When we get an interrupt for a channel program, it is not
necessarily the final interrupt; for example, the issuing
guest may request an intermediate interrupt by specifying
the program-controlled-interrupt flag on a ccw.
We must not switch the state to idle if the interrupt is not
yet final; even more importantly, we must not free the translated
channel program if the interrupt is not yet final, or the host
can crash during cp rewind.
The CPUID flag ARCH_CAPABILITIES is unconditioinally exposed to host
userspace for all x86 hosts, i.e. KVM advertises ARCH_CAPABILITIES
regardless of hardware support under the pretense that KVM fully
emulates MSR_IA32_ARCH_CAPABILITIES. Unfortunately, only VMX hosts
handle accesses to MSR_IA32_ARCH_CAPABILITIES (despite KVM_GET_MSRS
also reporting MSR_IA32_ARCH_CAPABILITIES for all hosts).
Move the MSR_IA32_ARCH_CAPABILITIES handling to common x86 code so
that it's emulated on AMD hosts.
Fixes: 1eaafe91a0df4 ("kvm: x86: IA32_ARCH_CAPABILITIES is always supported") Cc: stable@vger.kernel.org Reported-by: Xiaoyao Li <xiaoyao.li@linux.intel.com> Cc: Jim Mattson <jmattson@google.com> Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
KVM's API requires thats ioctls must be issued from the same process
that created the VM. In other words, userspace can play games with a
VM's file descriptors, e.g. fork(), SCM_RIGHTS, etc..., but only the
creator can do anything useful. Explicitly reject device ioctls that
are issued by a process other than the VM's creator, and update KVM's
API documentation to extend its requirements to device ioctls.
Fixes: 852b6d57dc7f ("kvm: add device control API") Cc: <stable@vger.kernel.org> Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The SMT disable 'nosmt' command line argument is not working properly when
CONFIG_HOTPLUG_CPU is disabled. The teardown of the sibling CPUs which are
required to be brought up due to the MCE issues, cannot work. The CPUs are
then kept in a half dead state.
As the 'nosmt' functionality has become popular due to the speculative
hardware vulnerabilities, the half torn down state is not a proper solution
to the problem.
Enforce CONFIG_HOTPLUG_CPU=y when SMP is enabled so the full operation is
possible.
Reported-by: Tianyu Lan <Tianyu.Lan@microsoft.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Konrad Wilk <konrad.wilk@oracle.com> Cc: Josh Poimboeuf <jpoimboe@redhat.com> Cc: Mukesh Ojha <mojha@codeaurora.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Rik van Riel <riel@surriel.com> Cc: Andy Lutomirski <luto@kernel.org> Cc: Micheal Kelley <michael.h.kelley@microsoft.com> Cc: "K. Y. Srinivasan" <kys@microsoft.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Borislav Petkov <bp@alien8.de> Cc: K. Y. Srinivasan <kys@microsoft.com> Cc: stable@vger.kernel.org Link: https://lkml.kernel.org/r/20190326163811.598166056@linutronix.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Tianyu reported a crash in a CPU hotplug teardown callback when booting a
kernel which has CONFIG_HOTPLUG_CPU disabled with the 'nosmt' boot
parameter.
It turns out that the SMP=y CONFIG_HOTPLUG_CPU=n case has been broken
forever in case that a bringup callback fails. Unfortunately this issue was
not recognized when the CPU hotplug code was reworked, so the shortcoming
just stayed in place.
When a bringup callback fails, the CPU hotplug code rolls back the
operation and takes the CPU offline.
The 'nosmt' command line argument uses a bringup failure to abort the
bringup of SMT sibling CPUs. This partial bringup is required due to the
MCE misdesign on Intel CPUs.
With CONFIG_HOTPLUG_CPU=y the rollback works perfectly fine, but
CONFIG_HOTPLUG_CPU=n lacks essential mechanisms to exercise the low level
teardown of a CPU including the synchronizations in various facilities like
RCU, NOHZ and others.
As a consequence the teardown callbacks which must be executed on the
outgoing CPU within stop machine with interrupts disabled are executed on
the control CPU in interrupt enabled and preemptible context causing the
kernel to crash and burn. The pre state machine code has a different
failure mode which is more subtle and resulting in a less obvious use after
free crash because the control side frees resources which are still in use
by the undead CPU.
But this is not a x86 only problem. Any architecture which supports the
SMP=y HOTPLUG_CPU=n combination suffers from the same issue. It's just less
likely to be triggered because in 99.99999% of the cases all bringup
callbacks succeed.
The easy solution of making HOTPLUG_CPU mandatory for SMP is not working on
all architectures as the following architectures have either no hotplug
support at all or not all subarchitectures support it:
Crashing the kernel in such a situation is not an acceptable state
either.
Implement a minimal rollback variant by limiting the teardown to the point
where all regular teardown callbacks have been invoked and leave the CPU in
the 'dead' idle state. This has the following consequences:
- the CPU is brought down to the point where the stop_machine takedown
would happen.
- the CPU stays there forever and is idle
- The CPU is cleared in the CPU active mask, but not in the CPU online
mask which is a legit state.
- Interrupts are not forced away from the CPU
- All facilities which only look at online mask would still see it, but
that is the case during normal hotplug/unplug operations as well. It's
just a (way) longer time frame.
This will expose issues, which haven't been exposed before or only seldom,
because now the normally transient state of being non active but online is
a permanent state. In testing this exposed already an issue vs. work queues
where the vmstat code schedules work on the almost dead CPU which ends up
in an unbound workqueue and triggers 'preemtible context' warnings. This is
not a problem of this change, it merily exposes an already existing issue.
Still this is better than crashing fully without a chance to debug it.
This is mainly thought as workaround for those architectures which do not
support HOTPLUG_CPU. All others should enforce HOTPLUG_CPU for SMP.
