| Age | Commit message (Collapse) | Author |
|---|
|
This patch addresses a race condition for an ODP MR that can result in a
CQE with an error on the UMR QP.
During the __mlx5_ib_dereg_mr() flow, the following sequence of calls
occurs:
mlx5_revoke_mr()
mlx5r_umr_revoke_mr()
mlx5r_umr_post_send_wait()
At this point, the lkey is freed from the hardware's perspective.
However, concurrently, mlx5_ib_invalidate_range() might be triggered by
another task attempting to invalidate a range for the same freed lkey.
This task will:
- Acquire the umem_odp->umem_mutex lock.
- Call mlx5r_umr_update_xlt() on the UMR QP.
- Since the lkey has already been freed, this can lead to a CQE error,
causing the UMR QP to enter an error state [1].
To resolve this race condition, the umem_odp->umem_mutex lock is now also
acquired as part of the mlx5_revoke_mr() scope. Upon successful revoke,
we set umem_odp->private which points to that MR to NULL, preventing any
further invalidation attempts on its lkey.
[1] From dmesg:
infiniband rocep8s0f0: dump_cqe:277:(pid 0): WC error: 6, Message: memory bind operation error
cqe_dump: 00000000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
cqe_dump: 00000010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
cqe_dump: 00000020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
cqe_dump: 00000030: 00 00 00 00 08 00 78 06 25 00 11 b9 00 0e dd d2
WARNING: CPU: 15 PID: 1506 at drivers/infiniband/hw/mlx5/umr.c:394 mlx5r_umr_post_send_wait+0x15a/0x2b0 [mlx5_ib]
Modules linked in: ip6table_mangle ip6table_natip6table_filter ip6_tables iptable_mangle xt_conntrack xt_MASQUERADE nf_conntrack_netlink nfnetlink xt_addrtype iptable_nat nf_nat br_netfilter rpcsec_gss_krb5 auth_rpcgss oid_registry overlay rpcrdma rdma_ucm ib_iser libiscsi scsi_transport_iscsi rdma_cm iw_cm ib_umad ib_ipoib ib_cm mlx5_ib ib_uverbs ib_core fuse mlx5_core
CPU: 15 UID: 0 PID: 1506 Comm: ibv_rc_pingpong Not tainted 6.12.0-rc7+ #1626
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014
RIP: 0010:mlx5r_umr_post_send_wait+0x15a/0x2b0 [mlx5_ib]
[..]
Call Trace:
<TASK>
mlx5r_umr_update_xlt+0x23c/0x3e0 [mlx5_ib]
mlx5_ib_invalidate_range+0x2e1/0x330 [mlx5_ib]
__mmu_notifier_invalidate_range_start+0x1e1/0x240
zap_page_range_single+0xf1/0x1a0
madvise_vma_behavior+0x677/0x6e0
do_madvise+0x1a2/0x4b0
__x64_sys_madvise+0x25/0x30
do_syscall_64+0x6b/0x140
entry_SYSCALL_64_after_hwframe+0x76/0x7e
Fixes: e6fb246ccafb ("RDMA/mlx5: Consolidate MR destruction to mlx5_ib_dereg_mr()")
Cc: stable@vger.kernel.org
Link: https://patch.msgid.link/r/68a1e007c25b2b8fe5d625f238cc3b63e5341f77.1737290229.git.leon@kernel.org
Signed-off-by: Yishai Hadas <yishaih@nvidia.com>
Reviewed-by: Artemy Kovalyov <artemyko@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
|
|
We don't need all the helpers inlined here.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
|
|
The iommu_hwpt_pgfault is used to report IO page fault data to userspace,
but iommufd_fault_fops_read was never zeroing its padding. This leaks the
content of the kernel stack memory to userspace.
Also, the iommufd uAPI requires explicit padding and use of __aligned_u64
to ensure ABI compatibility's with 32 bit.
pahole result, before:
struct iommu_hwpt_pgfault {
__u32 flags; /* 0 4 */
__u32 dev_id; /* 4 4 */
__u32 pasid; /* 8 4 */
__u32 grpid; /* 12 4 */
__u32 perm; /* 16 4 */
/* XXX 4 bytes hole, try to pack */
__u64 addr; /* 24 8 */
__u32 length; /* 32 4 */
__u32 cookie; /* 36 4 */
/* size: 40, cachelines: 1, members: 8 */
/* sum members: 36, holes: 1, sum holes: 4 */
/* last cacheline: 40 bytes */
};
pahole result, after:
struct iommu_hwpt_pgfault {
__u32 flags; /* 0 4 */
__u32 dev_id; /* 4 4 */
__u32 pasid; /* 8 4 */
__u32 grpid; /* 12 4 */
__u32 perm; /* 16 4 */
__u32 __reserved; /* 20 4 */
__u64 addr __attribute__((__aligned__(8))); /* 24 8 */
__u32 length; /* 32 4 */
__u32 cookie; /* 36 4 */
/* size: 40, cachelines: 1, members: 9 */
/* forced alignments: 1 */
/* last cacheline: 40 bytes */
} __attribute__((__aligned__(8)));
Fixes: c714f15860fc ("iommufd: Add fault and response message definitions")
Link: https://patch.msgid.link/r/20250120195051.2450-1-nicolinc@nvidia.com
Cc: stable@vger.kernel.org
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Nicolin Chen <nicolinc@nvidia.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull misc x86 updates from Borislav Petkov:
- The first part of a restructuring of AMD's representation of a
northbridge which is legacy now, and the creation of the new AMD node
concept which represents the Zen architecture of having a collection
of I/O devices within an SoC. Those nodes comprise the so-called data
fabric on Zen.
