]> git.ipfire.org Git - thirdparty/kernel/linux.git/log
thirdparty/kernel/linux.git
3 weeks agocgroup: Increment nr_dying_subsys_* from rmdir context
Petr Malat [Thu, 23 Apr 2026 09:48:26 +0000 (04:48 -0500)] 
cgroup: Increment nr_dying_subsys_* from rmdir context

Incrementing nr_dying_subsys_* in offline_css(), which is executed by
cgroup_offline_wq worker, leads to a race where user can see the value
to be 0 if he reads cgroup.stat after calling rmdir and before the worker
executes. This makes the user wrongly expect resources released by the
removed cgroup to be available for a new assignment.

Increment nr_dying_subsys_* from kill_css(), which is called from the
cgroup_rmdir() context.

Fixes: ab0312526867 ("cgroup: Show # of subsystem CSSes in cgroup.stat")
Signed-off-by: Petr Malat <oss@malat.biz>
Signed-off-by: Tejun Heo <tj@kernel.org>
4 weeks agocgroup/cpuset: record DL BW alloc CPU for attach rollback
Guopeng Zhang [Fri, 17 Apr 2026 03:37:41 +0000 (11:37 +0800)] 
cgroup/cpuset: record DL BW alloc CPU for attach rollback

cpuset_can_attach() allocates DL bandwidth only when migrating
deadline tasks to a disjoint CPU mask, but cpuset_cancel_attach()
rolls back based only on nr_migrate_dl_tasks. This makes the DL
bandwidth alloc/free paths asymmetric: rollback can call dl_bw_free()
even when no dl_bw_alloc() was done.

Rollback also needs to undo the reservation against the same CPU/root
domain that was charged. Record the CPU used by dl_bw_alloc() and use
that state in cpuset_cancel_attach(). If no allocation happened,
dl_bw_cpu stays at -1 and rollback skips dl_bw_free(). If allocation
did happen, bandwidth is returned to the same CPU/root domain.

Successful attach paths are unchanged. This only fixes failed attach
rollback accounting.

Fixes: 2ef269ef1ac0 ("cgroup/cpuset: Free DL BW in case can_attach() fails")
Signed-off-by: Guopeng Zhang <zhangguopeng@kylinos.cn>
Reviewed-by: Waiman Long <longman@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
4 weeks agocgroup/rdma: fix integer overflow in rdmacg_try_charge()
cuitao [Tue, 14 Apr 2026 01:53:27 +0000 (09:53 +0800)] 
cgroup/rdma: fix integer overflow in rdmacg_try_charge()

The expression `rpool->resources[index].usage + 1` is computed in int
arithmetic before being assigned to s64 variable `new`. When usage equals
INT_MAX (the default "max" value), the addition overflows to INT_MIN.
This negative value then passes the `new > max` check incorrectly,
allowing a charge that should be rejected and corrupting usage to
negative.

Fix by casting usage to s64 before the addition so the arithmetic is
done in 64-bit.

Fixes: 39d3e7584a68 ("rdmacg: Added rdma cgroup controller")
Signed-off-by: cuitao <cuitao@kylinos.cn>
Reviewed-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
4 weeks agosched/psi: fix race between file release and pressure write
Edward Adam Davis [Tue, 14 Apr 2026 06:15:43 +0000 (14:15 +0800)] 
sched/psi: fix race between file release and pressure write

A potential race condition exists between pressure write and cgroup file
release regarding the priv member of struct kernfs_open_file, which
triggers the uaf reported in [1].

Consider the following scenario involving execution on two separate CPUs:

   CPU0 CPU1
   ==== ====
vfs_rmdir()
kernfs_iop_rmdir()
cgroup_rmdir()
cgroup_kn_lock_live()
cgroup_destroy_locked()
cgroup_addrm_files()
cgroup_rm_file()
kernfs_remove_by_name()
kernfs_remove_by_name_ns()
 vfs_write() __kernfs_remove()
 new_sync_write() kernfs_drain()
 kernfs_fop_write_iter() kernfs_drain_open_files()
 cgroup_file_write() kernfs_release_file()
 pressure_write() cgroup_file_release()
 ctx = of->priv;
kfree(ctx);
  of->priv = NULL;
cgroup_kn_unlock()
 cgroup_kn_lock_live()
 cgroup_get(cgrp)
 cgroup_kn_unlock()
 if (ctx->psi.trigger)  // here, trigger uaf for ctx, that is of->priv

The cgroup_rmdir() is protected by the cgroup_mutex, it also safeguards
the memory deallocation of of->priv performed within cgroup_file_release().
However, the operations involving of->priv executed within pressure_write()
are not entirely covered by the protection of cgroup_mutex. Consequently,
if the code in pressure_write(), specifically the section handling the
ctx variable executes after cgroup_file_release() has completed, a uaf
vulnerability involving of->priv is triggered.

Therefore, the issue can be resolved by extending the scope of the
cgroup_mutex lock within pressure_write() to encompass all code paths
involving of->priv, thereby properly synchronizing the race condition
occurring between cgroup_file_release() and pressure_write().

And, if an live kn lock can be successfully acquired while executing
the pressure write operation, it indicates that the cgroup deletion
process has not yet reached its final stage; consequently, the priv
pointer within open_file cannot be NULL. Therefore, the operation to
retrieve the ctx value must be moved to a point *after* the live kn
lock has been successfully acquired.

In another situation, specifically after entering cgroup_kn_lock_live()
but before acquiring cgroup_mutex, there exists a different class of
race condition:

CPU0: write memory.pressure               CPU1: write cgroup.pressure=0
===========================   =============================

kernfs_fop_write_iter()
 kernfs_get_active_of(of)
 pressure_write()
   cgroup_kn_lock_live(memory.pressure)
     cgroup_tryget(cgrp)
     kernfs_break_active_protection(kn)
     ... blocks on cgroup_mutex

                                        cgroup_pressure_write()
                                        cgroup_kn_lock_live(cgroup.pressure)
                                        cgroup_file_show(memory.pressure, false)
                                          kernfs_show(false)
                                            kernfs_drain_open_files()
                                              cgroup_file_release(of)
                                                kfree(ctx)
                                                  of->priv = NULL
                                        cgroup_kn_unlock()

   ... acquires cgroup_mutex
   ctx = of->priv;        // may now be NULL
   if (ctx->psi.trigger)  // NULL dereference

Consequently, there is a possibility that of->priv is NULL, the pressure
write needs to check for this.

Now that the scope of the cgroup_mutex has been expanded, the original
explicit cgroup_get/put operations are no longer necessary, this is
because acquiring/releasing the live kn lock inherently executes a
cgroup get/put operation.

[1]
BUG: KASAN: slab-use-after-free in pressure_write+0xa4/0x210 kernel/cgroup/cgroup.c:4011
Call Trace:
 pressure_write+0xa4/0x210 kernel/cgroup/cgroup.c:4011
 cgroup_file_write+0x36f/0x790 kernel/cgroup/cgroup.c:4311
 kernfs_fop_write_iter+0x3b0/0x540 fs/kernfs/file.c:352

Allocated by task 9352:
 cgroup_file_open+0x90/0x3a0 kernel/cgroup/cgroup.c:4256
 kernfs_fop_open+0x9eb/0xcb0 fs/kernfs/file.c:724
 do_dentry_open+0x83d/0x13e0 fs/open.c:949

Freed by task 9353:
 cgroup_file_release+0xd6/0x100 kernel/cgroup/cgroup.c:4283
 kernfs_release_file fs/kernfs/file.c:764 [inline]
 kernfs_drain_open_files+0x392/0x720 fs/kernfs/file.c:834
 kernfs_drain+0x470/0x600 fs/kernfs/dir.c:525

Fixes: 0e94682b73bf ("psi: introduce psi monitor")
Reported-by: syzbot+33e571025d88efd1312c@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=33e571025d88efd1312c
Tested-by: syzbot+33e571025d88efd1312c@syzkaller.appspotmail.com
Signed-off-by: Edward Adam Davis <eadavis@qq.com>
Reviewed-by: Chen Ridong <chenridong@huaweicloud.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
4 weeks agoMerge tag 'mips_7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux
Linus Torvalds [Fri, 17 Apr 2026 15:53:23 +0000 (08:53 -0700)] 
Merge tag 'mips_7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux

Pull MIPS updates from Thomas Bogendoerfer:

 - Support for Mobileye EyeQ6Lplus

 - Cleanups and fixes

* tag 'mips_7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux: (30 commits)
  MIPS/mtd: Handle READY GPIO in generic NAND platform data
  MIPS/input: Move RB532 button to GPIO descriptors
  MIPS: validate DT bootargs before appending them
  MIPS: Alchemy: Remove unused forward declaration
  MAINTAINERS: Mobileye: Add EyeQ6Lplus files
  MIPS: config: add eyeq6lplus_defconfig
  MIPS: Add Mobileye EyeQ6Lplus evaluation board dts
  MIPS: Add Mobileye EyeQ6Lplus SoC dtsi
  clk: eyeq: Add Mobileye EyeQ6Lplus OLB
  clk: eyeq: Adjust PLL accuracy computation
  clk: eyeq: Skip post-divisor when computing PLL frequency
  pinctrl: eyeq5: Add Mobileye EyeQ6Lplus OLB
  pinctrl: eyeq5: Use match data
  reset: eyeq: Add Mobileye EyeQ6Lplus OLB
  MIPS: Add Mobileye EyeQ6Lplus support
  dt-bindings: soc: mobileye: Add EyeQ6Lplus OLB
  dt-bindings: mips: Add Mobileye EyeQ6Lplus SoC
  MIPS: dts: loongson64g-package: Switch to Loongson UART driver
  mips: pci-mt7620: rework initialization procedure
  mips: pci-mt7620: add more register init values
  ...