Fixes: 2e1a3483ce74 ("cpu/hotplug: Split out the state walk into functions") Reported-by: Tianyu Lan <Tianyu.Lan@microsoft.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Tested-by: Tianyu Lan <Tianyu.Lan@microsoft.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Konrad Wilk <konrad.wilk@oracle.com> Cc: Josh Poimboeuf <jpoimboe@redhat.com> Cc: Mukesh Ojha <mojha@codeaurora.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Rik van Riel <riel@surriel.com> Cc: Andy Lutomirski <luto@kernel.org> Cc: Micheal Kelley <michael.h.kelley@microsoft.com> Cc: "K. Y. Srinivasan" <kys@microsoft.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Borislav Petkov <bp@alien8.de> Cc: K. Y. Srinivasan <kys@microsoft.com> Cc: stable@vger.kernel.org Link: https://lkml.kernel.org/r/20190326163811.503390616@linutronix.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A TSC packet can slip past MTC packets so that the timestamp appears to
go backwards. One estimate is that can be up to about 40 CPU cycles,
which is certainly less than 0x1000 TSC ticks, but accept slippage an
order of magnitude more to be on the safe side.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: stable@vger.kernel.org Fixes: 79b58424b821c ("perf tools: Add Intel PT support for decoding MTC packets") Link: http://lkml.kernel.org/r/20190325135135.18348-1-adrian.hunter@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Our MIPS 1004Kc SoCs were seeing random userspace crashes with SIGILL
and SIGSEGV that could not be traced back to a userspace code bug. They
had all the magic signs of an I/D cache coherency issue.
Now recently we noticed that the /proc/sys/vm/compact_memory interface
was quite efficient at provoking this class of userspace crashes.
Studying the code in mm/migrate.c there is a distinction made between
migrating a page that is mapped at the instant of migration and one that
is not mapped. Our problem turned out to be the non-mapped pages.
For the non-mapped page the code performs a copy of the page content and
all relevant meta-data of the page without doing the required D-cache
maintenance. This leaves dirty data in the D-cache of the CPU and on
the 1004K cores this data is not visible to the I-cache. A subsequent
page-fault that triggers a mapping of the page will happily serve the
process with potentially stale code.
What about ARM then, this bug should have seen greater exposure? Well
ARM became immune to this flaw back in 2010, see commit c01778001a4f
("ARM: 6379/1: Assume new page cache pages have dirty D-cache").
My proposed fix moves the D-cache maintenance inside move_to_new_page to
make it common for both cases.
Link: http://lkml.kernel.org/r/20190315083502.11849-1-larper@axis.com Fixes: 97ee0524614 ("flush cache before installing new page at migraton") Signed-off-by: Lars Persson <larper@axis.com> Reviewed-by: Paul Burton <paul.burton@mips.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the kernel is compiled with preemption enabled, the URB completion
handler can run in parallel with the work responsible for waking up the
tty layer. If the URB handler sets the EVENT_TTY_WAKEUP bit during the
call to tty_port_tty_wakeup() to signal that there is room for additional
input, it will be cleared at the end of this call. As a result, TX traffic
on the upper layer will be blocked.
This can be seen with a kernel configured with CONFIG_PREEMPT, and a fast
modem connected with PPP running over a USB CDC-ACM port.
Use test_and_clear_bit() instead, which ensures that each wakeup requested
by the URB completion code will trigger a call to tty_port_tty_wakeup().
A suspended SS port in U3 link state will go to U0 when resumed, but
can almost immediately after that enter U1 or U2 link power save
states before host controller driver reads the port status.
Host controller driver only checks for U0 state, and might miss
the finished resume, leaving flags unclear and skip notifying usb
code of the wake.
Add U1 and U2 to the possible link states when checking for finished
port resume.
When plugging BUFFALO LUA4-U3-AGT USB3.0 to Gigabit Ethernet LAN
Adapter, warning messages filled up dmesg.
[ 101.098287] xhci-hcd ee000000.usb: WARN Successful completion on short TX for slot 1 ep 4: needs XHCI_TRUST_TX_LENGTH quirk?
[ 101.117463] xhci-hcd ee000000.usb: WARN Successful completion on short TX for slot 1 ep 4: needs XHCI_TRUST_TX_LENGTH quirk?
[ 101.136513] xhci-hcd ee000000.usb: WARN Successful completion on short TX for slot 1 ep 4: needs XHCI_TRUST_TX_LENGTH quirk?
Adding the XHCI_TRUST_TX_LENGTH quirk resolves the issue.
There are cases where multiple device tree nodes point to the
same phy node by means of the "phys" property, but we should
only consider those nodes that are marked as available rather
than just any node.
Fixes: 98bfb3946695 ("usb: of: add an api to get dr_mode by the phy node") Cc: stable@vger.kernel.org # v4.4+ Signed-off-by: Fabrizio Castro <fabrizio.castro@bp.renesas.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In f_hidg_write() the write_spinlock is acquired before calling
usb_ep_queue() which causes a deadlock when dummy_hcd is being used.
This is because dummy_queue() callbacks into f_hidg_req_complete() which
tries to acquire the same spinlock. This is (part of) the backtrace when
the deadlock occurs:
0xffffffffc06b1410 in f_hidg_req_complete
0xffffffffc06a590a in usb_gadget_giveback_request
0xffffffffc06cfff2 in dummy_queue
0xffffffffc06a4b96 in usb_ep_queue
0xffffffffc06b1eb6 in f_hidg_write
0xffffffff8127730b in __vfs_write
0xffffffff812774d1 in vfs_write
0xffffffff81277725 in SYSC_write
Fix this by releasing the write_spinlock before calling usb_ep_queue()
Reviewed-by: James Bottomley <James.Bottomley@HansenPartnership.com> Tested-by: James Bottomley <James.Bottomley@HansenPartnership.com> Cc: stable@vger.kernel.org # 4.11+ Fixes: 749494b6bdbb ("usb: gadget: f_hid: fix: Move IN request allocation to set_alt()") Signed-off-by: Radoslav Gerganov <rgerganov@vmware.com> Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
While only the first PHY supports mode switching, the remaining PHYs
work in USB host mode. They should support set_mode with mode=USB_HOST
instead of failing. This is especially needed now that the USB core does
set_mode for all USB ports, which was added in commit b97a31348379 ("usb:
core: comply to PHY framework").
Make set_mode with mode=USB_HOST a no-op instead of failing for the
non-OTG USB PHYs.
Fixes: 6ba43c291961 ("phy-sun4i-usb: Add support for phy_set_mode") Signed-off-by: Chen-Yu Tsai <wens@csie.org> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Current code test wrong value so it does not verify if the written
data is correctly read back. Fix it.
Also make it return -EPERM if read value does not match written bit,
just like it done for adnp_gpio_direction_output().
If drm_gem_handle_create() fails in vgem_gem_create(), then the
drm_vgem_gem_object is freed twice: once when the reference is dropped
by drm_gem_object_put_unlocked(), and again by __vgem_gem_destroy().
