If REQ_NOWAIT is set, then do a non-blocking allocation if the operation
is a write and we need to insert a new page. Currently REQ_NOWAIT cannot
be set as the queue isn't marked as supporting nowait, this change is in
preparation for allowing that.
radix_tree_preload() warns on attempting to call it with an allocation
mask that doesn't allow blocking. While that warning could arguably
be removed, we need to handle radix insertion failures anyway as they
are more likely if we cannot block to get memory.
Remove legacy BUG_ON()'s and turn them into proper errors instead, one
for the allocation failure and one for finding a page that doesn't
match the correct index.
By default, non-mq drivers do not support nowait. This causes io_uring
to use a slower path as the driver cannot be trust not to block. brd
can safely set the nowait flag, as worst case all it does is a NOIO
allocation.
For io_uring, this makes a substantial difference. Before:
47894e0fa6a5 ("virt/sev-guest: Prevent IV reuse in the SNP guest driver")
changed the behavior associated with the return value when the caller
does not supply a large enough certificate buffer. Prior to the commit a
value of -EIO was returned. Now, 0 is returned. This breaks the
established ABI with the user.
Change the code to detect the buffer size error and return -EIO.
Fixes: 47894e0fa6a5 ("virt/sev-guest: Prevent IV reuse in the SNP guest driver") Reported-by: Larry Dewey <larry.dewey@amd.com> Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com> Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de> Tested-by: Larry Dewey <larry.dewey@amd.com> Cc: <stable@kernel.org> Link: https://lore.kernel.org/r/2afbcae6daf13f7ad5a4296692e0a0fe1bc1e4ee.1677083979.git.thomas.lendacky@amd.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When plain IBRS is enabled (not enhanced IBRS), the logic in
spectre_v2_user_select_mitigation() determines that STIBP is not needed.
The IBRS bit implicitly protects against cross-thread branch target
injection. However, with legacy IBRS, the IBRS bit is cleared on
returning to userspace for performance reasons which leaves userspace
threads vulnerable to cross-thread branch target injection against which
STIBP protects.
Exclude IBRS from the spectre_v2_in_ibrs_mode() check to allow for
enabling STIBP (through seccomp/prctl() by default or always-on, if
selected by spectre_v2_user kernel cmdline parameter).
The AMD side of the loader has always claimed to support mixed
steppings. But somewhere along the way, it broke that by assuming that
the cached patch blob is a single one instead of it being one per
*node*.
So turn it into a per-node one so that each node can stash the blob
relevant for it.
[ NB: Fixes tag is not really the exactly correct one but it is good
enough. ]
Fixes: fe055896c040 ("x86/microcode: Merge the early microcode loader") Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de> Cc: <stable@kernel.org> # 2355370cd941 ("x86/microcode/amd: Remove load_microcode_amd()'s bsp parameter") Cc: <stable@kernel.org> # a5ad92134bd1 ("x86/microcode/AMD: Add a @cpu parameter to the reloading functions") Link: https://lore.kernel.org/r/20230130161709.11615-4-bp@alien8.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When arch_prepare_optimized_kprobe calculating jump destination address,
it copies original instructions from jmp-optimized kprobe (see
__recover_optprobed_insn), and calculated based on length of original
instruction.
arch_check_optimized_kprobe does not check KPROBE_FLAG_OPTIMATED when
checking whether jmp-optimized kprobe exists.
As a result, setup_detour_execution may jump to a range that has been
overwritten by jump destination address, resulting in an inval opcode error.
For example, assume that register two kprobes whose addresses are
<func+9> and <func+11> in "func" function.
The original code of "func" function is as follows:
2. Register the kprobe for <func+9>, assume that is kp2, corresponding optimized_kprobe is op2.
register_kprobe(kp2)
register_aggr_kprobe
alloc_aggr_kprobe
__prepare_optimized_kprobe
arch_prepare_optimized_kprobe
__recover_optprobed_insn // copy original bytes from kp1->optinsn.copied_insn,
// jump address = <func+14>
3. disable kp1:
disable_kprobe(kp1)
__disable_kprobe
...
if (p == orig_p || aggr_kprobe_disabled(orig_p)) {
ret = disarm_kprobe(orig_p, true) // add op1 in unoptimizing_list, not unoptimized
orig_p->flags |= KPROBE_FLAG_DISABLED; // op1->flags == KPROBE_FLAG_OPTIMATED | KPROBE_FLAG_DISABLED
...
4. unregister kp2
__unregister_kprobe_top
...
if (!kprobe_disabled(ap) && !kprobes_all_disarmed) {
optimize_kprobe(op)
...
if (arch_check_optimized_kprobe(op) < 0) // because op1 has KPROBE_FLAG_DISABLED, here not return
return;
p->kp.flags |= KPROBE_FLAG_OPTIMIZED; // now op2 has KPROBE_FLAG_OPTIMIZED
}
commit f66c0447cca1 ("kprobes: Set unoptimized flag after unoptimizing code")
modified the update timing of the KPROBE_FLAG_OPTIMIZED, a optimized_kprobe
may be in the optimizing or unoptimizing state when op.kp->flags
has KPROBE_FLAG_OPTIMIZED and op->list is not empty.
The __recover_optprobed_insn check logic is incorrect, a kprobe in the
unoptimizing state may be incorrectly determined as unoptimizing.
As a result, incorrect instructions are copied.
The optprobe_queued_unopt function needs to be exported for invoking in
arch directory.
Link: https://lore.kernel.org/all/20230216034247.32348-2-yangjihong1@huawei.com/ Fixes: f66c0447cca1 ("kprobes: Set unoptimized flag after unoptimizing code") Cc: stable@vger.kernel.org Signed-off-by: Yang Jihong <yangjihong1@huawei.com> Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org> Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Disable SVM and more importantly force GIF=1 when halting a CPU or
rebooting the machine. Similar to VMX, SVM allows software to block
INITs via CLGI, and thus can be problematic for a crash/reboot. The
window for failure is smaller with SVM as INIT is only blocked while
GIF=0, i.e. between CLGI and STGI, but the window does exist.
Fixes: fba4f472b33a ("x86/reboot: Turn off KVM when halting a CPU") Cc: stable@vger.kernel.org Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Link: https://lore.kernel.org/r/20221130233650.1404148-5-seanjc@google.com Signed-off-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Disable SVM on all CPUs via NMI shootdown during an emergency reboot.
Like VMX, SVM can block INIT, e.g. if the emergency reboot is triggered
between CLGI and STGI, and thus can prevent bringing up other CPUs via
INIT-SIPI-SIPI.
Disable virtualization in crash_nmi_callback() and rework the
emergency_vmx_disable_all() path to do an NMI shootdown if and only if a
shootdown has not already occurred. NMI crash shootdown fundamentally
can't support multiple invocations as responding CPUs are deliberately
put into halt state without unblocking NMIs. But, the emergency reboot
path doesn't have any work of its own, it simply cares about disabling
virtualization, i.e. so long as a shootdown occurred, emergency reboot
doesn't care who initiated the shootdown, or when.
