There is a KASAN slab-out-of-bounds:
BUG: KASAN: slab-out-of-bounds in _copy_from_iter_full+0x783/0xaa0
Read of size 80 at addr ffff88810c35e180 by task mount.cifs/539
It can be reproduced by the following step:
1. samba configured with: server max protocol = SMB2_10
2. mount -o vers=default
When parse the mount version parameter, the 'ops' and 'vals'
was setted to smb30, if negotiate result is smb21, just
update the 'ops' to smb21, but the 'vals' is still smb30.
When add lease context, the iov_base is allocated with smb21
ops, but the iov_len is initiallited with the smb30. Because
the iov_len is longer than iov_base, when send the message,
copy array out of bounds.
we need to keep the 'ops' and 'vals' consistent.
Fixes: 9764c02fcbad ("SMB3: Add support for multidialect negotiate (SMB2.1 and later)") Fixes: d5c7076b772a ("smb3: add smb3.1.1 to default dialect list") Signed-off-by: ZhangXiaoxu <[email protected]> Signed-off-by: Steve French <[email protected]> CC: Stable <[email protected]> Reviewed-by: Pavel Shilovsky <[email protected]>
Mao Han [Wed, 10 Apr 2019 08:16:43 +0000 (16:16 +0800)]
perf evsel: Use hweight64() instead of hweight_long(attr.sample_regs_user)
On 32-bits platform with more than 32 registers, the 64 bits mask is
truncate to the lower 32 bits and the return value of hweight_long will
always smaller than 32. When kernel outputs more than 32 registers, but
the user perf program only counts 32, there will be a data mismatch
result to overflow check fail.
tools lib traceevent: Fix missing equality check for strcmp
There was a missing comparison with 0 when checking if type is "s64" or
"u64". Therefore, the body of the if-statement was entered if "type" was
"u64" or not "s64", which made the first strcmp() redundant since if
type is "u64", it's not "s64".
If type is "s64", the body of the if-statement is not entered but since
the remainder of the function consists of if-statements which will not
be entered if type is "s64", we will just return "val", which is
correct, albeit at the cost of a few more calls to strcmp(), i.e., it
will behave just as if the if-statement was entered.
If type is neither "s64" or "u64", the body of the if-statement will be
entered incorrectly and "val" returned. This means that any type that is
checked after "s64" and "u64" is handled the same way as "s64" and
"u64", i.e., the limiting of "val" to fit in for example "s8" is never
reached.
This was introduced in the kernel tree when the sources were copied from
trace-cmd in commit f7d82350e597 ("tools/events: Add files to create
libtraceevent.a"), and in the trace-cmd repo in 1cdbae6035cei
("Implement typecasting in parser") when the function was introduced,
i.e., it has always behaved the wrong way.
The ALSA proc helper manages the child nodes in a linked list, but its
addition and deletion is done without any lock. This leads to a
corruption if they are operated concurrently. Usually this isn't a
problem because the proc entries are added sequentially in the driver
probe procedure itself. But the card registrations are done often
asynchronously, and the crash could be actually reproduced with
syzkaller.
This patch papers over it by protecting the link addition and deletion
with the parent's mutex. There is "access" mutex that is used for the
file access, and this can be reused for this purpose as well.
Vitaly Kuznetsov [Wed, 27 Mar 2019 14:12:20 +0000 (15:12 +0100)]
KVM: x86: avoid misreporting level-triggered irqs as edge-triggered in tracing
In __apic_accept_irq() interface trig_mode is int and actually on some code
paths it is set above u8:
kvm_apic_set_irq() extracts it from 'struct kvm_lapic_irq' where trig_mode
is u16. This is done on purpose as e.g. kvm_set_msi_irq() sets it to
(1 << 15) & e->msi.data
kvm_apic_local_deliver sets it to reg & (1 << 15).
Fix the immediate issue by making 'tm' into u16. We may also want to adjust
__apic_accept_irq() interface and use proper sizes for vector, level,
trig_mode but this is not urgent.
Add a simple test for SMM, based on VMX. The test implements its own
sync between the guest and the host as using our ucall library seems to
be too cumbersome: SMI handler is happening in real-address mode.
This patch also fixes KVM_SET_NESTED_STATE to happen after
KVM_SET_VCPU_EVENTS, in fact it places it last. This is because
KVM needs to know whether the processor is in SMM or not.
Paolo Bonzini [Thu, 11 Apr 2019 13:51:19 +0000 (15:51 +0200)]
selftests: kvm: fix for compilers that do not support -no-pie
-no-pie was added to GCC at the same time as their configuration option
--enable-default-pie. Compilers that were built before do not have
-no-pie, but they also do not need it. Detect the option at build
time.
Paolo Bonzini [Thu, 11 Apr 2019 13:57:14 +0000 (15:57 +0200)]
selftests: kvm/evmcs_test: complete I/O before migrating guest state
Starting state migration after an IO exit without first completing IO
may result in test failures. We already have two tests that need this
(this patch in fact fixes evmcs_test, similar to what was fixed for
state_test in commit 0f73bbc851ed, "KVM: selftests: complete IO before
migrating guest state", 2019-03-13) and a third is coming. So, move the
code to vcpu_save_state, and while at it do not access register state
until after I/O is complete.
KVM: x86: Always use 32-bit SMRAM save state for 32-bit kernels
Invoking the 64-bit variation on a 32-bit kenrel will crash the guest,
trigger a WARN, and/or lead to a buffer overrun in the host, e.g.
rsm_load_state_64() writes r8-r15 unconditionally, but enum kvm_reg and
thus x86_emulate_ctxt._regs only define r8-r15 for CONFIG_X86_64.
KVM allows userspace to report long mode support via CPUID, even though
the guest is all but guaranteed to crash if it actually tries to enable
long mode. But, a pure 32-bit guest that is ignorant of long mode will
happily plod along.
SMM complicates things as 64-bit CPUs use a different SMRAM save state
area. KVM handles this correctly for 64-bit kernels, e.g. uses the
legacy save state map if userspace has hid long mode from the guest,
but doesn't fare well when userspace reports long mode support on a
32-bit host kernel (32-bit KVM doesn't support 64-bit guests).
Since the alternative is to crash the guest, e.g. by not loading state
or explicitly requesting shutdown, unconditionally use the legacy SMRAM
save state map for 32-bit KVM. If a guest has managed to get far enough
to handle SMIs when running under a weird/buggy userspace hypervisor,
then don't deliberately crash the guest since there are no downsides
(from KVM's perspective) to allow it to continue running.
KVM: x86: Don't clear EFER during SMM transitions for 32-bit vCPU
Neither AMD nor Intel CPUs have an EFER field in the legacy SMRAM save
state area, i.e. don't save/restore EFER across SMM transitions. KVM
somewhat models this, e.g. doesn't clear EFER on entry to SMM if the
guest doesn't support long mode. But during RSM, KVM unconditionally
clears EFER so that it can get back to pure 32-bit mode in order to
start loading CRs with their actual non-SMM values.
