Skip searching the software IO TLB if a device has never used it, making
sure these devices are not affected by the introduction of multiple IO TLB
memory pools.

Additional memory barrier is required to ensure that the new value of the
flag is visible to other CPUs after mapping a new bounce buffer. For
efficiency, the flag check should be inlined, and then the memory barrier
must be moved to is_swiotlb_buffer(). However, it can replace the existing
barrier in swiotlb_find_pool(), because all callers use is_swiotlb_buffer()
first to verify that the buffer address belongs to the software IO TLB.
Signed-off-by: Petr Tesarik <petr.tesarik.ext@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

When swiotlb_find_slots() cannot find suitable slots, schedule the
allocation of a new memory pool. It is not possible to allocate the pool
immediately, because this code may run in interrupt context, which is not
suitable for large memory allocations. This means that the memory pool will
be available too late for the currently requested mapping, but the stress
on the software IO TLB allocator is likely to continue, and subsequent
allocations will benefit from the additional pool eventually.

Keep all memory pools for an allocator in an RCU list to avoid locking on
the read side. For modifications, add a new spinlock to struct io_tlb_mem.

The spinlock also protects updates to the total number of slabs (nslabs in
struct io_tlb_mem), but not reads of the value. Readers may therefore
encounter a stale value, but this is not an issue:

- swiotlb_tbl_map_single() and is_swiotlb_active() only check for non-zero
  value. This is ensured by the existence of the default memory pool,
  allocated at boot.

- The exact value is used only for non-critical purposes (debugfs, kernel
  messages).
Signed-off-by: Petr Tesarik <petr.tesarik.ext@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

The value returned by default_swiotlb_limit() should be constant, because
it is used to decide whether DMA can be used. To allow allocating memory
pools on the fly, use the maximum possible physical address rather than the
highest address used by the default pool.

For swiotlb_init_remap(), this is either an arch-specific limit used by
memblock_alloc_low(), or the highest directly mapped physical address if
the initialization flags include SWIOTLB_ANY. For swiotlb_init_late(), the
highest address is determined by the GFP flags.
Signed-off-by: Petr Tesarik <petr.tesarik.ext@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Try to allocate a transient memory pool if no suitable slots can be found
and the respective SWIOTLB is allowed to grow. The transient pool is just
enough big for this one bounce buffer. It is inserted into a per-device
list of transient memory pools, and it is freed again when the bounce
buffer is unmapped.

Transient memory pools are kept in an RCU list. A memory barrier is
required after adding a new entry, because any address within a transient
buffer must be immediately recognized as belonging to the SWIOTLB, even if
it is passed to another CPU.

Deletion does not require any synchronization beyond RCU ordering
guarantees. After a buffer is unmapped, its physical addresses may no
longer be passed to the DMA API, so the memory range of the corresponding
stale entry in the RCU list never matches. If the memory range gets
allocated again, then it happens only after a RCU quiescent state.

Since bounce buffers can now be allocated from different pools, add a
parameter to swiotlb_alloc_pool() to let the caller know which memory pool
is used. Add swiotlb_find_pool() to find the memory pool corresponding to
an address. This function is now also used by is_swiotlb_buffer(), because
a simple boundary check is no longer sufficient.

The logic in swiotlb_alloc_tlb() is taken from __dma_direct_alloc_pages(),
simplified and enhanced to use coherent memory pools if needed.

Note that this is not the most efficient way to provide a bounce buffer,
but when a DMA buffer can't be mapped, something may (and will) actually
break. At that point it is better to make an allocation, even if it may be
an expensive operation.
Signed-off-by: Petr Tesarik <petr.tesarik.ext@huawei.com>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Add a config option (CONFIG_SWIOTLB_DYNAMIC) to enable or disable dynamic
allocation of additional bounce buffers.

If this option is set, mark the default SWIOTLB as able to grow and
restricted DMA pools as unable.