This has at least one practical advantage of not having to add a PCI
ID each time a new data fabric PCI device releases. Eventually, the
lot more uniform provider of data fabric functionality amd_node.c
will be used by all the drivers which need it
- Smaller cleanups
* tag 'x86_misc_for_v6.14_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/amd_node: Use defines for SMN register offsets
x86/amd_node: Remove dependency on AMD_NB
x86/amd_node: Update __amd_smn_rw() error paths
x86/amd_nb: Move SMN access code to a new amd_node driver
x86/amd_nb, hwmon: (k10temp): Simplify amd_pci_dev_to_node_id()
x86/amd_nb: Simplify function 3 search
x86/amd_nb: Use topology info to get AMD node count
x86/amd_nb: Simplify root device search
x86/amd_nb: Simplify function 4 search
x86: Start moving AMD node functionality out of AMD_NB
x86/amd_nb: Clean up early_is_amd_nb()
x86/amd_nb: Restrict init function to AMD-based systems
x86/mtrr: Rename mtrr_overwrite_state() to guest_force_mtrr_state()
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 cpuid updates from Borislav Petkov:
- Remove the less generic CPU matching infra around struct x86_cpu_desc
and use the generic struct x86_cpu_id thing
- Remove magic naked numbers for CPUID functions and use proper defines
of the prefix CPUID_LEAF_*. Consolidate some of the crazy use around
the tree
- Smaller cleanups and improvements
* tag 'x86_cpu_for_v6.14_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/cpu: Make all all CPUID leaf names consistent
x86/fpu: Remove unnecessary CPUID level check
x86/fpu: Move CPUID leaf definitions to common code
x86/tsc: Remove CPUID "frequency" leaf magic numbers.
x86/tsc: Move away from TSC leaf magic numbers
x86/cpu: Move TSC CPUID leaf definition
x86/cpu: Refresh DCA leaf reading code
x86/cpu: Remove unnecessary MwAIT leaf checks
x86/cpu: Use MWAIT leaf definition
x86/cpu: Move MWAIT leaf definition to common header
x86/cpu: Remove 'x86_cpu_desc' infrastructure
x86/cpu: Move AMD erratum 1386 table over to 'x86_cpu_id'
x86/cpu: Replace PEBS use of 'x86_cpu_desc' use with 'x86_cpu_id'
x86/cpu: Expose only stepping min/max interface
x86/cpu: Introduce new microcode matching helper
x86/cpufeature: Document cpu_feature_enabled() as the default to use
x86/paravirt: Remove the WBINVD callback
x86/cpufeatures: Free up unused feature bits
|
|
BTREE_ITER_cached_nofill has some tricky corner cases; it's used
internally for iterators that aren't walking the key cache, but need to
be coherent with the key cache.
It tells traverse to look up and lock the key cache entry if present,
but don't create one if it doesn't exist.
That means we have to have a BTREE_ITER_UPTODATE path (because after
traverse the path has to be UPTODATE, or we pop assertions) that doesn't
point to anything (which is the less bad option, taken by the previous
fix).
The previous fix for this path missed an issue that can happen in
bch2_trans_peek_key_cache(): we can't set should_be_locked on a path
that doesn't point to anything and doesn't hold locks.
Fixes: bd5b09727f3d ("bcachefs: Don't set btree_path to updtodate if we don't fill")
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
|
|
Document the PCIe controller on the IPQ5424 platform using the
IPQ9574 bindings as a fallback, since the PCIe on the IPQ5424
is similar to IPQ9574.
Link: https://lore.kernel.org/r/20241213134950.234946-2-quic_mmanikan@quicinc.com
Signed-off-by: Manikanta Mylavarapu <quic_mmanikan@quicinc.com>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Acked-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
|
|
Qcom PCIe RC controllers are capable of generating 'global' SPI interrupt
to the host CPU. This interrupt can be used by the device driver to handle
PCIe link specific events such as Link up and Link down, which give the
driver a chance to start bus enumeration on its own when link is up and
initiate link training if link goes to a bad state. The PCIe driver can
still work without this interrupt but it will provide a nice user
experience when device gets plugged and removed.