If "crash_kexec_post_notifiers" is specified on the kernel command line,
panic() will invoke crash_smp_send_stop() and result in a second call to
nmi_shootdown_cpus() during native_machine_emergency_restart().
Invoke the callback _before_ disabling virtualization, as the current
VMCS needs to be cleared before doing VMXOFF. Note, this results in a
subtle change in ordering between disabling virtualization and stopping
Intel PT on the responding CPUs. While VMX and Intel PT do interact,
VMXOFF and writes to MSR_IA32_RTIT_CTL do not induce faults between one
another, which is all that matters when panicking.
Harden nmi_shootdown_cpus() against multiple invocations to try and
capture any such kernel bugs via a WARN instead of hanging the system
during a crash/dump, e.g. prior to the recent hardening of
register_nmi_handler(), re-registering the NMI handler would trigger a
double list_add() and hang the system if CONFIG_BUG_ON_DATA_CORRUPTION=y.
Extract the disabling logic to a common helper to deduplicate code, and
to prepare for doing the shootdown in the emergency reboot path if SVM
is supported.
Note, prior to commit ed72736183c4 ("x86/reboot: Force all cpus to exit
VMX root if VMX is supported"), nmi_shootdown_cpus() was subtly protected
against a second invocation by a cpu_vmx_enabled() check as the kdump
handler would disable VMX if it ran first.
Fixes: ed72736183c4 ("x86/reboot: Force all cpus to exit VMX root if VMX is supported") Cc: stable@vger.kernel.org Reported-by: Guilherme G. Piccoli <gpiccoli@igalia.com> Cc: Vitaly Kuznetsov <vkuznets@redhat.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Link: https://lore.kernel.org/all/20220427224924.592546-2-gpiccoli@igalia.com Tested-by: Guilherme G. Piccoli <gpiccoli@igalia.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Link: https://lore.kernel.org/r/20221130233650.1404148-2-seanjc@google.com Signed-off-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Set GIF=1 prior to disabling SVM to ensure that INIT is recognized if the
kernel is disabling SVM in an emergency, e.g. if the kernel is about to
jump into a crash kernel or may reboot without doing a full CPU RESET.
If GIF is left cleared, the new kernel (or firmware) will be unabled to
awaken APs. Eat faults on STGI (due to EFER.SVME=0) as it's possible
that SVM could be disabled via NMI shootdown between reading EFER.SVME
and executing STGI.
Use $(KHDR_INCLUDES) as lookup path for kernel headers. This prevents
building against kernel headers from the build environment in scenarios
where kernel headers are installed into a specific output directory
(O=...).
This "(unknown) (section: .init.data)" all refer to svm_x86_ops.
Tag svm_hv_hardware_setup() with __init to fix a modpost warning as the
non-stub implementation accesses __initdata (svm_x86_ops), i.e. would
generate a use-after-free if svm_hv_hardware_setup() were actually invoked
post-init. The helper is only called from svm_hardware_setup(), which is
also __init, i.e. lack of __init is benign other than the modpost warning.
Fixes: 1e0c7d40758b ("KVM: SVM: hyper-v: Remote TLB flush for SVM") Signed-off-by: Randy Dunlap <rdunlap@infradead.org> Cc: Vineeth Pillai <viremana@linux.microsoft.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: kvm@vger.kernel.org Cc: stable@vger.kernel.org Reviewed-by: Sean Christopherson <seanjc@google.com> Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20230222073315.9081-1-rdunlap@infradead.org> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
KVM_SEV_SEND_UPDATE_DATA and KVM_SEV_RECEIVE_UPDATE_DATA have an integer
overflow issue. Params.guest_len and offset are both 32 bits wide, with a
large params.guest_len the check to confirm a page boundary is not
crossed can falsely pass:
/* Check if we are crossing the page boundary *
offset = params.guest_uaddr & (PAGE_SIZE - 1);
if ((params.guest_len + offset > PAGE_SIZE))
Add an additional check to confirm that params.guest_len itself is not
greater than PAGE_SIZE.
Note, this isn't a security concern as overflow can happen if and only if
params.guest_len is greater than 0xfffff000, and the FW spec says these
commands fail with lengths greater than 16KB, i.e. the PSP will detect
KVM's goof.
Fixes: 15fb7de1a7f5 ("KVM: SVM: Add KVM_SEV_RECEIVE_UPDATE_DATA command") Fixes: d3d1af85e2c7 ("KVM: SVM: Add KVM_SEND_UPDATE_DATA command") Reported-by: Andy Nguyen <theflow@google.com> Suggested-by: Thomas Lendacky <thomas.lendacky@amd.com> Signed-off-by: Peter Gonda <pgonda@google.com> Cc: David Rientjes <rientjes@google.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Sean Christopherson <seanjc@google.com> Cc: kvm@vger.kernel.org Cc: stable@vger.kernel.org Cc: linux-kernel@vger.kernel.org Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com> Link: https://lore.kernel.org/r/20230207171354.4012821-1-pgonda@google.com Signed-off-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Move the VMCB updates from avic_refresh_apicv_exec_ctrl() into
avic_set_virtual_apic_mode() and invert the dependency being said
functions to avoid calling avic_vcpu_{load,put}() and
avic_set_pi_irte_mode() when "only" setting the virtual APIC mode.
avic_set_virtual_apic_mode() is invoked from common x86 with preemption
enabled, which makes avic_vcpu_{load,put}() unhappy. Luckily, calling
those and updating IRTE stuff is unnecessary as the only reason
avic_set_virtual_apic_mode() is called is to handle transitions between
xAPIC and x2APIC that don't also toggle APICv activation. And if
activation doesn't change, there's no need to fiddle with the physical
APIC ID table or update IRTE.
The "full" refresh is guaranteed to be called if activation changes in
this case as the only call to the "set" path is:
Emulate ICR writes on AVIC IPI failures due to invalid targets using the
same logic as failures due to invalid types. AVIC acceleration fails if
_any_ of the targets are invalid, and crucially VM-Exits before sending
IPIs to targets that _are_ valid. In logical mode, the destination is a
bitmap, i.e. a single IPI can target multiple logical IDs. Doing nothing
causes KVM to drop IPIs if at least one target is valid and at least one
target is invalid.
Fixes: 18f40c53e10f ("svm: Add VMEXIT handlers for AVIC") Cc: stable@vger.kernel.org Reviewed-by: Paolo Bonzini <pbonzini@redhat.com> Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230106011306.85230-5-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Flush the TLB when activating AVIC as the CPU can insert into the TLB
while AVIC is "locally" disabled. KVM doesn't treat "APIC hardware
disabled" as VM-wide AVIC inhibition, and so when a vCPU has its APIC
hardware disabled, AVIC is not guaranteed to be inhibited. As a result,
KVM may create a valid NPT mapping for the APIC base, which the CPU can
cache as a non-AVIC translation.
Note, Intel handles this in vmx_set_virtual_apic_mode().