However, if the address of the default memory pool is explicitly queried,
make the default SWIOTLB also unable to grow. This is currently used to set
up PCI BAR movable regions on some Octeon MIPS boards which may not be able
to use a SWIOTLB pool elsewhere in physical memory. See octeon_pci_setup()
for more details.

If a remap function is specified, it must be also called on any dynamically
allocated pools, but there are some issues:

- The remap function may block, so it should not be called from an atomic
  context.
- There is no corresponding unremap() function if the memory pool is
  freed.
- The only in-tree implementation (xen_swiotlb_fixup) requires that the
  number of slots in the memory pool is a multiple of SWIOTLB_SEGSIZE.

Keep it simple for now and disable growing the SWIOTLB if a remap function
was specified.
Signed-off-by: Petr Tesarik <petr.tesarik.ext@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Carve out memory pool specific fields from struct io_tlb_mem. The original
struct now contains shared data for the whole allocator, while the new
struct io_tlb_pool contains data that is specific to one memory pool of
(potentially) many.
Signed-off-by: Petr Tesarik <petr.tesarik.ext@huawei.com>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Add some kernel-doc comments and move the existing documentation of struct
io_tlb_slot to its correct location. The latter was forgotten in commit
942a8186 ("swiotlb: move struct io_tlb_slot to swiotlb.c").

Use the opportunity to give swiotlb_do_find_slots() a more descriptive name
and make it clear how it differs from swiotlb_find_slots().
Signed-off-by: Petr Tesarik <petr.tesarik.ext@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

SWIOTLB implementation details should not be exposed to the rest of the
kernel. This will allow to make changes to the implementation without
modifying non-swiotlb code.

To avoid breaking existing users, provide helper functions for the few
required fields.

As a bonus, using a helper function to initialize struct device allows to
get rid of an #ifdef in driver core.
Signed-off-by: Petr Tesarik <petr.tesarik.ext@huawei.com>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>

If swiotlb is allocated, immediately return 0, so callers do not have to
check io_tlb_default_mem.nslabs explicitly.
Signed-off-by: Petr Tesarik <petr.tesarik.ext@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

In the process of parsing the DTS, check whether the memory region
specified by the DTS CMA node area overlaps with the kernel text
memory space reserved by memblock before calling
early_init_fdt_scan_reserved_mem.
Signed-off-by: Binglei Wang <l3b2w1@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

The kernel parameter 'cma_pernuma=' only supports reserving the same
size of CMA area for each node. We need to reserve different sizes of
CMA area for specified nodes if these devices belong to different nodes.

Adding another kernel parameter 'numa_cma=' to reserve CMA area for
the specified node. If we want to use one of these parameters, we need to
enable DMA_NUMA_CMA.

At the same time, print the node id in cma_declare_contiguous_nid() if
CONFIG_NUMA is enabled.
Signed-off-by: Yajun Deng <yajun.deng@linux.dev>
Signed-off-by: Christoph Hellwig <hch@lst.de>

In the commit b7176c26 ("dma-contiguous: provide the ability to
reserve per-numa CMA"), Barry adds DMA_PERNUMA_CMA for ARM64.

But this feature is architecture independent, so support per-numa CMA
for all architectures, and enable it by default if NUMA.
Signed-off-by: Yajun Deng <yajun.deng@linux.dev>
Tested-by: Yicong Yang <yangyicong@hisilicon.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

This function has a __weak definition and an override that is only used on
freescale powerpc chips. The powerpc definition however does not see the
declaration that is in a .c file:

arch/powerpc/kernel/dma-mask.c:7:6: error: no previous prototype for 'arch_dma_set_mask' [-Werror=missing-prototypes]

Move it into the linux/dma-map-ops.h header where the other arch_dma_* functions
are declared.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Drivers have no business looking at dma-mapping or swiotlb internals.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Juergen Gross <jgross@suse.com>

Remove the dangerous late initialization of xen-swiotlb in
pci_xen_swiotlb_init_late and instead just always initialize
xen-swiotlb in the boot code if CONFIG_XEN_PCIDEV_FRONTEND is
enabled and Xen PV PCI is possible.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Juergen Gross <jgross@suse.com>

When running as a Xen PV guests passed through PCI devices only have a
chance to work if the Xen supplied memory map has some PCI space
reserved.