Document the interrupt as optional for SM8550 and SM8650 platforms.
Link: https://lore.kernel.org/r/20241126-topic-sm8x50-pcie-global-irq-v1-1-4049cfccd073@linaro.org
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
|
|
Convert device tree binding document mobiveil-pcie.txt to YAML format
and merge layerscape-pcie-gen4.txt into this file.
Additional changes:
- interrupt-names: "aer", "pme", "intr", which align order in examples.
- reg-names: reorder as csr_axi_slave, config_axi_slave to match
layerscape-pcie-gen4 and existing Layerscape DTS users.
Fix below CHECK_DTBS warning:
arch/arm64/boot/dts/freescale/fsl-lx2160a-qds.dtb: /soc/pcie@3400000: failed to match any schema with compatible: ['fsl,lx2160a-pcie']
Link: https://lore.kernel.org/r/20241211171318.4129818-1-Frank.Li@nxp.com
Signed-off-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
[kwilczynski: commit log]
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
[bhelgaas: fix typos, whitespace, consistent bus-range usage]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
|
Fix a sparse warning due to the invalid return type from poll ops,
which is __poll_t.
Fixes: 46757a3e7d50 ("ALSA: FCP: Add Focusrite Control Protocol driver")
Link: https://patch.msgid.link/20250121170032.7236-4-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
|
|
Fix a wrong conversion macro used for resp->opcode, which is __le32.
Fixes: 46757a3e7d50 ("ALSA: FCP: Add Focusrite Control Protocol driver")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202501212331.SaePSmsA-lkp@intel.com/
Link: https://patch.msgid.link/20250121170032.7236-3-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
|
|
The cached level meter values are returned from the USB core as
__le32, hence declare properly.
Fixes: 46757a3e7d50 ("ALSA: FCP: Add Focusrite Control Protocol driver")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202501212331.SaePSmsA-lkp@intel.com/
Link: https://patch.msgid.link/20250121170032.7236-2-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 SEV updates from Borislav Petkov:
- A segmented Reverse Map table (RMP) is a across-nodes distributed
table of sorts which contains per-node descriptors of each node-local
4K page, denoting its ownership (hypervisor, guest, etc) in the realm
of confidential computing. Add support for such a table in order to
improve referential locality when accessing or modifying RMP table
entries
- Add support for reading the TSC in SNP guests by removing any
interference or influence the hypervisor might have, with the goal of
making a confidential guest even more independent from the hypervisor
* tag 'x86_sev_for_v6.14_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/sev: Add the Secure TSC feature for SNP guests
x86/tsc: Init the TSC for Secure TSC guests
x86/sev: Mark the TSC in a secure TSC guest as reliable
x86/sev: Prevent RDTSC/RDTSCP interception for Secure TSC enabled guests
x86/sev: Prevent GUEST_TSC_FREQ MSR interception for Secure TSC enabled guests
x86/sev: Change TSC MSR behavior for Secure TSC enabled guests
x86/sev: Add Secure TSC support for SNP guests
x86/sev: Relocate SNP guest messaging routines to common code
x86/sev: Carve out and export SNP guest messaging init routines
virt: sev-guest: Replace GFP_KERNEL_ACCOUNT with GFP_KERNEL
virt: sev-guest: Remove is_vmpck_empty() helper
x86/sev/docs: Document the SNP Reverse Map Table (RMP)
x86/sev: Add full support for a segmented RMP table
x86/sev: Treat the contiguous RMP table as a single RMP segment
x86/sev: Map only the RMP table entries instead of the full RMP range
x86/sev: Move the SNP probe routine out of the way
x86/sev: Require the RMPREAD instruction after Zen4
x86/sev: Add support for the RMPREAD instruction
x86/sev: Prepare for using the RMPREAD instruction to access the RMP
|
|
Add Microchip parts to the Device ID table so the driver supports PCI100x
devices.
Add a new macro to quirk the Microchip Switchtec PCI100x parts to allow DMA
access via NTB to work when the IOMMU is turned on.
PCI100x family has 6 variants; each variant is designed for different
application usages, different port counts and lane counts:
PCI1001 has 1 x4 upstream port and 3 x4 downstream ports
PCI1002 has 1 x4 upstream port and 4 x2 downstream ports
PCI1003 has 2 x4 upstream ports, 2 x2 upstream ports, and 2 x2
downstream ports
PCI1004 has 4 x4 upstream ports
PCI1005 has 1 x4 upstream port and 6 x2 downstream ports
PCI1006 has 6 x2 upstream ports and 2 x2 downstream ports
[Historical note: these parts use PCI_VENDOR_ID_EFAR (0x1055), from EFAR
Microsystems, which was acquired in 1996 by Standard Microsystems Corp,
which was acquired by Microchip Technology in 2012. The PCI-SIG confirms
that Vendor ID 0x1055 is assigned to Microchip even though it's not
visible via https://pcisig.com/membership/member-companies]
Link: https://lore.kernel.org/r/20250120095524.243103-1-Saladi.Rakeshbabu@microchip.com
Signed-off-by: Rakesh Babu Saladi <Saladi.Rakeshbabu@microchip.com>
[bhelgaas: Vendor ID history]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-By: Logan Gunthorpe <logang@deltatee.com>
|
|
LAYOUTSTATS and LAYOUTERROR should be marked MOVEABLE for when we
need to move tasks off a non-functional transport.
