
block: ublk: move ublk_chr_class destroying after devices are removed

The 'ublk_chr_class' is needed when deleting ublk char devices in
ublk_exit(), so move it after devices(idle) are removed.

Fixes the following warning reported by Harris, James R:

[  859.178950] sysfs group 'power' not found for kobject 'ublkc0'
[  859.178962] WARNING: CPU: 3 PID: 1109 at fs/sysfs/group.c:278 sysfs_remove_group+0x9c/0xb0

Reported-by: "Harris, James R" <james.r.harris@intel.com>
Fixes: 71f28f3136af ("ublk_drv: add io_uring based userspace block driver")
Link: https://lore.kernel.org/linux-block/Y9JlFmSgDl3+zy3N@T590/T/#t
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Tested-by: Jim Harris <james.r.harris@intel.com>
Link: https://lore.kernel.org/r/20230126115346.263344-1-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
21 months agoPartially revert "perf/arm-cmn: Optimise DTC counter accesses"
Robin Murphy [Mon, 23 Jan 2023 18:30:38 +0000 (18:30 +0000)]
Partially revert "perf/arm-cmn: Optimise DTC counter accesses"

It turns out the optimisation implemented by commit 4f2c3872dde5 is
totally broken, since all the places that consume hw->dtcs_used for
events other than cycle count are still not expecting it to be sparsely
populated, and fail to read all the relevant DTC counters correctly if
so.

If implemented correctly, the optimisation potentially saves up to 3
register reads per event update, which is reasonably significant for
events targeting a single node, but still not worth a massive amount of
additional code complexity overall. Getting it right within the current
design looks a fair bit more involved than it was ever intended to be,
so let's just make a functional revert which restores the old behaviour
while still backporting easily.

Fixes: 4f2c3872dde5 ("perf/arm-cmn: Optimise DTC counter accesses")
Reported-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Link: https://lore.kernel.org/r/b41bb4ed7283c3d8400ce5cf5e6ec94915e6750f.1674498637.git.robin.murphy@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
21 months agonet: mdio-mux-meson-g12a: force internal PHY off on mux switch
Jerome Brunet [Tue, 24 Jan 2023 10:11:57 +0000 (11:11 +0100)]
net: mdio-mux-meson-g12a: force internal PHY off on mux switch

Force the internal PHY off then on when switching to the internal path.
This fixes problems where the PHY ID is not properly set.

Fixes: 7090425104db ("net: phy: add amlogic g12a mdio mux support")
Suggested-by: Qi Duan <qi.duan@amlogic.com>
Co-developed-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20230124101157.232234-1-jbrunet@baylibre.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agodocs: networking: Fix bridge documentation URL
Ivan Vecera [Tue, 24 Jan 2023 14:51:26 +0000 (15:51 +0100)]
docs: networking: Fix bridge documentation URL

Current documentation URL [1] is no longer valid.

[1] https://www.linuxfoundation.org/collaborate/workgroups/networking/bridge

Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Link: https://lore.kernel.org/r/20230124145127.189221-1-ivecera@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agotsnep: Fix TX queue stop/wake for multiple queues
Gerhard Engleder [Tue, 24 Jan 2023 19:14:40 +0000 (20:14 +0100)]
tsnep: Fix TX queue stop/wake for multiple queues

netif_stop_queue() and netif_wake_queue() act on TX queue 0. This is ok
as long as only a single TX queue is supported. But support for multiple
TX queues was introduced with 762031375d5c and I missed to adapt stop
and wake of TX queues.

Use netif_stop_subqueue() and netif_tx_wake_queue() to act on specific
TX queue.

Fixes: 762031375d5c ("tsnep: Support multiple TX/RX queue pairs")
Signed-off-by: Gerhard Engleder <gerhard@engleder-embedded.com>
Link: https://lore.kernel.org/r/20230124191440.56887-1-gerhard@engleder-embedded.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agonet/tg3: resolve deadlock in tg3_reset_task() during EEH
David Christensen [Tue, 24 Jan 2023 18:53:39 +0000 (13:53 -0500)]
net/tg3: resolve deadlock in tg3_reset_task() during EEH

During EEH error injection testing, a deadlock was encountered in the tg3
driver when tg3_io_error_detected() was attempting to cancel outstanding
reset tasks:

crash> foreach UN bt
...
PID: 159    TASK: c0000000067c6000  CPU: 8   COMMAND: "eehd"
...
 #5 [c00000000681f990] __cancel_work_timer at c00000000019fd18
 #6 [c00000000681fa30] tg3_io_error_detected at c00800000295f098 [tg3]
 #7 [c00000000681faf0] eeh_report_error at c00000000004e25c
...