4 weeks agoMerge tag 'alpha-for-v7.1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/lindh...
Linus Torvalds [Fri, 17 Apr 2026 15:34:43 +0000 (08:34 -0700)] 
Merge tag 'alpha-for-v7.1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/lindholm/alpha

Pull alpha updates from Magnus Lindholm:
 "One fix to silence pgprot_modify() compiler warnings, and one patch
  adding SECCOMP/SECCOMP_FILTER support together with the syscall and
  ptrace fixes needed for it"

* tag 'alpha-for-v7.1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/lindholm/alpha:
  alpha: Define pgprot_modify to silence tautological comparison warnings
  alpha: add support for SECCOMP and SECCOMP_FILTER

4 weeks agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Fri, 17 Apr 2026 14:18:03 +0000 (07:18 -0700)] 
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm

Pull kvm updates from Paolo Bonzini:
 "Arm:

   - Add support for tracing in the standalone EL2 hypervisor code,
     which should help both debugging and performance analysis. This
     uses the new infrastructure for 'remote' trace buffers that can be
     exposed by non-kernel entities such as firmware, and which came
     through the tracing tree

   - Add support for GICv5 Per Processor Interrupts (PPIs), as the
     starting point for supporting the new GIC architecture in KVM

   - Finally add support for pKVM protected guests, where pages are
     unmapped from the host as they are faulted into the guest and can
     be shared back from the guest using pKVM hypercalls. Protected
     guests are created using a new machine type identifier. As the
     elusive guestmem has not yet delivered on its promises, anonymous
     memory is also supported

     This is only a first step towards full isolation from the host; for
     example, the CPU register state and DMA accesses are not yet
     isolated. Because this does not really yet bring fully what it
     promises, it is hidden behind CONFIG_ARM_PKVM_GUEST +
     'kvm-arm.mode=protected', and also triggers TAINT_USER when a VM is
     created. Caveat emptor

   - Rework the dreaded user_mem_abort() function to make it more
     maintainable, reducing the amount of state being exposed to the
     various helpers and rendering a substantial amount of state
     immutable

   - Expand the Stage-2 page table dumper to support NV shadow page
     tables on a per-VM basis

   - Tidy up the pKVM PSCI proxy code to be slightly less hard to
     follow

   - Fix both SPE and TRBE in non-VHE configurations so that they do not
     generate spurious, out of context table walks that ultimately lead
     to very bad HW lockups

   - A small set of patches fixing the Stage-2 MMU freeing in error
     cases

   - Tighten-up accepted SMC immediate value to be only #0 for host
     SMCCC calls

   - The usual cleanups and other selftest churn

  LoongArch:

   - Use CSR_CRMD_PLV for kvm_arch_vcpu_in_kernel()

   - Add DMSINTC irqchip in kernel support

  RISC-V:

   - Fix steal time shared memory alignment checks

   - Fix vector context allocation leak

   - Fix array out-of-bounds in pmu_ctr_read() and pmu_fw_ctr_read_hi()

   - Fix double-free of sdata in kvm_pmu_clear_snapshot_area()

   - Fix integer overflow in kvm_pmu_validate_counter_mask()

   - Fix shift-out-of-bounds in make_xfence_request()

   - Fix lost write protection on huge pages during dirty logging

   - Split huge pages during fault handling for dirty logging

   - Skip CSR restore if VCPU is reloaded on the same core

   - Implement kvm_arch_has_default_irqchip() for KVM selftests

   - Factored-out ISA checks into separate sources

   - Added hideleg to struct kvm_vcpu_config

   - Factored-out VCPU config into separate sources

   - Support configuration of per-VM HGATP mode from KVM user space

  s390:

   - Support for ESA (31-bit) guests inside nested hypervisors

   - Remove restriction on memslot alignment, which is not needed
     anymore with the new gmap code

   - Fix LPSW/E to update the bear (which of course is the breaking
     event address register)

  x86:

   - Shut up various UBSAN warnings on reading module parameter before
     they were initialized

   - Don't zero-allocate page tables that are used for splitting
     hugepages in the TDP MMU, as KVM is guaranteed to set all SPTEs in
     the page table and thus write all bytes

   - As an optimization, bail early when trying to unsync 4KiB mappings
     if the target gfn can just be mapped with a 2MiB hugepage

  x86 generic:

   - Copy single-chunk MMIO write values into struct kvm_vcpu (more
     precisely struct kvm_mmio_fragment) to fix use-after-free stack
     bugs where KVM would dereference stack pointer after an exit to
     userspace

   - Clean up and comment the emulated MMIO code to try to make it
     easier to maintain (not necessarily "easy", but "easier")

   - Move VMXON+VMXOFF and EFER.SVME toggling out of KVM (not *all* of
     VMX and SVM enabling) as it is needed for trusted I/O

   - Advertise support for AVX512 Bit Matrix Multiply (BMM) instructions

   - Immediately fail the build if a required #define is missing in one
     of KVM's headers that is included multiple times

   - Reject SET_GUEST_DEBUG with -EBUSY if there's an already injected
     exception, mostly to prevent syzkaller from abusing the uAPI to
     trigger WARNs, but also because it can help prevent userspace from
     unintentionally crashing the VM

   - Exempt SMM from CPUID faulting on Intel, as per the spec

   - Misc hardening and cleanup changes

  x86 (AMD):

   - Fix and optimize IRQ window inhibit handling for AVIC; make it
     per-vCPU so that KVM doesn't prematurely re-enable AVIC if multiple
     vCPUs have to-be-injected IRQs

   - Clean up and optimize the OSVW handling, avoiding a bug in which
     KVM would overwrite state when enabling virtualization on multiple
     CPUs in parallel. This should not be a problem because OSVW should
     usually be the same for all CPUs

   - Drop a WARN in KVM_MEMORY_ENCRYPT_REG_REGION where KVM complains
     about a "too large" size based purely on user input

   - Clean up and harden the pinning code for KVM_MEMORY_ENCRYPT_REG_REGION

   - Disallow synchronizing a VMSA of an already-launched/encrypted
     vCPU, as doing so for an SNP guest will crash the host due to an
     RMP violation page fault

   - Overhaul KVM's APIs for detecting SEV+ guests so that VM-scoped
     queries are required to hold kvm->lock, and enforce it by lockdep.
     Fix various bugs where sev_guest() was not ensured to be stable for
     the whole duration of a function or ioctl

   - Convert a pile of kvm->lock SEV code to guard()

   - Play nicer with userspace that does not enable
     KVM_CAP_EXCEPTION_PAYLOAD, for which KVM needs to set CR2 and DR6
     as a response to ioctls such as KVM_GET_VCPU_EVENTS (even if the
     payload would end up in EXITINFO2 rather than CR2, for example).
     Only set CR2 and DR6 when consumption of the payload is imminent,
     but on the other hand force delivery of the payload in all paths
     where userspace retrieves CR2 or DR6

   - Use vcpu->arch.cr2 when updating vmcb12's CR2 on nested #VMEXIT
     instead of vmcb02->save.cr2. The value is out of sync after a
     save/restore or after a #PF is injected into L2

   - Fix a class of nSVM bugs where some fields written by the CPU are
     not synchronized from vmcb02 to cached vmcb12 after VMRUN, and so
     are not up-to-date when saved by KVM_GET_NESTED_STATE

   - Fix a class of bugs where the ordering between KVM_SET_NESTED_STATE
     and KVM_SET_{S}REGS could cause vmcb02 to be incorrectly
     initialized after save+restore

   - Add a variety of missing nSVM consistency checks

   - Fix several bugs where KVM failed to correctly update VMCB fields
     on nested #VMEXIT

   - Fix several bugs where KVM failed to correctly synthesize #UD or
     #GP for SVM-related instructions

   - Add support for save+restore of virtualized LBRs (on SVM)

   - Refactor various helpers and macros to improve clarity and
     (hopefully) make the code easier to maintain

   - Aggressively sanitize fields when copying from vmcb12, to guard
     against unintentionally allowing L1 to utilize yet-to-be-defined
     features

   - Fix several bugs where KVM botched rAX legality checks when
     emulating SVM instructions. There are remaining issues in that KVM
     doesn't handle size prefix overrides for 64-bit guests

   - Fail emulation of VMRUN/VMLOAD/VMSAVE if mapping vmcb12 fails
     instead of somewhat arbitrarily synthesizing #GP (i.e. don't double
     down on AMD's architectural but sketchy behavior of generating #GP
     for "unsupported" addresses)

   - Cache all used vmcb12 fields to further harden against TOCTOU bugs

  x86 (Intel):

   - Drop obsolete branch hint prefixes from the VMX instruction macros

   - Use ASM_INPUT_RM() in __vmcs_writel() to coerce clang into using a
     register input when appropriate

   - Code cleanups

  guest_memfd:

   - Don't mark guest_memfd folios as accessed, as guest_memfd doesn't
     support reclaim, the memory is unevictable, and there is no storage
     to write back to

  LoongArch selftests:

   - Add KVM PMU test cases

  s390 selftests:

   - Enable more memory selftests

  x86 selftests:

   - Add support for Hygon CPUs in KVM selftests

   - Fix a bug in the MSR test where it would get false failures on
     AMD/Hygon CPUs with exactly one of RDPID or RDTSCP

   - Add an MADV_COLLAPSE testcase for guest_memfd as a regression test
     for a bug where the kernel would attempt to collapse guest_memfd
     folios against KVM's will"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (373 commits)
  KVM: x86: use inlines instead of macros for is_sev_*guest
  x86/virt: Treat SVM as unsupported when running as an SEV+ guest
  KVM: SEV: Goto an existing error label if charging misc_cg for an ASID fails
  KVM: SVM: Move lock-protected allocation of SEV ASID into a separate helper
  KVM: SEV: use mutex guard in snp_handle_guest_req()
  KVM: SEV: use mutex guard in sev_mem_enc_unregister_region()
  KVM: SEV: use mutex guard in sev_mem_enc_ioctl()
  KVM: SEV: use mutex guard in snp_launch_update()
  KVM: SEV: Assert that kvm->lock is held when querying SEV+ support
  KVM: SEV: Document that checking for SEV+ guests when reclaiming memory is "safe"
  KVM: SEV: Hide "struct kvm_sev_info" behind CONFIG_KVM_AMD_SEV=y
  KVM: SEV: WARN on unhandled VM type when initializing VM
  KVM: LoongArch: selftests: Add PMU overflow interrupt test
  KVM: LoongArch: selftests: Add basic PMU event counting test
  KVM: LoongArch: selftests: Add cpucfg read/write helpers
  LoongArch: KVM: Add DMSINTC inject msi to vCPU
  LoongArch: KVM: Add DMSINTC device support
  LoongArch: KVM: Make vcpu_is_preempted() as a macro rather than function
  LoongArch: KVM: Move host CSR_GSTAT save and restore in context switch
  LoongArch: KVM: Move host CSR_EENTRY save and restore in context switch
  ...