Add a flag xen_pv_pci_possible which will be set in early boot in case
the memory map has at least one area with the type E820_TYPE_RESERVED.
Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Pull spi fixes from Mark Brown:
 "A bunch of fixes for the Qualcomm QSPI driver, fixing multiple issues
  with the newly added DMA mode - it had a number of issues exposed when
  tested in a wider range of use cases, both race condition style issues
  and issues with different inputs to those that had been used in test"

* tag 'spi-fix-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
  spi: spi-qcom-qspi: Add mem_ops to avoid PIO for badly sized reads
  spi: spi-qcom-qspi: Fallback to PIO for xfers that aren't multiples of 4 bytes
  spi: spi-qcom-qspi: Add DMA_CHAIN_DONE to ALL_IRQS
  spi: spi-qcom-qspi: Call dma_wmb() after setting up descriptors
  spi: spi-qcom-qspi: Use GFP_ATOMIC flag while allocating for descriptor
  spi: spi-qcom-qspi: Ignore disabled interrupts' status in isr

Merge tag 'regulator-fix-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator

Pull regulator fixes from Mark Brown:
 "A couple of small fixes for the the mt6358 driver, fixing error
  reporting and a bootstrapping issue"

* tag 'regulator-fix-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
  regulator: mt6358: Fix incorrect VCN33 sync error message
  regulator: mt6358: Sync VCN33_* enable status after checking ID

Pull USB fixes from Greg KH:
 "Here are a set of USB driver fixes for 6.5-rc4. Include in here are:

   - new USB serial device ids

   - dwc3 driver fixes for reported issues

   - typec driver fixes for reported problems

   - gadget driver fixes

   - reverts of some problematic USB changes that went into -rc1

  All of these have been in linux-next with no reported problems"

* tag 'usb-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (24 commits)
  usb: misc: ehset: fix wrong if condition
  usb: dwc3: pci: skip BYT GPIO lookup table for hardwired phy
  usb: cdns3: fix incorrect calculation of ep_buf_size when more than one config
  usb: gadget: call usb_gadget_check_config() to verify UDC capability
  usb: typec: Use sysfs_emit_at when concatenating the string
  usb: typec: Iterate pds array when showing the pd list
  usb: typec: Set port->pd before adding device for typec_port
  usb: typec: qcom: fix return value check in qcom_pmic_typec_probe()
  Revert "usb: gadget: tegra-xudc: Fix error check in tegra_xudc_powerdomain_init()"
  Revert "usb: xhci: tegra: Fix error check"
  USB: gadget: Fix the memory leak in raw_gadget driver
  usb: gadget: core: remove unbalanced mutex_unlock in usb_gadget_activate
  Revert "usb: dwc3: core: Enable AutoRetry feature in the controller"
  Revert "xhci: add quirk for host controllers that don't update endpoint DCS"
  USB: quirks: add quirk for Focusrite Scarlett
  usb: xhci-mtk: set the dma max_seg_size
  MAINTAINERS: drop invalid usb/cdns3 Reviewer e-mail
  usb: dwc3: don't reset device side if dwc3 was configured as host-only
  usb: typec: ucsi: move typec_set_mode(TYPEC_STATE_SAFE) to ucsi_unregister_partner()
  usb: ohci-at91: Fix the unhandle interrupt when resume
  ...