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com> Cc: stable@vger.kernel.org Signed-off-by: Sean Christopherson <seanjc@google.com> Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20230106011306.85230-4-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Truncate the vcpu_id, a.k.a. x2APIC ID, to an 8-bit value when comparing
it against the xAPIC ID to avoid false positives (sort of) on systems
with >255 CPUs, i.e. with IDs that don't fit into a u8. The intent of
APIC_ID_MODIFIED is to inhibit APICv/AVIC when the xAPIC is changed from
it's original value,
The mismatch isn't technically a false positive, as architecturally the
xAPIC IDs do end up being aliased in this scenario, and neither APICv
nor AVIC correctly handles IPI virtualization when there is aliasing.
However, KVM already deliberately does not honor the aliasing behavior
that results when an x2APIC ID gets truncated to an xAPIC ID. I.e. the
resulting APICv/AVIC behavior is aligned with KVM's existing behavior
when KVM's x2APIC hotplug hack is effectively enabled.
If/when KVM provides a way to disable the hotplug hack, APICv/AVIC can
piggyback whatever logic disables the optimized APIC map (which is what
provides the hotplug hack), i.e. so that KVM's optimized map and APIC
virtualization yield the same behavior.
For now, fix the immediate problem of APIC virtualization being disabled
for large VMs, which is a much more pressing issue than ensuring KVM
honors architectural behavior for APIC ID aliasing.
Fixes: 3743c2f02517 ("KVM: x86: inhibit APICv/AVIC on changes to APIC ID or APIC base") Reported-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com> Cc: stable@vger.kernel.org Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230106011306.85230-7-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Don't inhibit APICv/AVIC due to an xAPIC ID mismatch if the APIC is
hardware disabled. The ID cannot be consumed while the APIC is disabled,
and the ID is guaranteed to be set back to the vcpu_id when the APIC is
hardware enabled (architectural behavior correctly emulated by KVM).
Fixes: 3743c2f02517 ("KVM: x86: inhibit APICv/AVIC on changes to APIC ID or APIC base") Cc: stable@vger.kernel.org Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230106011306.85230-6-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When emulating a x2APIC write in response to an APICv/AVIC trap, get the
the written value from the vAPIC page without checking that reads are
allowed for the target register. AVIC can generate trap-like VM-Exits on
writes to EOI, and so KVM needs to get the written value from the backing
page without running afoul of EOI's write-only behavior.
Alternatively, EOI could be special cased to always write '0', e.g. so
that the sanity check could be preserved, but x2APIC on AMD is actually
supposed to disallow non-zero writes (not emulated by KVM), and the
sanity check was a byproduct of how the KVM code was written, i.e. wasn't
added to guard against anything in particular.
Fixes: 70c8327c11c6 ("KVM: x86: Bug the VM if an accelerated x2APIC trap occurs on a "bad" reg") Fixes: 1bd9dfec9fd4 ("KVM: x86: Do not block APIC write for non ICR registers") Reported-by: Alejandro Jimenez <alejandro.j.jimenez@oracle.com> Cc: stable@vger.kernel.org Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20230106011306.85230-2-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Purge the "highest ISR" cache when updating APICv state on a vCPU. The
cache must not be used when APICv is active as hardware may emulate EOIs
(and other operations) without exiting to KVM.
This fixes a bug where KVM will effectively block IRQs in perpetuity due
to the "highest ISR" never getting reset if APICv is activated on a vCPU
while an IRQ is in-service. Hardware emulates the EOI and KVM never gets
a chance to update its cache.
Register /dev/kvm, i.e. expose KVM to userspace, only after all other
setup has completed. Once /dev/kvm is exposed, userspace can start
invoking KVM ioctls, creating VMs, etc... If userspace creates a VM
before KVM is done with its configuration, bad things may happen, e.g.
KVM will fail to properly migrate vCPU state if a VM is created before
KVM has registered preemption notifiers.
Cc: stable@vger.kernel.org Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20221130230934.1014142-2-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
KVM enables 'Enlightened VMCS' and 'Enlightened MSR Bitmap' when running as
a nested hypervisor on top of Hyper-V. When MSR bitmap is updated,
evmcs_touch_msr_bitmap function uses current_vmcs per-cpu variable to mark
that the msr bitmap was changed.
vmx_vcpu_create() modifies the msr bitmap via vmx_disable_intercept_for_msr
-> vmx_msr_bitmap_l01_changed which in the end calls this function. The
function checks for current_vmcs if it is null but the check is
insufficient because current_vmcs is not initialized. Because of this, the
code might incorrectly write to the structure pointed by current_vmcs value
left by another task. Preemption is not disabled, the current task can be
preempted and moved to another CPU while current_vmcs is accessed multiple
times from evmcs_touch_msr_bitmap() which leads to crash.
The manipulation of MSR bitmaps by callers happens only for vmcs01 so the
solution is to use vmx->vmcs01.vmcs instead of current_vmcs.
Destroy and free the target coalesced MMIO device if unregistering said
device fails. As clearly noted in the code, kvm_io_bus_unregister_dev()
does not destroy the target device.
Don't update recovery_cp when curr_resync is MD_RESYNC_ACTIVE, otherwise
md may skip the resync of the first 3 sectors if the resync procedure is
interrupted before the first calling of ->sync_request() as shown below:
// e.g., set array as idle
set_bit(MD_RECOVERY_INTR, &&mddev_recovery)
// resync loop
// check INTR before calling sync_request
!test_bit(MD_RECOVERY_INTR, &mddev->recovery
// resync interrupted
// update recovery_cp from 0 to 3
// the resync of three 3 sectors will be skipped
mddev->recovery_cp = 3
Fixes: eac58d08d493 ("md: Use enum for overloaded magic numbers used by mddev->curr_resync") Cc: stable@vger.kernel.org # 6.0+ Signed-off-by: Hou Tao <houtao1@huawei.com> Reviewed-by: Logan Gunthorpe <logang@deltatee.com> Signed-off-by: Song Liu <song@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When we append new block just after the end of preallocated extent, the
code in inode_getblk() wrongly determined we're going to use the
preallocated extent which resulted in adding block into a wrong logical
offset in the file. Sequence like this manifests it:
The code that determined the use of preallocated extent is actually
stale because udf_do_extend_file() does not create preallocation anymore
so after calling that function we are sure there's no usable
preallocation. Just remove the faulty condition.
CC: stable@vger.kernel.org Fixes: 16d055656814 ("udf: Discard preallocation before extending file with a hole") Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When UDF filesystem is corrupted, hidden system inodes can be linked
into directory hierarchy which is an avenue for further serious
corruption of the filesystem and kernel confusion as noticed by syzbot
fuzzed images. Refuse to access system inodes linked into directory
hierarchy and vice versa.
CC: stable@vger.kernel.org Reported-by: syzbot+38695a20b8addcbc1084@syzkaller.appspotmail.com Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
System files in UDF filesystem have link count 0. To not confuse VFS we
fudge the link count to be 1 when reading such inodes however we forget
to restore the link count of 0 when writing such inodes. Fix that.