4 weeks agox86/CPU: Fix FPDSS on Zen1
Borislav Petkov (AMD) [Tue, 7 Apr 2026 09:40:03 +0000 (11:40 +0200)] 
x86/CPU: Fix FPDSS on Zen1

Zen1's hardware divider can leave, under certain circumstances, partial
results from previous operations.  Those results can be leaked by
another, attacker thread.

Fix that with a chicken bit.

Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4 weeks agoMerge tag 'for-linus-iommufd' of git://git.kernel.org/pub/scm/linux/kernel/git/jgg...
Linus Torvalds [Fri, 17 Apr 2026 04:21:55 +0000 (21:21 -0700)] 
Merge tag 'for-linus-iommufd' of git://git.kernel.org/pub/scm/linux/kernel/git/jgg/iommufd

Pull iommufd updates from Jason Gunthorpe:
 "Several fixes:

   - Add missing static const

   - Correct type 1 emulation for VFIO_CHECK_EXTENSION when no-iommu is
     turned on

   - Fix selftest memory leak and syzkaller splat

   - Fix missed -EFAULT in fault reporting write() fops

   - Fix a race where map/unmap with the internal IOVA allocator can
     unmap things it should not"

* tag 'for-linus-iommufd' of git://git.kernel.org/pub/scm/linux/kernel/git/jgg/iommufd:
  iommufd: Fix a race with concurrent allocation and unmap
  iommufd/selftest: Remove MOCK_IOMMUPT_AMDV1 format
  iommufd: Fix return value of iommufd_fault_fops_write()
  iommufd: update outdated comment for renamed iommufd_hw_pagetable_alloc()
  iommufd/selftest: Fix page leaks in mock_viommu_{init,destroy}
  iommufd: vfio compatibility extension check for noiommu mode
  iommufd: Constify struct dma_buf_attach_ops

4 weeks agoMerge tag 'for-linus-fwctl' of git://git.kernel.org/pub/scm/linux/kernel/git/fwctl...
Linus Torvalds [Fri, 17 Apr 2026 04:15:56 +0000 (21:15 -0700)] 
Merge tag 'for-linus-fwctl' of git://git.kernel.org/pub/scm/linux/kernel/git/fwctl/fwctl

Pull fwctl updates from Jason Gunthorpe:

 - New fwctl driver for Broadcom RDMA NICs

 - Bug fix for non-modular builds

* tag 'for-linus-fwctl' of git://git.kernel.org/pub/scm/linux/kernel/git/fwctl/fwctl:
  fwctl: Fix class init ordering to avoid NULL pointer dereference on device removal
  fwctl/bnxt_fwctl: Add documentation entries
  fwctl/bnxt_fwctl: Add bnxt fwctl device
  fwctl/bnxt_en: Create an aux device for fwctl
  fwctl/bnxt_en: Refactor aux bus functions to be more generic
  fwctl/bnxt_en: Move common definitions to include/linux/bnxt/

4 weeks agoMerge tag 'soc-arm-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Fri, 17 Apr 2026 03:45:14 +0000 (20:45 -0700)] 
Merge tag 'soc-arm-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc

Pull SoC ARM code updates from Arnd Bergmann:
 "These are again very minimal updates:

   - A workaround for firmware on Google Nexus 10

   - A fix for early debugging on OMAP1

   - A rework for Microchip SoC configuration

   - Cleanups on OMAP2 an R-Car-Gen2"

* tag 'soc-arm-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc:
  ARM: omap2: dead code cleanup in kconfig for ARCH_OMAP4
  ARM: OMAP1: Fix DEBUG_LL and earlyprintk on OMAP16XX
  arm64: Kconfig: provide a top-level switch for Microchip platforms
  ARM: shmobile: rcar-gen2: Use of_phandle_args_equal() helper
  ARM: omap: fix all kernel-doc warnings
  ARM: omap2: Replace scnprintf with strscpy in omap3_cpuinfo
  ARM: samsung: exynos5250: Allow CPU1 to boot

4 weeks agoMerge tag 'soc-defconfig-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Fri, 17 Apr 2026 03:40:20 +0000 (20:40 -0700)] 
Merge tag 'soc-defconfig-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc

Pull SoC defconfig updates from Arnd Bergmann:
 "As usual, we enable a number of additional device drivers as loadable
  modules, to support the added platforms. The largest change this time
  is for OMAP2/3, which were not that well supported in the generic
  arm32 defconfig.

  The Tegra SoC platforms are now enabled by default in Kconfig when
  ARCH_TEGRA is enabled, which means the defconfig change is done at the
  same time as the Kconfig change here"

* tag 'soc-defconfig-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (25 commits)
  arch/arm: Drop CONFIG_FIRMWARE_EDID from defconfig files
  arm64: defconfig: Enable DP83TG720 PHY driver
  arm64: tegra: defconfig: Drop redundant ARCH_TEGRA_foo_SOC
  ARM: tegra: defconfig: Drop redundant ARCH_TEGRA_foo_SOC
  arm64: defconfig: enable pci-pwrctrl-generic as module
  arm64: defconfig: Enable Lontium LT8713sx driver
  arm64: defconfig: Enable Qualcomm Eliza SoC display clock controller
  arm64: defconfig: enable IPQ5210 RDP504 base configs
  arm64: defconfig: Enable Milos LPASS LPI pinctrl driver
  arm64: defconfig: Enable Kaanapali clock controllers
  arm64: defconfig: Enable configs for Arduino VENTUNO Q
  arm64: defconfig: Enable Qualcomm Eliza basic resource providers
  arm64: defconfig: Enable S5KJN1 camera sensor
  arm64: defconfig: Enable configurations for Toradex Aquila AM69
  arm64: defconfig: remove SENSORS_SA67MCU
  arm64: defconfig: Enable Qualcomm WCD937x headphone codec as module
  arm64: defconfig: Enable QCOMTEE module for QTEE-enabled Qualcomm SoCs
  ARM: shmobile: defconfig: Refresh for v7.0-rc1
  arm: multi_v7_defconfig: Enable more OMAP 3/4 related configs
  ARM: multi_v7_defconfig: omap2plus_defconfig: Enable ITE IT66121 driver
  ...

4 weeks agoMerge tag 'soc-drivers-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Fri, 17 Apr 2026 03:34:34 +0000 (20:34 -0700)] 
Merge tag 'soc-drivers-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc

Pull SoC driver updates from Arnd Bergmann:
 "The driver updates again are all over the place with many minor fixes
  going into platform specific code. The most notable changes are:

   - Support for Microchip pic64gx system controllers
   - Work on cleaning up devicetree bindings for SoC drivers, and
     converting them into the new format
   - Lots of smaller changes for Qualcomm SoC drivers, including support
     for a number of newly supported chips
   - reset controller API cleanups and a new driver for Cix Sky1
   - Reworks of the Tegra PMC and CBB drivers, along with a change to
     how individual Tegra SoCs get selected in Kconfig and BPMP firmware
     driver updates including a refresh of the ABI header to match the
     version used by firmware
   - STM32 updates to the firewall bus driver and support for the debug
     bus through OP-TEE
   - SCMI firmware driver improvements for reliability, in particular
     for dealing with broken firmware interrupts
   - Memory driver updates for Tegra, and a patch to remove the unused
     Baikal T1 driver"

* tag 'soc-drivers-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (193 commits)
  firmware: arm_ffa: Use the correct buffer size during RXTX_MAP
  firmware: qcom: scm: Allow QSEECOM on Lenovo IdeaCentre Mini X
  clk: spear: fix resource leak in clk_register_vco_pll()
  reset: rzv2h-usb2phy: Add support for VBUS mux controller registration
  reset: rzv2h-usb2phy: Convert to regmap API
  dt-bindings: reset: renesas,rzv2h-usb2phy: Document RZ/G3E USB2PHY reset
  dt-bindings: reset: renesas,rzv2h-usb2phy: Add '#mux-state-cells' property
  soc: microchip: add mpfs gpio interrupt mux driver
  dt-bindings: soc: microchip: document PolarFire SoC's gpio interrupt mux
  gpio: mpfs: Add interrupt support
  soc: qcom: ubwc: add helpers to get programmable values
  soc: qcom: ubwc: add helper to get min_acc length
  firmware: qcom: scm: Register gunyah watchdog device
  soc: qcom: socinfo: Add SoC ID for SA8650P
  dt-bindings: arm: qcom,ids: Add SoC ID for SA8650P
  firmware: qcom: scm: Allow QSEECOM on Mahua CRD
  soc: qcom: wcnss: simplify allocation of req
  soc: qcom: pd-mapper: Add support for Eliza
  soc: qcom: aoss: compare against normalized cooling state
  soc: qcom: llcc: fix v1 SB syndrome register offset
  ...

