Damien Le Moal [Thu, 9 Jan 2020 05:03:55 +0000 (14:03 +0900)]
null_blk: Fix zone write handling
null_zone_write() only allows writing empty and implicitly opened zones.
Writing to closed and explicitly opened zones must also be allowed and
the zone condition must be transitioned to implicit open if the zone
is not explicitly opened already.
Fixes: da644b2cc1a4 ("null_blk: add zone open, close, and finish support") Signed-off-by: Damien Le Moal <[email protected]> Signed-off-by: Jens Axboe <[email protected]>
Ian Abbott [Tue, 14 Jan 2020 18:25:32 +0000 (18:25 +0000)]
staging: comedi: ni_routes: allow partial routing information
This patch fixes a regression on setting up asynchronous commands to use
external trigger sources when board-specific routing information is
missing.
`ni_find_device_routes()` (called via `ni_assign_device_routes()`) finds
the table of register values for the device family and the set of valid
routes for the specific board. If both are found,
`tables->route_values` is set to point to the table of register values
for the device family and `tables->valid_routes` is set to point to the
list of valid routes for the specific board. If either is not found,
both `tables->route_values` and `tables->valid_routes` are left set at
their initial null values (initialized by `ni_assign_device_routes()`)
and the function returns `-ENODATA`.
Returning an error results in some routing functionality being disabled.
Unfortunately, leaving `table->route_values` set to `NULL` also breaks
the setting up of asynchronous commands that are configured to use
external trigger sources. Calls to `ni_check_trigger_arg()` or
`ni_check_trigger_arg_roffs()` while checking the asynchronous command
set-up would result in a null pointer dereference if
`table->route_values` is `NULL`. The null pointer dereference is fixed
in another patch, but it now results in failure to set up the
asynchronous command. That is a regression from the behavior prior to
commit 347e244884c3 ("staging: comedi: tio: implement global tio/ctr
routing") and commit 56d0b826d39f ("staging: comedi: ni_mio_common:
implement new routing for TRIG_EXT").
Change `ni_find_device_routes()` to set `tables->route_values` and/or
`tables->valid_routes` to valid information even if the other one can
only be set to `NULL` due to missing information. The function will
still return an error in that case. This should result in
`tables->valid_routes` being valid for all currently supported device
families even if the board-specific routing information is missing.
That should be enough to fix the regression on setting up asynchronous
commands to use external triggers for boards with missing routing
information.
Ian Abbott [Tue, 14 Jan 2020 18:25:31 +0000 (18:25 +0000)]
staging: comedi: ni_routes: fix null dereference in ni_find_route_source()
In `ni_find_route_source()`, `tables->route_values` gets dereferenced.
However it is possible that `tables->route_values` is `NULL`, leading to
a null pointer dereference. `tables->route_values` will be `NULL` if
the call to `ni_assign_device_routes()` during board initialization
returned an error due to missing device family routing information or
missing board-specific routing information. For example, there is
currently no board-specific routing information provided for the
PCIe-6251 board and several other boards, so those are affected by this
bug.
The bug is triggered when `ni_find_route_source()` is called via
`ni_check_trigger_arg()` or `ni_check_trigger_arg_roffs()` when checking
the arguments for setting up asynchronous commands. Fix it by returning
`-EINVAL` if `tables->route_values` is `NULL`.
Even with this fix, setting up asynchronous commands to use external
trigger sources for boards with missing routing information will still
fail gracefully. Since `ni_find_route_source()` only depends on the
device family routing information, it would be better if that was made
available even if the board-specific routing information is missing.
That will be addressed by another patch.
David S. Miller [Wed, 15 Jan 2020 12:16:30 +0000 (04:16 -0800)]
Merge branch 'mlxsw-Various-fixes'
Ido Schimmel says:
====================
mlxsw: Various fixes
This patch set contains various fixes for mlxsw.
Patch #1 splits the init() callback between Spectrum-2 and Spectrum-3 in
order to avoid enforcing the same firmware version for both ASICs, as
this can't possibly work. Without this patch the driver cannot boot with
the Spectrum-3 ASIC.
Patches #2-#3 fix a long standing race condition that was recently
exposed while testing the driver on an emulator, which is very slow
compared to the actual hardware. The problem is explained in detail in
the commit messages.
Patch #4 fixes a selftest.
Patch #5 prevents offloaded qdiscs from presenting a non-zero backlog to
the user when the netdev is down. This is done by clearing the cached
backlog in the driver when the netdev goes down.
Patch #6 fixes qdisc statistics (backlog and tail drops) to also take
into account the multicast traffic classes.
v2:
* Patches #2-#3: use skb_cow_head() instead of skb_unshare() as
suggested by Jakub. Remove unnecessary check regarding headroom
* Patches #5-#6: new
====================
Petr Machata [Wed, 15 Jan 2020 11:53:49 +0000 (13:53 +0200)]
mlxsw: spectrum_qdisc: Include MC TCs in Qdisc counters
mlxsw configures Spectrum in such a way that BUM traffic is passed not
through its nominal traffic class TC, but through its MC counterpart TC+8.
However, when collecting statistics, Qdiscs only look at the nominal TC and
ignore the MC TC.
Add two helpers to compute the value for logical TC from the constituents,
one for backlog, the other for tail drops. Use them throughout instead of
going through the xstats pointer directly.
Counters for TX bytes and packets are deduced from packet priority
counters, and therefore already include BUM traffic. wred_drop counter is
irrelevant on MC TCs, because RED is not enabled on them.
Petr Machata [Wed, 15 Jan 2020 11:53:48 +0000 (13:53 +0200)]
mlxsw: spectrum: Wipe xstats.backlog of down ports
Per-port counter cache used by Qdiscs is updated periodically, unless the
port is down. The fact that the cache is not updated for down ports is no
problem for most counters, which are relative in nature. However, backlog
is absolute in nature, and if there is a non-zero value in the cache around
the time that the port goes down, that value just stays there. This value
then leaks to offloaded Qdiscs that report non-zero backlog even if
there (obviously) is no traffic.
The HW does not keep backlog of a downed port, so do likewise: as the port
goes down, wipe the backlog value from xstats.