CC: stable@vger.kernel.org Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When write to inline file fails (or happens only partly), we still
updated length of inline data as if the whole write succeeded. Fix the
update of length of inline data to happen only if the write succeeds.
Reported-by: syzbot+0937935b993956ba28ab@syzkaller.appspotmail.com CC: stable@vger.kernel.org Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When merging very long extents we try to push as much length as possible
to the first extent. However this is unnecessarily complicated and not
really worth the trouble. Furthermore there was a bug in the logic
resulting in corrupting extents in the file as syzbot reproducer shows.
So just don't bother with the merging of extents that are too long
together.
CC: stable@vger.kernel.org Reported-by: syzbot+60f291a24acecb3c2bd5@syzkaller.appspotmail.com Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When a file expansion failed because we didn't have enough space for
indirect extents make sure we truncate extents created so far so that we
don't leave extents beyond EOF.
CC: stable@vger.kernel.org Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Update ptrace tests according to all potential Yama security policies.
This is required to make such tests pass even if Yama is enabled.
Tests are not skipped but they now check both Landlock and Yama boundary
restrictions at run time to keep a maximum test coverage (i.e. positive
and negative testing).
Signed-off-by: Jeff Xu <jeffxu@google.com> Link: https://lore.kernel.org/r/20230114020306.1407195-2-jeffxu@google.com Cc: stable@vger.kernel.org
[mic: Add curly braces around EXPECT_EQ() to make it build, and improve
commit message] Co-developed-by: Mickaël Salaün <mic@digikod.net> Signed-off-by: Mickaël Salaün <mic@digikod.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
overlayfs may be disabled in the kernel configuration, causing related
tests to fail. Check that overlayfs is supported at runtime, so we can
skip layout2_overlay.* accordingly.
This fixes three issues on move extents ioctl without auto defrag:
a) In ocfs2_find_victim_alloc_group(), we have to convert bits to block
first in case of global bitmap.
b) In ocfs2_probe_alloc_group(), when finding enough bits in block
group bitmap, we have to back off move_len to start pos as well,
otherwise it may corrupt filesystem.
c) In ocfs2_ioctl_move_extents(), set me_threshold both for non-auto
and auto defrag paths. Otherwise it will set move_max_hop to 0 and
finally cause unexpectedly ENOSPC error.
Currently there are no tools triggering the above issues since
defragfs.ocfs2 enables auto defrag by default. Tested with manually
changing defragfs.ocfs2 to run non auto defrag path.
Link: https://lkml.kernel.org/r/20230220050526.22020-1-heming.zhao@suse.com Signed-off-by: Heming Zhao <heming.zhao@suse.com> Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: Gang He <ghe@suse.com> Cc: Jun Piao <piaojun@huawei.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This bug has the same root cause of 'commit 7f27ec978b0e ("ocfs2: call
ocfs2_journal_access_di() before ocfs2_journal_dirty() in
ocfs2_write_end_nolock()")'. For this bug, jbd2_journal_restart() is
called by ocfs2_split_extent() during defragmenting.
How to fix
For ocfs2_split_extent() can handle journal operations totally by itself.
Caller doesn't need to call journal access/dirty pair, and caller only
needs to call journal start/stop pair. The fix method is to remove
journal access/dirty from __ocfs2_move_extent().
The discussion for this patch:
https://oss.oracle.com/pipermail/ocfs2-devel/2023-February/000647.html
Link: https://lkml.kernel.org/r/20230217003717.32469-1-heming.zhao@suse.com Signed-off-by: Heming Zhao <heming.zhao@suse.com> Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com> Cc: Mark Fasheh <mark@fasheh.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Changwei Ge <gechangwei@live.cn> Cc: Gang He <ghe@suse.com> Cc: Jun Piao <piaojun@huawei.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When writing a page from an encrypted file that is using
filesystem-layer encryption (not inline encryption), f2fs encrypts the
pagecache page into a bounce page, then writes the bounce page.
It also passes the bounce page to wbc_account_cgroup_owner(). That's
incorrect, because the bounce page is a newly allocated temporary page
that doesn't have the memory cgroup of the original pagecache page.
This makes wbc_account_cgroup_owner() not account the I/O to the owner
of the pagecache page as it should.
Fix this by always passing the pagecache page to
wbc_account_cgroup_owner().
If the storage gives a corrupted node block due to short power failure and
reset, f2fs stops the entire operations by setting the checkpoint failure flag.
Let's give more chances to live by re-issuing IOs for a while in such critical
path.
When converting an inline directory to a regular one, f2fs is leaking
uninitialized memory to disk because it doesn't initialize the entire
directory block. Fix this by zero-initializing the block.
This bug was introduced by commit 4ec17d688d74 ("f2fs: avoid unneeded
initializing when converting inline dentry"), which didn't consider the
security implications of leaking uninitialized memory to disk.
This was found by running xfstest generic/435 on a KMSAN-enabled kernel.
Fixes: 4ec17d688d74 ("f2fs: avoid unneeded initializing when converting inline dentry") Cc: <stable@vger.kernel.org> # v4.3+ Signed-off-by: Eric Biggers <ebiggers@google.com> Reviewed-by: Chao Yu <chao@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch moves to send a ack back for receiving a FIN message only
when we are in valid states. In other cases and there might be a sender
waiting for a ack we just let it timeout at the senders time and
hopefully all other cleanups will remove the FIN message on their
sending queue. As an example we should never send out an ACK being in
LAST_ACK state or we cannot assume a working socket communication when
we are in CLOSED state.
Cc: stable@vger.kernel.org Fixes: 489d8e559c65 ("fs: dlm: add reliable connection if reconnect") Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch moves the send fin handling, which should appear in a specific
state change, into the state change handling while the per node
state_lock is held. I experienced issues with other messages because
we changed the state and a fin message was sent out in a different state.
Cc: stable@vger.kernel.org Fixes: 489d8e559c65 ("fs: dlm: add reliable connection if reconnect") Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Similar to the stop tx flag, the rx flag should warn about a dlm message
being received at DLM_FIN state change, when we are assuming no other
dlm application messages. If we receive a FIN message and we are in the
state DLM_FIN_WAIT2 we call midcomms_node_reset() which puts the
midcomms node into DLM_CLOSED state. Afterwards we should not set the
DLM_NODE_FLAG_STOP_RX flag any more. This patch changes the setting
DLM_NODE_FLAG_STOP_RX in those state changes when we receive a FIN
message and we assume there will be no other dlm application messages
received until we hit DLM_CLOSED state.
Cc: stable@vger.kernel.org Fixes: 489d8e559c65 ("fs: dlm: add reliable connection if reconnect") Signed-off-by: Alexander Aring <aahringo@redhat.com> Signed-off-by: David Teigland <teigland@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When a file or a directory is deleted, the hint for the cluster of
its parent directory in its in-memory inode is set as DIR_DELETED.
Therefore, DIR_DELETED must be one of invalid cluster numbers. According
to the exFAT specification, a volume can have at most 2^32-11 clusters.