Clear EFER only when it will be written when loading the non-SMM state
so as to preserve bits that can theoretically be set on 32-bit vCPUs,
e.g. KVM always emulates EFER_SCE.
And because CR4.PAE is cleared only to play nice with EFER, wrap that
code in the long mode check as well. Note, this may result in a
compiler warning about cr4 being consumed uninitialized. Re-read CR4
even though it's technically unnecessary, as doing so allows for more
readable code and RSM emulation is not a performance critical path.
KVM: x86: clear SMM flags before loading state while leaving SMM
RSM emulation is currently broken on VMX when the interrupted guest has
CR4.VMXE=1. Stop dancing around the issue of HF_SMM_MASK being set when
loading SMSTATE into architectural state, e.g. by toggling it for
problematic flows, and simply clear HF_SMM_MASK prior to loading
architectural state (from SMRAM save state area).
Prepare for clearing HF_SMM_MASK prior to loading state from the SMRAM
save state map, i.e. kvm_smm_changed() needs to be called after state
has been loaded and so cannot be done automatically when setting
hflags from RSM.
KVM: x86: Load SMRAM in a single shot when leaving SMM
RSM emulation is currently broken on VMX when the interrupted guest has
CR4.VMXE=1. Rather than dance around the issue of HF_SMM_MASK being set
when loading SMSTATE into architectural state, ideally RSM emulation
itself would be reworked to clear HF_SMM_MASK prior to loading non-SMM
architectural state.
Ostensibly, the only motivation for having HF_SMM_MASK set throughout
the loading of state from the SMRAM save state area is so that the
memory accesses from GET_SMSTATE() are tagged with role.smm. Load
all of the SMRAM save state area from guest memory at the beginning of
RSM emulation, and load state from the buffer instead of reading guest
memory one-by-one.
This paves the way for clearing HF_SMM_MASK prior to loading state,
and also aligns RSM with the enter_smm() behavior, which fills a
buffer and writes SMRAM save state in a single go.
Liran Alon [Mon, 25 Mar 2019 19:09:17 +0000 (21:09 +0200)]
KVM: nVMX: Expose RDPMC-exiting only when guest supports PMU
Issue was discovered when running kvm-unit-tests on KVM running as L1 on
top of Hyper-V.
When vmx_instruction_intercept unit-test attempts to run RDPMC to test
RDPMC-exiting, it is intercepted by L1 KVM which it's EXIT_REASON_RDPMC
handler raise #GP because vCPU exposed by Hyper-V doesn't support PMU.
Instead of unit-test expectation to be reflected with EXIT_REASON_RDPMC.
The reason vmx_instruction_intercept unit-test attempts to run RDPMC
even though Hyper-V doesn't support PMU is because L1 expose to L2
support for RDPMC-exiting. Which is reasonable to assume that is
supported only in case CPU supports PMU to being with.
Above issue can easily be simulated by modifying
vmx_instruction_intercept config in x86/unittests.cfg to run QEMU with
"-cpu host,+vmx,-pmu" and run unit-test.
To handle issue, change KVM to expose RDPMC-exiting only when guest
supports PMU.
Liran Alon [Mon, 25 Mar 2019 19:10:17 +0000 (21:10 +0200)]
KVM: x86: Raise #GP when guest vCPU do not support PMU
Before this change, reading a VMware pseduo PMC will succeed even when
PMU is not supported by guest. This can easily be seen by running
kvm-unit-test vmware_backdoors with "-cpu host,-pmu" option.
In this case, host_xcr0 is 0x2ff, guest vcpu xcr0 is 0xff. After schedule
out, host cpu has guest xcr0 loaded (0xff).
In __switch_to {
switch_fpu_finish
copy_kernel_to_fpregs
XRSTORS
If any bit i in XSTATE_BV[i] == 1 and xcr0[i] == 0, XRSTORS will
generate #GP (In this case, bit 9). Then ex_handler_fprestore kicks in
and tries to reinitialize fpu by restoring init fpu state. Same story as
last #GP, except we get DOUBLE FAULT this time.
KVM: x86: svm: make sure NMI is injected after nmi_singlestep
I noticed that apic test from kvm-unit-tests always hangs on my EPYC 7401P,
the hanging test nmi-after-sti is trying to deliver 30000 NMIs and tracing
shows that we're sometimes able to deliver a few but never all.
When we're trying to inject an NMI we may fail to do so immediately for
various reasons, however, we still need to inject it so enable_nmi_window()
arms nmi_singlestep mode. #DB occurs as expected, but we're not checking
for pending NMIs before entering the guest and unless there's a different
event to process, the NMI will never get delivered.
Make KVM_REQ_EVENT request on the vCPU from db_interception() to make sure
pending NMIs are checked and possibly injected.
Only clear the valid bit when invalidate logical APIC id entry.
The current logic clear the valid bit, but also set the rest of
the bits (including reserved bits) to 1.
that the change coud potentially cause an extra IPI to be sent to
the destination vcpu because the AVIC hardware already set the IRR bit
before the incomplete IPI #VMEXIT with id=1 (target vcpu is not running).
Since writting to ICR and ICR2 will also set the IRR. If something triggers
the destination vcpu to get scheduled before the emulation finishes, then
this could result in an additional IPI.
Also, the issue mentioned in the commit bb218fbcfaaa was misdiagnosed.
Ben Gardon [Mon, 8 Apr 2019 18:07:30 +0000 (11:07 -0700)]
kvm: mmu: Fix overflow on kvm mmu page limit calculation
KVM bases its memory usage limits on the total number of guest pages
across all memslots. However, those limits, and the calculations to
produce them, use 32 bit unsigned integers. This can result in overflow
if a VM has more guest pages that can be represented by a u32. As a
result of this overflow, KVM can use a low limit on the number of MMU
pages it will allocate. This makes KVM unable to map all of guest memory
at once, prompting spurious faults.
Tested: Ran all kvm-unit-tests on an Intel Haswell machine. This patch
introduced no new failures.
Paolo Bonzini [Mon, 15 Apr 2019 13:57:19 +0000 (15:57 +0200)]
KVM: nVMX: always use early vmcs check when EPT is disabled
The remaining failures of vmx.flat when EPT is disabled are caused by
incorrectly reflecting VMfails to the L1 hypervisor. What happens is
that nested_vmx_restore_host_state corrupts the guest CR3, reloading it
with the host's shadow CR3 instead, because it blindly loads GUEST_CR3
from the vmcs01.
For simplicity let's just always use hardware VMCS checks when EPT is
disabled. This way, nested_vmx_restore_host_state is not reached at
all (or at least shouldn't be reached).
sc16is7xx: move label 'err_spi' to correct section
err_spi is used when SERIAL_SC16IS7XX_SPI is enabled, so make
the label only available under SERIAL_SC16IS7XX_SPI option.