Mausezahn does not recognize "own" as a keyword on source IP address. As a
result, the MC stream is not running at all, and therefore no UC
degradation can be observed even in principle.
Fix the invocation, and tighten the test: due to the minimum shaper
configured at the MC TCs, we always expect about 20% degradation. Fail the
test if it is lower.
Ido Schimmel [Wed, 15 Jan 2020 11:53:46 +0000 (13:53 +0200)]
mlxsw: switchx2: Do not modify cloned SKBs during xmit
The driver needs to prepend a Tx header to each packet it is
transmitting. The header includes information such as the egress port
and traffic class.
The addition of the header requires the driver to modify the SKB's
header and therefore it must not be shared. Otherwise, we risk hitting
various race conditions.
For example, when a packet is flooded (cloned) by the bridge driver to
two switch ports swp1 and swp2:
t0 - mlxsw_sp_port_xmit() is called for swp1. Tx header is prepended with
swp1's port number
t1 - mlxsw_sp_port_xmit() is called for swp2. Tx header is prepended with
swp2's port number, overwriting swp1's port number
t2 - The device processes data buffer from t0. Packet is transmitted via
swp2
t3 - The device processes data buffer from t1. Packet is transmitted via
swp2
Usually, the device is fast enough and transmits the packet before its
Tx header is overwritten, but this is not the case in emulated
environments.
Fix this by making sure the SKB's header is writable by calling
skb_cow_head(). Since the function ensures we have headroom to push the
Tx header, the check further in the function can be removed.
v2:
* Use skb_cow_head() instead of skb_unshare() as suggested by Jakub
* Remove unnecessary check regarding headroom
Ido Schimmel [Wed, 15 Jan 2020 11:53:45 +0000 (13:53 +0200)]
mlxsw: spectrum: Do not modify cloned SKBs during xmit
The driver needs to prepend a Tx header to each packet it is
transmitting. The header includes information such as the egress port
and traffic class.
The addition of the header requires the driver to modify the SKB's
header and therefore it must not be shared. Otherwise, we risk hitting
various race conditions.
For example, when a packet is flooded (cloned) by the bridge driver to
two switch ports swp1 and swp2:
t0 - mlxsw_sp_port_xmit() is called for swp1. Tx header is prepended with
swp1's port number
t1 - mlxsw_sp_port_xmit() is called for swp2. Tx header is prepended with
swp2's port number, overwriting swp1's port number
t2 - The device processes data buffer from t0. Packet is transmitted via
swp2
t3 - The device processes data buffer from t1. Packet is transmitted via
swp2
Usually, the device is fast enough and transmits the packet before its
Tx header is overwritten, but this is not the case in emulated
environments.
Fix this by making sure the SKB's header is writable by calling
skb_cow_head(). Since the function ensures we have headroom to push the
Tx header, the check further in the function can be removed.
v2:
* Use skb_cow_head() instead of skb_unshare() as suggested by Jakub
* Remove unnecessary check regarding headroom
Ido Schimmel [Wed, 15 Jan 2020 11:53:44 +0000 (13:53 +0200)]
mlxsw: spectrum: Do not enforce same firmware version for multiple ASICs
In commit a72afb6879bb ("mlxsw: Enforce firmware version for
Spectrum-2") I added a required firmware version for Spectrum-2, but
missed the fact that mlxsw_sp2_init() is used by both Spectrum-2 and
Spectrum-3. This means that the same firmware version will be used for
both, which is wrong.
Fix this by creating a new init() callback for Spectrum-3.
- D+ goes high, Host starts running by remote wakeup
- D+ is not stable, goes low
- Host requests GetPortStatus at (*1) and gets the following hub status:
- Current Connect Status bit is 0
- Connect Status Change bit is 1
- D+ stabilizes, goes high
- Host requests ClearPortFeature and thus Connect Status Change bit is
cleared at (*2)
- After waiting 100 ms, Host starts the Interrupt Transfer at (*3)
- Since the Connect Status Change bit is 0, Hub returns NAK.
In this case, port_event() is not called in hub_event() and Host cannot
recognize device. To solve this issue, flag change_bits even if only
Connect Status Change bit is 1 when got in the first GetPortStatus.
This issue occurs rarely because it only if D+ changes during a very
short time between GetPortStatus and ClearPortFeature. However, it is
fatal if it occurs in embedded system.
David S. Miller [Wed, 15 Jan 2020 12:12:00 +0000 (04:12 -0800)]
Merge tag 'mac80211-for-net-2020-01-15' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211
Johannes Berg says:
====================
A few fixes:
* -O3 enablement fallout, thanks to Arnd who ran this
* fixes for a few leaks, thanks to Felix
* channel 12 regulatory fix for custom regdomains
* check for a crash reported by syzbot
(NULL function is called on drivers that don't have it)
* fix TKIP replay protection after setup with some APs
(from Jouni)
* restrict obtaining some mesh data to avoid WARN_ONs
* fix deadlocks with auto-disconnect (socket owner)
* fix radar detection events with multiple devices
====================
The reason is that a CPU's thermal interrupt is enabled prior to
executing its hotplug onlining callback which will initialize the
throttling workqueues.
Such a race can lead to therm_throt_process() accessing an uninitialized
therm_work, leading to the above BUG at a very early bootup stage.
Therefore, unmask the thermal interrupt vector only after having setup
the workqueues completely.
Kevin Hao [Tue, 14 Jan 2020 08:28:18 +0000 (16:28 +0800)]
Revert "gpio: thunderx: Switch to GPIOLIB_IRQCHIP"
This reverts commit a7fc89f9d5fcc10a5474cfe555f5a9e5df8b0f1f because
there are some bugs in this commit, and we don't have a simple way to
fix these bugs. So revert this commit to make the thunderx gpio work
on the stable kernel at least. We will switch to GPIOLIB_IRQCHIP
for thunderx gpio by following patches.
Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.4.0-rc7+ #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Use READ_ONCE() and WRITE_ONCE() to annotate this expected race.