This was hit by syzkaller using fault injection.
Fix it by skipping the second free.
Reported-by: syzbot+e73f2fb5ed5a5df36d33@syzkaller.appspotmail.com Fixes: af33a9190d02 ("drm/vgem: Enable dmabuf import interfaces") Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Laura Abbott <labbott@redhat.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: stable@vger.kernel.org Signed-off-by: Eric Biggers <ebiggers@google.com> Acked-by: Laura Abbott <labbott@redhat.com> Signed-off-by: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190226214451.195123-1-ebiggers@kernel.org Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A new dir entry can be created in get_subdir and its 'header->parent' is
set to NULL. Only after insert_header success, it will be set to 'dir',
otherwise 'header->parent' is set to NULL and drop_sysctl_table is called.
However in err handling path of get_subdir, drop_sysctl_table also be
called on 'new->header' regardless its value of parent pointer. Then
put_links is called, which triggers NULL-ptr deref when access member of
header->parent.
In fact we have multiple error paths which call drop_sysctl_table() there,
upon failure on insert_links() we also call drop_sysctl_table().And even
in the successful case on __register_sysctl_table() we still always call
drop_sysctl_table().This patch fix it.
Link: http://lkml.kernel.org/r/20190314085527.13244-1-yuehaibing@huawei.com Fixes: 0e47c99d7fe25 ("sysctl: Replace root_list with links between sysctl_table_sets") Signed-off-by: YueHaibing <yuehaibing@huawei.com> Reported-by: Hulk Robot <hulkci@huawei.com> Acked-by: Luis Chamberlain <mcgrof@kernel.org> Cc: Kees Cook <keescook@chromium.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Alexei Starovoitov <ast@kernel.org> Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Eric W. Biederman <ebiederm@xmission.com> Cc: <stable@vger.kernel.org> [3.4+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The Quectel EM12 is a Cat. 12 LTE modem. It behaves in the exactly the
same way as the EP06 (including the dynamic configuration behavior), so
the same checks on reserved interfaces, etc. are needed.
Signed-off-by: Kristian Evensen <kristian.evensen@gmail.com> Cc: stable <stable@vger.kernel.org> Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The SIMCom SIM5218 and compatible devices have 5 USB interfaces, only 4
of which are serial ports. The fifth is a network interface supported
by the qmi-wwan driver. Furthermore, the serial ports do not support
modem control signals. Add driver_info flags to reflect this.
The write_parport_reg_nonblock() helper takes a reference to the struct
mos_parport, but failed to release it in a couple of error paths after
allocation failures, leading to a memory leak.
Johan said that move the kref_get() and mos_parport assignment to the
end of urbtrack initialisation is a better way, so move it. and
mos_parport do not used until urbtrack initialisation.
Signed-off-by: Lin Yi <teroincn@163.com> Fixes: b69578df7e98 ("USB: usbserial: mos7720: add support for parallel port on moschip 7715") Cc: stable <stable@vger.kernel.org> # 2.6.35 Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Lorenz Messtechnik has a device that is controlled by the cp210x driver,
so add the device id to the driver. The device id was provided by
Silicon-Labs for the devices from this vendor.
We disable transmission interrupt (clear SCSCR_TIE) after all data has been transmitted
(if uart_circ_empty(xmit)). While transmitting, if the data is still in the tty buffer,
re-enable the SCSCR_TIE bit, which was done at sci_start_tx().
This is unnecessary processing, wasting CPU operation if the data transmission length is large.
And further, transmit end, FIFO empty bits disabling have also been performed in the step above.
`ni_cdio_cmdtest()` validates Comedi asynchronous commands for the DIO
subdevice (subdevice 2) of supported National Instruments M-series
cards. It is called when handling the `COMEDI_CMD` and `COMEDI_CMDTEST`
ioctls for this subdevice. There are two causes for a possible
divide-by-zero error when validating that the `stop_arg` member of the
passed-in command is not too large.
The first cause for the divide-by-zero is that calls to
`comedi_bytes_per_scan()` are only valid once the command has been
copied to `s->async->cmd`, but that copy is only done for the
`COMEDI_CMD` ioctl. For the `COMEDI_CMDTEST` ioctl, it will use
whatever was left there by the previous `COMEDI_CMD` ioctl, if any.
(This is very likely, as it is usual for the application to use
`COMEDI_CMDTEST` before `COMEDI_CMD`.) If there has been no previous,
valid `COMEDI_CMD` for this subdevice, then `comedi_bytes_per_scan()`
will return 0, so the subsequent division in `ni_cdio_cmdtest()` of
`s->async->prealloc_bufsz / comedi_bytes_per_scan(s)` will be a
divide-by-zero error. To fix this error, call a new function
`comedi_bytes_per_scan_cmd(s, cmd)`, based on the existing
`comedi_bytes_per_scan(s)` but using a specified `struct comedi_cmd` for
its calculations. (Also refactor `comedi_bytes_per_scan()` to call the
new function.)
Once the first cause for the divide-by-zero has been fixed, the second
cause is that `comedi_bytes_per_scan_cmd()` can legitimately return 0 if
the `scan_end_arg` member of the `struct comedi_cmd` being tested is 0.
Fix it by only performing the division (and validating that `stop_arg`
is no more than the maximum value) if `comedi_bytes_per_scan_cmd()`
returns a non-zero value.
The problem was reported on the COMEDI mailing list here:
https://groups.google.com/forum/#!topic/comedi_list/4t9WlHzMhKM
Reported-by: Ivan Vasilyev <grabesstimme@gmail.com> Tested-by: Ivan Vasilyev <grabesstimme@gmail.com> Fixes: f164cbf98fa8 ("staging: comedi: ni_mio_common: add finite regeneration to dio output") Cc: <stable@vger.kernel.org> # 4.6+ Cc: Spencer E. Olson <olsonse@umich.edu> Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Suppose more than one non-NPIV FCP device is active on the same channel.
Send I/O to storage and have some of the pending I/O run into a SCSI
command timeout, e.g. due to bit errors on the fibre. Now the error
situation stops. However, we saw FCP requests continue to timeout in the
channel. The abort will be successful, but the subsequent TUR fails.
Scsi_eh starts. The LUN reset fails. The target reset fails. The host
reset only did an FCP device recovery. However, for non-NPIV FCP devices,
this does not close and reopen ports on the SAN-side if other non-NPIV FCP
device(s) share the same open ports.