However, DIR_DELETED is wrongly defined as 0xFFFF0321, which could be
a valid cluster number. To fix it, let's redefine DIR_DELETED as
0xFFFFFFF7, the bad cluster number.
Fixes: 1acf1a564b60 ("exfat: add in-memory and on-disk structures and headers") Cc: stable@vger.kernel.org # v5.7+ Reported-by: Yuezhang Mo <Yuezhang.Mo@sony.com> Signed-off-by: Sungjong Seo <sj1557.seo@samsung.com> Signed-off-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the position is not aligned with the dentry size, the return
value of readdir() will be NULL and errno is 0, which means the
end of the directory stream is reached.
If the position is aligned with dentry size, but there is no file
or directory at the position, exfat_readdir() will continue to
get dentry from the next dentry. So the dentry gotten by readdir()
may not be at the position.
After this commit, if the position is not aligned with the dentry
size, round the position up to the dentry size and continue to get
the dentry.
Fixes: ca06197382bd ("exfat: add directory operations") Cc: stable@vger.kernel.org # v5.7+ Reported-by: Wang Yugui <wangyugui@e16-tech.com> Signed-off-by: Yuezhang Mo <Yuezhang.Mo@sony.com> Reviewed-by: Andy Wu <Andy.Wu@sony.com> Reviewed-by: Aoyama Wataru <wataru.aoyama@sony.com> Reviewed-by: Sungjong Seo <sj1557.seo@samsung.com> Signed-off-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since seekdir() does not check whether the position is valid, the
position may exceed the size of the directory. We found that for
a directory with discontinuous clusters, if the position exceeds
the size of the directory and the excess size is greater than or
equal to the cluster size, exfat_readdir() will return -EIO,
causing a file system error and making the file system unavailable.
The current hfsplus_put_super first calls hfs_btree_close on
sbi->ext_tree, then invokes iput on sbi->hidden_dir, resulting in an
use-after-free issue in hfsplus_release_folio.
As shown in hfsplus_fill_super, the error handling code also calls iput
before hfs_btree_close.
To fix this error, we move all iput calls before hfsplus_btree_close.
Note that this patch is tested on Syzbot.
Link: https://lkml.kernel.org/r/20230226124948.3175736-1-mudongliangabcd@gmail.com Reported-by: syzbot+57e3e98f7e3b80f64d56@syzkaller.appspotmail.com Tested-by: Dongliang Mu <mudongliangabcd@gmail.com> Signed-off-by: Dongliang Mu <mudongliangabcd@gmail.com> Cc: Bart Van Assche <bvanassche@acm.org> Cc: Jens Axboe <axboe@kernel.dk> Cc: Muchun Song <songmuchun@bytedance.com> Cc: Roman Gushchin <roman.gushchin@linux.dev> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
/* Dispose of resources used by a node */
void hfs_bnode_put(struct hfs_bnode *node)
{
if (node) {
<skipped>
BUG_ON(!atomic_read(&node->refcnt)); <- we have issue here!!!!
<skipped>
}
}
By tracing the refcnt, I found the node is created by hfs_bmap_alloc()
with refcnt 1. Then the node is used by hfs_btree_write(). There is a
missing of hfs_bnode_get() after find the node. The issue happened in
following path:
<alloc>
hfs_bmap_alloc
hfs_bnode_find
__hfs_bnode_create <- allocate a new node with refcnt 1.
hfs_bnode_put <- decrease the refcnt
<write>
hfs_btree_write
hfs_bnode_find
__hfs_bnode_create
hfs_bnode_findhash <- find the node without refcnt increased.
hfs_bnode_put <- trigger the BUG_ON() since refcnt is 0.
Link: https://lkml.kernel.org/r/20221212021627.3766829-1-liushixin2@huawei.com Reported-by: syzbot+5b04b49a7ec7226c7426@syzkaller.appspotmail.com Signed-off-by: Liu Shixin <liushixin2@huawei.com> Cc: Fabio M. De Francesco <fmdefrancesco@gmail.com> Cc: Viacheslav Dubeyko <slava@dubeyko.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Just like for task_work, set the task mode to TASK_RUNNING before doing
any potential resume work. We're not holding any locks at this point,
but we may have already set the task state to TASK_INTERRUPTIBLE in
preparation for going to sleep waiting for events. Ensure that we set it
back to TASK_RUNNING if we have work to process, to avoid warnings on
calling blocking operations with !TASK_RUNNING.
Fixes: b5d3ae202fbf ("io_uring: handle TIF_NOTIFY_RESUME when checking for task_work") Reported-by: kernel test robot <oliver.sang@intel.com> Link: https://lore.kernel.org/oe-lkp/202302062208.24d3e563-oliver.sang@intel.com Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
During rcutorture shutdown, the rcu_torture_cleanup() function calls
torture_cleanup_begin(), which sets the fullstop global variable to
FULLSTOP_RMMOD. This causes the rcutorture threads for readers and
fakewriters to exit all of their "while" loops and start shutting down.
They then call torture_kthread_stopping(), which in turn waits for
kthread_stop() to be called. However, rcu_torture_cleanup() has
not yet called kthread_stop() on those threads, and before it gets a
chance to do so, multiple instances of torture_kthread_stopping() invoke
schedule_timeout_interruptible(1) in a tight loop. Tracing confirms that
TIMER_SOFTIRQ can then continuously execute timer callbacks. If that
TIMER_SOFTIRQ preempts the task executing rcu_torture_cleanup(), that
task might never invoke kthread_stop().
This commit improves this situation by increasing the timeout passed to
schedule_timeout_interruptible() from one jiffy to 1/20th of a second.
This change prevents TIMER_SOFTIRQ from monopolizing its CPU, thus
allowing rcu_torture_cleanup() to carry out the needed kthread_stop()
invocations. Testing has shown 100 runs of TREE07 passing reliably,
as oppose to the tens-of-percent failure rates seen beforehand.
Cc: Paul McKenney <paulmck@kernel.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Zhouyi Zhou <zhouzhouyi@gmail.com> Cc: <stable@vger.kernel.org> # 6.0.x Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org> Tested-by: Zhouyi Zhou <zhouzhouyi@gmail.com> Reviewed-by: Davidlohr Bueso <dave@stgolabs.net> Signed-off-by: Paul E. McKenney <paulmck@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
ksmbd allowed the actual frame length to be smaller than the rfc1002
length. If allowed, it is possible to allocates a large amount of memory
that can be limited by credit management and can eventually cause memory
exhaustion problem. This patch do not allow it except SMB2 Negotiate
request which will be validated when message handling proceeds.
Also, Allow a message that padded to 8byte boundary.