Signed-off-by: Olga Kornievskaia <okorniev@redhat.com>
Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>
|
|
OFFLOAD_CANCEL should be marked MOVEABLE for when we need to move
tasks off a non-functional transport.
Fixes: c975c2092657 ("NFS send OFFLOAD_CANCEL when COPY killed")
Signed-off-by: Olga Kornievskaia <okorniev@redhat.com>
Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>
|
|
We need to include sequence size in the compound.
Fixes: 0491567b51ef ("NFS: add COPY_NOTIFY operation")
Signed-off-by: Olga Kornievskaia <okorniev@redhat.com>
Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>
|
|
Refactor: This struct can be used unchanged for the new
OFFLOAD_STATUS implementation, so give it a more generic name.
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>
|
|
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>
|
|
RFC 7862 permits the callback service to respond to a CB_OFFLOAD
operation with NFS4ERR_DELAY. Use that instead of
NFS4ERR_SERVERFAULT for temporary memory allocation failure, as that
is more consistent with how other operations report memory
allocation failure.
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>
|
|
Having the NFS_FSCACHE option depend on the NETFS_SUPPORT options makes
selecting NFS_FSCACHE impossible unless another option that additionally
selects NETFS_SUPPORT is already selected.
As a result, for example, being able to reach and select the NFS_FSCACHE
option requires the CEPH_FS or CIFS option to be selected beforehand, which
obviously doesn't make much sense.
Let's correct this by making the NFS_FSCACHE option actually select the
NETFS_SUPPORT option, instead of depending on it.
Fixes: 915cd30cdea8 ("netfs, fscache: Combine fscache with netfs")
Cc: stable@vger.kernel.org
Reported-by: Diederik de Haas <didi.debian@cknow.org>
Signed-off-by: Dragan Simic <dsimic@manjaro.org>
Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>
|
|
nfs4_stat_to_errno() expects a NFSv4 error code as an argument and
returns a POSIX errno.
The problem is LOCALIO is passing nfs4_stat_to_errno() the POSIX errno
return values from filp->f_op->read_iter(), filp->f_op->write_iter()
and vfs_fsync_range().
So the POSIX errno that nfs_local_pgio_done() and
nfs_local_commit_done() are passing to nfs4_stat_to_errno() are
failing to match any NFSv4 error code, which results in
nfs4_stat_to_errno() defaulting to returning -EREMOTEIO. This causes
assertions in upper layers due to -EREMOTEIO not being a valid NFSv4
error code.
Fix this by updating nfs_local_pgio_done() and nfs_local_commit_done()
to use the new nfs_localio_errno_to_nfs4_stat() to map a POSIX errno
to an NFSv4 error code.
Care was taken to factor out nfs4_errtbl_common[] to avoid duplicating
the same NFS error to errno table. nfs4_errtbl_common[] is checked
first by both nfs4_stat_to_errno and nfs_localio_errno_to_nfs4_stat
before they check their own more specialized tables (nfs4_errtbl[] and
nfs4_errtbl_localio[] respectively).
While auditing the associated error mapping tables, the (ab)use of -1
for the last table entry was removed in favor of using ARRAY_SIZE to
iterate the nfs_errtbl[] and nfs4_errtbl[]. And 'errno_NFSERR_IO' was
removed because it caused needless obfuscation.
Fixes: 70ba381e1a431 ("nfs: add LOCALIO support")
Reported-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 microcode loader updates from Borislav Petkov:
- A bunch of minor cleanups
* tag 'x86_microcode_for_v6.14_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/microcode/AMD: Remove ret local var in early_apply_microcode()
x86/microcode/AMD: Have __apply_microcode_amd() return bool
x86/microcode/AMD: Make __verify_patch_size() return bool
x86/microcode/AMD: Remove bogus comment from parse_container()
x86/microcode/AMD: Return bool from find_blobs_in_containers()
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 resource control updates from Borislav Petkov:
- Extend resctrl with the capability of total memory bandwidth
monitoring, thus accomodating systems which support only total but
not local memory bandwidth monitoring. Add the respective new mount
options
- The usual cleanups
* tag 'x86_cache_for_v6.14_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/resctrl: Document the new "mba_MBps_event" file
x86/resctrl: Add write option to "mba_MBps_event" file
x86/resctrl: Add "mba_MBps_event" file to CTRL_MON directories
x86/resctrl: Make mba_sc use total bandwidth if local is not supported
x86/resctrl: Compute memory bandwidth for all supported events
x86/resctrl: Modify update_mba_bw() to use per CTRL_MON group event
x86/resctrl: Prepare for per-CTRL_MON group mba_MBps control
x86/resctrl: Introduce resctrl_file_fflags_init() to initialize fflags
x86/resctrl: Use kthread_run_on_cpu()
|
|
In f2fs_new_inode(), if we fail to get a new inode, we go iput(), followed by
f2fs_evict_inode(). If the inode is not marked as bad, it'll try to call
f2fs_remove_inode_page() which tries to read the inode block given node id.