In order to resolve the continuing FCP request timeouts, we need to
explicitly close and reopen ports on the SAN-side.
This was missing since the beginning of zfcp in v2.6.0 history commit ea127f975424 ("[PATCH] s390 (7/7): zfcp host adapter.").
Note: The FSF requests for forced port reopen could run into FSF request
timeouts due to other reasons. This would trigger an internal FCP device
recovery. Pending forced port reopen recoveries would get dismissed. So
some ports might not get fully reopened during this host reset handler.
However, subsequent I/O would trigger the above described escalation and
eventually all ports would be forced reopen to resolve any continuing FCP
request timeouts due to earlier bit errors.
Signed-off-by: Steffen Maier <maier@linux.ibm.com> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Cc: <stable@vger.kernel.org> #3.0+ Reviewed-by: Jens Remus <jremus@linux.ibm.com> Reviewed-by: Benjamin Block <bblock@linux.ibm.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
An already deleted SCSI device can exist on the Scsi_Host and remain there
because something still holds a reference. A new SCSI device with the same
H:C:T:L and FCP device, target port WWPN, and FCP LUN can be created. When
we try to unblock an rport, we still find the deleted SCSI device and
return early because the zfcp_scsi_dev of that SCSI device is not
ZFCP_STATUS_COMMON_UNBLOCKED. Hence we miss to unblock the rport, even if
the new proper SCSI device would be in good state.
Therefore, skip deleted SCSI devices when iterating the sdevs of the shost.
[cf. __scsi_device_lookup{_by_target}() or scsi_device_get()]
The following abbreviated trace sequence can indicate such problem:
Area : REC
Tag : ersfs_3
LUN : 0x4045400300000000
WWPN : 0x50050763031bd327
LUN status : 0x40000000 not ZFCP_STATUS_COMMON_UNBLOCKED
Ready count : n not incremented yet
Running count : 0x00000000
ERP want : 0x01
ERP need : 0xc1 ZFCP_ERP_ACTION_NONE
Area : REC
Tag : ersfs_3
LUN : 0x4045400300000000
WWPN : 0x50050763031bd327
LUN status : 0x41000000
Ready count : n+1
Running count : 0x00000000
ERP want : 0x01
ERP need : 0x01
...
Area : REC
Level : 4 only with increased trace level
Tag : ertru_l
LUN : 0x4045400300000000
WWPN : 0x50050763031bd327
LUN status : 0x40000000
Request ID : 0x0000000000000000
ERP status : 0x01800000
ERP step : 0x1000
ERP action : 0x01
ERP count : 0x00
NOT followed by a trace record with tag "scpaddy"
for WWPN 0x50050763031bd327.
Signed-off-by: Steffen Maier <maier@linux.ibm.com> Fixes: 6f2ce1c6af37 ("scsi: zfcp: fix rport unblock race with LUN recovery") Cc: <stable@vger.kernel.org> #2.6.32+ Reviewed-by: Jens Remus <jremus@linux.ibm.com> Reviewed-by: Benjamin Block <bblock@linux.ibm.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit a83da8a4509d ("scsi: sd: Optimal I/O size should be a multiple
of physical block size") split one conditional into several separate
statements in an effort to provide more accurate warning messages when
a device reports a nonsensical value. However, this reorganization
accidentally dropped the precondition of the reported value being
larger than zero. This lead to a warning getting emitted on devices
that do not report an optimal I/O size at all.
Remain silent if a device does not report an optimal I/O size.
Fixes: a83da8a4509d ("scsi: sd: Optimal I/O size should be a multiple of physical block size") Cc: Randy Dunlap <rdunlap@infradead.org> Cc: <stable@vger.kernel.org> Reported-by: Hussam Al-Tayeb <ht990332@gmx.com> Tested-by: Hussam Al-Tayeb <ht990332@gmx.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The scsi_end_request() function calls scsi_cmd_to_driver() indirectly and
hence needs the disk->private_data pointer. Avoid that that pointer is
cleared before all affected I/O requests have finished. This patch avoids
that the following crash occurs:
Cc: Christoph Hellwig <hch@lst.de> Cc: Ming Lei <ming.lei@redhat.com> Cc: Hannes Reinecke <hare@suse.com> Cc: Johannes Thumshirn <jthumshirn@suse.de> Cc: Jason Yan <yanaijie@huawei.com> Cc: <stable@vger.kernel.org> Signed-off-by: Bart Van Assche <bvanassche@acm.org> Reported-by: Jason Yan <yanaijie@huawei.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
ocfs2_reflink_inodes_lock() can swap the inode1/inode2 variables so that
we always grab cluster locks in order of increasing inode number.
Unfortunately, we forget to swap the inode record buffer head pointers
when we've done this, which leads to incorrect bookkeepping when we're
trying to make the two inodes have the same refcount tree.
This has the effect of causing filesystem shutdowns if you're trying to
reflink data from inode 100 into inode 97, where inode 100 already has a
refcount tree attached and inode 97 doesn't. The reflink code decides
to copy the refcount tree pointer from 100 to 97, but uses inode 97's
inode record to open the tree root (which it doesn't have) and blows up.
This issue causes filesystem shutdowns and metadata corruption!
Link: http://lkml.kernel.org/r/20190312214910.GK20533@magnolia Fixes: 29ac8e856cb369 ("ocfs2: implement the VFS clone_range, copy_range, and dedupe_range features") Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Joseph Qi <jiangqi903@gmail.com> Cc: Mark Fasheh <mfasheh@versity.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Junxiao Bi <junxiao.bi@oracle.com> Cc: Joseph Qi <joseph.qi@huawei.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
syzbot is hitting lockdep warning [1] due to trying to open a fifo
during an execve() operation. But we don't need to open non regular
files during an execve() operation, for all files which we will need are
the executable file itself and the interpreter programs like /bin/sh and
ld-linux.so.2 .
Since the manpage for execve(2) says that execve() returns EACCES when
the file or a script interpreter is not a regular file, and the manpage
for uselib(2) says that uselib() can return EACCES, and we use
FMODE_EXEC when opening for execve()/uselib(), we can bail out if a non
regular file is requested with FMODE_EXEC set.