Fixes: e2f34481b24d ("cifsd: add server-side procedures for SMB3") Cc: stable@vger.kernel.org Signed-off-by: Namjae Jeon <linkinjeon@kernel.org> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A non-first waiter can potentially spin in the for loop of
rwsem_down_write_slowpath() without sleeping but fail to acquire the
lock even if the rwsem is free if the following sequence happens:
Non-first RT waiter First waiter Lock holder
------------------- ------------ -----------
Acquire wait_lock
rwsem_try_write_lock():
Set handoff bit if RT or
wait too long
Set waiter->handoff_set
Release wait_lock
Acquire wait_lock
Inherit waiter->handoff_set
Release wait_lock
Clear owner
Release lock
if (waiter.handoff_set) {
rwsem_spin_on_owner(();
if (OWNER_NULL)
goto trylock_again;
}
trylock_again:
Acquire wait_lock
rwsem_try_write_lock():
if (first->handoff_set && (waiter != first))
return false;
Release wait_lock
A non-first waiter cannot really acquire the rwsem even if it mistakenly
believes that it can spin on OWNER_NULL value. If that waiter happens
to be an RT task running on the same CPU as the first waiter, it can
block the first waiter from acquiring the rwsem leading to live lock.
Fix this problem by making sure that a non-first waiter cannot spin in
the slowpath loop without sleeping.
Fixes: d257cc8cb8d5 ("locking/rwsem: Make handoff bit handling more consistent") Signed-off-by: Waiman Long <longman@redhat.com> Signed-off-by: Ingo Molnar <mingo@kernel.org> Tested-by: Mukesh Ojha <quic_mojha@quicinc.com> Reviewed-by: Mukesh Ojha <quic_mojha@quicinc.com> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20230126003628.365092-2-longman@redhat.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Async discard does not acquire the block group reference count while it
holds a reference on the discard list. This is generally OK, as the
paths which destroy block groups tend to try to synchronize on
cancelling async discard work. However, relying on cancelling work
requires careful analysis to be sure it is safe from races with
unpinning scheduling more work.
While I am unable to find a race with unpinning in the current code for
either the unused bgs or relocation paths, I believe we have one in an
older version of auto relocation in a Meta internal build. This suggests
that this is in fact an error prone model, and could be fragile to
future changes to these bg deletion paths.
To make this ownership more clear, add a refcount for async discard. If
work is queued for a block group, its refcount should be incremented,
and when work is completed or canceled, it should be decremented.
CC: stable@vger.kernel.org # 5.15+ Signed-off-by: Boris Burkov <boris@bur.io> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the function mpi3mr_get_all_tgt_info(), devmap_info points to
alltgt_info->dmi then there is no need to memcpy() data from devmap_info to
alltgt_info->dmi. Remove the unnecessary memcpy(). This also allows to
remove the local variable 'rval' and the goto label 'out'.
Link: https://lore.kernel.org/r/20230214005019.1897251-3-shinichiro.kawasaki@wdc.com Cc: stable@vger.kernel.org Fixes: f5e6d5a34376 ("scsi: mpi3mr: Add support for driver commands") Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> Acked-by: Sathya Prakash Veerichetty <sathya.prakash@broadcom.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The function mpi3mr_get_all_tgt_info() has four issues:
1) It calculates valid entry length in alltgt_info assuming the header part
of the struct mpi3mr_device_map_info would equal to sizeof(u32). The
correct size is sizeof(u64).
2) When it calculates the valid entry length kern_entrylen, it excludes one
entry by subtracting 1 from num_devices.
3) It copies num_device by calling memcpy(). Substitution is enough.
4) It does not specify the calculated length to sg_copy_from_buffer().
Instead, it specifies the payload length which is larger than the
alltgt_info size. It causes "BUG: KASAN: slab-out-of-bounds".
Fix the issues by using the correct header size, removing the subtraction
from num_devices, replacing the memcpy() with substitution and specifying
the correct length to sg_copy_from_buffer().
Link: https://lore.kernel.org/r/20230214005019.1897251-2-shinichiro.kawasaki@wdc.com Cc: stable@vger.kernel.org Fixes: f5e6d5a34376 ("scsi: mpi3mr: Add support for driver commands") Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> Acked-by: Sathya Prakash Veerichetty <sathya.prakash@broadcom.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit c1af985d27da ("scsi: mpi3mr: Add Event acknowledgment logic")
introduced an array mrioc->evtack_cmds but initialization of the array
elements was missed. They are just zero cleared. The function
mpi3mr_complete_evt_ack() refers host_tag field of the elements. Due to the
zero value of the host_tag field, the function calls clear_bit() for
mrico->evtack_cmds_bitmap with wrong bit index. This results in memory
access to invalid address and "BUG: KASAN: use-after-free". This BUG was
observed at eHBA-9600 firmware update to version 8.3.1.0. To fix it, add
the missing initialization of mrioc->evtack_cmds.
Link: https://lore.kernel.org/r/20230214005019.1897251-5-shinichiro.kawasaki@wdc.com Cc: stable@vger.kernel.org Fixes: c1af985d27da ("scsi: mpi3mr: Add Event acknowledgment logic") Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com> Reviewed-by: Damien Le Moal <damien.lemoal@opensource.wdc.com> Acked-by: Sathya Prakash Veerichetty <sathya.prakash@broadcom.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If we did not get a lease we can still return a single use cfid to the caller.
The cfid will not have has_lease set and will thus not be shared with any
other concurrent users and will be freed immediately when the caller
drops the handle.
This avoids extra roundtrips for servers that do not support directory leases
where they would first fail to get a cfid with a lease and then fallback
to try a normal SMB2_open()
Signed-off-by: Ronnie Sahlberg <lsahlber@redhat.com> Cc: stable@vger.kernel.org Reviewed-by: Bharath SM <bharathsm@microsoft.com> Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some servers may return that we got a lease in rsp->OplockLevel
but then in the lease context contradict this and say we got no lease
at all. Thus we need to check the context if we have a lease.
Additionally, If we do not get a lease we need to make sure we close
the handle before we return an error to the caller.
Signed-off-by: Ronnie Sahlberg <lsahlber@redhat.com> Cc: stable@vger.kernel.org Reviewed-by: Bharath SM <bharathsm@microsoft.com> Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The aim of using encryption on a connection is to keep
the data confidential, so we must not use plaintext rdma offload
for that data!
It seems that current windows servers and ksmbd would allow
this, but that's no reason to expose the users data in plaintext!
And servers hopefully reject this in future.
Note modern windows servers support signed or encrypted offload,
see MS-SMB2 2.2.3.1.6 SMB2_RDMA_TRANSFORM_CAPABILITIES, but we don't
support that yet.
Signed-off-by: Stefan Metzmacher <metze@samba.org> 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: stable@vger.kernel.org Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We should have the logic to decide if we want rdma offload
in a single spot in order to advance it in future.
Signed-off-by: Stefan Metzmacher <metze@samba.org> 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: stable@vger.kernel.org Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This will simplify the following changes and makes it easy to get
in passed in from the caller in future.
Signed-off-by: Stefan Metzmacher <metze@samba.org> 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: stable@vger.kernel.org Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The client was sending rfc1002 session request packet with a wrong
length field set, therefore failing to mount shares against old SMB
servers over port 139.
Fix this by calculating the correct length as specified in rfc1002.