Felix Fietkau [Mon, 13 Jan 2020 18:21:07 +0000 (19:21 +0100)]
cfg80211: fix page refcount issue in A-MSDU decap
The fragments attached to a skb can be part of a compound page. In that case,
page_ref_inc will increment the refcount for the wrong page. Fix this by
using get_page instead, which calls page_ref_inc on the compound head and
also checks for overflow.
Arnd Bergmann [Tue, 7 Jan 2020 20:07:35 +0000 (21:07 +0100)]
wireless: wext: avoid gcc -O3 warning
After the introduction of CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE_O3,
the wext code produces a bogus warning:
In function 'iw_handler_get_iwstats',
inlined from 'ioctl_standard_call' at net/wireless/wext-core.c:1015:9,
inlined from 'wireless_process_ioctl' at net/wireless/wext-core.c:935:10,
inlined from 'wext_ioctl_dispatch.part.8' at net/wireless/wext-core.c:986:8,
inlined from 'wext_handle_ioctl':
net/wireless/wext-core.c:671:3: error: argument 1 null where non-null expected [-Werror=nonnull]
memcpy(extra, stats, sizeof(struct iw_statistics));
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from arch/x86/include/asm/string.h:5,
net/wireless/wext-core.c: In function 'wext_handle_ioctl':
arch/x86/include/asm/string_64.h:14:14: note: in a call to function 'memcpy' declared here
The problem is that ioctl_standard_call() sometimes calls the handler
with a NULL argument that would cause a problem for iw_handler_get_iwstats.
However, iw_handler_get_iwstats never actually gets called that way.
Marking that function as noinline avoids the warning and leads
to slightly smaller object code as well.
Jouni Malinen [Tue, 7 Jan 2020 15:35:45 +0000 (17:35 +0200)]
mac80211: Fix TKIP replay protection immediately after key setup
TKIP replay protection was skipped for the very first frame received
after a new key is configured. While this is potentially needed to avoid
dropping a frame in some cases, this does leave a window for replay
attacks with group-addressed frames at the station side. Any earlier
frame sent by the AP using the same key would be accepted as a valid
frame and the internal RSC would then be updated to the TSC from that
frame. This would allow multiple previously transmitted group-addressed
frames to be replayed until the next valid new group-addressed frame
from the AP is received by the station.
Fix this by limiting the no-replay-protection exception to apply only
for the case where TSC=0, i.e., when this is for the very first frame
protected using the new key, and the local RSC had not been set to a
higher value when configuring the key (which may happen with GTK).
Orr Mazor [Sun, 22 Dec 2019 14:55:31 +0000 (14:55 +0000)]
cfg80211: Fix radar event during another phy CAC
In case a radar event of CAC_FINISHED or RADAR_DETECTED
happens during another phy is during CAC we might need
to cancel that CAC.
If we got a radar in a channel that another phy is now
doing CAC on then the CAC should be canceled there.
If, for example, 2 phys doing CAC on the same channels,
or on comptable channels, once on of them will finish his
CAC the other might need to cancel his CAC, since it is no
longer relevant.
To fix that the commit adds an callback and implement it in
mac80211 to end CAC.
This commit also adds a call to said callback if after a radar
event we see the CAC is no longer relevant
Ganapathi Bhat [Fri, 20 Dec 2019 10:14:32 +0000 (10:14 +0000)]
wireless: fix enabling channel 12 for custom regulatory domain
Commit e33e2241e272 ("Revert "cfg80211: Use 5MHz bandwidth by
default when checking usable channels"") fixed a broken
regulatory (leaving channel 12 open for AP where not permitted).
Apply a similar fix to custom regulatory domain processing.
Al Viro [Wed, 15 Jan 2020 03:09:57 +0000 (22:09 -0500)]
fix autofs regression caused by follow_managed() changes
we need to reload ->d_flags after the call of ->d_manage() - the thing
might've been called with dentry still negative and have the damn thing
turned positive while we'd waited.
Al Viro [Fri, 10 Jan 2020 22:17:19 +0000 (17:17 -0500)]
reimplement path_mountpoint() with less magic
... and get rid of a bunch of bugs in it. Background:
the reason for path_mountpoint() is that umount() really doesn't
want attempts to revalidate the root of what it's trying to umount.
The thing we want to avoid actually happen from complete_walk();
solution was to do something parallel to normal path_lookupat()
and it both went overboard and got the boilerplate subtly
(and not so subtly) wrong.
A better solution is to do pretty much what the normal path_lookupat()
does, but instead of complete_walk() do unlazy_walk(). All it takes
to avoid that ->d_weak_revalidate() call... mountpoint_last() goes
away, along with everything it got wrong, and so does the magic around
LOOKUP_NO_REVAL.
Another source of bugs is that when we traverse mounts at the final
location (and we need to do that - umount . expects to get whatever's
overmounting ., if any, out of the lookup) we really ought to take
care of ->d_manage() - as it is, manual umount of autofs automount
in progress can lead to unpleasant surprises for the daemon. Easily
solved by using handle_lookup_down() instead of follow_mount().
Jens Axboe [Wed, 15 Jan 2020 05:09:06 +0000 (22:09 -0700)]
io_uring: be consistent in assigning next work from handler
If we pass back dependent work in case of links, we need to always
ensure that we call the link setup and work prep handler. If not, we
might be missing some setup for the next work item.
Lorenzo Bianconi [Tue, 14 Jan 2020 10:21:16 +0000 (11:21 +0100)]
net: mvneta: fix dma sync size in mvneta_run_xdp
Page pool API will start syncing (if requested) starting from
page->dma_addr + pool->p.offset. Fix dma sync length in
mvneta_run_xdp since we do not need to account xdp headroom
Fixes: 07e13edbb6a6 ("net: mvneta: get rid of huge dma sync in mvneta_rx_refill") Signed-off-by: Lorenzo Bianconi <[email protected]> Acked-by: Jesper Dangaard Brouer <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Johan Hovold [Tue, 14 Jan 2020 08:27:29 +0000 (09:27 +0100)]
r8152: add missing endpoint sanity check
Add missing endpoint sanity check to probe in order to prevent a
NULL-pointer dereference (or slab out-of-bounds access) when retrieving
the interrupt-endpoint bInterval on ndo_open() in case a device lacks
the expected endpoints.