Pull tty/serial fixes from Greg KH:
 "Here are some small TTY and serial driver fixes for 6.5-rc4 for some
  reported problems. Included in here is:

   - TIOCSTI fix for braille readers

   - documentation fix for minor numbers

   - MAINTAINERS update for new serial files in -rc1

   - minor serial driver fixes for reported problems

  All of these have been in linux-next with no reported problems"

* tag 'tty-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  serial: 8250_dw: Preserve original value of DLF register
  tty: serial: sh-sci: Fix sleeping in atomic context
  serial: sifive: Fix sifive_serial_console_setup() section
  Documentation: devices.txt: reconcile serial/ucc_uart minor numers
  MAINTAINERS: Update TTY layer for lists and recently added files
  tty: n_gsm: fix UAF in gsm_cleanup_mux
  TIOCSTI: always enable for CAP_SYS_ADMIN

Pull staging driver fixes from Greg KH:
 "Here are three small staging driver fixes for 6.5-rc4 that resolve
  some reported problems. These fixes are:

   - fix for an old bug in the r8712 driver

   - fbtft driver fix for a spi device

   - potential overflow fix in the ks7010 driver

  All of these have been in linux-next with no reported problems"

* tag 'staging-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  staging: ks7010: potential buffer overflow in ks_wlan_set_encode_ext()
  staging: fbtft: ili9341: use macro FBTFT_REGISTER_SPI_DRIVER
  staging: r8712: Fix memory leak in _r8712_init_xmit_priv()

Pull char driver and Documentation fixes from Greg KH:
 "Here is a char driver fix and some documentation updates for 6.5-rc4
  that contain the following changes:

   - sram/genalloc bugfix for reported problem

   - security-bugs.rst update based on recent discussions

   - embargoed-hardware-issues minor cleanups and then partial revert
     for the project/company lists

  All of these have been in linux-next for a while with no reported
  problems, and the documentation updates have all been reviewed by the
  relevant developers"

* tag 'char-misc-6.5-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  misc/genalloc: Name subpools by of_node_full_name()
  Documentation: embargoed-hardware-issues.rst: add AMD to the list
  Documentation: embargoed-hardware-issues.rst: clean out empty and unused entries
  Documentation: security-bugs.rst: clarify CVE handling
  Documentation: security-bugs.rst: update preferences when dealing with the linux-distros group

Merge tag 'probes-fixes-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace

Pull probe fixes from Masami Hiramatsu:

 - probe-events: add NULL check for some BTF API calls which can return
   error code and NULL.

 - ftrace selftests: check fprobe and kprobe event correctly. This fixes
   a miss condition of the test command.

 - kprobes: do not allow probing functions that start with "__cfi_" or
   "__pfx_" since those are auto generated for kernel CFI and not
   executed.

* tag 'probes-fixes-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  kprobes: Prohibit probing on CFI preamble symbol
  selftests/ftrace: Fix to check fprobe event eneblement
  tracing/probes: Fix to add NULL check for BTF APIs

Pull kvm fixes from Paolo Bonzini:
 "x86:

   - Do not register IRQ bypass consumer if posted interrupts not
     supported

   - Fix missed device interrupt due to non-atomic update of IRR

   - Use GFP_KERNEL_ACCOUNT for pid_table in ipiv

   - Make VMREAD error path play nice with noinstr

   - x86: Acquire SRCU read lock when handling fastpath MSR writes

   - Support linking rseq tests statically against glibc 2.35+

   - Fix reference count for stats file descriptors

   - Detect userspace setting invalid CR0

  Non-KVM:

   - Remove coccinelle script that has caused multiple confusion
     ("debugfs, coccinelle: check for obsolete DEFINE_SIMPLE_ATTRIBUTE()
     usage", acked by Greg)"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (21 commits)
  KVM: selftests: Expand x86's sregs test to cover illegal CR0 values
  KVM: VMX: Don't fudge CR0 and CR4 for restricted L2 guest
  KVM: x86: Disallow KVM_SET_SREGS{2} if incoming CR0 is invalid
  Revert "debugfs, coccinelle: check for obsolete DEFINE_SIMPLE_ATTRIBUTE() usage"
  KVM: selftests: Verify stats fd is usable after VM fd has been closed
  KVM: selftests: Verify stats fd can be dup()'d and read
  KVM: selftests: Verify userspace can create "redundant" binary stats files
  KVM: selftests: Explicitly free vcpus array in binary stats test
  KVM: selftests: Clean up stats fd in common stats_test() helper
  KVM: selftests: Use pread() to read binary stats header
  KVM: Grab a reference to KVM for VM and vCPU stats file descriptors
  selftests/rseq: Play nice with binaries statically linked against glibc 2.35+
  Revert "KVM: SVM: Skip WRMSR fastpath on VM-Exit if next RIP isn't valid"
  KVM: x86: Acquire SRCU read lock when handling fastpath MSR writes
  KVM: VMX: Use vmread_error() to report VM-Fail in "goto" path
  KVM: VMX: Make VMREAD error path play nice with noinstr
  KVM: x86/irq: Conditionally register IRQ bypass consumer again
  KVM: X86: Use GFP_KERNEL_ACCOUNT for pid_table in ipiv
  KVM: x86: check the kvm_cpu_get_interrupt result before using it
  KVM: x86: VMX: set irr_pending in kvm_apic_update_irr
  ...

Pull locking fix from Borislav Petkov:

 - Fix a rtmutex race condition resulting from sharing of the sort key
   between the lock waiters and the PI chain tree (->pi_waiters) of a
   task by giving each tree their own sort key

* tag 'locking_urgent_for_v6.5_rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  locking/rtmutex: Fix task->pi_waiters integrity

Pull x86 fixes from Borislav Petkov:

 - AMD's automatic IBRS doesn't enable cross-thread branch target
   injection protection (STIBP) for user processes. Enable STIBP on such
   systems.

 - Do not delete (but put the ref instead) of AMD MCE error thresholding
   sysfs kobjects when destroying them in order not to delete the kernfs
   pointer prematurely

 - Restore annotation in ret_from_fork_asm() in order to fix kthread
   stack unwinding from being marked as unreliable and thus breaking
   livepatching

* tag 'x86_urgent_for_v6.5_rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/cpu: Enable STIBP on AMD if Automatic IBRS is enabled
  x86/MCE/AMD: Decrement threshold_bank refcount when removing threshold blocks
  x86: Fix kthread unwind

Pull irq fixes from Borislav Petkov:

 - Work around an erratum on GIC700, where a race between a CPU handling
   a wake-up interrupt, a change of affinity, and another CPU going to
   sleep can result in a lack of wake-up event on the next interrupt

 - Fix the locking required on a VPE for GICv4

 - Enable Rockchip 3588001 erratum workaround for RK3588S

 - Fix the irq-bcm6345-l1 assumtions of the boot CPU always be the first
   CPU in the system

* tag 'irq_urgent_for_v6.5_rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/gic-v3: Workaround for GIC-700 erratum 2941627
  irqchip/gic-v3: Enable Rockchip 3588001 erratum workaround for RK3588S
  irqchip/gic-v4.1: Properly lock VPEs when doing a directLPI invalidation
  irq-bcm6345-l1: Do not assume a fixed block to cpu mapping

Pull smb client fixes from Steve French:
 "Four small SMB3 client fixes:

   - two reconnect fixes (to address the case where non-default
     iocharset gets incorrectly overridden at reconnect with the
     default charset)

   - fix for NTLMSSP_AUTH request setting a flag incorrectly)

   - Add missing check for invalid tlink (tree connection) in ioctl"

* tag '6.5-rc3-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: add missing return value check for cifs_sb_tlink
  smb3: do not set NTLMSSP_VERSION flag for negotiate not auth request
  cifs: fix charset issue in reconnection
  fs/nls: make load_nls() take a const parameter

Pull tracing fixes from Steven Rostedt:

 - Fix to /sys/kernel/tracing/per_cpu/cpu*/stats read and entries.