Otherwise, the below warning appears.
drivers/tty/serial/sc16is7xx.c:1523:1: warning: label ‘err_spi’ defined but not used [-Wunused-label]
err_spi:
^~~~~~~
serial: sh-sci: Fix HSCIF RX sampling point calculation
There are several issues with the formula used for calculating the
deviation from the intended rate:
1. While min_err and last_stop are signed, srr and baud are unsigned.
Hence the signed values are promoted to unsigned, which will lead
to a bogus value of deviation if min_err is negative,
2. Srr is the register field value, which is one less than the actual
sampling rate factor,
3. The divisions do not use rounding.
Fix this by casting unsigned variables to int, adding one to srr, and
using a single DIV_ROUND_CLOSEST().
Commit 008258d995a6 ("clocksource/drivers/timer-ti-dm: Make
omap_dm_timer_set_load_start() static") made omap_dm_time_set_load_start
static because its prototype was not defined in a header. Unfortunately,
this causes a build warning on multi_v7_defconfig because this function
is not used anywhere in this translation unit:
drivers/clocksource/timer-ti-dm.c:589:12: error: unused function
'omap_dm_timer_set_load_start' [-Werror,-Wunused-function]
In fact, omap_dm_timer_set_load_start hasn't been used anywhere since
commit f190be7f39a5 ("staging: tidspbridge: remove driver") and the
prototype was removed in commit 592ea6bd1fad ("clocksource: timer-ti-dm:
Make unexported functions static"), which is probably where this should
have happened.
Fixes: 592ea6bd1fad ("clocksource: timer-ti-dm: Make unexported functions static") Fixes: 008258d995a6 ("clocksource/drivers/timer-ti-dm: Make omap_dm_timer_set_load_start() static") Signed-off-by: Nathan Chancellor <[email protected]> Acked-by: Tony Lindgren <[email protected]> Signed-off-by: Daniel Lezcano <[email protected]>
staging: erofs: fix unexpected out-of-bound data access
Unexpected out-of-bound data will be read in erofs_read_raw_page
after commit 07173c3ec276 ("block: enable multipage bvecs") since
one iovec could have multiple pages.
Let's fix as what Ming's pointed out in the previous email [1].
Ian Abbott [Mon, 15 Apr 2019 11:52:30 +0000 (12:52 +0100)]
staging: comedi: vmk80xx: Fix possible double-free of ->usb_rx_buf
`vmk80xx_alloc_usb_buffers()` is called from `vmk80xx_auto_attach()` to
allocate RX and TX buffers for USB transfers. It allocates
`devpriv->usb_rx_buf` followed by `devpriv->usb_tx_buf`. If the
allocation of `devpriv->usb_tx_buf` fails, it frees
`devpriv->usb_rx_buf`, leaving the pointer set dangling, and returns an
error. Later, `vmk80xx_detach()` will be called from the core comedi
module code to clean up. `vmk80xx_detach()` also frees both
`devpriv->usb_rx_buf` and `devpriv->usb_tx_buf`, but
`devpriv->usb_rx_buf` may have already been freed, leading to a
double-free error. Fix it by removing the call to
`kfree(devpriv->usb_rx_buf)` from `vmk80xx_alloc_usb_buffers()`, relying
on `vmk80xx_detach()` to free the memory.
Ian Abbott [Mon, 15 Apr 2019 11:10:14 +0000 (12:10 +0100)]
staging: comedi: vmk80xx: Fix use of uninitialized semaphore
If `vmk80xx_auto_attach()` returns an error, the core comedi module code
will call `vmk80xx_detach()` to clean up. If `vmk80xx_auto_attach()`
successfully allocated the comedi device private data,
`vmk80xx_detach()` assumes that a `struct semaphore limit_sem` contained
in the private data has been initialized and uses it. Unfortunately,
there are a couple of places where `vmk80xx_auto_attach()` can return an
error after allocating the device private data but before initializing
the semaphore, so this assumption is invalid. Fix it by initializing
the semaphore just after allocating the private data in
`vmk80xx_auto_attach()` before any other errors can be returned.
I believe this was the cause of the following syzbot crash report
<https://syzkaller.appspot.com/bug?extid=54c2f58f15fe6876b6ad>:
usb 1-1: config 0 has no interface number 0
usb 1-1: New USB device found, idVendor=10cf, idProduct=8068, bcdDevice=e6.8d
usb 1-1: New USB device strings: Mfr=0, Product=0, SerialNumber=0
usb 1-1: config 0 descriptor??
vmk80xx 1-1:0.117: driver 'vmk80xx' failed to auto-configure device.
INFO: trying to register non-static key.
the code is fine but needs lockdep annotation.
turning off the locking correctness validator.
CPU: 0 PID: 12 Comm: kworker/0:1 Not tainted 5.1.0-rc4-319354-g9a33b36 #3
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Workqueue: usb_hub_wq hub_event
Call Trace:
__dump_stack lib/dump_stack.c:77 [inline]
dump_stack+0xe8/0x16e lib/dump_stack.c:113
assign_lock_key kernel/locking/lockdep.c:786 [inline]
register_lock_class+0x11b8/0x1250 kernel/locking/lockdep.c:1095
__lock_acquire+0xfb/0x37c0 kernel/locking/lockdep.c:3582
lock_acquire+0x10d/0x2f0 kernel/locking/lockdep.c:4211
__raw_spin_lock_irqsave include/linux/spinlock_api_smp.h:110 [inline]
_raw_spin_lock_irqsave+0x44/0x60 kernel/locking/spinlock.c:152
down+0x12/0x80 kernel/locking/semaphore.c:58
vmk80xx_detach+0x59/0x100 drivers/staging/comedi/drivers/vmk80xx.c:829
comedi_device_detach+0xed/0x800 drivers/staging/comedi/drivers.c:204
comedi_device_cleanup.part.0+0x68/0x140 drivers/staging/comedi/comedi_fops.c:156
comedi_device_cleanup drivers/staging/comedi/comedi_fops.c:187 [inline]
comedi_free_board_dev.part.0+0x16/0x90 drivers/staging/comedi/comedi_fops.c:190
comedi_free_board_dev drivers/staging/comedi/comedi_fops.c:189 [inline]
comedi_release_hardware_device+0x111/0x140 drivers/staging/comedi/comedi_fops.c:2880
comedi_auto_config.cold+0x124/0x1b0 drivers/staging/comedi/drivers.c:1068
usb_probe_interface+0x31d/0x820 drivers/usb/core/driver.c:361
really_probe+0x2da/0xb10 drivers/base/dd.c:509
driver_probe_device+0x21d/0x350 drivers/base/dd.c:671
__device_attach_driver+0x1d8/0x290 drivers/base/dd.c:778
bus_for_each_drv+0x163/0x1e0 drivers/base/bus.c:454
__device_attach+0x223/0x3a0 drivers/base/dd.c:844
bus_probe_device+0x1f1/0x2a0 drivers/base/bus.c:514
device_add+0xad2/0x16e0 drivers/base/core.c:2106
usb_set_configuration+0xdf7/0x1740 drivers/usb/core/message.c:2021
generic_probe+0xa2/0xda drivers/usb/core/generic.c:210
usb_probe_device+0xc0/0x150 drivers/usb/core/driver.c:266
really_probe+0x2da/0xb10 drivers/base/dd.c:509
driver_probe_device+0x21d/0x350 drivers/base/dd.c:671
__device_attach_driver+0x1d8/0x290 drivers/base/dd.c:778
bus_for_each_drv+0x163/0x1e0 drivers/base/bus.c:454
__device_attach+0x223/0x3a0 drivers/base/dd.c:844
bus_probe_device+0x1f1/0x2a0 drivers/base/bus.c:514
device_add+0xad2/0x16e0 drivers/base/core.c:2106
usb_new_device.cold+0x537/0xccf drivers/usb/core/hub.c:2534
hub_port_connect drivers/usb/core/hub.c:5089 [inline]
hub_port_connect_change drivers/usb/core/hub.c:5204 [inline]
port_event drivers/usb/core/hub.c:5350 [inline]
hub_event+0x138e/0x3b00 drivers/usb/core/hub.c:5432
process_one_work+0x90f/0x1580 kernel/workqueue.c:2269
worker_thread+0x9b/0xe20 kernel/workqueue.c:2415
kthread+0x313/0x420 kernel/kthread.c:253
ret_from_fork+0x3a/0x50 arch/x86/entry/entry_64.S:352
has an unintended side-effect of also suppressing all AUX records with no flags
and non-zero size, so all the regular records in the full trace mode.