PID: 290    TASK: c000000036e5f800  CPU: 6   COMMAND: "kworker/6:1"
...
 #4 [c00000003721fbc0] rtnl_lock at c000000000c940d8
 #5 [c00000003721fbe0] tg3_reset_task at c008000002969358 [tg3]
 #6 [c00000003721fc60] process_one_work at c00000000019e5c4
...

PID: 296    TASK: c000000037a65800  CPU: 21  COMMAND: "kworker/21:1"
...
 #4 [c000000037247bc0] rtnl_lock at c000000000c940d8
 #5 [c000000037247be0] tg3_reset_task at c008000002969358 [tg3]
 #6 [c000000037247c60] process_one_work at c00000000019e5c4
...

PID: 655    TASK: c000000036f49000  CPU: 16  COMMAND: "kworker/16:2"
...:1

 #4 [c0000000373ebbc0] rtnl_lock at c000000000c940d8
 #5 [c0000000373ebbe0] tg3_reset_task at c008000002969358 [tg3]
 #6 [c0000000373ebc60] process_one_work at c00000000019e5c4
...

Code inspection shows that both tg3_io_error_detected() and
tg3_reset_task() attempt to acquire the RTNL lock at the beginning of
their code blocks.  If tg3_reset_task() should happen to execute between
the times when tg3_io_error_deteced() acquires the RTNL lock and
tg3_reset_task_cancel() is called, a deadlock will occur.

Moving tg3_reset_task_cancel() call earlier within the code block, prior
to acquiring RTNL, prevents this from happening, but also exposes another
deadlock issue where tg3_reset_task() may execute AFTER
tg3_io_error_detected() has executed:

crash> foreach UN bt
PID: 159    TASK: c0000000067d2000  CPU: 9   COMMAND: "eehd"
...
 #4 [c000000006867a60] rtnl_lock at c000000000c940d8
 #5 [c000000006867a80] tg3_io_slot_reset at c0080000026c2ea8 [tg3]
 #6 [c000000006867b00] eeh_report_reset at c00000000004de88
...
PID: 363    TASK: c000000037564000  CPU: 6   COMMAND: "kworker/6:1"
...
 #3 [c000000036c1bb70] msleep at c000000000259e6c
 #4 [c000000036c1bba0] napi_disable at c000000000c6b848
 #5 [c000000036c1bbe0] tg3_reset_task at c0080000026d942c [tg3]
 #6 [c000000036c1bc60] process_one_work at c00000000019e5c4
...

This issue can be avoided by aborting tg3_reset_task() if EEH error
recovery is already in progress.

Fixes: db84bf43ef23 ("tg3: tg3_reset_task() needs to use rtnl_lock to synchronize")
Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Link: https://lore.kernel.org/r/20230124185339.225806-1-drc@linux.vnet.ibm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
21 months agoksmbd: downgrade ndr version error message to debug
Namjae Jeon [Tue, 24 Jan 2023 15:09:02 +0000 (00:09 +0900)]
ksmbd: downgrade ndr version error message to debug

When user switch samba to ksmbd, The following message flood is coming
when accessing files. Samba seems to changs dos attribute version to v5.
This patch downgrade ndr version error message to debug.

$ dmesg
...
[68971.766914] ksmbd: v5 version is not supported
[68971.779808] ksmbd: v5 version is not supported
[68971.871544] ksmbd: v5 version is not supported
[68971.910135] ksmbd: v5 version is not supported
...

Cc: stable@vger.kernel.org
Fixes: e2f34481b24d ("cifsd: add server-side procedures for SMB3")
Signed-off-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
21 months agoksmbd: limit pdu length size according to connection status
Namjae Jeon [Tue, 24 Jan 2023 15:13:20 +0000 (00:13 +0900)]
ksmbd: limit pdu length size according to connection status

Stream protocol length will never be larger than 16KB until session setup.
After session setup, the size of requests will not be larger than
16KB + SMB2 MAX WRITE size. This patch limits these invalidly oversized
requests and closes the connection immediately.