But, there's no block address allocated yet, which gives a chance to access
a wrong block address, if the block device has some garbage data in NAT table.
We need to make sure NAT table should have zero data for all the unallocated
node ids, but also would be better to take this unnecessary path as well.
Let's mark the faild inode as bad.
Fixes: 0abd675e97e6 ("f2fs: support plain user/group quota")
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 CPU speculation update from Borislav Petkov:
- Add support for AMD hardware which is not affected by SRSO on the
user/kernel attack vector and advertise it to guest userspace
* tag 'x86_bugs_for_v6.14_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
KVM: x86: Advertise SRSO_USER_KERNEL_NO to userspace
x86/bugs: Add SRSO_USER_KERNEL_NO support
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras
Pull EDAC updates from Borislav Petkov:
- Remove the EDAC PowerPC Cell driver due to the removal of the IBM
Cell blades support
- Add a new EDAC driver for Loongson SoCs which reports single-bit
correctable errors
- Extend the SKX and i10NM EDAC drivers to support UV systems which can
have more than 8 nodes
- Add Intel Clearwater Forest server support to i10nm_edac
- Minor fix
* tag 'edac_updates_for_v6.14' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
EDAC/cell: Remove powerpc Cell driver
EDAC: Add an EDAC driver for the Loongson memory controller
EDAC: Fix typos in comments
EDAC/{i10nm,skx,skx_common}: Support UV systems
EDAC/i10nm: Add Intel Clearwater Forest server support
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 RAS updates from Borislav Petkov:
- Remove the shared threshold bank hack on AMD and streamline and
simplify it
- Cleanup and sanitize MCA code
* tag 'ras_core_for_v6.14_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/mce/amd: Remove shared threshold bank plumbing
x86/mce: Remove the redundant mce_hygon_feature_init()
x86/mce: Convert family/model mixed checks to VFM-based checks
x86/mce: Break up __mcheck_cpu_apply_quirks()
x86/mce: Make four functions return bool
x86/mce/threshold: Remove the redundant this_cpu_dec_return()
x86/mce: Make several functions return bool
|
|
A BAR size is always a power of two. buf_size = min(SZ_1M, bar_size).
If the BAR size is <= 1MB, there will be one iteration, no remainder.
If the BAR size is > 1MB, there will be more than one iteration, but the
size will always be evenly divisible by 1MB, so no remainder.
Link: https://lore.kernel.org/r/20250109094556.1724663-2-18255117159@163.com
Suggested-by: Niklas Cassel <cassel@kernel.org>
Signed-off-by: Hans Zhang <18255117159@163.com>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Niklas Cassel <cassel@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
|
|
Add a more advanced BAR test that writes all BARs in one go, and then reads
them back and verifies that the value matches the BAR number bitwise OR'ed
with offset, this allows us to verify:
- The BAR number was what we intended to read
- The offset was what we intended to read
This allows us to detect potential address translation issues on the EP.
Reading back the BAR directly after writing will not allow us to detect the
case where inbound address translation on the endpoint incorrectly causes
multiple BARs to be redirected to the same memory region (within the EP).
Link: https://lore.kernel.org/r/20241116032045.2574168-2-cassel@kernel.org
Signed-off-by: Niklas Cassel <cassel@kernel.org>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
|
|
The test BAR is on the EP side is allocated using pci_epf_alloc_space(),
which allocates the backing memory using dma_alloc_coherent(), which will
return zeroed memory regardless of __GFP_ZERO was set or not.
This means that running a new version of pci-endpoint-test.c (host side)
with an old version of pci-epf-test.c (EP side) will not see any
capabilities being set (as intended), so this is backwards compatible.
Additionally, the EP side always allocates at least 128 bytes for the test
BAR (excluding the MSI-X table), this means that adding another register at
offset 0x30 is still within the 128 available bytes.
For now, we only add the CAP_UNALIGNED_ACCESS capability.
If CAP_UNALIGNED_ACCESS is set, that means that the EP side supports
reading/writing to an address without any alignment requirements.