This breaks some use cases for people.
c60f83b813e5 ("perf, pt, coresight: Fix address filters for vmas with non-zero offset")
changes the address filtering logic to communicate filter ranges to the PMU driver
via a single address range object, instead of having the driver do the final bit of
math.
That change forgets to take into account kernel filters, which are not calculated
the same way as DSO based filters.
Fix that by passing the kernel filters the same way as file-based filters.
This doesn't require any additional changes in the drivers.
Paolo Bonzini [Mon, 15 Apr 2019 13:16:17 +0000 (15:16 +0200)]
KVM: nVMX: allow tests to use bad virtual-APIC page address
As mentioned in the comment, there are some special cases where we can simply
clear the TPR shadow bit from the CPU-based execution controls in the vmcs02.
Handle them so that we can remove some XFAILs from vmx.flat.
I found out (the hard way) that under some .config options (notably L1_CACHE_SHIFT=7)
and compiler combinations this on-stack alignment leads to a 320 byte
stack usage, which then triggers a KASAN stack warning elsewhere.
Using 320 bytes of stack space for a 40 byte structure is ludicrous and
clearly not right.
Jian-Hong Pan [Fri, 12 Apr 2019 08:01:53 +0000 (16:01 +0800)]
x86/reboot, efi: Use EFI reboot for Acer TravelMate X514-51T
Upon reboot, the Acer TravelMate X514-51T laptop appears to complete the
shutdown process, but then it hangs in BIOS POST with a black screen.
The problem is intermittent - at some points it has appeared related to
Secure Boot settings or different kernel builds, but ultimately we have
not been able to identify the exact conditions that trigger the issue to
come and go.
Besides, the EFI mode cannot be disabled in the BIOS of this model.
However, after extensive testing, we observe that using the EFI reboot
method reliably avoids the issue in all cases.
So add a boot time quirk to use EFI reboot on such systems.
kprobes: Fix error check when reusing optimized probes
The following commit introduced a bug in one of our error paths:
819319fc9346 ("kprobes: Return error if we fail to reuse kprobe instead of BUG_ON()")
it missed to handle the return value of kprobe_optready() as
error-value. In reality, the kprobe_optready() returns a bool
result, so "true" case must be passed instead of 0.
This causes some errors on kprobe boot-time selftests on ARM:
[ ] Beginning kprobe tests...
[ ] Probe ARM code
[ ] kprobe
[ ] kretprobe
[ ] ARM instruction simulation
[ ] Check decoding tables
[ ] Run test cases
[ ] FAIL: test_case_handler not run
[ ] FAIL: Test andge r10, r11, r14, asr r7
[ ] FAIL: Scenario 11
...
[ ] FAIL: Scenario 7
[ ] Total instruction simulation tests=1631, pass=1433 fail=198
[ ] kprobe tests failed
This can happen if an optimized probe is unregistered and next
kprobe is registered on same address until the previous probe
is not reclaimed.
If this happens, a hidden aggregated probe may be kept in memory,
and no new kprobe can probe same address. Also, in that case
register_kprobe() will return "1" instead of minus error value,
which can mislead caller logic.
locking/lockdep: Make lockdep_unregister_key() honor 'debug_locks' again
If lockdep_register_key() and lockdep_unregister_key() are called with
debug_locks == false then the following warning is reported:
WARNING: CPU: 2 PID: 15145 at kernel/locking/lockdep.c:4920 lockdep_unregister_key+0x1ad/0x240
That warning is reported because lockdep_unregister_key() ignores the
value of 'debug_locks' and because the behavior of lockdep_register_key()
depends on whether or not 'debug_locks' is set. Fix this inconsistency
by making lockdep_unregister_key() take 'debug_locks' again into
account.
Sami Tolvanen [Mon, 15 Apr 2019 16:49:56 +0000 (09:49 -0700)]
x86/build/lto: Fix truncated .bss with -fdata-sections
With CONFIG_LD_DEAD_CODE_DATA_ELIMINATION=y, we compile the kernel with
-fdata-sections, which also splits the .bss section.
The new section, with a new .bss.* name, which pattern gets missed by the
main x86 linker script which only expects the '.bss' name. This results
in the discarding of the second part and a too small, truncated .bss
section and an unhappy, non-working kernel.
Use the common BSS_MAIN macro in the linker script to properly capture
and merge all the generated BSS sections.
Jaesoo Lee [Wed, 10 Apr 2019 00:02:22 +0000 (17:02 -0700)]
scsi: core: set result when the command cannot be dispatched
When SCSI blk-mq is enabled, there is a bug in handling errors in
scsi_queue_rq. Specifically, the bug is not setting result field of
scsi_request correctly when the dispatch of the command has been
failed. Since the upper layer code including the sg_io ioctl expects to
receive any error status from result field of scsi_request, the error is
silently ignored and this could cause data corruptions for some
applications.
Merge tag 'libnvdimm-fixes-5.1-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm
Pull libnvdimm fixes from Dan Williams:
"I debated holding this back for the v5.2 merge window due to the size
of the "zero-key" changes, but affected users would benefit from
having the fixes sooner. It did not make sense to change the zero-key
semantic in isolation for the "secure-erase" command, but instead
include it for all security commands.