   If a resize shrinks the buffer it clears the read count to notify
   readers that they need to reset. But the read count is also used for
   accounting and this causes the numbers to be off. Instead, create a
   separate variable to use to notify readers to reset.

 - Fix the ref counts of the "soft disable" mode. The wrong value was
   used for testing if soft disable mode should be enabled or disable,
   but instead, just change the logic to do the enable and disable in
   place when the SOFT_MODE is set or cleared.

 - Several kernel-doc fixes

 - Removal of unused external declarations

* tag 'trace-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  tracing: Fix warning in trace_buffered_event_disable()
  ftrace: Remove unused extern declarations
  tracing: Fix kernel-doc warnings in trace_seq.c
  tracing: Fix kernel-doc warnings in trace_events_trigger.c
  tracing/synthetic: Fix kernel-doc warnings in trace_events_synth.c
  ring-buffer: Fix kernel-doc warnings in ring_buffer.c
  ring-buffer: Fix wrong stat of cpu_buffer->read

Commit a2225d93 ("autofs: remove left-over autofs4 stubs")
promised the removal of the fs/autofs/Kconfig fragment for AUTOFS4_FS
within a couple of releases, but five years later this still has not
happened yet, and AUTOFS4_FS is still enabled in 63 defconfigs.

Get rid of it mechanically:

   git grep -l CONFIG_AUTOFS4_FS -- '*defconfig' |
       xargs sed -i 's/AUTOFS4_FS/AUTOFS_FS/'

Also just remove the AUTOFS4_FS config option stub.  Anybody who hasn't
regenerated their config file in the last five years will need to just
get the new name right when they do.
Signed-off-by: Sven Joachim <svenjoac@gmx.de>
Acked-by: Ian Kent <raven@themaw.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge tag 'loongarch-fixes-6.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson

Pull LoongArch fixes from Huacai Chen:
 "Some bug fixes for build system, builtin cmdline handling, bpf and
  {copy, clear}_user, together with a trivial cleanup"

* tag 'loongarch-fixes-6.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson:
  LoongArch: Cleanup __builtin_constant_p() checking for cpu_has_*
  LoongArch: BPF: Fix check condition to call lu32id in move_imm()
  LoongArch: BPF: Enable bpf_probe_read{, str}() on LoongArch
  LoongArch: Fix return value underflow in exception path
  LoongArch: Fix CMDLINE_EXTEND and CMDLINE_BOOTLOADER handling
  LoongArch: Fix module relocation error with binutils 2.41
  LoongArch: Only fiddle with CHECKFLAGS if `need-compiler'

Add coverage to x86's set_sregs_test to verify KVM rejects vendor-agnostic
illegal CR0 values, i.e. CR0 values whose legality doesn't depend on the
current VMX mode.  KVM historically has neglected to reject bad CR0s from
userspace, i.e. would happily accept a completely bogus CR0 via
KVM_SET_SREGS{2}.

Punt VMX specific subtests to future work, as they would require quite a
bit more effort, and KVM gets coverage for CR0 checks in general through
other means, e.g. KVM-Unit-Tests.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230613203037.1968489-4-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Stuff CR0 and/or CR4 to be compliant with a restricted guest if and only
if KVM itself is not configured to utilize unrestricted guests, i.e. don't
stuff CR0/CR4 for a restricted L2 that is running as the guest of an
unrestricted L1.  Any attempt to VM-Enter a restricted guest with invalid
CR0/CR4 values should fail, i.e. in a nested scenario, KVM (as L0) should
never observe a restricted L2 with incompatible CR0/CR4, since nested
VM-Enter from L1 should have failed.