Fixes: d7173623bf0b ("cifs: use ALIGN() and round_up() macros") Cc: stable@vger.kernel.org Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com> Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Migration mode is a VM attribute which enables tracking of changes in
storage attributes (PGSTE). It assumes dirty tracking is enabled on all
memslots to keep a dirty bitmap of pages with changed storage attributes.
When enabling migration mode, we currently check that dirty tracking is
enabled for all memslots. However, userspace can disable dirty tracking
without disabling migration mode.
Since migration mode is pointless with dirty tracking disabled, disable
migration mode whenever userspace disables dirty tracking on any slot.
Also update the documentation to clarify that dirty tracking must be
enabled when enabling migration mode, which is already enforced by the
code in kvm_s390_vm_start_migration().
Also highlight in the documentation for KVM_S390_GET_CMMA_BITS that it
can now fail with -EINVAL when dirty tracking is disabled while
migration mode is on. Move all the error codes to a table so this stays
readable.
To disable migration mode, slots_lock should be held, which is taken
in kvm_set_memory_region() and thus held in
kvm_arch_prepare_memory_region().
Restructure the prepare code a bit so all the sanity checking is done
before disabling migration mode. This ensures migration mode isn't
disabled when some sanity check fails.
Recent test_kprobe_missed kprobes kunit test uncovers the following
problem. Once kprobe is triggered from another kprobe (kprobe reenter),
all future kprobes on this cpu are considered as kprobe reenter, thus
pre_handler and post_handler are not being called and kprobes are counted
as "missed".
Commit b9599798f953 ("[S390] kprobes: activation and deactivation")
introduced a simpler scheme for kprobes (de)activation and status
tracking by using push_kprobe/pop_kprobe, which supposed to work for
both initial kprobe entry as well as kprobe reentry and helps to avoid
handling those two cases differently. The problem is that a sequence of
calls in case of kprobes reenter:
push_kprobe() <- NULL (current_kprobe)
push_kprobe() <- kprobe1 (current_kprobe)
pop_kprobe() -> kprobe1 (current_kprobe)
pop_kprobe() -> kprobe1 (current_kprobe)
leaves "kprobe1" as "current_kprobe" on this cpu, instead of setting it
to NULL. In fact push_kprobe/pop_kprobe can only store a single state
(there is just one prev_kprobe in kprobe_ctlblk). Which is a hack but
sufficient, there is no need to have another prev_kprobe just to store
NULL. To make a simple and backportable fix simply reset "prev_kprobe"
when kprobe is poped from this "stack". No need to worry about
"kprobe_status" in this case, because its value is only checked when
current_kprobe != NULL.
Recent test_kprobe_missed kprobes kunit test uncovers the following error
(reported when CONFIG_DEBUG_ATOMIC_SLEEP is enabled):
BUG: sleeping function called from invalid context at kernel/locking/mutex.c:580
in_atomic(): 0, irqs_disabled(): 1, non_block: 0, pid: 662, name: kunit_try_catch
preempt_count: 0, expected: 0
RCU nest depth: 0, expected: 0
no locks held by kunit_try_catch/662.
irq event stamp: 280
hardirqs last enabled at (279): [<00000003e60a3d42>] __do_pgm_check+0x17a/0x1c0
hardirqs last disabled at (280): [<00000003e3bd774a>] kprobe_exceptions_notify+0x27a/0x318
softirqs last enabled at (0): [<00000003e3c5c890>] copy_process+0x14a8/0x4c80
softirqs last disabled at (0): [<0000000000000000>] 0x0
CPU: 46 PID: 662 Comm: kunit_try_catch Tainted: G N 6.2.0-173644-g44c18d77f0c0 #2
Hardware name: IBM 3931 A01 704 (LPAR)
Call Trace:
[<00000003e60a3a00>] dump_stack_lvl+0x120/0x198
[<00000003e3d02e82>] __might_resched+0x60a/0x668
[<00000003e60b9908>] __mutex_lock+0xc0/0x14e0
[<00000003e60bad5a>] mutex_lock_nested+0x32/0x40
[<00000003e3f7b460>] unregister_kprobe+0x30/0xd8
[<00000003e51b2602>] test_kprobe_missed+0xf2/0x268
[<00000003e51b5406>] kunit_try_run_case+0x10e/0x290
[<00000003e51b7dfa>] kunit_generic_run_threadfn_adapter+0x62/0xb8
[<00000003e3ce30f8>] kthread+0x2d0/0x398
[<00000003e3b96afa>] __ret_from_fork+0x8a/0xe8
[<00000003e60ccada>] ret_from_fork+0xa/0x40
The reason for this error report is that kprobes handling code failed
to restore irqs.
The problem is that when kprobe is triggered from another kprobe
post_handler current sequence of enable_singlestep / disable_singlestep
is the following:
enable_singlestep <- original kprobe (saves kprobe_saved_imask)
enable_singlestep <- kprobe triggered from post_handler (clobbers kprobe_saved_imask)
disable_singlestep <- kprobe triggered from post_handler (restores kprobe_saved_imask)
disable_singlestep <- original kprobe (restores wrong clobbered kprobe_saved_imask)
There is just one kprobe_ctlblk per cpu and both calls saves and
loads irq mask to kprobe_saved_imask. To fix the problem simply move
resume_execution (which calls disable_singlestep) before calling
post_handler. This also fixes the problem that post_handler is called
with pt_regs which were not yet adjusted after single-stepping.
When debugging vmlinux with QEMU + GDB, the following GDB error may
occur:
(gdb) c
Continuing.
Warning:
Cannot insert breakpoint -1.
Cannot access memory at address 0xffffffffffff95c0
Command aborted.
(gdb)
The reason is that, when .interp section is present, GDB tries to
locate the file specified in it in memory and put a number of
breakpoints there (see enable_break() function in gdb/solib-svr4.c).
Sometimes GDB finds a bogus location that matches its heuristics,
fails to set a breakpoint and stops. This makes further debugging
impossible.
The .interp section contains misleading information anyway (vmlinux
does not need ld.so), so fix by discarding it.
Commit f05f62d04271f ("s390/vmem: get rid of memory segment list")
reshuffled the call to vmem_add_mapping() in __segment_load(), which now
overwrites rc after it was set to contain the segment type code.
As result, __segment_load() will now always return 0 on success, which
corresponds to the segment type code SEG_TYPE_SW, i.e. a writeable
segment. This results in a kernel crash when loading a read-only segment
as dcssblk block device, and trying to write to it.
Instead of reshuffling code again, make sure to return the segment type
on success, and also describe this rather delicate and unexpected logic
in the function comment. Also initialize new segtype variable with
invalid value, to prevent possible future confusion.
Heming reported a BUG when using io_uring doing link-cp on ocfs2. [1]
Do the following steps can reproduce this BUG:
mount -t ocfs2 /dev/vdc /mnt/ocfs2
cp testfile /mnt/ocfs2/
./link-cp /mnt/ocfs2/testfile /mnt/ocfs2/testfile.1
umount /mnt/ocfs2
Then umount will fail, and it outputs:
umount: /mnt/ocfs2: target is busy.