The short background on the need for these changes is that some NVDIMM
platforms enable security with a default zero-key rather than let the
OS specify the initial key. This makes the security enabling that
landed in v5.0 unusable for some users.
Summary:
- Compatibility fix for nvdimm-security implementations with a
default zero-key.
- Miscellaneous small fixes for out-of-bound accesses, cleanup after
initialization failures, and missing debug messages"
* tag 'libnvdimm-fixes-5.1-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
tools/testing/nvdimm: Retain security state after overwrite
libnvdimm/pmem: fix a possible OOB access when read and write pmem
libnvdimm/security, acpi/nfit: unify zero-key for all security commands
libnvdimm/security: provide fix for secure-erase to use zero-key
libnvdimm/btt: Fix a kmemdup failure check
libnvdimm/namespace: Fix a potential NULL pointer dereference
acpi/nfit: Always dump _DSM output payload
Merge tag 'fsdax-fix-5.1-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm
Pull fsdax fix from Dan Williams:
"A single filesystem-dax fix. It has been lingering in -next for a long
while and there are no other fsdax fixes on the horizon:
- Avoid a crash scenario with architectures like powerpc that require
'pgtable_deposit' for the zero page"
* tag 'fsdax-fix-5.1-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
fs/dax: Deposit pagetable even when installing zero page
Jonathan Lemon [Sun, 14 Apr 2019 21:21:29 +0000 (14:21 -0700)]
route: Avoid crash from dereferencing NULL rt->from
When __ip6_rt_update_pmtu() is called, rt->from is RCU dereferenced, but is
never checked for null - rt6_flush_exceptions() may have removed the entry.
When a bond is enslaved to another bond, bond_netdev_event() only
handles the event as if the bond is a master, and skips treating the
bond as a slave.
This leads to a refcount leak on the slave, since we don't remove the
adjacency to its master and the master holds a reference on the slave.
Reproducer:
ip link add bondL type bond
ip link add bondU type bond
ip link set bondL master bondU
ip link del bondL
No "Fixes:" tag, this code is older than git history.
The reverted bugfix will cause another issue.
Reported by syzbot+6024817a931b2830bc93@syzkaller.appspotmail.com.
See https://syzkaller.appspot.com/x/log.txt?x=1737671b200000 for
details.
io_uring: fix possible deadlock between io_uring_{enter,register}
If we have multiple threads, one doing io_uring_enter() while the other
is doing io_uring_register(), we can run into a deadlock between the
two. io_uring_register() must wait for existing users of the io_uring
instance to exit. But it does so while holding the io_uring mutex.
Callers of io_uring_enter() may need this mutex to make progress (and
eventually exit). If we wait for users to exit in io_uring_register(),
we can't do so with the io_uring mutex held without potentially risking
a deadlock.
Drop the io_uring mutex while waiting for existing callers to exit. This
is safe and guaranteed to make forward progress, since we already killed
the percpu ref before doing so. Hence later callers of io_uring_enter()
will be rejected.
KVM: x86/mmu: Fix an inverted list_empty() check when zapping sptes
A recently introduced helper for handling zap vs. remote flush
incorrectly bails early, effectively leaking defunct shadow pages.
Manifests as a slab BUG when exiting KVM due to the shadow pages
being alive when their associated cache is destroyed.
==========================================================================
BUG kvm_mmu_page_header: Objects remaining in kvm_mmu_page_header on ...
--------------------------------------------------------------------------
Disabling lock debugging due to kernel taint
INFO: Slab 0x00000000fc436387 objects=26 used=23 fp=0x00000000d023caee ...
CPU: 6 PID: 4315 Comm: rmmod Tainted: G B 5.1.0-rc2+ #19
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
Call Trace:
dump_stack+0x46/0x5b
slab_err+0xad/0xd0
? on_each_cpu_mask+0x3c/0x50
? ksm_migrate_page+0x60/0x60
? on_each_cpu_cond_mask+0x7c/0xa0
? __kmalloc+0x1ca/0x1e0
__kmem_cache_shutdown+0x13a/0x310
shutdown_cache+0xf/0x130
kmem_cache_destroy+0x1d5/0x200
kvm_mmu_module_exit+0xa/0x30 [kvm]
kvm_arch_exit+0x45/0x60 [kvm]
kvm_exit+0x6f/0x80 [kvm]
vmx_exit+0x1a/0x50 [kvm_intel]
__x64_sys_delete_module+0x153/0x1f0
? exit_to_usermode_loop+0x88/0xc0
do_syscall_64+0x4f/0x100
entry_SYSCALL_64_after_hwframe+0x44/0xa9
Fixes: a21136345cb6f ("KVM: x86/mmu: Split remote_flush+zap case out of kvm_mmu_flush_or_zap()") Signed-off-by: Sean Christopherson <[email protected]> Signed-off-by: Paolo Bonzini <[email protected]>
which is correct because GPL is not a valid identifier. Of course this
could have been caught by checkpatch.pl _before_ submitting or merging the
patch.
WARNING: 'SPDX-License-Identifier: GPL' is not supported in LICENSES/...
#19: FILE: drivers/power/supply/goldfish_battery.c:1:
+// SPDX-License-Identifier: GPL
Which is absolutely hillarious as the commit introducing this wreckage says
in the changelog:
There was a checkpatch complain:
"Missing or malformed SPDX-License-Identifier tag".
Oh well. Replacing a checkpatch warning by a different checkpatch warning
is a really useful exercise.
Use the proper GPL-2.0 identifier which is what the boiler plate in the
file had originally.
Fixes: e75e3a125b40 ("drivers: power: supply: goldfish_battery: Put an SPDX tag") Signed-off-by: Thomas Gleixner <[email protected]> Signed-off-by: Greg Kroah-Hartman <[email protected]>
s390/pkey: add one more argument space for debug feature entry
The debug feature entries have been used with up to 5 arguents
(including the pointer to the format string) but there was only
space reserved for 4 arguemnts. So now the registration does
reserve space for 5 times a long value.
This fixes a sometime appearing weired value as the last
value of an debug feature entry like this:
... pkey_sec2protkey zcrypt_send_cprb (cardnr=10 domain=12)
failed with errno -2143346254
David Francis [Fri, 29 Mar 2019 17:23:15 +0000 (13:23 -0400)]
drm/amd/display: If one stream full updates, full update all planes
[Why]
On some compositors, with two monitors attached, VT terminal
switch can cause a graphical issue by the following means:
There are two streams, one for each monitor. Each stream has one
plane
current state:
M1:S1->P1
M2:S2->P2
The user calls for a terminal switch and a commit is made to
change both planes to linear swizzle mode. In atomic check,
a new dc_state is constructed with new planes on each stream
new state:
M1:S1->P3
M2:S2->P4
In commit tail, each stream is committed, one at a time. The first
stream (S1) updates properly, triggerring a full update and replacing
the state
current state:
M1:S1->P3
M2:S2->P4
The update for S2 comes in, but dc detects that there is no difference
between the stream and plane in the new and current states, and so
triggers a fast update. The fast update does not program swizzle,
so the second monitor is corrupted
[How]
Add a flag to dc_plane_state that forces full updates
When a stream undergoes a full update, set this flag on all changed
planes, then clear it on the current stream
Subsequent streams will get full updates as a result
Jann Horn reported that he can overflow the page ref count with
sufficient memory (and a filesystem that is intentionally extremely
slow).