4 weeks agoMerge tag 'soc-dt-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Fri, 17 Apr 2026 03:28:48 +0000 (20:28 -0700)] 
Merge tag 'soc-dt-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc

Pull SoC devicetree updates from Arnd Bergmann:
 "A number of SoC platforms are adding modernized variants of their
  already supported chips time, with a total of 12 new SoCs, and two
  older SoC getting removed:

   - Qualcomm Glymur is a compute SoC using 18 Oryon-2 CPU cores
   - Qualcomm Mahua is a variant of Glymur with only 12 CPU cores, but
     largely identical.
   - Qualcomm Eliza is an embeded platform for mobile phone (SM7750) and
     IOT (QC7790S/M) workloads
   - Qualcomm IPQ5210 is a wireless networking SoC using Cortex-A53
     cores
   - Qualcomm apq8084 and ipq806x had only rudimentary support but no
     actual products using them, so they are now gone.
   - Axis ARTPEC-9 is a follow-up to the ARTPEC-8 embedded SoC, using
     the Samsung SoC platform but now with Cortex-A55 cores
   - ARM Zena is a virtual platform in FVP using Cortex-A720AE cores,
     with additional versions planned to be merged in the future.
   - ARM corstone-1000-a320 is a reference platform for IOT, using
     low-end Cortex-A320 cores
   - Microchip LAN9691 is an updated 64-bit variant of the arm32 lan966x
     series of networking SoCs
   - Microchip PIC64GX is an embedded RISC-V chip using SIFIVE U54 CPU
     cores
   - Rockchip RV1103B is the low-end 32-bit single-core vision processor
   - Renesas RZ/G3L (r9a08g046) is an industrial embedded chip using
     Cortex-A55 cores, similar to the G3E and G3S variants we already
     supported.
   - NXP S32N79 is an automotive SoC using Cortex-A78AE cores, a
     significant upgrade from the older S32V and S32G series

  These all come with at least one reference board or an initial product
  using these, in total there are 67 newly added boards. The ones for
  already supported SoCs are:

   - Two more Aspeed BMC based boards
   - Three older tablets based on 32-bit OMAP4 and Exynos5 SoCs
   - One Set-top-box based on Allwinner H6
   - 22 additional industrial/embedded boards using 64-bit NXP i.MX8M or
     i.MX9 SoCs
   - 20 Qualcomm SoC based machines across all possible markets:
     workstation, gaming, laptop, phone, networking, reference, ...
   - Three more Rockchips rk35xx based boards
   - Four variants of the Toradex Verdin using TI AM62

  Other notable bits are:

   - A cleanup for the 32-bit Tegra paz00 board moved the last board
     specific code on Tegra into equivalent dts syntax.
   - There continues to be a significant number of fixes for static
     checking of dtc syntax, but it feels like this is slowing down,
     hopefully getting into a state where most known issues are
     addressed
   - Additional hardware support for many existing boards across SoC
     families, notably Qualcomm, Broadcom, i.MX2, i.MX6, Rockchips,
     STM32, Mediatek, Tegra, TI and Microchip"

* tag 'soc-dt-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (841 commits)
  arm64: dts: ti: k3: Use memory-region-names for r5f
  ARM: dts: imx: Add DT overlays for DH i.MX6 DHCOM SoM and boards
  ARM: dts: imx6sx: remove fallback compatible string fsl,imx28-lcdif
  ARM: dts: imx25: rename node name tcq to touchscreen
  ARM: dts: imx: b850v3: Disable unused usdhc4
  ARM: dts: imx: b850v3: Define GPIO line names
  ARM: dts: imx: b850v3: Use alphabetical sorting
  ARM: dts: imx: bx50v3: Configure phy-mode to eliminate a warning
  ARM: dts: imx: bx50v3: Configure switch PHY max-speed to 100Mbps
  ARM: dts: imx7ulp: Add CPU clock and OPP table support
  ARM: dts: imx7-mba7: Deassert BOOT_EN after boot
  ARM: dts: tqma7: add boot phase properties
  ARM: dts: imx7s: add boot phase properties
  ARM: dts: tqma6ul[l]: correct spelling of TQ-Systems
  ARM: dts: mba6ulx: add boot phase properties
  ARM: dts: imx6ul[l]-tqma6ul[l]: add boot phase properties
  ARM: dts: imx6ul/imx6ull: add boot phase properties
  ARM: dts: imx6qdl-mba6: add boot phase properties
  ARM: dts: imx6qdl-tqma6: add boot phase properties
  ARM: dts: imx6qdl: add boot phase properties
  ...

4 weeks agoMerge tag 'mm-nonmm-stable-2026-04-15-04-20' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Fri, 17 Apr 2026 03:11:56 +0000 (20:11 -0700)] 
Merge tag 'mm-nonmm-stable-2026-04-15-04-20' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm

Pull non-MM updates from Andrew Morton:

 - "pid: make sub-init creation retryable" (Oleg Nesterov)

   Make creation of init in a new namespace more robust by clearing away
   some historical cruft which is no longer needed. Also some
   documentation fixups

 - "selftests/fchmodat2: Error handling and general" (Mark Brown)

   Fix and a cleanup for the fchmodat2() syscall selftest

 - "lib: polynomial: Move to math/ and clean up" (Andy Shevchenko)

 - "hung_task: Provide runtime reset interface for hung task detector"
   (Aaron Tomlin)

   Give administrators the ability to zero out
   /proc/sys/kernel/hung_task_detect_count

 - "tools/getdelays: use the static UAPI headers from
   tools/include/uapi" (Thomas Weißschuh)

   Teach getdelays to use the in-kernel UAPI headers rather than the
   system-provided ones

 - "watchdog/hardlockup: Improvements to hardlockup" (Mayank Rungta)

   Several cleanups and fixups to the hardlockup detector code and its
   documentation

 - "lib/bch: fix undefined behavior from signed left-shifts" (Josh Law)

   A couple of small/theoretical fixes in the bch code

 - "ocfs2/dlm: fix two bugs in dlm_match_regions()" (Junrui Luo)

 - "cleanup the RAID5 XOR library" (Christoph Hellwig)

   A quite far-reaching cleanup to this code. I can't do better than to
   quote Christoph:

     "The XOR library used for the RAID5 parity is a bit of a mess right
      now. The main file sits in crypto/ despite not being cryptography
      and not using the crypto API, with the generic implementations
      sitting in include/asm-generic and the arch implementations
      sitting in an asm/ header in theory. The latter doesn't work for
      many cases, so architectures often build the code directly into
      the core kernel, or create another module for the architecture
      code.

      Change this to a single module in lib/ that also contains the
      architecture optimizations, similar to the library work Eric
      Biggers has done for the CRC and crypto libraries later. After
      that it changes to better calling conventions that allow for
      smarter architecture implementations (although none is contained
      here yet), and uses static_call to avoid indirection function call
      overhead"

 - "lib/list_sort: Clean up list_sort() scheduling workarounds"
   (Kuan-Wei Chiu)

   Clean up this library code by removing a hacky thing which was added
   for UBIFS, which UBIFS doesn't actually need

 - "Fix bugs in extract_iter_to_sg()" (Christian Ehrhardt)

   Fix a few bugs in the scatterlist code, add in-kernel tests for the
   now-fixed bugs and fix a leak in the test itself

 - "kdump: Enable LUKS-encrypted dump target support in ARM64 and
   PowerPC" (Coiby Xu)

   Enable support of the LUKS-encrypted device dump target on arm64 and
   powerpc

 - "ocfs2: consolidate extent list validation into block read callbacks"
   (Joseph Qi)

   Cleanup, simplify, and make more robust ocfs2's validation of extent
   list fields (Kernel test robot loves mounting corrupted fs images!)

* tag 'mm-nonmm-stable-2026-04-15-04-20' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm: (127 commits)
  ocfs2: validate group add input before caching
  ocfs2: validate bg_bits during freefrag scan
  ocfs2: fix listxattr handling when the buffer is full
  doc: watchdog: fix typos etc
  update Sean's email address
  ocfs2: use get_random_u32() where appropriate
  ocfs2: split transactions in dio completion to avoid credit exhaustion
  ocfs2: remove redundant l_next_free_rec check in __ocfs2_find_path()
  ocfs2: validate extent block list fields during block read
  ocfs2: remove empty extent list check in ocfs2_dx_dir_lookup_rec()
  ocfs2: validate dx_root extent list fields during block read
  ocfs2: fix use-after-free in ocfs2_fault() when VM_FAULT_RETRY
  ocfs2: handle invalid dinode in ocfs2_group_extend
  .get_maintainer.ignore: add Askar
  ocfs2: validate bg_list extent bounds in discontig groups
  checkpatch: exclude forward declarations of const structs
  tools/accounting: handle truncated taskstats netlink messages
  taskstats: set version in TGID exit notifications
  ocfs2/heartbeat: fix slot mapping rollback leaks on error paths
  arm64,ppc64le/kdump: pass dm-crypt keys to kdump kernel
  ...

4 weeks agoMerge tag 'v7.1-rc1-part2-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Fri, 17 Apr 2026 02:14:55 +0000 (19:14 -0700)] 
Merge tag 'v7.1-rc1-part2-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull smb client updates from Steve French:

 - Fix integer underflow in encrypted read

 - Four debug patches, adding a few tracepoints

 - Minor update to MAINTAINERS file (preferred server URL for cifs)

 - Remove the BUG_ON() calls in d_mark_tmpfile_name

* tag 'v7.1-rc1-part2-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  MAINTAINERS: change git.samba.org to https
  smb: client: fix integer underflow in receive_encrypted_read()
  smb: client: add tracepoints for deferred handle caching
  smb: client: add oplock level to smb3_open_done tracepoint
  smb: client: add tracepoint for local lock conflicts
  smb: client: add tracepoints for lock operations
  vfs: get rid of BUG_ON() in d_mark_tmpfile_name()

4 weeks agoMerge tag 'v7.1-rc-part1-smbdirect-fixes' of git://git.samba.org/ksmbd
Linus Torvalds [Thu, 16 Apr 2026 15:25:04 +0000 (08:25 -0700)] 
Merge tag 'v7.1-rc-part1-smbdirect-fixes' of git://git.samba.org/ksmbd

Pull smbdirect updates from Steve French:
 "Move smbdirect server and client code to common directory:

   - temporary use of smbdirect_all_c_files.c to allow micro steps

   - factor out common functions into a smbdirect.ko.