Linus Torvalds [Tue, 14 Jan 2020 21:33:14 +0000 (13:33 -0800)]
Merge tag 'nfs-for-5.5-2' of git://git.linux-nfs.org/projects/anna/linux-nfs
Pull NFS client bugfixes from Anna Schumaker:
"Three NFS over RDMA fixes for bugs Chuck found that can be hit during
device removal:
- Fix create_qp crash on device unload
- Fix completion wait during device removal
- Fix oops in receive handler after device removal"
* tag 'nfs-for-5.5-2' of git://git.linux-nfs.org/projects/anna/linux-nfs:
xprtrdma: Fix oops in Receive handler after device removal
xprtrdma: Fix completion wait during device removal
xprtrdma: Fix create_qp crash on device unload
Ming Lei [Sat, 11 Jan 2020 12:57:43 +0000 (20:57 +0800)]
block: fix get_max_segment_size() overflow on 32bit arch
Commit 429120f3df2d starts to take account of segment's start dma address
when computing max segment size, and data type of 'unsigned long'
is used to do that. However, the segment mask may be 0xffffffff, so
the figured out segment size may be overflowed in case of zero physical
address on 32bit arch.
Fix the issue by returning queue_max_segment_size() directly when that
happens.
Sunil Muthuswamy [Tue, 14 Jan 2020 00:52:14 +0000 (00:52 +0000)]
hv_sock: Remove the accept port restriction
Currently, hv_sock restricts the port the guest socket can accept
connections on. hv_sock divides the socket port namespace into two parts
for server side (listening socket), 0-0x7FFFFFFF & 0x80000000-0xFFFFFFFF
(there are no restrictions on client port namespace). The first part
(0-0x7FFFFFFF) is reserved for sockets where connections can be accepted.
The second part (0x80000000-0xFFFFFFFF) is reserved for allocating ports
for the peer (host) socket, once a connection is accepted.
This reservation of the port namespace is specific to hv_sock and not
known by the generic vsock library (ex: af_vsock). This is problematic
because auto-binds/ephemeral ports are handled by the generic vsock
library and it has no knowledge of this port reservation and could
allocate a port that is not compatible with hv_sock (and legitimately so).
The issue hasn't surfaced so far because the auto-bind code of vsock
(__vsock_bind_stream) prior to the change 'VSOCK: bind to random port for
VMADDR_PORT_ANY' would start walking up from LAST_RESERVED_PORT (1023) and
start assigning ports. That will take a large number of iterations to hit
0x7FFFFFFF. But, after the above change to randomize port selection, the
issue has started coming up more frequently.
There has really been no good reason to have this port reservation logic
in hv_sock from the get go. Reserving a local port for peer ports is not
how things are handled generally. Peer ports should reflect the peer port.
This fixes the issue by lifting the port reservation, and also returns the
right peer port. Since the code converts the GUID to the peer port (by
using the first 4 bytes), there is a possibility of conflicts, but that
seems like a reasonable risk to take, given this is limited to vsock and
that only applies to all local sockets.
Eric Dumazet [Mon, 13 Jan 2020 17:27:11 +0000 (09:27 -0800)]
net: usb: lan78xx: limit size of local TSO packets
lan78xx_tx_bh() makes sure to not exceed MAX_SINGLE_PACKET_SIZE
bytes in the aggregated packets it builds, but does
nothing to prevent large GSO packets being submitted.
Pierre-Francois reported various hangs when/if TSO is enabled.
For localy generated packets, we can use netif_set_gso_max_size()
to limit the size of TSO packets.
Note that forwarded packets could still hit the issue,
so a complete fix might require implementing .ndo_features_check
for this driver, forcing a software segmentation if the size
of the TSO packet exceeds MAX_SINGLE_PACKET_SIZE.
Vladis Dronov [Mon, 13 Jan 2020 13:00:09 +0000 (14:00 +0100)]
ptp: free ptp device pin descriptors properly
There is a bug in ptp_clock_unregister(), where ptp_cleanup_pin_groups()
first frees ptp->pin_{,dev_}attr, but then posix_clock_unregister() needs
them to destroy a related sysfs device.
These functions can not be just swapped, as posix_clock_unregister() frees
ptp which is needed in the ptp_cleanup_pin_groups(). Fix this by calling
ptp_cleanup_pin_groups() in ptp_clock_release(), right before ptp is freed.
This makes this patch fix an UAF bug in a patch which fixes an UAF bug.
Wayne Lin [Mon, 6 Jan 2020 10:21:58 +0000 (18:21 +0800)]
drm/dp_mst: clear time slots for ports invalid
[Why]
When change the connection status in a MST topology, mst device
which detect the event will send out CONNECTION_STATUS_NOTIFY messgae.
e.g. src-mst-mst-sst => src-mst (unplug) mst-sst
Currently, under the above case of unplugging device, ports which have
been allocated payloads and are no longer in the topology still occupy
time slots and recorded in proposed_vcpi[] of topology manager.
If we don't clean up the proposed_vcpi[], when code flow goes to try to
update payload table by calling drm_dp_update_payload_part1(), we will
fail at checking port validation due to there are ports with proposed
time slots but no longer in the mst topology. As the result of that, we
will also stop updating the DPCD payload table of down stream port.
[How]
While handling the CONNECTION_STATUS_NOTIFY message, add a detection to
see if the event indicates that a device is unplugged to an output port.
If the detection is true, then iterrate over all proposed_vcpi[] to
see whether a port of the proposed_vcpi[] is still in the topology or
not. If the port is invalid, set its num_slots to 0.
Thereafter, when try to update payload table by calling
drm_dp_update_payload_part1(), we can successfully update the DPCD
payload table of down stream port and clear the proposed_vcpi[] to NULL.
