The fact that blk_mq_destroy_queue also drops a queue reference leads
to various places having to grab an extra reference.  Move the call to
blk_put_queue into the callers to allow removing the extra references.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NSagi Grimberg <sagi@grimberg.me>
Reviewed-by: NChaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: NKeith Busch <kbusch@kernel.org>
Link: https://lore.kernel.org/r/20221018135720.670094-2-hch@lst.de
[axboe: fix fabrics_q vs admin_q conflict in nvme core.c]
Signed-off-by: NJens Axboe <axboe@kernel.dk>

This reverts commit 72a95859.

It broke reboots on big-endian MIPS and MIPS64 malta QEMU instances,
which use the syscon driver.  Little-endian is not effected, which means
likely it's important to handle regmap_get_val_endian() in this function
after all.

Fixes: 72a95859 ("mfd: syscon: Remove repetition of the regmap_get_val_endian()")
Reviewed-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Lee Jones <lee@kernel.org>
Signed-off-by: NJason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Also update the documentation accordingly.
Signed-off-by: NWilken Gottwalt <wilken.gottwalt@posteo.net>
Link: https://lore.kernel.org/r/Y0FghqQCHG/cX5Jz@monster.localdomainSigned-off-by: NGuenter Roeck <linux@roeck-us.net>

The generic EFI stub can be instructed to avoid SetVirtualAddressMap(),
and simply run with the firmware's 1:1 mapping. In this case, it
populates the virtual address fields of the runtime regions in the
memory map with the physical address of each region, so that the mapping
code has to be none the wiser. Only if SetVirtualAddressMap() fails, the
virtual addresses are wiped and the kernel code knows that the regions
cannot be mapped.

However, wiping amounts to setting it to zero, and if a runtime region
happens to live at physical address 0, its valid 1:1 mapped virtual
address could be mistaken for a wiped field, resulting on loss of access
to the EFI services at runtime.

So let's only assume that VA == 0 means 'no runtime services' if the
region in question does not live at PA 0x0.
Signed-off-by: NArd Biesheuvel <ardb@kernel.org>

The linker script symbol definition that captures the size of the
compressed payload inside the zboot decompressor (which is exposed via
the image header) refers to '.' for the end of the region, which does
not give the correct result as the expression is not placed at the end
of the payload. So use the symbol name explicitly.
Signed-off-by: NArd Biesheuvel <ardb@kernel.org>

To stop the bots from sending sparse warnings to me and the list about
efi_main() not having a prototype, decorate it with asmlinkage so that
it is clear that it is called from assembly, and therefore needs to
remain external, even if it is never declared in a header file.
Signed-off-by: NArd Biesheuvel <ardb@kernel.org>

Commit bbc6d2c6 ("efi: vars: Switch to new wrapper layer")
refactored the efivars layer so that the 'business logic' related to
which UEFI variables affect the boot flow in which way could be moved
out of it, and into the efivarfs driver.

This inadvertently broke setting variables on firmware implementations
that lack the QueryVariableInfo() boot service, because we no longer
tolerate a EFI_UNSUPPORTED result from check_var_size() when calling
efivar_entry_set_get_size(), which now ends up calling check_var_size()
a second time inadvertently.

If QueryVariableInfo() is missing, we support writes of up to 64k -
let's move that logic into check_var_size(), and drop the redundant
call.

Cc: <stable@vger.kernel.org> # v6.0
Fixes: bbc6d2c6 ("efi: vars: Switch to new wrapper layer")
Signed-off-by: NArd Biesheuvel <ardb@kernel.org>

Amadeusz reports KASAN use-after-free errors introduced by commit
3881ee0b ("efi: avoid efivars layer when loading SSDTs from
variables"). The problem appears to be that the memory that holds the
new ACPI table is now freed unconditionally, instead of only when the
ACPI core reported a failure to load the table.

So let's fix this, by omitting the kfree() on success.