Fixes: 0626e6641f6b ("cifsd: add server handler for central processing and tranport layers")
Cc: stable@vger.kernel.org
Reported-by: zdi-disclosures@trendmicro.com # ZDI-CAN-18259
Signed-off-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
21 months agocxl/pmem: Fix nvdimm unregistration when cxl_pmem driver is absent
Dan Williams [Sat, 21 Jan 2023 00:26:12 +0000 (16:26 -0800)]
cxl/pmem: Fix nvdimm unregistration when cxl_pmem driver is absent

The cxl_pmem.ko module houses the driver for both cxl_nvdimm_bridge
objects and cxl_nvdimm objects. When the core creates a cxl_nvdimm it
arranges for it to be autoremoved when the bridge goes down. However, if
the bridge never initialized because the cxl_pmem.ko module never
loaded, it sets up a the following crash scenario:

    BUG: kernel NULL pointer dereference, address: 0000000000000478
    [..]
    RIP: 0010:cxl_nvdimm_probe+0x99/0x140 [cxl_pmem]
    [..]
    Call Trace:
     <TASK>
     cxl_bus_probe+0x17/0x50 [cxl_core]
     really_probe+0xde/0x380
     __driver_probe_device+0x78/0x170
     driver_probe_device+0x1f/0x90
     __driver_attach+0xd2/0x1c0
     bus_for_each_dev+0x79/0xc0
     bus_add_driver+0x1b1/0x200
     driver_register+0x89/0xe0
     cxl_pmem_init+0x50/0xff0 [cxl_pmem]

It turns out the recent rework to simplify nvdimm probing obviated the
need to unregister cxl_nvdimm objects at cxl_nvdimm_bridge ->remove()
time. Leave the cxl_nvdimm device registered until the hosting
cxl_memdev departs. The alternative is that the cxl_memdev needs to be
reattached whenever the cxl_nvdimm_bridge attach state cycles, which is
awkward and unnecessary.

The only requirement is to make sure that when the cxl_nvdimm_bridge
goes away any dependent cxl_nvdimm objects are shutdown. Handle that in
unregister_nvdimm_bus().

With these registration entanglements removed there is no longer a need
to pre-load the cxl_pmem module in cxl_acpi.

Fixes: cb9cfff82f6a ("cxl/acpi: Simplify cxl_nvdimm_bridge probing")
Reported-by: Gregory Price <gregory.price@memverge.com>
Debugged-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Tested-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/167426077263.3955046.9695309346988027311.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/port: Link the 'parent_dport' in portX/ and endpointX/ sysfs
Dan Williams [Wed, 25 Jan 2023 23:32:57 +0000 (15:32 -0800)]
cxl/port: Link the 'parent_dport' in portX/ and endpointX/ sysfs

Similar to the justification in:

1b58b4cac6fc ("cxl/port: Record parent dport when adding ports")

...userspace wants to know the routing information for ports for
calculating the memdev order for region creation among other things.
Cache the information the kernel discovers at enumeration time in a
'parent_dport' attribute to save userspace the time of trawling sysfs
to recover the same information.

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/167124082375.1626103.6047000000121298560.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agocxl/region: Clarify when a cxld->commit() callback is mandatory
Dan Williams [Sat, 17 Dec 2022 01:33:38 +0000 (17:33 -0800)]
cxl/region: Clarify when a cxld->commit() callback is mandatory

Both cxl_switch_decoders() and cxl_endpoint_decoders() are considered by
cxl_region_decode_commit(). Flag cases where cxl_switch_decoders with
multiple targets, or cxl_endpoint_decoders do not have a commit callback
set. The switch case is unlikely to happen since switches are only
enumerated by the CXL core, but the endpoint case may support decoders
defined by drivers outside of drivers/cxl, like accerator drivers.

Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/167124081824.1626103.1555704405392757219.stgit@dwillia2-xfh.jf.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
21 months agotools/testing/cxl: require 64-bit
Luis Chamberlain [Mon, 19 Dec 2022 19:50:50 +0000 (11:50 -0800)]
tools/testing/cxl: require 64-bit

size_t is limited to 32-bits and so the gen_pool_alloc() using
the size of SZ_64G would map to 0, triggering a low allocation
which is not expected. Force the dependency on 64-bit for cxl_test
as that is what it was designed for.

This issue was found by build test reports when converting this
driver as a proper upstream driver.

Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
Link: https://lore.kernel.org/r/20221219195050.325959-1-mcgrof@kernel.org
Signed-off-by: Dan Williams <dan.j.williams@intel.com>