Thus, if CAP_UNALIGNED_ACCESS is set, make sure that the host side does
not add any extra padding to the buffers that we allocate (which was only
done in order to get the buffers to satisfy certain alignment requirements
by the endpoint controller).
Link: https://lore.kernel.org/r/20241203063851.695733-6-cassel@kernel.org
Signed-off-by: Niklas Cassel <cassel@kernel.org>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
|
|
The test BAR is on the EP side is allocated using pci_epf_alloc_space(),
which allocates the backing memory using dma_alloc_coherent(), which will
return zeroed memory regardless of __GFP_ZERO was set or not.
This means that running a new version of pci-endpoint-test.c (host side)
with an old version of pci-epf-test.c (EP side) will not see any
capabilities being set (as intended), so this is backwards compatible.
Additionally, the EP side always allocates at least 128 bytes for the test
BAR (excluding the MSI-X table), this means that adding another register at
offset 0x30 is still within the 128 available bytes.
For now, we only add the CAP_UNALIGNED_ACCESS capability.
Set CAP_UNALIGNED_ACCESS if the EPC driver can handle any address (because
it implements the .align_addr callback).
Link: https://lore.kernel.org/r/20241203063851.695733-5-cassel@kernel.org
Signed-off-by: Niklas Cassel <cassel@kernel.org>
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
|
|
Currently, if DMA MEMCPY test is requested by the host, and if the endpoint
DMA controller supports DMA_PRIVATE, the test will fail. This is not
correct since there is no check for DMA_MEMCPY capability and the DMA
controller can support both DMA_PRIVATE and DMA_MEMCPY.
Fix the check and also reword the error message.
Link: https://lore.kernel.org/r/20250116171650.33585-2-manivannan.sadhasivam@linaro.org
Fixes: 8353813c88ef ("PCI: endpoint: Enable DMA tests for endpoints with DMA capabilities")
Reported-by: Niklas Cassel <cassel@kernel.org>
Closes: https://lore.kernel.org/linux-pci/Z3QtEihbiKIGogWA@ryzen
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Tested-by: Niklas Cassel <cassel@kernel.org>
Reviewed-by: Niklas Cassel <cassel@kernel.org>
|
|
If dma_chan_tx allocation fails, set dma_chan_rx to NULL after it is
freed.
Link: https://lore.kernel.org/r/20241227160841.92382-1-khalfella@gmail.com
Fixes: 8353813c88ef ("PCI: endpoint: Enable DMA tests for endpoints with DMA capabilities")
Signed-off-by: Mohamed Khalfella <khalfella@gmail.com>
[kwilczynski: commit log]
Signed-off-by: Krzysztof Wilczyński <kwilczynski@kernel.org>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Niklas Cassel <cassel@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
|
|
The delete on last close functionality can now only be tested properly
by using multiple threads to hold open the variable files and testing
what happens as they complete.
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
|
|
Changing voltage might ignore slew rate and cause a current surge.
With current implementation the driver will get the regulator to change
the voltage range used during run time. According to communication I
have had with Texas Instruments, this is not intended, since the
Dynamic Voltage Scaling in the hardware is only designed to work
within a voltage range. The current implementation will therefore
ignore the slew rate that is defined in devicetree when the voltage
range is changed during use.
The current implementation will always select a voltage in the most
accurate range that can reach that voltage even though multiple ranges
are able to reach that voltage. There are 4 Voltage ranges with the
following reach:
0b00: 0.4-0.71875V (1.25mV step size)
0b01: 0.4-1.0375V (2.5mV)
0b10: 0.4-1.675V (5mV)
0b11: 0.8-3.3V (10mV)
This in practice means that a change from below to above 0.71875V will
use the smallest range(0b00) for the values below and the second
smallest range(0b01) for the voltages above (Up to 1.675V). I have
timed how long it takes to go from below 0.71875V to above. The
increase was 100mV which, with the slew rate set to 1250µV/µs. This
in theory should take 80µs to do. With the current implementation, it
takes 10µs on my hardware. Doing the same test with the slew rate set
to 5000µV/µs, which should take 20µs, also only takes 10µs to do on
my hardware. Not only is this not in line with the technical
specification for the regulator. It also causes a current surge. Which
when calculating the output current, as described in the technical
specification, compared to what I could observe on my hardware the real
output is ~1A higher (~1.2A) than what I calculated it to be(~0.2A).
I tested also transitioning from a bigger to a smaller range, and the
results were the same.
Instead, let's limit the voltage range to a single one, which is in
line with the intended use of the regulator. This is done by looking
up the minimum and maximum requested voltage specified in devicetree.