Cc: <stable@vger.kernel.org> # v6.0
Link: https://lore.kernel.org/all/a101a10a-4fbb-5fae-2e3c-76cf96ed8fbd@linux.intel.com/
Fixes: 3881ee0b ("efi: avoid efivars layer when loading SSDTs from variables")
Reported-by: NAmadeusz Sławiński <amadeuszx.slawinski@linux.intel.com>
Signed-off-by: NArd Biesheuvel <ardb@kernel.org>

The zboot decompressor series introduced a feature to sign the PE/COFF
kernel image for secure boot as part of the kernel build. This was
necessary because there are actually two images that need to be signed:
the kernel with the EFI stub attached, and the decompressor application.

This is a bit of a burden, because it means that the images must be
signed on the the same system that performs the build, and this is not
realistic for distros.

During the next cycle, we will introduce changes to the zboot code so
that the inner image no longer needs to be signed. This means that the
outer PE/COFF image can be handled as usual, and be signed later in the
release process.

Let's remove the associated Kconfig options now so that they don't end
up in a LTS release while already being deprecated.
Signed-off-by: NArd Biesheuvel <ardb@kernel.org>

A splat from kmem_cache_destroy() was seen with a kernel prior to
commit ee2653bb ("iommu/vt-d: Remove domain and devinfo mempool")
when there was a failure in init_dmars(), because the iommu_domain
cache still had objects. While the mempool code is now gone, there
still is a leak of the si_domain memory if init_dmars() fails. So
clean up si_domain in the init_dmars() error path.

Cc: Lu Baolu <baolu.lu@linux.intel.com>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: Will Deacon <will@kernel.org>
Cc: Robin Murphy <robin.murphy@arm.com>
Fixes: 86080ccc ("iommu/vt-d: Allocate si_domain in init_dmars()")
Signed-off-by: NJerry Snitselaar <jsnitsel@redhat.com>
Link: https://lore.kernel.org/r/20221010144842.308890-1-jsnitsel@redhat.comSigned-off-by: NLu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: NJoerg Roedel <jroedel@suse.de>

Commit 5f64ce54 ("iommu/vt-d: Duplicate iommu_resv_region objects
per device list") converted rcu_lock in get_resv_regions to
dmar_global_lock to allow sleeping in iommu_alloc_resv_region(). This
introduced possible recursive locking if get_resv_regions is called from
within a section where intel_iommu_init() already holds dmar_global_lock.

Especially, after commit 57365a04 ("iommu: Move bus setup to IOMMU
device registration"), below lockdep splats could always be seen.

 ============================================
 WARNING: possible recursive locking detected
 6.0.0-rc4+ #325 Tainted: G          I
 --------------------------------------------
 swapper/0/1 is trying to acquire lock:
 ffffffffa8a18c90 (dmar_global_lock){++++}-{3:3}, at:
 intel_iommu_get_resv_regions+0x25/0x270

 but task is already holding lock:
 ffffffffa8a18c90 (dmar_global_lock){++++}-{3:3}, at:
 intel_iommu_init+0x36d/0x6ea

 ...

 Call Trace:
  <TASK>
  dump_stack_lvl+0x48/0x5f
  __lock_acquire.cold.73+0xad/0x2bb
  lock_acquire+0xc2/0x2e0
  ? intel_iommu_get_resv_regions+0x25/0x270
  ? lock_is_held_type+0x9d/0x110
  down_read+0x42/0x150
  ? intel_iommu_get_resv_regions+0x25/0x270
  intel_iommu_get_resv_regions+0x25/0x270
  iommu_create_device_direct_mappings.isra.28+0x8d/0x1c0
  ? iommu_get_dma_cookie+0x6d/0x90
  bus_iommu_probe+0x19f/0x2e0
  iommu_device_register+0xd4/0x130
  intel_iommu_init+0x3e1/0x6ea
  ? iommu_setup+0x289/0x289
  ? rdinit_setup+0x34/0x34
  pci_iommu_init+0x12/0x3a
  do_one_initcall+0x65/0x320
  ? rdinit_setup+0x34/0x34
  ? rcu_read_lock_sched_held+0x5a/0x80
  kernel_init_freeable+0x28a/0x2f3
  ? rest_init+0x1b0/0x1b0
  kernel_init+0x1a/0x130
  ret_from_fork+0x1f/0x30
  </TASK>

This rolls back dmar_global_lock to rcu_lock in get_resv_regions to avoid
the lockdep splat.