Changes since v1:(https://patchwork.kernel.org/patch/11275801/)
* Invert the conditional to reduce the indenting
Chuck Lever [Fri, 3 Jan 2020 16:52:22 +0000 (11:52 -0500)]
xprtrdma: Fix oops in Receive handler after device removal
Since v5.4, a device removal occasionally triggered this oops:
Dec 2 17:13:53 manet kernel: BUG: unable to handle page fault for address: 0000000c00000219
Dec 2 17:13:53 manet kernel: #PF: supervisor read access in kernel mode
Dec 2 17:13:53 manet kernel: #PF: error_code(0x0000) - not-present page
Dec 2 17:13:53 manet kernel: PGD 0 P4D 0
Dec 2 17:13:53 manet kernel: Oops: 0000 [#1] SMP
Dec 2 17:13:53 manet kernel: CPU: 2 PID: 468 Comm: kworker/2:1H Tainted: G W 5.4.0-00050-g53717e43af61 #883
Dec 2 17:13:53 manet kernel: Hardware name: Supermicro SYS-6028R-T/X10DRi, BIOS 1.1a 10/16/2015
Dec 2 17:13:53 manet kernel: Workqueue: ib-comp-wq ib_cq_poll_work [ib_core]
Dec 2 17:13:53 manet kernel: RIP: 0010:rpcrdma_wc_receive+0x7c/0xf6 [rpcrdma]
Dec 2 17:13:53 manet kernel: Code: 6d 8b 43 14 89 c1 89 45 78 48 89 4d 40 8b 43 2c 89 45 14 8b 43 20 89 45 18 48 8b 45 20 8b 53 14 48 8b 30 48 8b 40 10 48 8b 38 <48> 8b 87 18 02 00 00 48 85 c0 75 18 48 8b 05 1e 24 c4 e1 48 85 c0
Dec 2 17:13:53 manet kernel: RSP: 0018:ffffc900035dfe00 EFLAGS: 00010246
Dec 2 17:13:53 manet kernel: RAX: ffff888467290000 RBX: ffff88846c638400 RCX: 0000000000000048
Dec 2 17:13:53 manet kernel: RDX: 0000000000000048 RSI: 00000000f942e000 RDI: 0000000c00000001
Dec 2 17:13:53 manet kernel: RBP: ffff888467611b00 R08: ffff888464e4a3c4 R09: 0000000000000000
Dec 2 17:13:53 manet kernel: R10: ffffc900035dfc88 R11: fefefefefefefeff R12: ffff888865af4428
Dec 2 17:13:53 manet kernel: R13: ffff888466023000 R14: ffff88846c63f000 R15: 0000000000000010
Dec 2 17:13:53 manet kernel: FS: 0000000000000000(0000) GS:ffff88846fa80000(0000) knlGS:0000000000000000
Dec 2 17:13:53 manet kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Dec 2 17:13:53 manet kernel: CR2: 0000000c00000219 CR3: 0000000002009002 CR4: 00000000001606e0
Dec 2 17:13:53 manet kernel: Call Trace:
Dec 2 17:13:53 manet kernel: __ib_process_cq+0x5c/0x14e [ib_core]
Dec 2 17:13:53 manet kernel: ib_cq_poll_work+0x26/0x70 [ib_core]
Dec 2 17:13:53 manet kernel: process_one_work+0x19d/0x2cd
Dec 2 17:13:53 manet kernel: ? cancel_delayed_work_sync+0xf/0xf
Dec 2 17:13:53 manet kernel: worker_thread+0x1a6/0x25a
Dec 2 17:13:53 manet kernel: ? cancel_delayed_work_sync+0xf/0xf
Dec 2 17:13:53 manet kernel: kthread+0xf4/0xf9
Dec 2 17:13:53 manet kernel: ? kthread_queue_delayed_work+0x74/0x74
Dec 2 17:13:53 manet kernel: ret_from_fork+0x24/0x30
The proximal cause is that this rpcrdma_rep has a rr_rdmabuf that
is still pointing to the old ib_device, which has been freed. The
only way that is possible is if this rpcrdma_rep was not destroyed
by rpcrdma_ia_remove.
Debugging showed that was indeed the case: this rpcrdma_rep was
still in use by a completing RPC at the time of the device removal,
and thus wasn't on the rep free list. So, it was not found by
rpcrdma_reps_destroy().
The fix is to introduce a list of all rpcrdma_reps so that they all
can be found when a device is removed. That list is used to perform
only regbuf DMA unmapping, replacing that call to
rpcrdma_reps_destroy().
Meanwhile, to prevent corruption of this list, I've moved the
destruction of temp rpcrdma_rep objects to rpcrdma_post_recvs().
rpcrdma_xprt_drain() ensures that post_recvs (and thus rep_destroy) is
not invoked while rpcrdma_reps_unmap is walking rb_all_reps, thus
protecting the rb_all_reps list.
Fixes: b0b227f071a0 ("xprtrdma: Use an llist to manage free rpcrdma_reps") Signed-off-by: Chuck Lever <[email protected]> Signed-off-by: Anna Schumaker <[email protected]>
Chuck Lever [Fri, 3 Jan 2020 16:52:17 +0000 (11:52 -0500)]
xprtrdma: Fix completion wait during device removal
I've found that on occasion, "rmmod <dev>" will hang while if an NFS
is under load.
Ensure that ri_remove_done is initialized only just before the
transport is woken up to force a close. This avoids the completion
possibly getting initialized again while the CM event handler is
waiting for a wake-up.
Fixes: bebd031866ca ("xprtrdma: Support unplugging an HCA from under an NFS mount") Signed-off-by: Chuck Lever <[email protected]> Signed-off-by: Anna Schumaker <[email protected]>
The fix is to copy the qp_init_attr struct that was just created by
rpcrdma_ep_create() instead of using the one from the previous
connection instance.