   - convert cifs.ko to use smbdirect.ko

   - convert ksmbd.ko to use smbdirect.ko

   - let smbdirect.ko use global workqueues

   - move ib_client logic from ksmbd.ko into smbdirect.ko

   - remove smbdirect_all_c_files.c hack again

   - some locking and teardown related fixes on top"

* tag 'v7.1-rc-part1-smbdirect-fixes' of git://git.samba.org/ksmbd: (145 commits)
  smb: smbdirect: let smbdirect_connection_deregister_mr_io unlock while waiting
  smb: smbdirect: fix the logic in smbdirect_socket_destroy_sync() without an error
  smb: smbdirect: fix copyright header of smbdirect.h
  smb: smbdirect: change smbdirect_socket_parameters.{initiator_depth,responder_resources} to __u16
  smb: smbdirect: remove unused SMBDIRECT_USE_INLINE_C_FILES logic
  smb: server: no longer use smbdirect_socket_set_custom_workqueue()
  smb: client: no longer use smbdirect_socket_set_custom_workqueue()
  smb: smbdirect: introduce global workqueues
  smb: smbdirect: prepare use of dedicated workqueues for different steps
  smb: smbdirect: remove unused smbdirect_connection_mr_io_recovery_work()
  smb: smbdirect: wrap rdma_disconnect() in rdma_[un]lock_handler()
  smb: server: make use of smbdirect_netdev_rdma_capable_mode_type()
  smb: smbdirect: introduce smbdirect_netdev_rdma_capable_mode_type()
  smb: server: make use of smbdirect.ko
  smb: server: remove unused ksmbd_transport_ops.prepare()
  smb: server: make use of smbdirect_socket_{listen,accept}()
  smb: server: only use public smbdirect functions
  smb: server: make use of smbdirect_socket_create_accepting()/smbdirect_socket_release()
  smb: server: make use of smbdirect_{socket_init_accepting,connection_wait_for_connected}()
  smb: server: make use of smbdirect_connection_send_iter() and related functions
  ...

4 weeks agoMerge tag 'livepatching-for-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 16 Apr 2026 15:13:27 +0000 (08:13 -0700)] 
Merge tag 'livepatching-for-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/livepatching/livepatching

Pull livepatching updates from Petr Mladek:

 - Add two new selftests

* tag 'livepatching-for-7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/livepatching/livepatching:
  selftests/livepatch: add test for module function patching
  selftests: livepatch: test-ftrace: livepatch a traced function

4 weeks agoMerge tag 'm68k-for-v7.1-tag1' of git://git.kernel.org/pub/scm/linux/kernel/git/geert...
Linus Torvalds [Thu, 16 Apr 2026 15:11:01 +0000 (08:11 -0700)] 
Merge tag 'm68k-for-v7.1-tag1' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux-m68k

Pull m68k updates from Geert Uytterhoeven:

 - Add support for QEMU virt-ctrl, and use it for system reset
   and power off on the virt platform

 - defconfig updates

 - Miscellaneous fixes and improvements

* tag 'm68k-for-v7.1-tag1' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux-m68k:
  m68k: virt: Switch to qemu-virt-ctrl driver
  power: reset: Add QEMU virt-ctrl driver
  m68k: defconfig: Update defconfigs for v7.0-rc1
  m68k: emu: Replace unbounded sprintf() in nfhd_init_one()
  m68k: uapi: Add ucontext.h
  m68k: defconfig: hp300: Enable monochrome and 16-color linux logos
  m68k: q40: Remove commented out code

4 weeks agoMerge tag 'efi-next-for-v7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi
Linus Torvalds [Thu, 16 Apr 2026 15:06:25 +0000 (08:06 -0700)] 
Merge tag 'efi-next-for-v7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi

Pull EFI updates from Ard Biesheuvel:
 "Again not a busy cycle for EFI, just some minor tweaks and bug fixes:

   - Enable boot graphics resource table (BGRT) on Xen/x86

   - Correct a misguided assumption in the memory attributes table
     sanity check

   - Start tagging efi_mem_reserve()'d regions as MEMBLOCK_RSRV_KERN

   - Some other minor fixes and cleanups"

* tag 'efi-next-for-v7.1' of git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi:
  efi/capsule-loader: fix incorrect sizeof in phys array reallocation
  efi: Tag memblock reservations of boot services regions as RSRV_KERN
  memblock: Permit existing reserved regions to be marked RSRV_KERN
  efi/memattr: Fix thinko in table size sanity check
  efi: libstub: fix type of fdt 32 and 64bit variables
  efi: Drop unused efi_range_is_wc() function
  efi: Enable BGRT loading under Xen
  efi: make efi_mem_type() and efi_mem_attributes() work on Xen PV

4 weeks agoMerge tag 'vfio-v7.1-rc1' of https://github.com/awilliam/linux-vfio
Linus Torvalds [Thu, 16 Apr 2026 15:01:16 +0000 (08:01 -0700)] 
Merge tag 'vfio-v7.1-rc1' of https://github.com/awilliam/linux-vfio

Pull VFIO updates from Alex Williamson:

 - Update QAT vfio-pci variant driver for Gen 5, 420xx devices (Vijay
   Sundar Selvamani, Suman Kumar Chakraborty, Giovanni Cabiddu)

 - Fix vfio selftest MMIO DMA mapping selftest (Alex Mastro)

 - Conversions to const struct class in support of class_create()
   deprecation (Jori Koolstra)

 - Improve selftest compiler compatibility by avoiding initializer on
   variable-length array (Manish Honap)

 - Define new uAPI for drivers supporting migration to advise user-
   space of new initial data for reducing target startup latency.
   Implemented for mlx5 vfio-pci variant driver (Yishai Hadas)

 - Enable vfio selftests on aarch64, not just cross-compiles reporting
   arm64 (Ted Logan)

 - Update vfio selftest driver support to include additional DSA devices
   (Yi Lai)

 - Unconditionally include debugfs root pointer in vfio device struct,
   avoiding a build failure seen in hisi_acc variant driver without
   debugfs otherwise (Arnd Bergmann)

 - Add support for the s390 ISM (Internal Shared Memory) device via a
   new variant driver. The device is unique in the size of its BAR space
   (256TiB) and lack of mmap support (Julian Ruess)

 - Enforce that vfio-pci drivers implement a name in their ops structure
   for use in sequestering SR-IOV VFs (Alex Williamson)

 - Prune leftover group notifier code (Paolo Bonzini)

 - Fix Xe vfio-pci variant driver to avoid migration support as a
   dependency in the reset path and missing release call (MichaÅ‚
   Winiarski)

* tag 'vfio-v7.1-rc1' of https://github.com/awilliam/linux-vfio: (23 commits)
  vfio/xe: Add a missing vfio_pci_core_release_dev()
  vfio/xe: Reorganize the init to decouple migration from reset
  vfio: remove dead notifier code
  vfio/pci: Require vfio_device_ops.name
  MAINTAINERS: add VFIO ISM PCI DRIVER section
  vfio/ism: Implement vfio_pci driver for ISM devices
  vfio/pci: Rename vfio_config_do_rw() to vfio_pci_config_rw_single() and export it
  vfio: unhide vdev->debug_root
  vfio/qat: add support for Intel QAT 420xx VFs
  vfio: selftests: Support DMR and GNR-D DSA devices
  vfio: selftests: Build tests on aarch64
  vfio/mlx5: Add REINIT support to VFIO_MIG_GET_PRECOPY_INFO
  vfio/mlx5: consider inflight SAVE during PRE_COPY
  net/mlx5: Add IFC bits for migration state
  vfio: Adapt drivers to use the core helper vfio_check_precopy_ioctl
  vfio: Add support for VFIO_DEVICE_FEATURE_MIG_PRECOPY_INFOv2
  vfio: Define uAPI for re-init initial bytes during the PRE_COPY phase
  vfio: selftests: Fix VLA initialisation in vfio_pci_irq_set()
  vfio: uapi: fix comment typo
  vfio: mdev: replace mtty_dev->vd_class with a const struct class
  ...

4 weeks agoMerge branch 'for-7.1/module-function-test' into for-linus
Petr Mladek [Thu, 16 Apr 2026 08:33:43 +0000 (10:33 +0200)] 
Merge branch 'for-7.1/module-function-test' into for-linus

4 weeks agosmb: smbdirect: let smbdirect_connection_deregister_mr_io unlock while waiting
Stefan Metzmacher [Tue, 7 Apr 2026 14:46:30 +0000 (16:46 +0200)] 
smb: smbdirect: let smbdirect_connection_deregister_mr_io unlock while waiting

We should not hold a mutex locked during wait_for_completion()
holding a reference is enough.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Henrique Carvalho <henrique.carvalho@suse.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: smbdirect: fix the logic in smbdirect_socket_destroy_sync() without an error
Stefan Metzmacher [Tue, 7 Apr 2026 14:46:29 +0000 (16:46 +0200)] 
smb: smbdirect: fix the logic in smbdirect_socket_destroy_sync() without an error

If smbdirect_socket_destroy_sync() and sc->first_error was not set
we should set -ESHUTDOWN, that's a better condition
doing it only implicitly with the
sc->status < SMBDIRECT_SOCKET_DISCONNECTING check.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Henrique Carvalho <henrique.carvalho@suse.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: smbdirect: fix copyright header of smbdirect.h
Stefan Metzmacher [Tue, 7 Apr 2026 14:46:28 +0000 (16:46 +0200)] 
smb: smbdirect: fix copyright header of smbdirect.h

Everything in smbdirect.h was taken from my out of
tree prototype.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Henrique Carvalho <henrique.carvalho@suse.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: smbdirect: change smbdirect_socket_parameters.{initiator_depth,responder_resourc...
Stefan Metzmacher [Tue, 7 Apr 2026 14:46:27 +0000 (16:46 +0200)] 
smb: smbdirect: change smbdirect_socket_parameters.{initiator_depth,responder_resources} to __u16

We still limit this to U8_MAX as the rdma api only uses __u8
and that's also the limit for Infiniband and RoCE*,
while iWarp would be able to support larger values at
the protocol level.