Fixes: 57365a04 ("iommu: Move bus setup to IOMMU device registration")
Signed-off-by: NLu Baolu <baolu.lu@linux.intel.com>
Tested-by: NAlex Williamson <alex.williamson@redhat.com>
Link: https://lore.kernel.org/r/20220927053109.4053662-3-baolu.lu@linux.intel.comSigned-off-by: NJoerg Roedel <jroedel@suse.de>

Add gfp parameter to iommu_alloc_resv_region() for the callers to specify
the memory allocation behavior. Thus iommu_alloc_resv_region() could also
be available in critical contexts.
Signed-off-by: NLu Baolu <baolu.lu@linux.intel.com>
Tested-by: NAlex Williamson <alex.williamson@redhat.com>
Link: https://lore.kernel.org/r/20220927053109.4053662-2-baolu.lu@linux.intel.comSigned-off-by: NJoerg Roedel <jroedel@suse.de>

This fixes maybe_unused warnings/errors.

According to a comment during device tree removal, only ACPI is supported,
thus let's actually require it.

Fixes: be18c5ed ("i2c: mlxbf: remove device tree support")
Signed-off-by: NAdam Borowski <kilobyte@angband.pl>
Signed-off-by: NWolfram Sang <wsa@kernel.org>

Commit 16ce101d ("mm/memory.c: fix race when faulting a device private
page") changed the migrate_to_ram() callback to take a reference on the
device page to ensure it can't be freed while handling the fault. 
Unfortunately the corresponding update to Nouveau to accommodate this
change was inadvertently dropped from that patch causing GPU to CPU
migration to fail so add it here.

Link: https://lkml.kernel.org/r/20221019122934.866205-1-apopple@nvidia.com
Fixes: 16ce101d ("mm/memory.c: fix race when faulting a device private page")
Signed-off-by: NAlistair Popple <apopple@nvidia.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Cc: Lyude Paul <lyude@redhat.com>
Cc: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>

Reported-by: Nk2ci <kernel-bot@kylinos.cn>
Signed-off-by: NJiangshan Yi <yijiangshan@kylinos.cn>
Reviewed-by: Jean Delvare <jdelvare@suse.de> # for sis630
Signed-off-by: NWolfram Sang <wsa@kernel.org>

Commit 8795e182 ("PCI/portdrv: Don't disable AER reporting in get_port_device_capability()")
uncovered a bug in amdgpu that required a reordering of the driver
init sequence to avoid accessing a special register on the GPU
before it was properly set up leading to an PCI AER error.  This
reordering uncovered a different hw programming ordering dependency
in some APUs where the SDMA doorbells need to be programmed before
the GFX doorbells. To fix this, move the SDMA doorbell programming
back into the soc15 common code, but use the actual doorbell range
values directly rather than the values stored in the ring structure
since those will not be initialized at this point.

This is a partial revert, but with the doorbell assignment
fixed so the proper doorbell index is set before it's used.

Fixes: e3163bc8 ("drm/amdgpu: move nbio sdma_doorbell_range() into sdma code for vega")
Acked-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: skhan@linuxfoundation.org
Cc: stable@vger.kernel.org

Commit c347a787 (drbd: set ->bi_bdev in drbd_req_new) moved a
bio_set_dev call (which has since been removed) to "earlier", from
drbd_request_prepare to drbd_req_new.

The problem is that this accesses device->ldev->backing_bdev, which is
not NULL-checked at this point. When we don't have an ldev (i.e. when
the DRBD device is diskless), this leads to a null pointer deref.

So, only allocate the private_bio if we actually have a disk. This is
also a small optimization, since we don't clone the bio to only to
immediately free it again in the diskless case.