Admittedly it's not exactly easy. To have more than four billion
references to a page requires a minimum of 32GB of kernel memory just
for the pointers to the pages, much less any metadata to keep track of
those pointers. Jann needed a total of 140GB of memory and a specially
crafted filesystem that leaves all reads pending (in order to not ever
free the page references and just keep adding more).
Still, we have a fairly straightforward way to limit the two obvious
user-controllable sources of page references: direct-IO like page
references gotten through get_user_pages(), and the splice pipe page
duplication. So let's just do that.
* branch page-refs:
fs: prevent page refcount overflow in pipe_buf_get
mm: prevent get_user_pages() from overflowing page refcount
mm: add 'try_get_page()' helper function
mm: make page ref count overflow check tighter and more explicit
I've cc'ed some of the checksum fixes to Eric Dumazet and i would like to get
his feedback before you pull.
For -stable v4.19
('net/mlx5: FPGA, tls, idr remove on flow delete')
('net/mlx5: FPGA, tls, hold rcu read lock a bit longer')
For -stable v4.20
('net/mlx5e: Rx, Check ip headers sanity')
('Revert "net/mlx5e: Enable reporting checksum unnecessary also for L3 packets"')
('net/mlx5e: Rx, Fixup skb checksum for packets with tail padding')
For -stable v5.0
('net/mlx5e: Switch to Toeplitz RSS hash by default')
('net/mlx5e: Protect against non-uplink representor for encap')
('net/mlx5e: XDP, Avoid checksum complete when XDP prog is loaded')
====================
The connection between sighand->siglock and st->lock comes through seccomp,
which takes st->lock while holding sighand->siglock.
Make sure interrupts are disabled when __speculation_ctrl_update() is
invoked via prctl() -> speculation_ctrl_update(). Add a lockdep assert to
catch future offenders.
David S. Miller [Sun, 14 Apr 2019 20:59:49 +0000 (13:59 -0700)]
Merge branch 'qed-doorbell-overflow-recovery'
Denis Bolotin says:
====================
qed: Fix the Doorbell Overflow Recovery mechanism
This patch series fixes and improves the doorbell recovery mechanism.
The main goals of this series are to fix missing attentions from the
doorbells block (DORQ) or not handling them properly, and execute the
recovery from periodic handler instead of the attention handler.
Please consider applying the series to net.
====================
Separate the overflow handling from the hardware interrupt status analysis.
The interrupt status is a single register and is common for all PFs. The
first PF reading the register is not necessarily the one who overflowed.
All PFs must check their overflow status on every attention.
In this change we clear the sticky indication in the attention handler to
allow doorbells to be processed again as soon as possible, but running
the doorbell recovery is scheduled for the periodic handler to reduce the
time spent in the attention handler.
Checking the need for DORQ flush was changed to "db_bar_no_edpm" because
qed_edpm_enabled()'s result could change dynamically and might have
prevented a needed flush.
When the DORQ (doorbell block) is overflowed, all PFs get attentions at the
same time. If one PF finished handling the attention before another PF even
started, the second PF might miss the DORQ's attention bit and not handle
the attention at all.
If the DORQ attention is missed and the issue is not resolved, another
attention will not be sent, therefore each attention is treated as a
potential DORQ attention.
As a result, the attention callback is called more frequently so the debug
print was moved to reduce its quantity.
The number of periodic doorbell recovery handler schedules was reduced
because it was the previous way to mitigating the missed attention issue.
DB_REC_DRY_RUN (running doorbell recovery without sending doorbells) is
never used. DB_REC_ONCE (send a single doorbell from the doorbell recovery)
is not needed anymore because by running the periodic handler we make sure
we check the overflow status later instead.
This patch is needed because in the next patches, the only doorbell
recovery type being used is DB_REC_REAL_DEAL, and the fixes are much
cleaner without this enum.
Matthew Wilcox [Fri, 5 Apr 2019 21:02:10 +0000 (14:02 -0700)]
fs: prevent page refcount overflow in pipe_buf_get
Change pipe_buf_get() to return a bool indicating whether it succeeded
in raising the refcount of the page (if the thing in the pipe is a page).
This removes another mechanism for overflowing the page refcount. All
callers converted to handle a failure.
mm: prevent get_user_pages() from overflowing page refcount
If the page refcount wraps around past zero, it will be freed while
there are still four billion references to it. One of the possible
avenues for an attacker to try to make this happen is by doing direct IO
on a page multiple times. This patch makes get_user_pages() refuse to
take a new page reference if there are already more than two billion
references to the page.
This is the same as the traditional 'get_page()' function, but instead
of unconditionally incrementing the reference count of the page, it only
does so if the count was "safe". It returns whether the reference count
was incremented (and is marked __must_check, since the caller obviously
has to be aware of it).
Also like 'get_page()', you can't use this function unless you already
had a reference to the page. The intent is that you can use this
exactly like get_page(), but in situations where you want to limit the
maximum reference count.
The code currently does an unconditional WARN_ON_ONCE() if we ever hit
the reference count issues (either zero or negative), as a notification
that the conditional non-increment actually happened.
NOTE! The count access for the "safety" check is inherently racy, but
that doesn't matter since the buffer we use is basically half the range
of the reference count (ie we look at the sign of the count).
mm: make page ref count overflow check tighter and more explicit
We have a VM_BUG_ON() to check that the page reference count doesn't
underflow (or get close to overflow) by checking the sign of the count.
That's all fine, but we actually want to allow people to use a "get page
ref unless it's already very high" helper function, and we want that one
to use the sign of the page ref (without triggering this VM_BUG_ON).
Change the VM_BUG_ON to only check for small underflows (or _very_ close
to overflowing), and ignore overflows which have strayed into negative
territory.
x86/resctrl: Do not repeat rdtgroup mode initialization
When cache allocation is supported and the user creates a new resctrl
resource group, the allocations of the new resource group are
initialized to all regions that it can possibly use. At this time these
regions are all that are shareable by other resource groups as well as
regions that are not currently used. The new resource group's mode is
also initialized to reflect this initialization and set to "shareable".
The new resource group's mode is currently repeatedly initialized within
the loop that configures the hardware with the resource group's default
allocations.
Move the initialization of the resource group's mode outside the
hardware configuration loop. The resource group's mode is now
initialized only once as the final step to reflect that its configured
allocations are "shareable".
Since the fget/fput handling was reworked in commit 09bb839434bd, we
never call io_file_put() with state == NULL (and hence file != NULL)
anymore. Remove that case.