Since this deadlock followed by khungtaskd warnings is trivially
reproducible by a local unprivileged user, and syzbot's frequent crash
due to this deadlock defers finding other bugs, let's workaround this
deadlock until we get a chance to find a better solution.
Fix commit 56067812d5b0 ("kbuild: modversions: add infrastructure for
emitting relative CRCs") where CRCs are interpreted in host byte order
rather than proper kernel byte order. The bug is conditional on
CONFIG_MODULE_REL_CRCS.
For example, when loading a BE module into a BE kernel compiled with a LE
system, the error "disagrees about version of symbol module_layout" is
produced. A message such as "Found checksum D7FA6856 vs module 5668FAD7"
will be given with debug enabled, which indicates an obvious endian
problem within __kcrctab within the kernel image.
The general solution is to use the macro TO_NATIVE, as is done in
similar cases throughout modpost.c. With this correction it has been
verified that a BE kernel compiled with a LE system accepts BE modules.
This change has also been verified with a LE kernel compiled with a LE
system, in which case TO_NATIVE returns its value unmodified since the
byte orders match. This is by far the common case.
Currently PCM core sets each opened stream forcibly to SUSPENDED state
via snd_pcm_suspend_all() call, and the user-space is responsible for
re-triggering the resume manually either via snd_pcm_resume() or
prepare call. The scheme works fine usually, but there are corner
cases where the stream can't be resumed by that call: the streams
still in OPEN state before finishing hw_params. When they are
suspended, user-space cannot perform resume or prepare because they
haven't been set up yet. The only possible recovery is to re-open the
device, which isn't nice at all. Similarly, when a stream is in
DISCONNECTED state, it makes no sense to change it to SUSPENDED
state. Ditto for in SETUP state; which you can re-prepare directly.
So, this patch addresses these issues by filtering the PCM streams to
be suspended by checking the PCM state. When a stream is in either
OPEN, SETUP or DISCONNECTED as well as already SUSPENDED, the suspend
action is skipped.
To be noted, this problem was originally reported for the PCM runtime
PM on HD-audio. And, the runtime PM problem itself was already
addressed (although not intended) by the code refactoring commits 3d21ef0b49f8 ("ALSA: pcm: Suspend streams globally via device type PM
ops") and 17bc4815de58 ("ALSA: pci: Remove superfluous
snd_pcm_suspend*() calls"). These commits eliminated the
snd_pcm_suspend*() calls from the runtime PM suspend callback code
path, hence the racy OPEN state won't appear while runtime PM.
(FWIW, the race window is between snd_pcm_open_substream() and the
first power up in azx_pcm_open().)
Although the runtime PM issue was already "fixed", the same problem is
still present for the system PM, hence this patch is still needed.
And for stable trees, this patch alone should suffice for fixing the
runtime PM problem, too.
The PCM OSS emulation converts and transfers the data on the fly via
"plugins". The data is converted over the dynamically allocated
buffer for each plugin, and recently syzkaller caught OOB in this
flow.
Although the bisection by syzbot pointed out to the commit 65766ee0bf7f ("ALSA: oss: Use kvzalloc() for local buffer
allocations"), this is merely a commit to replace vmalloc() with
kvmalloc(), hence it can't be the cause. The further debug action
revealed that this happens in the case where a slave PCM doesn't
support only the stereo channels while the OSS stream is set up for a
mono channel. Below is a brief explanation:
At each OSS parameter change, the driver sets up the PCM hw_params
again in snd_pcm_oss_change_params_lock(). This is also the place
where plugins are created and local buffers are allocated. The
problem is that the plugins are created before the final hw_params is
determined. Namely, two snd_pcm_hw_param_near() calls for setting the
period size and periods may influence on the final result of channels,
rates, etc, too, while the current code has already created plugins
beforehand with the premature values. So, the plugin believes that
channels=1, while the actual I/O is with channels=2, which makes the
driver reading/writing over the allocated buffer size.
The fix is simply to move the plugin allocation code after the final
hw_params call.
Fix this by sanitizing dev before using it to index dp->synths.
Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].
Fix this by sanitizing info->stream before using it to index
rmidi->streams.
Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].
This belated patch implements Andrew Lunn's request of
"remove the phy_read() and phy_write() functions."
<https://lore.kernel.org/patchwork/comment/902734/>
While seemingly harmless, this causes the switch's user
port PHYs to get registered twice. This is because the
DSA subsystem will create a slave mdio-bus not knowing
that the qca8k_phy_(read|write) accessors operate on
the external mdio-bus. So the same "bus" gets effectively
duplicated.
Cc: stable@vger.kernel.org Fixes: 6b93fb46480a ("net-next: dsa: add new driver for qca8xxx family") Signed-off-by: Christian Lamparter <chunkeey@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Yauheni Kaliuta pointed out that PTR_TO_STACK store/load verifier test
was failing on powerpc64 BE, and rightfully indicated that the PPC_LD()
macro is not masking away the last two bits of the offset per the ISA,
resulting in the generation of 'lwa' instruction instead of the intended
'ld' instruction.
Segher also pointed out that we can't simply mask away the last two bits
as that will result in loading/storing from/to a memory location that
was not intended.
This patch addresses this by using ldx/stdx if the offset is not
word-aligned. We load the offset into a temporary register (TMP_REG_2)
and use that as the index register in a subsequent ldx/stdx. We fix
PPC_LD() macro to mask off the last two bits, but enhance PPC_BPF_LL()
and PPC_BPF_STL() to factor in the offset value and generate the proper
instruction sequence. We also convert all existing users of PPC_LD() and
PPC_STD() to use these macros. All existing uses of these macros have
been audited to ensure that TMP_REG_2 can be clobbered.
Fixes: 156d0e290e96 ("powerpc/ebpf/jit: Implement JIT compiler for extended BPF") Cc: stable@vger.kernel.org # v4.9+ Reported-by: Yauheni Kaliuta <yauheni.kaliuta@redhat.com> Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the current cpuidle implementation for i.MX6q, the CPU that sets
'WAIT_UNCLOCKED' and the CPU that returns to 'WAIT_CLOCKED' are always
the same. While the CPU that sets 'WAIT_UNCLOCKED' is in IDLE state of
"WAIT", if the other CPU wakes up and enters IDLE state of "WFI"
istead of "WAIT", this CPU can not wake up at expired time.