Fixes: c347a787 ("drbd: set ->bi_bdev in drbd_req_new")
Co-developed-by: NChristoph Böhmwalder <christoph.boehmwalder@linbit.com>
Signed-off-by: NChristoph Böhmwalder <christoph.boehmwalder@linbit.com>
Co-developed-by: NJoel Colledge <joel.colledge@linbit.com>
Signed-off-by: NJoel Colledge <joel.colledge@linbit.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20221020085205.129090-1-christoph.boehmwalder@linbit.comSigned-off-by: NJens Axboe <axboe@kernel.dk>

__le32 and __le64 types aren't portable and are not available on
FreeBSD (which uses the same uAPI).

Instead of attempting to always output little endian, just use native
endianness in the dumps. Tools can detect the endianness in use by
looking at the 'magic' field, but equally we don't expect big-endian to
be used with Mali (there are no known implementations out there).

Bug: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7252
Fixes: 730c2bf4 ("drm/panfrost: Add support for devcoredump")
Reviewed-by: NAlyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Signed-off-by: NSteven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221017104602.142992-3-steven.price@arm.com

Commit 35a3b82f ("drm/connector: Introduce drmm_connector_init")
introduced the function drmm_connector_init() with a parameter for an
optional ddc pointer to the i2c controller used to access the DDC bus.

However, the underlying call to __drm_connector_init() was always
setting it to NULL instead of passing the ddc argument around.

This resulted in unexpected null pointer dereference on platforms
expecting to get a DDC controller.

Fixes: 35a3b82f ("drm/connector: Introduce drmm_connector_init")
Reviewed-by: NThomas Zimmermann <tzimmermann@suse.de>
Link: https://lore.kernel.org/r/20221019143442.1798964-1-maxime@cerno.techSigned-off-by: NMaxime Ripard <maxime@cerno.tech>

The xrgb2101010 format conversion test (unlike for other formats) does
an endianness conversion on the results. However, it always converts
TEST_BUF_SIZE 32-bit integers, which results in reading from (and
writing to) more memory than in present in the result buffer. Instead,
use the buffer size, divided by sizeof(u32).

The issue could be reproduced with KASAN:
./tools/testing/kunit/kunit.py run --kunitconfig drivers/gpu/drm/tests \
	--kconfig_add CONFIG_KASAN=y --kconfig_add CONFIG_KASAN_VMALLOC=y \
	--kconfig_add CONFIG_KASAN_KUNIT_TEST=y \
	drm_format_helper_test.*xrgb2101010
Reported-by: NLinux Kernel Functional Testing <lkft@linaro.org>
Fixes: 45311431 ("drm/format-helper: Add KUnit tests for drm_fb_xrgb8888_to_xrgb2101010()")
Signed-off-by: NDavid Gow <davidgow@google.com>
Reviewed-by: NMaíra Canal <mairacanal@riseup.net>
Reviewed-by: NJavier Martinez Canillas <javierm@redhat.com>
Reviewed-by: NJosé Expósito <jose.exposito89@gmail.com>
Signed-off-by: NJavier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221019073239.3779180-1-davidgow@google.com

If the cable is disconnected the PHY seems to toggle between MDI and
MDI-X modes. With the MDI crossover status interrupt active this causes
roughly 10 interrupts per second.

As the crossover status isn't checked by the driver, the interrupt can
be disabled to reduce the interrupt load.

Fixes: 87461f7a ("net: phy: DP83822 initial driver submission")
Signed-off-by: NFelix Riemann <felix.riemann@sma.de>
Reviewed-by: NAndrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20221018104755.30025-1-svc.sw.rte.linux@sma.deSigned-off-by: NJakub Kicinski <kuba@kernel.org>

Eliminate the following coccicheck warning:
./drivers/block/ublk_drv.c:127:16-19: WARNING use flexible-array member instead
Signed-off-by: NYushan Zhou <katrinzhou@tencent.com>
Link: https://lore.kernel.org/r/20221018100132.355393-1-zys.zljxml@gmail.comReviewed-by: NMing Lei <ming.lei@redhat.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>

Inject fault while probing module, if device_register() fails,
but the refcount of kobject is not decreased to 0, the name
allocated in dev_set_name() is leaked. Fix this by calling
put_device(), so that name can be freed in callback function
kobject_cleanup().