Fixes: 98ef77d1aaa7 ("xprtrdma: Send Queue size grows after a reconnect") Signed-off-by: Chuck Lever <[email protected]> Signed-off-by: Anna Schumaker <[email protected]>
Linus Torvalds [Tue, 14 Jan 2020 18:22:10 +0000 (10:22 -0800)]
Merge branch 'parisc-5.5-3' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux
Pull parisc fixes from Helge Deller:
"A boot crash fix by Mike Rapoport and a printk fix by Krzysztof
Kozlowski"
* 'parisc-5.5-3' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
parisc: fix map_pages() to actually populate upper directory
parisc: Use proper printk format for resource_size_t
Linus Torvalds [Tue, 14 Jan 2020 18:17:15 +0000 (10:17 -0800)]
Merge tag 'asm-generic-5.5' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/playground
Pull asm-generic fixes from Arnd Bergmann:
"Here are two bugfixes from Mike Rapoport, both fixing compile-time
errors for the nds32 architecture that were recently introduced"
* tag 'asm-generic-5.5' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/playground:
nds32: fix build failure caused by page table folding updates
asm-generic/nds32: don't redefine cacheflush primitives
Linus Torvalds [Tue, 14 Jan 2020 18:14:06 +0000 (10:14 -0800)]
Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull SCSI fixes from James Bottomley:
"Two simple fixes in the upper drivers (so both fairly core), one in
enclosures, which fixes replugging a device into an enclosure slot and
one in the disk driver which fixes revalidating a drive with
protection information (PI) to make it a non-PI drive ... previously
we were still remembering the old PI state.
Both fixed issues are quite rare in the field"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
scsi: enclosure: Fix stale device oops with hot replug
scsi: sd: Clear sdkp->protection_type if disk is reformatted without PI
David Howells [Tue, 14 Jan 2020 16:16:54 +0000 (16:16 +0000)]
afs: Fix afs_lookup() to not clobber the version on a new dentry
Fix afs_lookup() to not clobber the version set on a new dentry by
afs_do_lookup() - especially as it's using the wrong version of the
version (we need to use the one given to us by whatever op the dir
contents correspond to rather than what's in the afs_vnode).
Fixes: 9dd0b82ef530 ("afs: Fix missing dentry data version updating") Signed-off-by: David Howells <[email protected]> Signed-off-by: Linus Torvalds <[email protected]>
David Howells [Tue, 14 Jan 2020 16:16:25 +0000 (16:16 +0000)]
afs: Fix use-after-loss-of-ref
afs_lookup() has a tracepoint to indicate the outcome of
d_splice_alias(), passing it the inode to retrieve the fid from.
However, the function gave up its ref on that inode when it called
d_splice_alias(), which may have failed and dropped the inode.
David Howells [Tue, 14 Jan 2020 16:06:14 +0000 (16:06 +0000)]
keys: Fix request_key() cache
When the key cached by request_key() and co. is cleaned up on exit(),
the code looks in the wrong task_struct, and so clears the wrong cache.
This leads to anomalies in key refcounting when doing, say, a kernel
build on an afs volume, that then trigger kasan to report a
use-after-free when the key is viewed in /proc/keys.
Fix this by making exit_creds() look in the passed-in task_struct rather
than in current (the task_struct cleanup code is deferred by RCU and
potentially run in another task).
Fixes: 7743c48e54ee ("keys: Cache result of request_key*() temporarily in task_struct") Signed-off-by: David Howells <[email protected]> Signed-off-by: Linus Torvalds <[email protected]>
Linus Torvalds [Tue, 14 Jan 2020 17:22:51 +0000 (09:22 -0800)]
Merge branch 'akpm' (patches from Andrew)
Merge misc fixes from Andrew Morton:
"11 mm fixes"
* emailed patches from Andrew Morton <[email protected]>:
mm: khugepaged: add trace status description for SCAN_PAGE_HAS_PRIVATE
mm: memcg/slab: call flush_memcg_workqueue() only if memcg workqueue is valid
mm/page-writeback.c: improve arithmetic divisions
mm/page-writeback.c: use div64_ul() for u64-by-unsigned-long divide
mm/page-writeback.c: avoid potential division by zero in wb_min_max_ratio()
mm, debug_pagealloc: don't rely on static keys too early
mm: memcg/slab: fix percpu slab vmstats flushing
mm/shmem.c: thp, shmem: fix conflict of above-47bit hint address and PMD alignment
mm/huge_memory.c: thp: fix conflict of above-47bit hint address and PMD alignment
mm/memory_hotplug: don't free usage map when removing a re-added early section
mm, thp: tweak reclaim/compaction effort of local-only and all-node allocations
Marek Vasut [Thu, 19 Dec 2019 21:32:19 +0000 (22:32 +0100)]
ASoC: sgtl5000: add multi endpoint support
Support multiple endpoints on SGTL5000 codec port when used in
of_graph context.
This patch allows to share the codec port between two CPU DAIs.
Example:
Custom STM32MP157C board uses SGTL5000 audio codec. This codec is
connected to two serial audio interfaces, which are configured
either as rx or tx.
From AsoC point of view the topolgy is the following:
// 2 CPU DAIs (SAI2A/B), 1 Codec (SGTL5000)
Playback: CPU-A-DAI(slave) -> (master)CODEC-DAI/port0
Record: CPU-B-DAI(slave) <- (master)CODEC-DAI/port0
In the DT two endpoints have to be associated to the codec port:
sgtl5000_port: port {
sgtl5000_tx_endpoint: endpoint@0 {
remote-endpoint = <&sai2a_endpoint>;
};
sgtl5000_rx_endpoint: endpoint@1 {
remote-endpoint = <&sai2b_endpoint>;
};
};
However, when the audio graph card parses the codec nodes, it expects
to find DAI interface indexes matching the endpoints indexes.
The current patch forces the use of DAI id 0 for both endpoints,
which allows to share the codec DAI between the two CPU DAIs
for playback and capture streams respectively.
Shengjiu Wang [Wed, 4 Dec 2019 12:00:19 +0000 (20:00 +0800)]
ASoC: fsl_asrc: Add support for imx8qm & imx8qxp
There are two asrc module in imx8qm & imx8qxp, each module has
different clock configuration, and the DMA type is EDMA.
So in this patch, we define the new clocks, refine the clock map,
and include struct fsl_asrc_soc_data for different soc usage.
The EDMA channel is fixed with each dma request, one dma request
corresponding to one dma channel. So we need to request dma
channel with dma request of asrc module.