Because, in the case of "WFI", the CPU must be waked up by the local
timer interrupt. But, while 'WAIT_UNCLOCKED' is set, the local timer
is stopped, when all CPUs execute "wfi" instruction. As a result, the
local timer interrupt is not fired.
In this situation, this CPU will wake up by IRQ different from local
timer. (e.g. broacast timer)
So, this fix changes CPU to return to 'WAIT_CLOCKED'.
The reason is that kunmap(p_page) was completely left out, so we never
did an unmap for the p_page and the loop unmapping the rbio page was
iterating over the wrong number of stripes: unmapping should be done
with nr_data instead of rbio->real_stripes.
When Filipe added the recursive directory logging stuff in 2f2ff0ee5e430 ("Btrfs: fix metadata inconsistencies after directory
fsync") he specifically didn't take the directory i_mutex for the
children directories that we need to log because of lockdep. This is
generally fine, but can lead to this WARN_ON() tripping if we happen to
run delayed deletion's in between our first search and our second search
of dir_item/dir_indexes for this directory. We expect this to happen,
so the WARN_ON() isn't necessary. Drop the WARN_ON() and add a comment
so we know why this case can happen.
CC: stable@vger.kernel.org # 4.4+ Reviewed-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Josef Bacik <josef@toxicpanda.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If we do a shrinking truncate against an inode which is already present
in the respective log tree and then rename it, as part of logging the new
name we end up logging an inode item that reflects the old size of the
file (the one which we previously logged) and not the new smaller size.
The decision to preserve the size previously logged was added by commit 1a4bcf470c886b ("Btrfs: fix fsync data loss after adding hard link to
inode") in order to avoid data loss after replaying the log. However that
decision is only needed for the case the logged inode size is smaller then
the current size of the inode, as explained in that commit's change log.
If the current size of the inode is smaller then the previously logged
size, we know a shrinking truncate happened and therefore need to use
that smaller size.
$ mount /dev/sdb /mnt
$ od -t x1 -A d /mnt/bar 0000000 ab ab ab ab ab ab ab ab ab ab ab ab ab ab ab ab
* 0008000
Once we rename the file, we log its name (and inode item), and because
the inode was already logged before in the current transaction, we log it
with a size of 8000 bytes because that is the size we previously logged
(with the first fsync). As part of the rename, besides logging the inode,
we do also sync the log, which is done since commit d4682ba03ef618
("Btrfs: sync log after logging new name"), so the next fsync against our
inode is effectively a no-op, since no new changes happened since the
rename operation. Even if did not sync the log during the rename
operation, the same problem (fize size of 8000 bytes instead of 3000
bytes) would be visible after replaying the log if the log ended up
getting synced to disk through some other means, such as for example by
fsyncing some other modified file. In the example above the fsync after
the rename operation is there just because not every filesystem may
guarantee logging/journalling the inode (and syncing the log/journal)
during the rename operation, for example it is needed for f2fs, but not
for ext4 and xfs.
Fix this scenario by, when logging a new name (which is triggered by
rename and link operations), using the current size of the inode instead
of the previously logged inode size.
A test case for fstests follows soon.
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=202695 CC: stable@vger.kernel.org # 4.4+ Reported-by: Seulbae Kim <seulbae@gatech.edu> Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When I updated the spectre_v2 reporting to handle software count cache
flush I got the logic wrong when there's no software count cache
enabled at all.
The result is that on systems with the software count cache flush
disabled we print:
Which correctly indicates that the count cache is disabled, but
incorrectly says the software count cache flush is enabled.
The root of the problem is that we are trying to handle all
combinations of options. But we know now that we only expect to see
the software count cache flush enabled if the other options are false.
So split the two cases, which simplifies the logic and fixes the bug.
We were also missing a space before "(hardware accelerated)".
The commit identified below adds MC_BTB_FLUSH macro only when
CONFIG_PPC_FSL_BOOK3E is defined. This results in the following error
on some configs (seen several times with kisskb randconfig_defconfig)
Fixed the following build warning:
powerpc-linux-gnu-ld: warning: orphan section `__btb_flush_fixup' from
`arch/powerpc/kernel/head_44x.o' being placed in section
`__btb_flush_fixup'.
In order to protect against speculation attacks on
indirect branches, the branch predictor is flushed at
kernel entry to protect for the following situations:
- userspace process attacking another userspace process
- userspace process attacking the kernel
Basically when the privillege level change (i.e.the kernel
is entered), the branch predictor state is flushed.
In order to protect against speculation attacks on
indirect branches, the branch predictor is flushed at
kernel entry to protect for the following situations:
- userspace process attacking another userspace process
- userspace process attacking the kernel
Basically when the privillege level change (i.e. the
kernel is entered), the branch predictor state is flushed.
In order to flush the branch predictor the guest kernel performs
writes to the BUCSR register which is hypervisor privilleged. However,
the branch predictor is flushed at each KVM entry, so the branch
predictor has been already flushed, so just return as soon as possible
to guest.
In order to protect against speculation attacks (Spectre
variant 2) on NXP PowerPC platforms, the branch predictor
should be flushed when the privillege level is changed.
This patch is adding the infrastructure to fixup at runtime
the code sections that are performing the branch predictor flush
depending on a boot arg parameter which is added later in a
separate patch.
Look for fw-features properties to determine the appropriate settings
for the count cache flush, and then call the generic powerpc code to
set it up based on the security feature flags.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use the existing hypercall to determine the appropriate settings for
the count cache flush, and then call the generic powerpc code to set
it up based on the security feature flags.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some CPU revisions support a mode where the count cache needs to be
flushed by software on context switch. Additionally some revisions may
have a hardware accelerated flush, in which case the software flush
sequence can be shortened.
If we detect the appropriate flag from firmware we patch a branch
into _switch() which takes us to a count cache flush sequence.
That sequence in turn may be patched to return early if we detect that
the CPU supports accelerating the flush sequence in hardware.
Add debugfs support for reporting the state of the flush, as well as
runtime disabling it.
And modify the spectre_v2 sysfs file to report the state of the
software flush.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add security feature flags to indicate the need for software to flush
the count cache on context switch, and for the presence of a hardware
assisted count cache flush.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add a macro and some helper C functions for patching single asm
instructions.
The gas macro means we can do something like:
1: nop
patch_site 1b, patch__foo
Which is less visually distracting than defining a GLOBAL symbol at 1,
and also doesn't pollute the symbol table which can confuse eg. perf.