unreferenced object 0xffff00c01aba2100 (size 128):
  comm "systemd-udevd", pid 1259, jiffies 4294903284 (age 294.152s)
  hex dump (first 32 bytes):
    68 6e 61 65 30 00 00 00 18 21 ba 1a c0 00 ff ff  hnae0....!......
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace:
    [<0000000034783f26>] slab_post_alloc_hook+0xa0/0x3e0
    [<00000000748188f2>] __kmem_cache_alloc_node+0x164/0x2b0
    [<00000000ab0743e8>] __kmalloc_node_track_caller+0x6c/0x390
    [<000000006c0ffb13>] kvasprintf+0x8c/0x118
    [<00000000fa27bfe1>] kvasprintf_const+0x60/0xc8
    [<0000000083e10ed7>] kobject_set_name_vargs+0x3c/0xc0
    [<000000000b87affc>] dev_set_name+0x7c/0xa0
    [<000000003fd8fe26>] hnae_ae_register+0xcc/0x190 [hnae]
    [<00000000fe97edc9>] hns_dsaf_ae_init+0x9c/0x108 [hns_dsaf]
    [<00000000c36ff1eb>] hns_dsaf_probe+0x548/0x748 [hns_dsaf]

Fixes: 6fe6611f ("net: add Hisilicon Network Subsystem hnae framework support")
Signed-off-by: NYang Yingliang <yangyingliang@huawei.com>
Reviewed-by: NLeon Romanovsky <leonro@nvidia.com>
Link: https://lore.kernel.org/r/20221018122451.1749171-1-yangyingliang@huawei.comSigned-off-by: NJakub Kicinski <kuba@kernel.org>

Inject fault while probing module, if device_register() fails,
but the refcount of kobject is not decreased to 0, the name
allocated in dev_set_name() is leaked. Fix this by calling
put_device(), so that name can be freed in callback function
kobject_cleanup().

unreferenced object 0xffff88810152ad20 (size 8):
  comm "modprobe", pid 252, jiffies 4294849206 (age 22.713s)
  hex dump (first 8 bytes):
    68 77 73 69 6d 30 00 ff                          hwsim0..
  backtrace:
    [<000000009c3504ed>] __kmalloc_node_track_caller+0x44/0x1b0
    [<00000000c0228a5e>] kvasprintf+0xb5/0x140
    [<00000000cff8c21f>] kvasprintf_const+0x55/0x180
    [<0000000055a1e073>] kobject_set_name_vargs+0x56/0x150
    [<000000000a80b139>] dev_set_name+0xab/0xe0

Fixes: f36a111a ("wwan_hwsim: WWAN device simulator")
Signed-off-by: NYang Yingliang <yangyingliang@huawei.com>
Reviewed-by: NLoic Poulain <loic.poulain@linaro.org>
Acked-by: NSergey Ryazanov <ryazanov.s.a@gmail.com>
Link: https://lore.kernel.org/r/20221018131607.1901641-1-yangyingliang@huawei.comSigned-off-by: NJakub Kicinski <kuba@kernel.org>

Filters on different vports are qualified by different implicit MACs and/or
VLANs, so shouldn't be considered equal even if their other match fields
are identical.

Fixes: 7c460d9b ("sfc: Extend and abstract efx_filter_spec to cover Huntington/EF10")
Co-developed-by: NEdward Cree <ecree.xilinx@gmail.com>
Signed-off-by: NEdward Cree <ecree.xilinx@gmail.com>
Signed-off-by: NPieter Jansen van Vuuren <pieter.jansen-van-vuuren@amd.com>
Reviewed-by: NMartin Habets <habetsm.xilinx@gmail.com>
Link: https://lore.kernel.org/r/20221018092841.32206-1-pieter.jansen-van-vuuren@amd.comSigned-off-by: NJakub Kicinski <kuba@kernel.org>

When we compile-in the CCI along with the imx412 driver and run on the RB5
we see that i2c_add_adapter() causes the probe of the imx412 driver to
happen.

This probe tries to perform an i2c xfer() and the xfer() in i2c-qcom-cci.c
fails on pm_runtime_get() because the i2c-qcom-cci.c::probe() function has
not completed to pm_runtime_enable(dev).

Fix this sequence by ensuring pm_runtime_xxx() calls happen prior to adding
the i2c adapter.