Jens Axboe [Fri, 18 Jan 2019 17:34:16 +0000 (10:34 -0700)]
bfq: update internal depth state when queue depth changes
A previous commit moved the shallow depth and BFQ depth map calculations
to be done at init time, moving it outside of the hotter IO path. This
potentially causes hangs if the users changes the depth of the scheduler
map, by writing to the 'nr_requests' sysfs file for that device.
Add a blk-mq-sched hook that allows blk-mq to inform the scheduler if
the depth changes, so that the scheduler can update its internal state.
io_uring: only test SQPOLL cpu after we've verified it
We currently call cpu_possible() even if we don't use the CPU. Move the
test under the SQ_AFF branch, which is the only place where we'll use
the value. Do the cpu_possible() test AFTER we've limited it to a max
of NR_CPUS. This avoids triggering the following warning:
WARNING: CPU: 1 PID: 7600 at include/linux/cpumask.h:121 cpu_max_bits_warn
if CONFIG_DEBUG_PER_CPU_MAPS is enabled.
While in there, also move the SQ thread idle period assignment inside
SETUP_SQPOLL, as we don't use it otherwise either.
Merge tag 'for-linus-20190412' of git://git.kernel.dk/linux-block
Pull block fixes from Jens Axboe:
"Set of fixes that should go into this round. This pull is larger than
I'd like at this time, but there's really no specific reason for that.
Some are fixes for issues that went into this merge window, others are
not. Anyway, this contains:
- Hardware queue limiting for virtio-blk/scsi (Dongli)
- Multi-page bvec fixes for lightnvm pblk
- Multi-bio dio error fix (Jason)
- Remove the cache hint from the io_uring tool side, since we didn't
move forward with that (me)
- Make io_uring SETUP_SQPOLL root restricted (me)
- Fix leak of page in error handling for pc requests (Jérôme)
- Fix BFQ regression introduced in this merge window (Paolo)
- Fix break logic for bio segment iteration (Ming)
- Fix NVMe cancel request error handling (Ming)
- NVMe pull request with two fixes (Christoph):
- fix the initial CSN for nvme-fc (James)
- handle log page offsets properly in the target (Keith)"
* tag 'for-linus-20190412' of git://git.kernel.dk/linux-block:
block: fix the return errno for direct IO
nvmet: fix discover log page when offsets are used
nvme-fc: correct csn initialization and increments on error
block: do not leak memory in bio_copy_user_iov()
lightnvm: pblk: fix crash in pblk_end_partial_read due to multipage bvecs
nvme: cancel request synchronously
blk-mq: introduce blk_mq_complete_request_sync()
scsi: virtio_scsi: limit number of hw queues by nr_cpu_ids
virtio-blk: limit number of hw queues by nr_cpu_ids
block, bfq: fix use after free in bfq_bfqq_expire
io_uring: restrict IORING_SETUP_SQPOLL to root
tools/io_uring: remove IOCQE_FLAG_CACHEHIT
block: don't use for-inside-for in bio_for_each_segment_all
Merge tag 'nfs-for-5.1-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Pull NFS client bugfixes from Trond Myklebust:
"Highlights include:
Stable fix:
- Fix a deadlock in close() due to incorrect draining of RDMA queues
Bugfixes:
- Revert "SUNRPC: Micro-optimise when the task is known not to be
sleeping" as it is causing stack overflows
- Fix a regression where NFSv4 getacl and fs_locations stopped
working
- Forbid setting AF_INET6 to "struct sockaddr_in"->sin_family.
- Fix xfstests failures due to incorrect copy_file_range() return
values"
* tag 'nfs-for-5.1-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
Revert "SUNRPC: Micro-optimise when the task is known not to be sleeping"
NFSv4.1 fix incorrect return value in copy_file_range
xprtrdma: Fix helper that drains the transport
NFS: Fix handling of reply page vector
NFS: Forbid setting AF_INET6 to "struct sockaddr_in"->sin_family.
Merge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux
Pull clk fixes from Stephen Boyd:
"Here's more than a handful of clk driver fixes for changes that came
in during the merge window:
- Fix the AT91 sama5d2 programmable clk prescaler formula
- A bunch of Amlogic meson clk driver fixes for the VPU clks
- A DMI quirk for Intel's Bay Trail SoC's driver to properly mark pmc
clks as critical only when really needed
- Stop overwriting CLK_SET_RATE_PARENT flag in mediatek's clk gate
implementation
- Use the right structure to test for a frequency table in i.MX's
PLL_1416x driver"
* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
clk: imx: Fix PLL_1416X not rounding rates
clk: mediatek: fix clk-gate flag setting
platform/x86: pmc_atom: Drop __initconst on dmi table
clk: x86: Add system specific quirk to mark clocks as critical
clk: meson: vid-pll-div: remove warning and return 0 on invalid config
clk: meson: pll: fix rounding and setting a rate that matches precisely
clk: meson-g12a: fix VPU clock parents
clk: meson: g12a: fix VPU clock muxes mask
clk: meson-gxbb: round the vdec dividers to closest
clk: at91: fix programmable clock for sama5d2
Merge tag 'pci-v5.1-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Pull PCI fixes from Bjorn Helgaas:
- Add a DMA alias quirk for another Marvell SATA device (Andre
Przywara)
- Fix a pciehp regression that broke safe removal of devices (Sergey
Miroshnichenko)
* tag 'pci-v5.1-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
PCI: pciehp: Ignore Link State Changes after powering off a slot
PCI: Add function 1 DMA alias quirk for Marvell 9170 SATA controller
Merge tag 'powerpc-5.1-5' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux
Pull powerpc fixes from Michael Ellerman:
"A minor build fix for 64-bit FLATMEM configs.
A fix for a boot failure on 32-bit powermacs.
My commit to fix CLOCK_MONOTONIC across Y2038 broke the 32-bit VDSO on
64-bit kernels, ie. compat mode, which is only used on big endian.
The rewrite of the SLB code we merged in 4.20 missed the fact that the
0x380 exception is also used with the Radix MMU to report out of range
accesses. This could lead to an oops if userspace tried to read from
addresses outside the user or kernel range.
Thanks to: Aneesh Kumar K.V, Christophe Leroy, Larry Finger, Nicholas
Piggin"
* tag 'powerpc-5.1-5' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
powerpc/mm: Define MAX_PHYSMEM_BITS for all 64-bit configs
powerpc/64s/radix: Fix radix segment exception handling
powerpc/vdso32: fix CLOCK_MONOTONIC on PPC64
powerpc/32: Fix early boot failure with RTAS built-in
Merge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Pull arm64 fixes from Will Deacon:
"The main thing is a fix to our FUTEX_WAKE_OP implementation which was
unbelievably broken, but did actually work for the one scenario that
GLIBC used to use.