While tracing umount, it blames mnt_get_count() not return as expected.
Do a deep investigation for fget()/fput() on related code flow, I've
finally found that fget() leaks since ocfs2 doesn't support nowait
buffered read.
io_issue_sqe
|-io_assign_file // do fget() first
|-io_read
|-io_iter_do_read
|-ocfs2_file_read_iter // return -EOPNOTSUPP
|-kiocb_done
|-io_rw_done
|-__io_complete_rw_common // set REQ_F_REISSUE
|-io_resubmit_prep
|-io_req_prep_async // override req->file, leak happens
This was introduced by commit a196c78b5443 in v5.18. Fix it by don't
re-assign req->file if it has already been assigned.
MSG_NOSIGNAL is not applicable for the receiving side, SIGPIPE is
generated when trying to write to a "broken pipe". AF_PACKET's
packet_recvmsg() does enforce this, giving back EINVAL when MSG_NOSIGNAL
is set - making it unuseable in io_uring's recvmsg.
Remove MSG_NOSIGNAL from io_recvmsg_prep().
Cc: stable@vger.kernel.org # v5.10+ Signed-off-by: David Lamparter <equinox@diac24.net> Cc: Eric Dumazet <edumazet@google.com> Cc: Jens Axboe <axboe@kernel.dk> Reviewed-by: Eric Dumazet <edumazet@google.com> Link: https://lore.kernel.org/r/20230224150123.128346-1-equinox@diac24.net Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If two or more mappings go back to back to each other they can be passed
into io_uring to be registered as a single registered buffer. That would
even work if mappings came from different sources, e.g. it's possible to
mix in this way anon pages and pages from shmem or hugetlb. That is not
a problem but it'd rather be less prone if we forbid such mixing.
If CONFIG_PREEMPT_NONE is set and the task_work chains are long, we
could be running into issues blocking others for too long. Add a
reschedule check in handle_tw_list(), and flush the ctx if we need to
reschedule.
If the kernel is configured with CONFIG_PREEMPT_NONE, we could be
sitting in a tight loop reaping events but not giving them a chance to
finish. This results in a trace ala:
If TIF_NOTIFY_RESUME is set, then we need to call resume_user_mode_work()
for PF_IO_WORKER threads. They never return to usermode, hence never get
a chance to process any items that are marked by this flag. Most notably
this includes the final put of files, but also any throttling markers set
by block cgroups.
We return POLLIN from io_uring_poll() depending on whether there are
CQEs for the userspace, and so we should use the user visible tail
pointer instead of a transient cached value.
Zero-length arrays are deprecated[1]. Replace struct io_uring_buf_ring's
"bufs" with a flexible array member. (How is the size of this array
verified?) Detected with GCC 13, using -fstrict-flex-arrays=3:
In function 'io_ring_buffer_select',
inlined from 'io_buffer_select' at io_uring/kbuf.c:183:10:
io_uring/kbuf.c:141:23: warning: array subscript 255 is outside the bounds of an interior zero-length array 'struct io_uring_buf[0]' [-Wzero-length-bounds]
141 | buf = &br->bufs[head];
| ^~~~~~~~~~~~~~~
In file included from include/linux/io_uring.h:7,
from io_uring/kbuf.c:10:
include/uapi/linux/io_uring.h: In function 'io_buffer_select':
include/uapi/linux/io_uring.h:628:41: note: while referencing 'bufs'
628 | struct io_uring_buf bufs[0];
| ^~~~
The resend_msg() function cannot fail, but there was error handling
around using it. Rework the handling of the error, and fix the out of
retries debug reporting that was wrong around this, too.
Make sure to disable the alarm before updating the four alarm time
registers to avoid spurious alarms during the update.
Note that the disable needs to be done outside of the ctrl_reg_lock
section to prevent a racing alarm interrupt from disabling the newly set
alarm when the lock is released.
Fixes: 9a9a54ad7aa2 ("drivers/rtc: add support for Qualcomm PMIC8xxx RTC") Cc: stable@vger.kernel.org # 3.1 Signed-off-by: Johan Hovold <johan+linaro@kernel.org> Reviewed-by: David Collins <quic_collinsd@quicinc.com> Link: https://lore.kernel.org/r/20230202155448.6715-2-johan+linaro@kernel.org Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
which is due to bio->bi_bdev being NULL. This can happen if we have two
tasks doing polled IO, and task B ends up completing IO from task A if
they are sharing a poll queue. If task B completes the IO and puts the
bio into our cache, then it can allocate that bio again before task A
is done polling for it. As that would necessitate a preempt between the
two tasks, it's enough to just be a bit more careful in checking for
whether or not bio->bi_bdev is NULL.
This isn't strictly needed in terms of correctness, but it does allow
polling to know if the bio has been put already by a different task
and hence avoid polling something that we don't need to.
If we're doing a large IO request which needs to be split into multiple
bios for issue, then we can run into the same situation as the below
marked commit fixes - parts will complete just fine, one or more parts
will fail to allocate a request. This will result in a partially
completed read or write request, where the caller gets EAGAIN even though
parts of the IO completed just fine.
Do the same for large bios as we do for splits - fail a NOWAIT request
with EAGAIN. This isn't technically fixing an issue in the below marked
patch, but for stable purposes, we should have either none of them or
both.
This depends on: 613b14884b85 ("block: handle bio_split_to_limits() NULL return")
Cc: stable@vger.kernel.org # 5.15+ Fixes: 9cea62b2cbab ("block: don't allow splitting of a REQ_NOWAIT bio") Link: https://github.com/axboe/liburing/issues/766 Reported-and-tested-by: Michael Kelley <mikelley@microsoft.com> Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The coreboot framebuffer doesn't support transparency, its 'reserved'
bit field is merely padding for byte/word alignment of pixel colors [1].
When trying to match the framebuffer to a simplefb format, the kernel
driver unnecessarily requires the format's transparency bit field to
exactly match this padding, even if the former is zero-width.
Due to a coreboot bug [2] (fixed upstream), some boards misreport the
reserved field's size as equal to its position (0x18 for both on a
'Lick' Chromebook), and the driver fails to probe where it would have
otherwise worked fine with e.g. the a8r8g8b8 or x8r8g8b8 formats.
Remove the transparency comparison with reserved bits. When the
bits-per-pixel and other color components match, transparency will
already be in a subset of the reserved field. Not forcing it to match
reserved bits allows the driver to work on the boards which misreport
the reserved field. It also enables using simplefb formats that don't
have transparency bits, although this doesn't currently happen due to
format support and ordering in linux/platform_data/simplefb.h.
fw_devlink can sometimes try to create a device link with the consumer
and supplier as the same device. These attempts will fail (correctly),
but are harmless. So, avoid printing an error for these cases. Also, add
more detail to the error message.
The referenced commit added a wrapper for drm_gem_shmem_get_pages_sgt(),
but in the process it accidentally changed the export type from GPL to
non-GPL. Switch it back to GPL.
projects / users / jedix / linux-maple.git / log