Fixes: e5175261 ("i2c: Add Qualcomm CCI I2C driver")
Reported-by: NVladimir Zapolskiy <vladimir.zapolskiy@linaro.org>
Reviewed-by: NVladimir Zapolskiy <vladimir.zapolskiy@linaro.org>
Tested-by: NVladimir Zapolskiy <vladimir.zapolskiy@linaro.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: NBryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: NRobert Foss <robert.foss@linaro.org>
Signed-off-by: NWolfram Sang <wsa@kernel.org>

The missing "platform" alias is required for the mgb4 v4l2 driver to load
the i2c controller driver when probing the HW.
Signed-off-by: NMartin Tůma <martin.tuma@digiteqautomotive.com>
Acked-by: NMichal Simek <michal.simek@amd.com>
Signed-off-by: NWolfram Sang <wsa@kernel.org>

Assigning the device's dma_range_map from the iterator variable after
the loop means it always points to the empty terminator at the end of
the map, which is not what we want. Similarly, freeing the iterator on
error when it points to somwhere in the middle of the allocated array
won't work either. Fix this.

Fixes: bf2ee8d0 ("ACPI: scan: Support multiple DMA windows with different offsets")
Signed-off-by: NRobin Murphy <robin.murphy@arm.com>
Reviewed-by: NJianmin Lv <lvjianmin@loongson.cn>
Tested-by: NJeremy Linton <jeremy.linton@arm.com>
Tested-by: NYicong Yang <yangyicong@hisilicon.com>
Reviewed-by: NLorenzo Pieralisi <lpieralisi@kernel.org>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

fix rxsc and txsc not getting freed before going out of scope

Fixes: c54ffc73 ("octeontx2-pf: mcs: Introduce MACSEC hardware offloading")
Signed-off-by: NManank Patel <pmanank200502@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit 63f534b8 ("ACPI: PCI: Rework acpi_get_pci_dev()") failed
to reference count the device returned by acpi_get_pci_dev() as
expected by its callers which in some cases may cause device objects
to be dropped prematurely.

Add the missing get_device() to acpi_get_pci_dev().

Fixes: 63f534b8 ("ACPI: PCI: Rework acpi_get_pci_dev()")
Reported-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

Make sure that we always have a CPU round trip to let the submission
code correctly decide if a TLB flush is necessary or not.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2113#note_1579296Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Acked-by: NLuben Tuikov <luben.tuikov@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221014081553.114899-2-christian.koenig@amd.com

The item passed into nvmet_subsys_attr_qid_max_show is not a member of
struct nvmet_port, it is part of nvmet_subsys.  Hence, don't try to
dereference it as struct nvme_ctrl pointer.

Fixes: 3e980f59 ("nvmet: Expose max queues to configfs")
Reported-by: NShinichiro Kawasaki <shinichiro.kawasaki@wdc.com>
Link: https://lore.kernel.org/r/20220913064203.133536-1-dwagner@suse.deSigned-off-by: NDaniel Wagner <dwagner@suse.de>
Reviewed-by: NHannes Reinecke <hare@suse.de>
Acked-by: NSagi Grimberg <sagi@grimberg.me>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

The keep alive timer needs to stay on nvmet_wq, and not
modified to reschedule on the system_wq.

This fixes a warning:
------------[ cut here ]------------
workqueue: WQ_MEM_RECLAIM
nvmet-wq:nvmet_rdma_release_queue_work [nvmet_rdma] is flushing
!WQ_MEM_RECLAIM events:nvmet_keep_alive_timer [nvmet]
WARNING: CPU: 3 PID: 1086 at kernel/workqueue.c:2628
check_flush_dependency+0x16c/0x1e0
Reported-by: NYi Zhang <yi.zhang@redhat.com>
Fixes: 8832cf92 ("nvmet: use a private workqueue instead of the system workqueue")
Signed-off-by: NSagi Grimberg <sagi@grimberg.me>
Reviewed-by: NChaitanya Kulkarni <kch@nvidia.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

Recent commit 52fde2c0 ("nvme: set dma alignment to dword") has
caused a regression on our platform.