These are obviously similar to our existing feature sections, but are
not automatically patched based on CPU/MMU features, rather they are
designed to be manually patched by C code at some arbitrary point.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In a subsequent patch we will enable building security.c for Book3E.
However the NXP platforms are not vulnerable to Meltdown, so make the
Meltdown vulnerability reporting PPC_BOOK3S_64 specific.
Signed-off-by: Diana Craciun <diana.craciun@nxp.com>
[mpe: Split out of larger patch] Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently we require platform code to call setup_barrier_nospec(). But
if we add an empty definition for the !CONFIG_PPC_BARRIER_NOSPEC case
then we can call it in setup_arch().
Add a config symbol to encode which platforms support the
barrier_nospec speculation barrier. Currently this is just Book3S 64
but we will add Book3E in a future patch.
When I added the spectre_v2 information in sysfs, I included the
availability of the ori31 speculation barrier.
Although the ori31 barrier can be used to mitigate v2, it's primarily
intended as a spectre v1 mitigation. Spectre v2 is mitigated by
hardware changes.
So rework the sysfs files to show the ori31 information in the
spectre_v1 file, rather than v2.
See commit 304ec1b05031 ("x86/uaccess: Use __uaccess_begin_nospec()
and uaccess_try_nospec") and b3bbfb3fb5d2 ("x86: Introduce
__uaccess_begin_nospec() and uaccess_try_nospec") for more detail.
In all cases we are ordering the load from the potentially
user-controlled pointer vs a previous branch based on an access_ok()
check or similar.
Base on a patch from Michal Suchanek.
Signed-off-by: Michal Suchanek <msuchanek@suse.de> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Based on the RFI patching. This is required to be able to disable the
speculation barrier.
Only one barrier type is supported and it does nothing when the
firmware does not enable it. Also re-patching modules is not supported
So the only meaningful thing that can be done is patching out the
speculation barrier at boot when the user says it is not wanted.
Signed-off-by: Michal Suchanek <msuchanek@suse.de> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In my latest patch I missed one rcu_read_unlock(), in case
device is down.
Fixes: 4477138fa0ae ("tun: properly test for IFF_UP") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Same reasons than the ones explained in commit 4179cb5a4c92
("vxlan: test dev->flags & IFF_UP before calling netif_rx()")
netif_rx_ni() or napi_gro_frags() must be called under a strict contract.
At device dismantle phase, core networking clears IFF_UP
and flush_all_backlogs() is called after rcu grace period
to make sure no incoming packet might be in a cpu backlog
and still referencing the device.
A similar protocol is used for gro layer.
Most drivers call netif_rx() from their interrupt handler,
and since the interrupts are disabled at device dismantle,
netif_rx() does not have to check dev->flags & IFF_UP
Virtual drivers do not have this guarantee, and must
therefore make the check themselves.
Fixes: 1bd4978a88ac ("tun: honor IFF_UP in tun_get_user()") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The bug that Stan reported is as follows. After a restart, a 16-bit NIC
may be incorrectly identified as a 32-bit NIC and stop working.
mac8390 slot.E: Memory length resource not found, probing
mac8390 slot.E: Farallon EtherMac II-C (type farallon)
mac8390 slot.E: MAC 00:00:c5:30:c2:99, IRQ 61, 32 KB shared memory at 0xfeed0000, 32-bit access.
The bug never arises after a cold start and only intermittently after a
warm start. (I didn't investigate why the bug is intermittent.)
It turns out that memcpy_toio() is deprecated and memcmp_withio() also
has issues. Replacing these calls with mmio accessors fixes the problem.
Reported-and-tested-by: Stan Johnson <userm57@yahoo.com> Fixes: 2964db0f5904 ("m68k: Mac DP8390 update") Signed-off-by: Finn Thain <fthain@telegraphics.com.au> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
TCP/UDP checksum validity was propagated to skb
only if IP checksum is valid.
But for IPv6 there is no validity as there is no checksum in IPv6.
This patch propagates TCP/UDP checksum validity regardless of IP checksum.
Fixes: 018423e90bee ("net: ethernet: aquantia: Add ring support code") Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com> Signed-off-by: Nikita Danilov <nikita.danilov@aquantia.com> Signed-off-by: Dmitry Bogdanov <dmitry.bogdanov@aquantia.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
sctp_hdr(skb) only works when skb->transport_header is set properly.
But in Netfilter, skb->transport_header for ipv6 is not guaranteed
to be right value for sctphdr. It would cause to fail to check the
checksum for sctp packets.
So fix it by using offset, which is always right in all places.
v1->v2:
- Fix the changelog.
Fixes: e6d8b64b34aa ("net: sctp: fix and consolidate SCTP checksumming code") Reported-by: Li Shuang <shuali@redhat.com> Signed-off-by: Xin Long <lucien.xin@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit ad6c9986bcb62 ("vxlan: Fix GRO cells race condition between
receive and link delete") fixed a race condition for the typical case a vxlan
device is dismantled from the current netns. But if a netns is dismantled,
vxlan_destroy_tunnels() is called to schedule a unregister_netdevice_queue()
of all the vxlan tunnels that are related to this netns.
In vxlan_destroy_tunnels(), gro_cells_destroy() is called and finished before
unregister_netdevice_queue(). This means that the gro_cells_destroy() call is
done too soon, for the same reasons explained in above commit.
So we need to fully respect the RCU rules, and thus must remove the
gro_cells_destroy() call or risk use after-free.
Fixes: 58ce31cca1ff ("vxlan: GRO support at tunnel layer") Signed-off-by: Suanming.Mou <mousuanming@huawei.com> Suggested-by: Eric Dumazet <eric.dumazet@gmail.com> Reviewed-by: Stefano Brivio <sbrivio@redhat.com> Reviewed-by: Zhiqiang Liu <liuzhiqiang26@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For the non-XDP case, commit 773225388dae15e72790 ("net: thunderx: Optimize
page recycling for XDP") added code to nicvf_free_rbdr() that, when releasing
the additional receive buffer page reference held for recycling, repeatedly
calls put_page() until the page's _refcount goes to zero. Which results in
the page being freed.
This is not okay if the page's _refcount was greater than 1 (in the non-XDP
case), because nicvf_free_rbdr() should not be subtracting more than what
nicvf_alloc_page() had previously added to the page's _refcount, which was
only 1 (in the non-XDP case).