- Fix terminally broken implementation of FUTEX_WAKE_OP atomics"
* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
arm64: futex: Fix FUTEX_WAKE_OP atomic ops with non-zero result value
arm64: backtrace: Don't bother trying to unwind the userspace stack
arm64/ftrace: fix inadvertent BUG() in trampoline check
The recent commit 98081ca62cba ("ALSA: hda - Record the current power
state before suspend/resume calls") made the HD-audio driver to store
the PM state in power_state field. This forgot, however, the
initialization at power up. Although the codec drivers usually don't
need to refer to this field in the normal operation, let's initialize
it properly for consistency.
Fixes: 98081ca62cba ("ALSA: hda - Record the current power state before suspend/resume calls") Signed-off-by: Takashi Iwai <[email protected]>
David Howells [Sat, 13 Apr 2019 07:37:37 +0000 (08:37 +0100)]
afs: Fix in-progess ops to ignore server-level callback invalidation
The in-kernel afs filesystem client counts the number of server-level
callback invalidation events (CB.InitCallBackState* RPC operations) that it
receives from the server. This is stored in cb_s_break in various
structures, including afs_server and afs_vnode.
If an inode is examined by afs_validate(), say, the afs_server copy is
compared, along with other break counters, to those in afs_vnode, and if
one or more of the counters do not match, it is considered that the
server's callback promise is broken. At points where this happens,
AFS_VNODE_CB_PROMISED is cleared to indicate that the status must be
refetched from the server.
afs_validate() issues an FS.FetchStatus operation to get updated metadata -
and based on the updated data_version may invalidate the pagecache too.
However, the break counters are also used to determine whether to note a
new callback in the vnode (which would set the AFS_VNODE_CB_PROMISED flag)
and whether to cache the permit data included in the YFSFetchStatus record
by the server.
The problem comes when the server sends us a CB.InitCallBackState op. The
first such instance doesn't cause cb_s_break to be incremented, but rather
causes AFS_SERVER_FL_NEW to be cleared - but thereafter, say some hours
after last use and all the volumes have been automatically unmounted and
the server has forgotten about the client[*], this *will* likely cause an
increment.
[*] There are other circumstances too, such as the server restarting or
needing to make space in its callback table.
Note that the server won't send us a CB.InitCallBackState op until we talk
to it again.
So what happens is:
(1) A mount for a new volume is attempted, a inode is created for the root
vnode and vnode->cb_s_break and AFS_VNODE_CB_PROMISED aren't set
immediately, as we don't have a nominated server to talk to yet - and
we may iterate through a few to find one.
(2) Before the operation happens, afs_fetch_status(), say, notes in the
cursor (fc.cb_break) the break counter sum from the vnode, volume and
server counters, but the server->cb_s_break is currently 0.
(3) We send FS.FetchStatus to the server. The server sends us back
CB.InitCallBackState. We increment server->cb_s_break.
(4) Our FS.FetchStatus completes. The reply includes a callback record.
(5) xdr_decode_AFSCallBack()/xdr_decode_YFSCallBack() check to see whether
the callback promise was broken by checking the break counter sum from
step (2) against the current sum.
This fails because of step (3), so we don't set the callback record
and, importantly, don't set AFS_VNODE_CB_PROMISED on the vnode.
This does not preclude the syscall from progressing, and we don't loop here
rechecking the status, but rather assume it's good enough for one round
only and will need to be rechecked next time.
(6) afs_validate() it triggered on the vnode, probably called from
d_revalidate() checking the parent directory.
(7) afs_validate() notes that AFS_VNODE_CB_PROMISED isn't set, so doesn't
update vnode->cb_s_break and assumes the vnode to be invalid.
(8) afs_validate() needs to calls afs_fetch_status(). Go back to step (2)
and repeat, every time the vnode is validated.
This primarily affects volume root dir vnodes. Everything subsequent to
those inherit an already incremented cb_s_break upon mounting.
The issue is that we assume that the callback record and the cached permit
information in a reply from the server can't be trusted after getting a
server break - but this is wrong since the server makes sure things are
done in the right order, holding up our ops if necessary[*].
[*] There is an extremely unlikely scenario where a reply from before the
CB.InitCallBackState could get its delivery deferred till after - at
which point we think we have a promise when we don't. This, however,
requires unlucky mass packet loss to one call.
AFS_SERVER_FL_NEW tries to paper over the cracks for the initial mount from
a server we've never contacted before, but this should be unnecessary.
It's also further insulated from the problem on an initial mount by
querying the server first with FS.GetCapabilities, which triggers the
CB.InitCallBackState.
Fix this by
(1) Remove AFS_SERVER_FL_NEW.
(2) In afs_calc_vnode_cb_break(), don't include cb_s_break in the
calculation.
(3) In afs_cb_is_broken(), don't include cb_s_break in the check.
Marc Dionne [Sat, 13 Apr 2019 07:37:37 +0000 (08:37 +0100)]
afs: Unlock pages for __pagevec_release()
__pagevec_release() complains loudly if any page in the vector is still
locked. The pages need to be locked for generic_error_remove_page(), but
that function doesn't actually unlock them.
David Howells [Sat, 13 Apr 2019 07:37:37 +0000 (08:37 +0100)]
afs: Differentiate abort due to unmarshalling from other errors
Differentiate an abort due to an unmarshalling error from an abort due to
other errors, such as ENETUNREACH. It doesn't make sense to set abort code
RXGEN_*_UNMARSHAL in such a case, so use RX_USER_ABORT instead.
get_seconds() has a limited range on 32-bit architectures and is
deprecated because of that. While AFS uses the same limits for
its inode timestamps on the wire protocol, let's just use the
simpler current_time() as we do for other file systems.
This will still zero out the 'tv_nsec' field of the timestamps
internally.
Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Ingo Molnar:
"Fix typos in user-visible resctrl parameters, and also fix assembly
constraint bugs that might result in miscompilation"
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/asm: Use stricter assembly constraints in bitops
x86/resctrl: Fix typos in the mba_sc mount option
Merge branch 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull scheduler fix from Ingo Molnar:
"Fix a NULL pointer dereference crash in certain environments"
* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
sched/fair: Do not re-read ->h_load_next during hierarchical load calculation
Merge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf fixes from Ingo Molnar:
"Six kernel side fixes: three related to NMI handling on AMD systems, a
race fix, a kexec initialization fix and a PEBS sampling fix"
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf/core: Fix perf_event_disable_inatomic() race
x86/perf/amd: Remove need to check "running" bit in NMI handler
x86/perf/amd: Resolve NMI latency issues for active PMCs
x86/perf/amd: Resolve race condition when disabling PMC
perf/x86/intel: Initialize TFA MSR
perf/x86/intel: Fix handling of wakeup_events for multi-entry PEBS
Merge branch 'locking-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull locking fix from Ingo Molnar:
"Fixes a crash when accessing /proc/lockdep"
* 'locking-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
locking/lockdep: Zap lock classes even with lock debugging disabled
projects / linux.git / log