It turned out that the nvme_get_log() method invocation caused the
nvme_hwmon_data structure instance corruption.  In particular the
nvme_hwmon_data.ctrl pointer was overwritten either with zeros or with
garbage.  After some research we discovered that the problem happened
even before the actual NVME DMA execution, but during the buffer mapping.
Since our platform is DMA-noncoherent, the mapping implied the cache-line
invalidations or write-backs depending on the DMA-direction parameter.
In case of the NVME SMART log getting the DMA was performed
from-device-to-memory, thus the cache-invalidation was activated during
the buffer mapping.  Since the log-buffer isn't cache-line aligned, the
cache-invalidation caused the neighbour data to be discarded.  The
neighbouring data turned to be the data surrounding the buffer in the
framework of the nvme_hwmon_data structure.

In order to fix that we need to make sure that the whole log-buffer is
defined within the cache-line-aligned memory region so the
cache-invalidation procedure wouldn't involve the adjacent data. One of
the option to guarantee that is to kmalloc the DMA-buffer [1]. Seeing the
rest of the NVME core driver prefer that method it has been chosen to fix
this problem too.

Note after a deeper researches we found out that the denoted commit wasn't
a root cause of the problem. It just revealed the invalidity by activating
the DMA-based NVME SMART log getting performed in the framework of the
NVME hwmon driver. The problem was here since the initial commit of the
driver.

[1] Documentation/core-api/dma-api-howto.rst

Fixes: 400b6a7b ("nvme: Add hardware monitoring support")
Signed-off-by: NSerge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

An NVMe controller works perfectly fine even when the hwmon
initialization fails.  Stop returning errors that do not come from a
controller reset from nvme_hwmon_init to handle this case consistently.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Reviewed-by: NSerge Semin <fancer.lancer@gmail.com>

Setting this flag on a scheduler fence prevents pipelining of jobs
depending on this fence. In other words we always insert a full CPU
round trip before dependent jobs are pushed to the pipeline.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2113#note_1579296Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Acked-by: NLuben Tuikov <luben.tuikov@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221014081553.114899-1-christian.koenig@amd.com

NVMe uses PRPs for data transfers and has no specific limit for a single
DMA segement.  Limiting the size will cause problems because the block
layer assumes PRP-ish devices using a virt boundary mask don't have a
segment limit.  And while this is true, we also really need to tell the
DMA mapping layer about it, otherwise dma-debug will trip over it.

Fixes: 5bd2927a ("nvme-apple: Add initial Apple SoC NVMe driver")
Suggested-by: NSven Peter <sven@svenpeter.dev>
Signed-off-by: NRussell King (Oracle) <rmk+kernel@armlinux.org.uk>
[hch: rewrote the commit message based on the PCIe commit]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Reviewed-by: NEric Curtin <ecurtin@redhat.com>
Reviewed-by: NSven Peter <sven@svenpeter.dev>

Kingston SSDs do support NVMe Write_Zeroes cmd but take long time to
process.  The firmware version is locked by these SSDs, we can not expect
firmware improvement, so disable Write_Zeroes cmd.
Signed-off-by: NXander Li <xander_li@kingston.com.tw>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

There is typo here so it releases the wrong variable.  "ctrl->admin_q"
was intended instead of "ctrl->fabrics_q".

Fixes: fe60e8c5 ("nvme: add common helpers to allocate and free tagsets")
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: NChaitanya Kulkarni <kch@nvidia.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>

It's possible that the driver will dereference a qcq that doesn't exist
when calling ionic_reconfigure_queues(), which causes a page fault BUG.

If a reduction in the number of queues is followed by a different
reconfig such as changing the ring size, the driver can hit a NULL
pointer when trying to clean up non-existent queues.

Fix this by checking to make sure both the qcqs array and qcq entry
exists bofore trying to use and free the entry.

Fixes: 101b40a0 ("ionic: change queue count with no reset")
Signed-off-by: NBrett Creeley <brett@pensando.io>
Signed-off-by: NShannon Nelson <snelson@pensando.io>
Link: https://lore.kernel.org/r/20221017233123.15869-1-snelson@pensando.ioSigned-off-by: NJakub Kicinski <kuba@kernel.org>