Signed-off-by: Jonas Andreasson <jonas.andreasson@axis.com>
Link: https://patch.msgid.link/20250121-tps-fix-v2-1-50cc4d0f1635@axis.com
Signed-off-by: Mark Brown <broonie@kernel.org>
|
|
In commit
9e2ab4b18ebd ("ASoC: rockchip: i2s-tdm: Fix inaccurate sampling rates"),
the set_sysclk callback was removed as considered unused as the mclk rate
can be set in the hw_params callback.
The difference between hw_params and set_sysclk is that the former is
called with the audio sampling rate set in the params (e.g.: 48000 Hz)
while the latter is called with a clock rate already computed with
sampling_rate * mclk-fs (e.g.: 48000 * 256)
For HDMI audio using the Rockchip I2S TDM driver, the mclk-fs value must
be set to 128 instead of the default 256, and that value is set in the
device tree at the machine driver level (like a simple-audio-card
compatible node).
Therefore, the i2s_tdm driver has no idea that another mclk-fs value can
be configured and simply computes the mclk rate in the hw_params callback
with DEFAULT_MCLK_FS * params_rate(params), which is wrong for HDMI
audio.
Re-add the set_sysclk callback so that the mclk rate is computed by the
machine driver which has the correct mclk-fs value set in its device tree
node.
Fixes: 9e2ab4b18ebd ("ASoC: rockchip: i2s-tdm: Fix inaccurate sampling rates")
Signed-off-by: Detlev Casanova <detlev.casanova@collabora.com>
Link: https://patch.msgid.link/20250117163102.65807-1-detlev.casanova@collabora.com
Signed-off-by: Mark Brown <broonie@kernel.org>
|
|
Checking for lockdep_assert_held(&ctx->uring_lock) in io_free_rsrc_node()
means that the assertion is only checked when the resource drops to zero
references.
Move the lockdep assertion up into the caller io_put_rsrc_node() so that it
instead happens on every reference count decrement.
Signed-off-by: Jann Horn <jannh@google.com>
Link: https://lore.kernel.org/r/20250120-uring-lockdep-assert-earlier-v1-1-68d8e071a4bb@google.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
io_uring_ctx parameter for io_rsrc_node_alloc() is unused for now.
This patch removes the parameter and fixes the callers accordingly.
Signed-off-by: Sidong Yang <sidong.yang@furiosa.ai>
Link: https://lore.kernel.org/r/20250115142033.658599-1-sidong.yang@furiosa.ai
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Make it always reference the returned file. It's safer, especially with
unregistrations happening under it. And it makes the api cleaner with no
conditional clean ups by the caller.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/0d0b13a63e8edd6b5d360fc821dcdb035cb6b7e0.1736995897.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
The locking in the buffer cloning code is somewhat complex because it goes
back and forth between locking the source ring and the destination ring.
Make it easier to reason about by locking both rings at the same time.
To avoid ABBA deadlocks, lock the rings in ascending kernel address order,
just like in lock_two_nondirectories().
Signed-off-by: Jann Horn <jannh@google.com>
Link: https://lore.kernel.org/r/20250115-uring-clone-refactor-v2-1-7289ba50776d@google.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Old email still works, and will indefinitely, but I'm switching to a new
one.
Signed-off-by: Corey Minyard <corey@minyard.net>
|
|
Signed-off-by: Shixiong Ou <oushixiong@kylinos.cn>
Signed-off-by: Helge Deller <deller@gmx.de>
|
|
devm_clk_get_optional() returns NULL for missing clocks and a PTR_ERR()
if there is a clock but we fail to get it, but currently we only handle
the latter case and do so as though the clock was missing. If we get an
error back we should handle that as an error since the clock exists but
we failed to get it, if we get NULL then the clock doesn't exist and we
should handle that.
Fixes: 4c6ac5446d06 ("spi: omap2-mcspi: Fix the IS_ERR() bug for devm_clk_get_optional_enabled()")
Reported-by: Lars Pedersen <lapeddk@gmail.com>
Link: https://patch.msgid.link/20250117-spi-fix-omap2-optional-v1-1-e77d4ac6db6e@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Tested-by: Lars Pedersen <lapeddk@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
|
|
Both kcryptd_io_read() and kcryptd_crypt_write_convert() will invoke
crypt_alloc_buffer() to allocate a new bio. Both of these two callers
initialize bi_iter.bi_sector for the new bio separatedly after
crypt_alloc_buffer() returns. However, kcryptd_crypt_write_convert()
will copy the bi_iter of the new bio into ctx.iter_out or ctx.iter_in.
Although it doesn't incur any harm now, it is better to fully initialize
bi_iter before it is used.
Therefore, initialize bi_iter.bi_sector in crypt_alloc_buffer() instead.
Signed-off-by: Hou Tao <houtao1@huawei.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
|
|
Both kcryptd_crypt_write_continue() and kcryptd_crypt_read_continue()
are running in the kworker context, it is OK to call cond_resched(),
Therefore, set atomic as false when invoking crypt_convert() under
kworker context.