And if KVM does observe an active, restricted L2 with incompatible state,
e.g. due to a KVM bug, fudging CR0/CR4 instead of letting VM-Enter fail
does more harm than good, as KVM will often neglect to undo the side
effects, e.g. won't clear rmode.vm86_active on nested VM-Exit, and thus
the damage can easily spill over to L1.  On the other hand, letting
VM-Enter fail due to bad guest state is more likely to contain the damage
to L2 as KVM relies on hardware to perform most guest state consistency
checks, i.e. KVM needs to be able to reflect a failed nested VM-Enter into
L1 irrespective of (un)restricted guest behavior.

Cc: Jim Mattson <jmattson@google.com>
Cc: stable@vger.kernel.org
Fixes: bddd82d1 ("KVM: nVMX: KVM needs to unset "unrestricted guest" VM-execution control in vmcs02 if vmcs12 doesn't set it")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230613203037.1968489-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Reject KVM_SET_SREGS{2} with -EINVAL if the incoming CR0 is invalid,
e.g. due to setting bits 63:32, illegal combinations, or to a value that
isn't allowed in VMX (non-)root mode.  The VMX checks in particular are
"fun" as failure to disallow Real Mode for an L2 that is configured with
unrestricted guest disabled, when KVM itself has unrestricted guest
enabled, will result in KVM forcing VM86 mode to virtual Real Mode for
L2, but then fail to unwind the related metadata when synthesizing a
nested VM-Exit back to L1 (which has unrestricted guest enabled).

Opportunistically fix a benign typo in the prototype for is_valid_cr4().

Cc: stable@vger.kernel.org
Reported-by: syzbot+5feef0b9ee9c8e9e5689@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/000000000000f316b705fdf6e2b4@google.comSigned-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230613203037.1968489-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Remove coccinelle's recommendation to use DEFINE_DEBUGFS_ATTRIBUTE()
instead of DEFINE_SIMPLE_ATTRIBUTE().  Regardless of whether or not the
"significant overhead" incurred by debugfs_create_file() is actually
meaningful, warnings from the script have led to a rash of low-quality
patches that have sowed confusion and consumed maintainer time for little
to no benefit.  There have been no less than four attempts to "fix" KVM,
and a quick search on lore shows that KVM is not alone.

This reverts commit 5103068e.

Link: https://lore.kernel.org/all/87tu2nbnz3.fsf@mpe.ellerman.id.au
Link: https://lore.kernel.org/all/c0b98151-16b6-6d8f-1765-0f7d46682d60@redhat.com
Link: https://lkml.kernel.org/r/20230706072954.4881-1-duminjie%40vivo.com
Link: https://lore.kernel.org/all/Y2FsbufV00jbyF0B@google.com
Link: https://lore.kernel.org/all/Y2ENJJ1YiSg5oHiy@orome
Link: https://lore.kernel.org/all/7560b350e7b23786ce712118a9a504356ff1cca4.camel@kernel.orgSuggested-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230726202920.507756-1-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Verify that VM and vCPU binary stats files are usable even after userspace
has put its last direct reference to the VM.  This is a regression test
for a UAF bug where KVM didn't gift the stats files a reference to the VM.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230711230131.648752-8-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Expand the binary stats test to verify that a stats fd can be dup()'d
and read, to (very) roughly simulate userspace passing around the file.
Adding the dup() test is primarily an intermediate step towards verifying
that userspace can read VM/vCPU stats before _and_ after userspace closes
its copy of the VM fd; the dup() test itself is only mildly interesting.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230711230131.648752-7-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Verify that KVM doesn't artificially limit KVM_GET_STATS_FD to a single
file per VM/vCPU.  There's no known use case for getting multiple stats
fds, but it should work, and more importantly creating multiple files will
make it easier to test that KVM correct manages VM refcounts for stats
files.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230711230131.648752-6-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Explicitly free the all-encompassing vcpus array in the binary stats test
so that the test is consistent with respect to freeing all dynamically
allocated resources (versus letting them be freed on exit).
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230711230131.648752-5-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>