As struct smbdirect_socket_parameters will be part
of the uapi for IPPROTO_SMBDIRECT in future, change it
now even if userspace sockets won't be supported yet.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: David Howells <dhowells@redhat.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Acked-by: Henrique Carvalho <henrique.carvalho@suse.com>
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: smbdirect: remove unused SMBDIRECT_USE_INLINE_C_FILES logic
Stefan Metzmacher [Thu, 30 Oct 2025 11:48:30 +0000 (12:48 +0100)] 
smb: smbdirect: remove unused SMBDIRECT_USE_INLINE_C_FILES logic

We always build as standalone module (or as part of the core kernel).

This also removes unused elements from struct smbdirect_socket
and unused exports.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: no longer use smbdirect_socket_set_custom_workqueue()
Stefan Metzmacher [Tue, 4 Nov 2025 16:35:46 +0000 (17:35 +0100)] 
smb: server: no longer use smbdirect_socket_set_custom_workqueue()

smbdirect.ko has global workqueues now, so we should use these
default once.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: no longer use smbdirect_socket_set_custom_workqueue()
Stefan Metzmacher [Tue, 4 Nov 2025 16:31:25 +0000 (17:31 +0100)] 
smb: client: no longer use smbdirect_socket_set_custom_workqueue()

smbdirect.ko has global workqueues now, so we should use these
default once.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: smbdirect: introduce global workqueues
Stefan Metzmacher [Tue, 4 Nov 2025 16:25:48 +0000 (17:25 +0100)] 
smb: smbdirect: introduce global workqueues

These will be used in future and callers should no
longer use smbdirect_socket_set_custom_workqueue().

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: smbdirect: prepare use of dedicated workqueues for different steps
Stefan Metzmacher [Tue, 4 Nov 2025 16:02:29 +0000 (17:02 +0100)] 
smb: smbdirect: prepare use of dedicated workqueues for different steps

This is a preparation in order to have global workqueues in
the smbdirect module instead of having the caller to
provide one.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: smbdirect: remove unused smbdirect_connection_mr_io_recovery_work()
Stefan Metzmacher [Tue, 4 Nov 2025 15:29:38 +0000 (16:29 +0100)] 
smb: smbdirect: remove unused smbdirect_connection_mr_io_recovery_work()

This would actually never be used as we only move to
SMBDIRECT_MR_ERROR when we directly call
smbdirect_socket_schedule_cleanup().

Doing an ib_dereg_mr/ib_alloc_mr dance on
working connection is not needed and
it's also pointless on a broken connection
as we don't reuse any ib_pd.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: smbdirect: wrap rdma_disconnect() in rdma_[un]lock_handler()
Stefan Metzmacher [Mon, 24 Nov 2025 13:49:55 +0000 (14:49 +0100)] 
smb: smbdirect: wrap rdma_disconnect() in rdma_[un]lock_handler()

This might not be needed, but it controls the order
of ib_drain_qp() and rdma_disconnect().

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_netdev_rdma_capable_mode_type()
Stefan Metzmacher [Tue, 9 Dec 2025 12:39:52 +0000 (13:39 +0100)] 
smb: server: make use of smbdirect_netdev_rdma_capable_mode_type()

This removes is basically the same logic.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: smbdirect: introduce smbdirect_netdev_rdma_capable_mode_type()
Stefan Metzmacher [Mon, 8 Dec 2025 19:56:45 +0000 (20:56 +0100)] 
smb: smbdirect: introduce smbdirect_netdev_rdma_capable_mode_type()

This is basically a copy of ksmbd_rdma_capable_netdev() in the
server, but this also prints a message when a device is renamed.

The differences are:
- It uses rdma_for_each_port() instead of implementing the
  same logic again.
- It returns RDMA_NODE_{UNSPECIFIED,IB_CA,RNIC} values instead of bool

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect.ko
Stefan Metzmacher [Fri, 18 Jul 2025 17:22:47 +0000 (19:22 +0200)] 
smb: server: make use of smbdirect.ko

This means we no longer inline the common smbdirect
.c files and use the exported functions from the
module instead.

Note the connection specific logging is still
redirect to ksmbd.ko functions via
smbdirect_socket_set_logging().

We still don't use real socket layer,
but we're very close...

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: remove unused ksmbd_transport_ops.prepare()
Stefan Metzmacher [Tue, 25 Nov 2025 16:44:31 +0000 (17:44 +0100)] 
smb: server: remove unused ksmbd_transport_ops.prepare()

This is no longer needed for smbdirect.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_socket_{listen,accept}()
Stefan Metzmacher [Fri, 14 Nov 2025 14:41:02 +0000 (15:41 +0100)] 
smb: server: make use of smbdirect_socket_{listen,accept}()

We no longer need the custom rdma listener.

The code logic is very similar to transport_tcp.c now
using a kernel thread that loops over smbdirect_socket_accept().

This is the first step in the direction of using IPPROTO_SMBDIRECT
sockets in future.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: only use public smbdirect functions
Stefan Metzmacher [Mon, 27 Oct 2025 20:39:19 +0000 (21:39 +0100)] 
smb: server: only use public smbdirect functions

Also remove a lot of unused includes...

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_socket_create_accepting()/smbdirect_socket_release()
Stefan Metzmacher [Thu, 23 Oct 2025 20:49:27 +0000 (22:49 +0200)] 
smb: server: make use of smbdirect_socket_create_accepting()/smbdirect_socket_release()

With this we no longer embed struct smbdirect_socket, which will allow
us to make it private in the following commits.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_{socket_init_accepting,connection_wait_for_connect...
Stefan Metzmacher [Sat, 20 Sep 2025 22:21:41 +0000 (00:21 +0200)] 
smb: server: make use of smbdirect_{socket_init_accepting,connection_wait_for_connected}()

This means we finally only use common functions in the server.

We still use the embedded struct smbdirect_socket and are
able to access internals, but the will be removed in the
next commits as well.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_send_iter() and related functions
Stefan Metzmacher [Fri, 17 Oct 2025 20:16:56 +0000 (22:16 +0200)] 
smb: server: make use of smbdirect_connection_send_iter() and related functions

This makes use of common code for sending messages, this will
allow to make more use of common code in the next commits.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: let smb_direct_post_send_data() return data_length
Stefan Metzmacher [Fri, 17 Oct 2025 18:20:02 +0000 (20:20 +0200)] 
smb: server: let smb_direct_post_send_data() return data_length

This make it easier moving to common code shared with the client.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: split out smb_direct_send_iter() out of smb_direct_writev()
Stefan Metzmacher [Fri, 17 Oct 2025 17:25:12 +0000 (19:25 +0200)] 
smb: server: split out smb_direct_send_iter() out of smb_direct_writev()

This will help to move to common code in future.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: let smbdirect_map_sges_from_iter() truncate the message boundary
Stefan Metzmacher [Fri, 17 Oct 2025 15:58:16 +0000 (17:58 +0200)] 
smb: server: let smbdirect_map_sges_from_iter() truncate the message boundary

smbdirect_map_sges_from_iter() already handles the case that only
a limited number of sges are available. Its return value
is data_length and the remaining bytes in the iter are
remaining_data_length.

This is now much easier and will allow us to share
more code with the client soon.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: inline smb_direct_create_header() into smb_direct_post_send_data()
Stefan Metzmacher [Fri, 17 Oct 2025 15:24:28 +0000 (17:24 +0200)] 
smb: server: inline smb_direct_create_header() into smb_direct_post_send_data()

The point is that ib_dma_map_single() is done first, but
the 'Fill in the packet header' will be done after
smbdirect_map_sges_from_iter().

This will simplify further changes in order to
share common code with the client.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: move iov_iter_kvec() out of smb_direct_post_send_data()
Stefan Metzmacher [Fri, 17 Oct 2025 14:51:42 +0000 (16:51 +0200)] 
smb: server: move iov_iter_kvec() out of smb_direct_post_send_data()

This will allow us to make the code more generic in order
to move it to common with the client.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_request_keep_alive()
Stefan Metzmacher [Fri, 17 Oct 2025 13:53:04 +0000 (15:53 +0200)] 
smb: server: make use of smbdirect_connection_request_keep_alive()

This will help to share more common code soon.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_grant_recv_credits()
Stefan Metzmacher [Fri, 17 Oct 2025 13:45:01 +0000 (15:45 +0200)] 
smb: server: make use of smbdirect_connection_grant_recv_credits()

This is already used by the client too and will
help to share more common code.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_recvmsg()
Stefan Metzmacher [Mon, 13 Oct 2025 16:49:30 +0000 (18:49 +0200)] 
smb: server: make use of smbdirect_connection_recvmsg()

This is basically the same logic, it just operates on iov_iter_kvec()
instead of a raw buffer pointer. This allows us to use common
code between client and server.

We keep returning -EINTR instead of -ERESTARTSYS if
wait_event_interruptible() fails. I don't if this is
required, but changing it is a task for another patch.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_socket_destroy_sync()
Stefan Metzmacher [Mon, 13 Oct 2025 14:36:07 +0000 (16:36 +0200)] 
smb: server: make use of smbdirect_socket_destroy_sync()

This is basically the same logic as before, but we now
use common code, which will also be used by the server soon.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of functions from smbdirect_rw.c
Stefan Metzmacher [Fri, 19 Sep 2025 08:23:24 +0000 (10:23 +0200)] 
smb: server: make use of functions from smbdirect_rw.c

The copied code only got new names, some indentation/formatting changes,
some variable names are changed too.