Shengjiu Wang [Wed, 4 Dec 2019 12:00:18 +0000 (20:00 +0800)]
ASoC: dt-bindings: fsl_asrc: add compatible string for imx8qm & imx8qxp
Add compatible string "fsl,imx8qm-asrc" for imx8qm platform,
"fsl,imx8qxp-asrc" for imx8qxp platform.
There are two asrc modules in imx8qm & imx8qxp, the clock mapping is
different for each other, so add new property "fsl,asrc-clk-map"
to distinguish them.
Marek Vasut [Fri, 20 Dec 2019 16:44:50 +0000 (17:44 +0100)]
ASoC: sgtl5000: Fix VDDA and VDDIO comparison
Comparing the voltage of VDDA and VDDIO to determine whether or not to
enable VDDC manual override is insufficient. This is a problem in case
the VDDA is supplied from different regulator than VDDIO, while both
report the same voltage to the regulator framework. In that case where
VDDA and VDDIO is supplied by different regulators, the VDDC manual
override must not be applied.
Marek Vasut [Fri, 20 Dec 2019 16:44:49 +0000 (17:44 +0100)]
regulator: core: Add regulator_is_equal() helper
Add regulator_is_equal() helper to compare whether two regulators are
the same. This is useful for checking whether two separate regulators
in a driver are actually the same supply.
ASoC: Intel: kbl_da7219_max98927: test return value on route add
Fix cppcheck warning:
[sound/soc/intel/boards/kbl_da7219_max98927.c:340] ->
[sound/soc/intel/boards/kbl_da7219_max98927.c:348]: (style) Variable
'ret' is reassigned a value before the old one has been used.
ASoC: SOF: Intel: hda-dai: fix compilation warning in pcm_prepare
Fix GCC warning with W=1, previous cleanup did not remove unnecessary
variable.
sound/soc/sof/intel/hda-dai.c: In function ‘hda_link_pcm_prepare’:
sound/soc/sof/intel/hda-dai.c:265:31: warning: variable ‘hda_stream’
set but not used [-Wunused-but-set-variable]
265 | struct sof_intel_hda_stream *hda_stream;
| ^~~~~~~~~~
ASoC: rt715: use dev_to_sdw_dev() instead of to_sdw_slave_device()
This patch fixup this error
CC sound/soc/codecs/rt715-sdw.o
linux/sound/soc/codecs/rt715-sdw.c: In function 'rt715_dev_resume':
linux/sound/soc/codecs/rt715-sdw.c:568:28: error: implicit declaration\
of function 'to_sdw_slave_device'; did you mean 'sdw_slave_modalias'?\
[-Werror=implicit-function-declaration]
struct sdw_slave *slave = to_sdw_slave_device(dev);
^~~~~~~~~~~~~~~~~~~
sdw_slave_modalias
linux/sound/soc/codecs/rt715-sdw.c:568:28: warning: initialization of\
'struct sdw_slave *' from 'int' makes pointer from integer without a\
cast [-Wint-conversion]
cc1: some warnings being treated as errors
Brendan Higgins [Fri, 13 Dec 2019 00:35:22 +0000 (16:35 -0800)]
lkdtm/bugs: fix build error in lkdtm_UNSET_SMEP
When building ARCH=um with CONFIG_UML_X86=y and CONFIG_64BIT=y we get
the build errors:
drivers/misc/lkdtm/bugs.c: In function ‘lkdtm_UNSET_SMEP’:
drivers/misc/lkdtm/bugs.c:288:8: error: implicit declaration of function ‘native_read_cr4’ [-Werror=implicit-function-declaration]
cr4 = native_read_cr4();
^~~~~~~~~~~~~~~
drivers/misc/lkdtm/bugs.c:290:13: error: ‘X86_CR4_SMEP’ undeclared (first use in this function); did you mean ‘X86_FEATURE_SMEP’?
if ((cr4 & X86_CR4_SMEP) != X86_CR4_SMEP) {
^~~~~~~~~~~~
X86_FEATURE_SMEP
drivers/misc/lkdtm/bugs.c:290:13: note: each undeclared identifier is reported only once for each function it appears in
drivers/misc/lkdtm/bugs.c:297:2: error: implicit declaration of function ‘native_write_cr4’; did you mean ‘direct_write_cr4’? [-Werror=implicit-function-declaration]
native_write_cr4(cr4);
^~~~~~~~~~~~~~~~
direct_write_cr4
So specify that this block of code should only build when
CONFIG_X86_64=y *AND* CONFIG_UML is unset.
Kees Cook [Thu, 2 Jan 2020 20:29:17 +0000 (12:29 -0800)]
lkdtm/bugs: Make double-fault test always available
Adjust the DOUBLE_FAULT test to always be available (so test harnesses
don't have to make exceptions more missing tests), and for the
arch-specific tests to "XFAIL" so that test harnesses can reason about
expected vs unexpected failures.
Arnd Bergmann [Fri, 13 Dec 2019 22:31:07 +0000 (15:31 -0700)]
coresight: etm4x: Fix unused function warning
Some of the newly added code in the etm4x driver is inside of an #ifdef,
and some other code is outside of it, leading to a harmless warning when
CONFIG_CPU_PM is disabled:
drivers/hwtracing/coresight/coresight-etm4x.c:68:13: error: 'etm4_os_lock' defined but not used [-Werror=unused-function]
static void etm4_os_lock(struct etmv4_drvdata *drvdata)
^~~~~~~~~~~~
To avoid the warning and simplify the the #ifdef checks, use
IS_ENABLED() instead, so the compiler can drop the unused functions
without complaining.
The driver was issuing synchronous uninterruptible control requests
without using a timeout. This could lead to the driver hanging
on open() or tiocmset() due to a malfunctioning (or malicious) device
until the device is physically disconnected.
The USB upper limit of five seconds per request should be more than
enough.
The altsetting sanity check in set_sync_ep_implicit_fb_quirk() was
checking for there to be at least one altsetting but then went on to
access the second one, which may not exist.
This could lead to random slab data being used to initialise the sync
endpoint in snd_usb_add_endpoint().