This can arise if a received packet is still being processed and the receive
buffer (i.e., skb->head) has not yet been freed via skb_free_head() when
nicvf_free_rbdr() is spinning through the aforementioned put_page() loop.
If this should occur, when the received packet finishes processing and
skb_free_head() is called, various problems can ensue. Exactly what, depends on
whether the page has already been reallocated or not, anything from "BUG: Bad
page state ... ", to "Unable to handle kernel NULL pointer dereference ..." or
"Unable to handle kernel paging request...".
So this patch changes nicvf_free_rbdr() to only call put_page() once for pages
held for recycling (in the non-XDP case).
Fixes: 773225388dae ("net: thunderx: Optimize page recycling for XDP") Signed-off-by: Dean Nelson <dnelson@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 773225388dae15e72790 ("net: thunderx: Optimize page recycling for XDP")
added code to nicvf_alloc_page() that inadvertently disables receive buffer
page recycling for the non-XDP case by always NULL'ng the page pointer.
This patch corrects two if-conditionals to allow for the recycling of non-XDP
mode pages by only setting the page pointer to NULL when the page is not ready
for recycling.
Fixes: 773225388dae ("net: thunderx: Optimize page recycling for XDP") Signed-off-by: Dean Nelson <dnelson@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When a dual stack tcp listener accepts an ipv4 flow,
it should not attempt to use an ipv6 header or tcp_v6_iif() helper.
Fixes: 1397ed35f22d ("ipv6: add flowinfo for tcp6 pkt_options for all cases") Fixes: df3687ffc665 ("ipv6: add the IPV6_FL_F_REFLECT flag to IPV6_FL_A_GET") Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As it stands if a shrink is delayed because of an outstanding
rehash, we will go into a rescheduling loop without ever doing
the rehash.
This patch fixes this by still carrying out the rehash and then
rescheduling so that we can shrink after the completion of the
rehash should it still be necessary.
The return value of EEXIST captures this case and other cases
(e.g., another thread expanded/rehashed the table at the same
time) where we should still proceed with the rehash.
Fixes: da20420f83ea ("rhashtable: Add nested tables") Reported-by: Josh Elsasser <jelsasser@appneta.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Tested-by: Josh Elsasser <jelsasser@appneta.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When using fanouts with AF_PACKET, the demux functions such as
fanout_demux_cpu will return an index in the fanout socket array, which
corresponds to the selected socket.
The ordering of this array depends on the order the sockets were added
to a given fanout group, so for FANOUT_CPU this means sockets are bound
to cpus in the order they are configured, which is OK.
However, when stopping then restarting the interface these sockets are
bound to, the sockets are reassigned to the fanout group in the reverse
order, due to the fact that they were inserted at the head of the
interface's AF_PACKET socket list.
This means that traffic that was directed to the first socket in the
fanout group is now directed to the last one after an interface restart.
In the case of FANOUT_CPU, traffic from CPU0 will be directed to the
socket that used to receive traffic from the last CPU after an interface
restart.
This commit introduces a helper to add a socket at the tail of a list,
then uses it to register AF_PACKET sockets.
Note that this changes the order in which sockets are listed in /proc and
with sock_diag.
Fixes: dc99f600698d ("packet: Add fanout support") Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com> Acked-by: Willem de Bruijn <willemb@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In netdev_queue_add_kobject and rx_queue_add_kobject,
if sysfs_create_group failed, kobject_put will call
netdev_queue_release to decrease dev refcont, however
dev_hold has not be called. So we will see this while
unregistering dev:
unregister_netdevice: waiting for bcsh0 to become free. Usage count = -1
Reported-by: Hulk Robot <hulkci@huawei.com> Fixes: d0d668371679 ("net: don't decrement kobj reference count on init failure") Signed-off-by: YueHaibing <yuehaibing@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When using 16K DMA buffers and ring mode, the DES3 refill is not working
correctly as the function is using a bogus pointer for checking the
private data. As a result stale pointers will remain in the RX descriptor
ring, so DMA will now likely overwrite/corrupt some already freed memory.
As simple reproducer, just receive some UDP traffic:
# ifconfig eth0 down; ifconfig eth0 mtu 9000; ifconfig eth0 up
# iperf3 -c 192.168.253.40 -u -b 0 -R
If you didn't crash by now check the RX descriptors to find non-contiguous
RX buffers:
# ping -s 8200 192.168.253.40
PING 192.168.253.40 (192.168.253.40) 8200(8228) bytes of data.
8208 bytes from 192.168.253.40: icmp_seq=1 ttl=64 time=1.00 ms
wrong data byte #8144 should be 0xd0 but was 0x88
Fix the wrong pointer. Also we must refill DES3 only if the DMA buffer
size is 16K.
Fixes: 54139cf3bb33 ("net: stmmac: adding multiple buffers for rx") Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com> Acked-by: Jose Abreu <joabreu@synopsys.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
rose_write_internal() uses a temp buffer of 100 bytes, but a manual
inspection showed that given arbitrary input, rose_create_facilities()
can fill up to 110 bytes.
Lets use a tailroom of 256 bytes for peace of mind, and remove
the bounce buffer : we can simply allocate a big enough skb
and adjust its length as needed.
syzbot report :
BUG: KASAN: stack-out-of-bounds in memcpy include/linux/string.h:352 [inline]
BUG: KASAN: stack-out-of-bounds in rose_create_facilities net/rose/rose_subr.c:521 [inline]
BUG: KASAN: stack-out-of-bounds in rose_write_internal+0x597/0x15d0 net/rose/rose_subr.c:116
Write of size 7 at addr ffff88808b1ffbef by task syz-executor.0/24854
Memory state around the buggy address: ffff88808b1ffa80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ffff88808b1ffb00: 00 00 00 00 00 00 00 00 f1 f1 f1 f1 00 00 00 03
>ffff88808b1ffb80: f2 f2 00 00 00 00 00 00 00 00 00 00 00 00 04 f3
^ ffff88808b1ffc00: f3 f3 f3 f3 00 00 00 00 00 00 00 00 00 00 00 00 ffff88808b1ffc80: 00 00 00 00 00 00 00 f1 f1 f1 f1 f1 f1 01 f2 01
Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since commit fc62814d690c ("net/packet: fix 4gb buffer limit due to overflow check")
one can now allocate packet ring buffers >= UINT_MAX. However, syzkaller
found that that triggers a warning:
projects / linux-stable / log