They also only use struct smbdirect_socket instead of
struct smb_direct_transport.

But the logic is still the same.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_socket_wait_for_credits()
Stefan Metzmacher [Fri, 19 Sep 2025 07:52:35 +0000 (09:52 +0200)] 
smb: server: make use of smbdirect_socket_wait_for_credits()

This will allow us to share more common code between client and
server soon.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_get_buf_page_count()
Stefan Metzmacher [Fri, 19 Sep 2025 07:41:16 +0000 (09:41 +0200)] 
smb: server: make use of smbdirect_get_buf_page_count()

This will allow us to move code into common code
between client and server soon.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_recv_io_refill[_work]()
Stefan Metzmacher [Thu, 18 Sep 2025 19:50:44 +0000 (21:50 +0200)] 
smb: server: make use of smbdirect_connection_recv_io_refill[_work]()

This is basically a copy of smb_direct_post_recv_credits(), but
there are several improvements compared to the existing function:

1. We calculate the number of missing posted buffers by getting the
   difference between recv_io.credits.target and recv_io.posted.count.

   Instead of the difference between recv_io.credits.target
   and recv_io.credits.count, because recv_io.credits.count is
   only updated once a message is send to the peer.

   It was not really a problem before, because we have
   a fixed number smbdirect_recv_io buffers, so the
   loop terminated when smbdirect_connection_get_recv_io()
   returns NULL.

   But using recv_io.posted.count makes it easier to
   understand.

2. In order to tell the peer about the newly posted buffer
   and grant the credits, we only trigger the send immediate
   when we're not granting only the last possible credit.

   This is mostly a difference relative to the servers
   smb_direct_post_recv_credits() implementation,
   which should avoid useless ping pong messages.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_post_recv_io()
Stefan Metzmacher [Thu, 18 Sep 2025 17:52:03 +0000 (19:52 +0200)] 
smb: server: make use of smbdirect_connection_post_recv_io()

The only difference is that smbdirect_connection_post_recv_io()
returns early if the connection is already broken.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_{create,destroy}_qp()
Stefan Metzmacher [Thu, 18 Sep 2025 16:48:40 +0000 (18:48 +0200)] 
smb: server: make use of smbdirect_connection_{create,destroy}_qp()

It's good a use common code for this and it will allow us
to share more code in the next steps.

Calling ib_drain_qp() twice is ok.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_negotiate_rdma_resources()
Stefan Metzmacher [Thu, 18 Sep 2025 16:38:14 +0000 (18:38 +0200)] 
smb: server: make use of smbdirect_connection_negotiate_rdma_resources()

It's good to have this logic in a central place, it will allow us
share more code soon.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_qp_event_handler()
Stefan Metzmacher [Mon, 15 Sep 2025 05:59:20 +0000 (07:59 +0200)] 
smb: server: make use of smbdirect_connection_qp_event_handler()

This is a copy of smb_direct_qpair_handler()...

It will allow more code to be moved to common functions
soon.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_map_sges_from_iter()
Stefan Metzmacher [Mon, 15 Sep 2025 02:39:13 +0000 (04:39 +0200)] 
smb: server: make use of smbdirect_map_sges_from_iter()

It will make it easier to move stuff into common code when
both client and server use smbdirect_map_sges_from_iter().

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_{create,destroy}_mem_pools()
Stefan Metzmacher [Fri, 12 Sep 2025 22:18:50 +0000 (00:18 +0200)] 
smb: server: make use of smbdirect_connection_{create,destroy}_mem_pools()

This were based on smb_direct_{create,destroy}_pools() in the server.

The main logical differences are the following:

We now don't use smbdirect_connection_get_recv_io() on cleanup,
instead it uses list_for_each_entry_safe()...

We don't generate warnings if smbdirect_recv_io payload
is copied into userspace buffers. This doesn't happen
in the server anyway.

And it uses list_add_tail() just to let me feel
better when looking at the code...

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_send_io_done()
Stefan Metzmacher [Thu, 11 Sep 2025 18:54:31 +0000 (20:54 +0200)] 
smb: server: make use of smbdirect_connection_send_io_done()

This also wakes up send_io.pending.dec_wait_queue, which
is currently always empty in the server, but that might
change in future. And we also don't spam the logs on IB_WC_WR_FLUSH_ERR.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_{alloc,free}_send_io()
Stefan Metzmacher [Thu, 11 Sep 2025 18:12:11 +0000 (20:12 +0200)] 
smb: server: make use of smbdirect_connection_{alloc,free}_send_io()

These are basically copies of smb_direct_{alloc,free}_sendmsg() just
a bit simpler and with the gfp_mask mask abstracted.

For now we still use KSMBD_DEFAULT_GFP, which includes
__GFP_RETRY_MAYFAIL.

The only difference is that we use ib_dma_unmap_page() for all sges,
this simplifies the logic and doesn't matter as
ib_dma_unmap_single() and ib_dma_unmap_page() both operate
on dma_addr_t and dma_unmap_single_attrs() is just an
alias for dma_unmap_page_attrs().
We already had such an inconsistency in the client
code where we use ib_dma_unmap_single(), while we mapped
using ib_dma_map_page().

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_frwr_is_supported()
Stefan Metzmacher [Thu, 28 Aug 2025 15:01:58 +0000 (17:01 +0200)] 
smb: server: make use of smbdirect_frwr_is_supported()

This is an exact copy of rdma_frwr_is_supported().

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_idle_timer_work()
Stefan Metzmacher [Wed, 27 Aug 2025 15:25:37 +0000 (17:25 +0200)] 
smb: server: make use of smbdirect_connection_idle_timer_work()

This is basically a copy of smb_direct_idle_connection_timer().
The only difference is that we had no logging before.

Note smbdirect_socket_prepare_create() already calls INIT_DELAYED_WORK().

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_reassembly_{append,first}_recv_io()
Stefan Metzmacher [Wed, 27 Aug 2025 14:57:47 +0000 (16:57 +0200)] 
smb: server: make use of smbdirect_connection_reassembly_{append,first}_recv_io()

These are basically copies of enqueue_reassembly() and
get_first_reassembly().  The only difference is that
sc->statistics.enqueue_reassembly_queue now updated.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_connection_{get,put}_recv_io()
Stefan Metzmacher [Tue, 26 Aug 2025 17:16:54 +0000 (19:16 +0200)] 
smb: server: make use of smbdirect_connection_{get,put}_recv_io()

These are basically copies of {get,put}_receive_buffer() in the client.
They are very similar to {get_free,put}_recvmsg() the only logical
difference is the updating of the sc->statistics.*.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_socket_schedule_cleanup()
Stefan Metzmacher [Tue, 26 Aug 2025 15:31:39 +0000 (17:31 +0200)] 
smb: server: make use of smbdirect_socket_schedule_cleanup()

This removes smb_direct_disconnect_rdma_connection() which is basically
the same as smbdirect_socket_schedule_cleanup().
And we pass more useful errors than -ECONNABORTED if we have them.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_socket_cleanup_work()
Stefan Metzmacher [Tue, 26 Aug 2025 15:20:30 +0000 (17:20 +0200)] 
smb: server: make use of smbdirect_socket_cleanup_work()

Note smbdirect_socket_prepare_create() already calls INIT_WORK()
with smbdirect_socket_cleanup_work.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_socket_wake_up_all()
Stefan Metzmacher [Tue, 2 Sep 2025 10:55:04 +0000 (12:55 +0200)] 
smb: server: make use of smbdirect_socket_wake_up_all()

This is a superset of smb_direct_disconnect_wake_up_all() and
calling wake_up_all(&sc->mr_io.ready.wait_queue); and
wake_up_all(&sc->mr_io.cleanup.wait_queue); in addition
should not matter as it's not used on the server anyway.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_socket_set_logging()
Stefan Metzmacher [Thu, 23 Oct 2025 12:40:48 +0000 (14:40 +0200)] 
smb: server: make use of smbdirect_socket_set_logging()

This will allow the logging to keep working as before,
when we move to common functions in the next commits.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: server: make use of smbdirect_socket_prepare_create()
Stefan Metzmacher [Tue, 26 Aug 2025 13:32:48 +0000 (15:32 +0200)] 
smb: server: make use of smbdirect_socket_prepare_create()

This prepares the use of functions from smbdirect_connection.c.

Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect.ko
Stefan Metzmacher [Fri, 18 Jul 2025 17:22:33 +0000 (19:22 +0200)] 
smb: client: make use of smbdirect.ko

This means we no longer inline the common smbdirect
.c files and use the exported functions from the
module instead.

Note the connection specific logging is still
redirect to cifs.ko functions via
smbdirect_socket_set_logging().

We still don't use real socket layer,
but we're very close...

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: David Howells <dhowells@redhat.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: only use public smbdirect functions
Stefan Metzmacher [Mon, 27 Oct 2025 20:29:48 +0000 (21:29 +0100)] 
smb: client: only use public smbdirect functions

Also remove a lot of unused includes...

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_socket_create_kern()/smbdirect_socket_release()
Stefan Metzmacher [Fri, 24 Oct 2025 15:41:03 +0000 (17:41 +0200)] 
smb: client: make use of smbdirect_socket_create_kern()/smbdirect_socket_release()

With this we no longer embed struct smbdirect_socket, which will allow
us to make it private in the following commits.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_socket_init_new() and smbdirect_connect_sync()
Stefan Metzmacher [Fri, 19 Sep 2025 20:54:36 +0000 (22:54 +0200)] 
smb: client: make use of smbdirect_socket_init_new() and smbdirect_connect_sync()

This means we finally only use common functions in the client.