Fixes: c75a8a7ae565 ("ALSA: snd-usb: add support for implicit feedback") Fixes: ca10a7ebdff1 ("ALSA: usb-audio: FT C400 sync playback EP to capture EP") Fixes: 5e35dc0338d8 ("ALSA: usb-audio: add implicit fb quirk for Behringer UFX1204") Fixes: 17f08b0d9aaf ("ALSA: usb-audio: add implicit fb quirk for Axe-Fx II") Fixes: 103e9625647a ("ALSA: usb-audio: simplify set_sync_ep_implicit_fb_quirk") Cc: stable <[email protected]> # 3.5 Signed-off-by: Johan Hovold <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Takashi Iwai <[email protected]>
Mike Rapoport [Wed, 8 Jan 2020 12:58:52 +0000 (14:58 +0200)]
parisc: fix map_pages() to actually populate upper directory
The commit d96885e277b5 ("parisc: use pgtable-nopXd instead of
4level-fixup") converted PA-RISC to use folded page tables, but it missed
the conversion of pgd_populate() to pud_populate() in maps_pages()
function. This caused the upper page table directory to remain empty and
the system would crash as a result.
Using pud_populate() that actually populates the page table instead of
dummy pgd_populate() fixes the issue.
parisc: Use proper printk format for resource_size_t
resource_size_t should be printed with its own size-independent format
to fix warnings when compiling on 64-bit platform (e.g. with
COMPILE_TEST):
arch/parisc/kernel/drivers.c: In function 'print_parisc_device':
arch/parisc/kernel/drivers.c:892:9: warning:
format '%p' expects argument of type 'void *',
but argument 4 has type 'resource_size_t {aka unsigned int}' [-Wformat=]
Tvrtko Ursulin [Fri, 10 Jan 2020 11:32:53 +0000 (11:32 +0000)]
drm/i915/pmu: Do not use colons or dashes in PMU names
We use PCI device path in the registered PMU name in order to distinguish
between multiple GPUs. But since tools/perf reserves a special meaning to
dash and colon characters we need to transliterate them to something else.
We choose an underscore.
v2:
* Use strreplace. (Chris)
* Dashes are not good either. (Chris)
ALSA: hda: patch_hdmi: remove warnings with empty body
make W=1 reports the following warnings, fix as suggested
sound/pci/hda/patch_hdmi.c: In function ‘hdmi_non_intrinsic_event’:
sound/pci/hda/patch_hdmi.c:824:3: warning: suggest braces around empty
body in an ‘if’ statement [-Wempty-body]
824 | ;
| ^
sound/pci/hda/patch_hdmi.c:826:3: warning: suggest braces around empty
body in an ‘if’ statement [-Wempty-body]
826 | ;
| ^
Johan Hovold [Mon, 13 Jan 2020 17:23:58 +0000 (18:23 +0100)]
NFC: pn533: fix bulk-message timeout
The driver was doing a synchronous uninterruptible bulk-transfer without
using a timeout. This could lead to the driver hanging on probe due to a
malfunctioning (or malicious) device until the device is physically
disconnected. While sleeping in probe the driver prevents other devices
connected to the same hub from being added to (or removed from) the bus.
An arbitrary limit of five seconds should be more than enough.
Fixes: dbafc28955fa ("NFC: pn533: don't send USB data off of the stack") Signed-off-by: Johan Hovold <[email protected]> Reviewed-by: Greg Kroah-Hartman <[email protected]> Signed-off-by: Jakub Kicinski <[email protected]>
Kristian Evensen [Mon, 13 Jan 2020 13:57:40 +0000 (14:57 +0100)]
qmi_wwan: Add support for Quectel RM500Q
RM500Q is a 5G module from Quectel, supporting both standalone and
non-standalone modes. The normal Quectel quirks apply (DTR and dynamic
interface numbers).
Linus Torvalds [Tue, 14 Jan 2020 02:40:57 +0000 (18:40 -0800)]
Merge tag 'Intel-CVE-2019-14615' from bundle by Akeem Abodunrin.
Merge Intel Gen9 graphics fix from Akeem Abodunrin:
"Insufficient control flow in certain data structures for some Intel
Processors with Intel Processor Graphics may allow an unauthenticated
user to potentially enable information disclosure via local access
This provides mitigation for Gen9 hardware. Note that Gen8 is not
impacted due to a previously implemented workaround.
The mitigation involves using an existing hardware feature to forcibly
clear down all EU state at each context switch"
* tag 'Intel-CVE-2019-14615' of emailed bundle from Akeem G Abodunrin <[email protected]>:
drm/i915/gen9: Clear residual context state on context switch
Milind Parab [Mon, 13 Jan 2020 03:30:43 +0000 (03:30 +0000)]
net: macb: fix for fixed-link mode
This patch fix the issue with fixed link. With fixed-link
device opening fails due to macb_phylink_connect not
handling fixed-link mode, in which case no MAC-PHY connection
is needed and phylink_connect return success (0), however
in current driver attempt is made to search and connect to
PHY even for fixed-link.
Jens Axboe [Tue, 14 Jan 2020 02:23:24 +0000 (19:23 -0700)]
io_uring: don't setup async context for read/write fixed
We don't need it, and if we have it, then the retry handler will attempt
to copy the non-existent iovec with the inline iovec, with a segment
count that doesn't make sense.
Yang Shi [Tue, 14 Jan 2020 00:29:36 +0000 (16:29 -0800)]
mm: khugepaged: add trace status description for SCAN_PAGE_HAS_PRIVATE
Commit 99cb0dbd47a1 ("mm,thp: add read-only THP support for (non-shmem)
FS") introduced a new khugepaged scan result: SCAN_PAGE_HAS_PRIVATE, but
the corresponding description for trace events were not added.
The root cause is that the IOMMU subsystem runs before the workqueue
subsystem, which the variable 'wq_online' is still 'false'. This leads
to the statement 'if (WARN_ON(!wq_online))' in flush_workqueue() is
'true'.
Since the variable 'memcg_kmem_cache_wq' is not allocated during the
time, it is unnecessary to call flush_memcg_workqueue(). This prevents
the WARNING message triggered by flush_workqueue().
projects / linux.git / log