Signed-off-by: Hou Tao <houtao1@huawei.com>
Reviewed-by: Ignat Korchagin <ignat@cloudflare.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
|
|
set_delayed() adjusts cfs_rq->h_nr_runnable for the hierarchy when an
entity is delayed irrespective of whether the entity corresponds to a
task or a cfs_rq.
Consider the following scenario:
root
/ \
A B (*) delayed since B is no longer eligible on root
| |
Task0 Task1 <--- dequeue_task_fair() - task blocks
When Task1 blocks (dequeue_entity() for task's se returns true),
dequeue_entities() will continue adjusting cfs_rq->h_nr_* for the
hierarchy of Task1. However, when the sched_entity corresponding to
cfs_rq B is delayed, set_delayed() will adjust the h_nr_runnable for the
hierarchy too leading to both dequeue_entity() and set_delayed()
decrementing h_nr_runnable for the dequeue of the same task.
A SCHED_WARN_ON() to inspect h_nr_runnable post its update in
dequeue_entities() like below:
cfs_rq->h_nr_runnable -= h_nr_runnable;
SCHED_WARN_ON(((int) cfs_rq->h_nr_runnable) < 0);
is consistently tripped when running wakeup intensive workloads like
hackbench in a cgroup.
This error is self correcting since cfs_rq are per-cpu and cannot
migrate. The entitiy is either picked for full dequeue or is requeued
when a task wakes up below it. Both those paths call clear_delayed()
which again increments h_nr_runnable of the hierarchy without
considering if the entity corresponds to a task or not.
h_nr_runnable will eventually reflect the correct value however in the
interim, the incorrect values can still influence PELT calculation which
uses se->runnable_weight or cfs_rq->h_nr_runnable.
Since only delayed tasks take the early return path in
dequeue_entities() and enqueue_task_fair(), adjust the
h_nr_runnable in {set,clear}_delayed() only when a task is delayed as
this path skips the h_nr_* update loops and returns early.
For entities corresponding to cfs_rq, the h_nr_* update loop in the
caller will do the right thing.
Fixes: 76f2f783294d ("sched/eevdf: More PELT vs DELAYED_DEQUEUE")
Signed-off-by: K Prateek Nayak <kprateek.nayak@amd.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Gautham R. Shenoy <gautham.shenoy@amd.com>
Tested-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Link: https://lkml.kernel.org/r/20250117105852.23908-1-kprateek.nayak@amd.com
|
|
Merged the 'fixes' branch into the 'for-next' branch to resolve
a conflict in alienware-wmi zone teardown code.
|
|
When testing on my tiled display, initially the tiled display is
detected correctly:
[90376.523692] xe 0000:67:00.0: [drm:drm_client_firmware_config.isra.0 [drm]] fallback: Not all outputs enabled
[90376.523713] xe 0000:67:00.0: [drm:drm_client_firmware_config.isra.0 [drm]] Enabled: 0, detected: 2
...
[90376.523967] xe 0000:67:00.0: [drm:drm_client_modeset_probe [drm]] [CRTC:82:pipe A] desired mode 1920x2160 set (1920,0)
[90376.524020] xe 0000:67:00.0: [drm:drm_client_modeset_probe [drm]] [CRTC:134:pipe B] desired mode 1920x2160 set (0,0)
But then, when modes have been set:
[90379.729525] xe 0000:67:00.0: [drm:drm_client_firmware_config.isra.0 [drm]] [CONNECTOR:287:DP-4] on [CRTC:82:pipe A]: 1920x2160
[90379.729640] xe 0000:67:00.0: [drm:drm_client_firmware_config.isra.0 [drm]] [CONNECTOR:289:DP-5] on [CRTC:134:pipe B]: 1920x2160
...
[90379.730036] xe 0000:67:00.0: [drm:drm_client_modeset_probe [drm]] [CRTC:82:pipe A] desired mode 1920x2160 set (0,0)
[90379.730124] xe 0000:67:00.0: [drm:drm_client_modeset_probe [drm]] [CRTC:134:pipe B] desired mode 1920x2160 set (0,0)
Call drm_client_get_tile_offsets() in drm_client_firmware_config() as
well, to ensure that the offset is set correctly.
This has to be done as a separate pass, as the tile order may not be
equal to the drm connector order.
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20250116142825.3933-2-dev@lankhorst.se
Signed-off-by: Maarten Lankhorst <dev@lankhorst.se>
Cc: <stable@vger.kernel.org>
|
|
Tiled displays have a different x/y offset to begin with. Instead of
attempting to remember this, just apply a delta instead.
This fixes the first tile being duplicated on other tiles when vt
switching.
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20250116142825.3933-1-dev@lankhorst.se
Signed-off-by: Maarten Lankhorst <dev@lankhorst.se>
Cc: <stable@vger.kernel.org>
|