We still use the embedded struct smbdirect_socket and are
able to access internals, but the will be removed in the
next commits as well.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: introduce and use smbd_debug_proc_show()
Stefan Metzmacher [Tue, 21 Oct 2025 19:07:26 +0000 (21:07 +0200)] 
smb: client: introduce and use smbd_debug_proc_show()

This will allow us to make struct smbdirect_socket private
in future.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: introduce and use smbd_mr_fill_buffer_descriptor()
Stefan Metzmacher [Tue, 21 Oct 2025 18:43:44 +0000 (20:43 +0200)] 
smb: client: introduce and use smbd_mr_fill_buffer_descriptor()

This will allow us to make struct smbdirect_mr_io private in future.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_send_{single_iter,immediate_work}()
Stefan Metzmacher [Fri, 17 Oct 2025 20:23:10 +0000 (22:23 +0200)] 
smb: client: make use of smbdirect_connection_send_{single_iter,immediate_work}()

With this the low level send functions are in common,
we'll have to do some more changes in generic smb code
in order to use smbdirect_connection_send_iter() instead
of looping around smbdirect_connection_send_single_iter().

David's cleanups will allow us to use
smbdirect_connection_send_iter().

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: David Howells <dhowells@redhat.com>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: let smbd_post_send_full_iter() get remaining_length and return data_length
Stefan Metzmacher [Fri, 17 Oct 2025 19:03:45 +0000 (21:03 +0200)] 
smb: client: let smbd_post_send_full_iter() get remaining_length and return data_length

This will simplify further changes in order to share
more common code in future.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: let smbd_post_send_iter() get remaining_length and return data_length
Stefan Metzmacher [Fri, 17 Oct 2025 18:46:29 +0000 (20:46 +0200)] 
smb: client: let smbd_post_send_iter() get remaining_length and return data_length

This lets the logic be like smb_direct_post_send_data(), so
we can share common code in the next steps.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: change smbd_post_send_empty() to void return
Stefan Metzmacher [Fri, 17 Oct 2025 18:33:10 +0000 (20:33 +0200)] 
smb: client: change smbd_post_send_empty() to void return

The caller doesn't check, so we better call
smbdirect_socket_schedule_cleanup() to handle the error.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_request_keep_alive()
Stefan Metzmacher [Fri, 17 Oct 2025 13:51:26 +0000 (15:51 +0200)] 
smb: client: make use of smbdirect_connection_request_keep_alive()

This will help to share more common code soon.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_grant_recv_credits()
Stefan Metzmacher [Fri, 17 Oct 2025 13:37:32 +0000 (15:37 +0200)] 
smb: client: make use of smbdirect_connection_grant_recv_credits()

This already calls atomic_add(new_credits, &sc->recv_io.credits.count),
so there's no need to do it in the caller anymore.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_recvmsg()
Stefan Metzmacher [Mon, 13 Oct 2025 16:42:08 +0000 (18:42 +0200)] 
smb: client: make use of smbdirect_connection_recvmsg()

This is basically the same as it was copied before...

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_socket_destroy_sync()
Stefan Metzmacher [Mon, 13 Oct 2025 14:33:20 +0000 (16:33 +0200)] 
smb: client: make use of smbdirect_socket_destroy_sync()

This is basically the same logic as before, but we now
use common code, which will also be used by the server soon.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of functions from smbdirect_mr.c
Stefan Metzmacher [Fri, 19 Sep 2025 06:35:51 +0000 (08:35 +0200)] 
smb: client: make use of functions from smbdirect_mr.c

The copied code only got new names, some indentation/formatting changes,
some variable names are changed too.

They also only use struct smbdirect_socket instead of
struct smbd_connection.

But the logic is still the same.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_recv_io_refill_work()
Stefan Metzmacher [Thu, 18 Sep 2025 19:39:25 +0000 (21:39 +0200)] 
smb: client: make use of smbdirect_connection_recv_io_refill_work()

This is basically a copy of smbd_post_send_credits(), but
there are several improvements compared to the existing function:

  We calculate the number of missing posted buffers by getting the
  difference between recv_io.credits.target and recv_io.posted.count.

  Instead of the difference between recv_io.credits.target
  and recv_io.credits.count, because recv_io.credits.count is
  only updated once a message is send to the peer.

  It was not really a problem before, because we have
  a fixed number smbdirect_recv_io buffers, so the
  loop terminated when smbdirect_connection_get_recv_io()
  returns NULL.

  But using recv_io.posted.count makes it easier to
  understand.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_post_recv_io()
Stefan Metzmacher [Wed, 17 Sep 2025 07:18:08 +0000 (09:18 +0200)] 
smb: client: make use of smbdirect_connection_post_recv_io()

The only difference is that smbdirect_connection_post_recv_io()
returns early if the connection is already broken.

And that the error code from ib_dma_mapping_error() (currently only -ENOMEM
is possible) is returned instead of -EIO.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: initialize recv_io->cqe.done = recv_done just once
Stefan Metzmacher [Wed, 17 Sep 2025 06:54:03 +0000 (08:54 +0200)] 
smb: client: initialize recv_io->cqe.done = recv_done just once

smbdirect_recv_io structures are pre-allocated so we can set the
callback function just once.

This will make it easy to move smbd_post_recv() to common code
soon.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_{create,destroy}_qp()
Stefan Metzmacher [Mon, 15 Sep 2025 23:53:23 +0000 (01:53 +0200)] 
smb: client: make use of smbdirect_connection_{create,destroy}_qp()

It's good a use common code for this and it will allow us
to share more code in the next steps.

Calling ib_drain_qp() twice is ok.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_negotiate_rdma_resources()
Stefan Metzmacher [Thu, 18 Sep 2025 16:37:40 +0000 (18:37 +0200)] 
smb: client: make use of smbdirect_connection_negotiate_rdma_resources()

It's good to have this logic in a central place, it will allow us
share more code soon.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_qp_event_handler()
Stefan Metzmacher [Mon, 15 Sep 2025 05:55:19 +0000 (07:55 +0200)] 
smb: client: make use of smbdirect_connection_qp_event_handler()

This is a copy of smbd_qp_async_error_upcall()...

It will allow more code to be moved to common functions
soon.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_map_sges_from_iter()
Stefan Metzmacher [Mon, 15 Sep 2025 00:26:47 +0000 (02:26 +0200)] 
smb: client: make use of smbdirect_map_sges_from_iter()

This is basically a copy of smb_extract_iter_to_rdma() moved
to common code.

Before we had the inconsistency we called ib_dma_unmap_single(),
while we mapped using ib_dma_map_page() in smb_set_sge().

Now ib_dma_unmap_page() is used for consistency.

It doesn't really matter as ib_dma_unmap_single() and
ib_dma_unmap_page() both operate
on dma_addr_t and dma_unmap_single_attrs() is just an
alias for dma_unmap_page_attrs().

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_{create,destroy}_mem_pools()
Stefan Metzmacher [Fri, 12 Sep 2025 22:10:06 +0000 (00:10 +0200)] 
smb: client: make use of smbdirect_connection_{create,destroy}_mem_pools()

The main logical differences are the following:

We now don't use smbdirect_connection_get_recv_io() on cleanup,
instead it uses list_for_each_entry_safe()...

For the smbdirect_recv_io payload we expose the whole payload including
the smbdirect_data_transfer header as documentation says data_offset = 0
and data_length != 0 would be valid, while the existing client code
requires data_offset >= 24.

The smbdirect_send_io cache includes header space for
sizeof(struct smbdirect_negotiate_resp) = 32 bytes
instead of sizeof(struct smbdirect_data_transfer) = 24 bytes.
If this ever becomes a problem, we can allocate separate
space for the smbdirect_negotiate_resp in the server.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_send_io_done()
Stefan Metzmacher [Thu, 11 Sep 2025 18:51:53 +0000 (20:51 +0200)] 
smb: client: make use of smbdirect_connection_send_io_done()

This handles freeing of siblings too, which is used on
the client yet, but that might follow later.

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_{alloc,free}_send_io()
Stefan Metzmacher [Thu, 11 Sep 2025 18:09:24 +0000 (20:09 +0200)] 
smb: client: make use of smbdirect_connection_{alloc,free}_send_io()

This simplifies the code and allows us to share more code in common
with the server.

The only difference is that we use ib_dma_unmap_page() for all sges,
this simplifies the logic and doesn't matter as
ib_dma_unmap_single() and ib_dma_unmap_page() both operate
on dma_addr_t and dma_unmap_single_attrs() is just an
alias for dma_unmap_page_attrs().

We already had such an inconsistency before
as we called ib_dma_unmap_single(), while we mapped
using ib_dma_map_page() in smb_set_sge().

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_frwr_is_supported()
Stefan Metzmacher [Thu, 28 Aug 2025 14:59:00 +0000 (16:59 +0200)] 
smb: client: make use of smbdirect_frwr_is_supported()

This an exact copy of frwr_is_supported().

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_idle_timer_work()
Stefan Metzmacher [Wed, 27 Aug 2025 15:22:44 +0000 (17:22 +0200)] 
smb: client: make use of smbdirect_connection_idle_timer_work()

This is basically a copy of idle_connection_timer().

Note smbdirect_socket_prepare_create() already calls INIT_DELAYED_WORK().

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
4 weeks agosmb: client: make use of smbdirect_connection_reassembly_{append,first}_recv_io()
Stefan Metzmacher [Wed, 27 Aug 2025 14:48:39 +0000 (16:48 +0200)] 
smb: client: make use of smbdirect_connection_reassembly_{append,first}_recv_io()

These are exact copies of enqueue_reassembly() and _get_first_reassembly().

Cc: Steve French <smfrench@gmail.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Long Li <longli@microsoft.com>
Cc: Namjae Jeon <linkinjeon@kernel.org>
Cc: linux-cifs@vger.kernel.org
Cc: samba-technical@lists.samba.org
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>