Thomas Gleixner [Wed, 3 Jan 2018 18:52:04 +0000 (19:52 +0100)]
x86/pti: Switch to kernel CR3 at early in entry_SYSCALL_compat()
The preparation for PTI which added CR3 switching to the entry code
misplaced the CR3 switch in entry_SYSCALL_compat().
With PTI enabled the entry code tries to access a per cpu variable after
switching to kernel GS. This fails because that variable is not mapped to
user space. This results in a double fault and in the worst case a kernel
crash.
Move the switch ahead of the access and clobber RSP which has been saved
already.
Linus Torvalds [Wed, 3 Jan 2018 19:03:07 +0000 (11:03 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace
Pull pid allocation bug fix from Eric Biederman:
"The replacement of the pid hash table and the pid bitmap with an idr
resulted in an implementation that now fails more often in low memory
situations. Allowing fuzzers to observe bad behavior from a memory
allocation failure during pid allocation.
This is a small change to fix this by making the kernel more robust in
the case of error. The non-error paths are left alone so the only
danger is to the already broken error path. I have manually injected
errors and verified that this new error handling works"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace:
pid: Handle failure to allocate the first pid in a pid namespace
Linus Torvalds [Wed, 3 Jan 2018 18:58:56 +0000 (10:58 -0800)]
Merge branch 'afs-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs
Pull afs/fscache fixes from David Howells:
- Fix the default return of fscache_maybe_release_page() when a cache
isn't in use - it prevents a filesystem from releasing pages. This
can cause a system to OOM.
- Fix a potential uninitialised variable in AFS.
- Fix AFS unlink's handling of the nlink count. It needs to use the
nlink manipulation functions so that inode structs of deleted inodes
actually get scheduled for destruction.
- Fix error handling in afs_write_end() so that the page gets unlocked
and put if we can't fill the unwritten portion.
* 'afs-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs:
afs: Fix missing error handling in afs_write_end()
afs: Fix unlink
afs: Potential uninitialized variable in afs_extract_data()
fscache: Fix the default for fscache_maybe_release_page()
David S. Miller [Wed, 3 Jan 2018 18:49:24 +0000 (13:49 -0500)]
Merge branch '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-queue
Jeff Kirsher says:
====================
Intel Wired LAN Driver Updates 2018-01-03
This series contains fixes for i40e and i40evf.
Amritha removes the UDP support for big buffer cloud filters since it is
not supported and having UDP enabled is a bug.
Alex fixes a bug in the __i40e_chk_linearize() which did not take into
account large (16K or larger) fragments that are split over 2 descriptors,
which could result in a transmit hang.
Jake fixes an issue where a devices own MAC address could be removed from
the unicast address list, so force a check on every address sync to ensure
removal does not happen.
Jiri Pirko fixes the return value when a filter configuration is not
supported, do not return "invalid" but return "not supported" so that
the core can react correctly.
====================
Neil Horman [Wed, 3 Jan 2018 18:09:23 +0000 (13:09 -0500)]
3c59x: fix missing dma_mapping_error check and bad ring refill logic
A few spots in 3c59x missed calls to dma_mapping_error checks, casuing
WARN_ONS to trigger. Clean those up. While we're at it, refactor the
refill code a bit so that if skb allocation or dma mapping fails, we
recycle the existing buffer. This prevents holes in the rx ring, and
makes for much simpler logic
Note: This is compile only tested. Ted, if you could run this and
confirm that it continues to work properly, I would appreciate it, as I
currently don't have access to this hardware
Kees Cook [Tue, 2 Jan 2018 23:21:33 +0000 (15:21 -0800)]
exec: Weaken dumpability for secureexec
This is a logical revert of commit e37fdb785a5f ("exec: Use secureexec
for setting dumpability")
This weakens dumpability back to checking only for uid/gid changes in
current (which is useless), but userspace depends on dumpability not
being tied to secureexec.
net: ena: fix error handling in ena_down() sequence
ENA admin command queue errors are not handled as part of ena_down().
As a result, in case of error admin queue transitions to non-running
state and aborts all subsequent commands including those coming from
ena_up(). Reset scheduled by the driver from the timer service
context would not proceed due to sharing rtnl with ena_up()/ena_down()
net: ena: unmask MSI-X only after device initialization is completed
Under certain conditions MSI-X interrupt might arrive right after it
was unmasked in ena_up(). There is a chance it would be processed by
the driver before device ENA_FLAG_DEV_UP flag is set. In such a case
the interrupt is ignored.
ENA device operates in auto-masked mode, therefore ignoring
interrupt leaves it masked for good.
Moving unmask of interrupt to be the last step in ena_up().
commit 96ac18f14a5a ("cxgb4: Add support for new flash parts")
removed initialization of adapter->params.sf_fw_start causing issues
while flashing firmware to card. We no longer need sf_fw_start
in adapter->params as we already have macros defined for FW flash
addresses.
Jacob Keller [Wed, 20 Dec 2017 16:04:36 +0000 (11:04 -0500)]
i40e: don't remove netdev->dev_addr when syncing uc list
In some circumstances, such as with bridging, it is possible that the
stack will add a devices own MAC address to its unicast address list.
If, later, the stack deletes this address, then the i40e driver will
receive a request to remove this address.
The driver stores its current MAC address as part of the MAC/VLAN hash
array, since it is convenient and matches exactly how the hardware
expects to be told which traffic to receive.
This causes a problem, since for more devices, the MAC address is stored
separately, and requests to delete a unicast address should not have the
ability to remove the filter for the MAC address.
Fix this by forcing a check on every address sync to ensure we do not
remove the device address.
There is a very narrow possibility of a race between .set_mac and
.set_rx_mode, if we don't change netdev->dev_addr before updating our
internal MAC list in .set_mac. This might be possible if .set_rx_mode is
going to remove MAC "XYZ" from the list, at the same time as .set_mac
changes our dev_addr to MAC "XYZ", we might possibly queue a delete,
then an add in .set_mac, then queue a delete in .set_rx_mode's
dev_uc_sync and then update netdev->dev_addr. We can avoid this by
moving the copy into dev_addr prior to the changes to the MAC filter
list.
A similar race on the other side does not cause problems, as if we're
changing our MAC form A to B, and we race with .set_rx_mode, it could
queue a delete from A, we'd update our address, and allow the delete.
This seems like a race, but in reality we're about to queue a delete of
A anyways, so it would not cause any issues.
A race in the initialization code is unlikely because the netdevice has
not yet been fully initialized and the stack should not be adding or
removing addresses yet.
Note that we don't (yet) need similar code for the VF driver because it
does not make use of __dev_uc_sync and __dev_mc_sync, but instead roles
its own method for handling updates to the MAC/VLAN list, which already
has code to protect against removal of the hardware address.
Alexander Duyck [Fri, 8 Dec 2017 18:55:04 +0000 (10:55 -0800)]
i40e/i40evf: Account for frags split over multiple descriptors in check linearize
The original code for __i40e_chk_linearize didn't take into account the
fact that if a fragment is 16K in size or larger it has to be split over 2
descriptors and the smaller of those 2 descriptors will be on the trailing
edge of the transmit. As a result we can get into situations where we didn't
catch requests that could result in a Tx hang.
This patch takes care of that by subtracting the length of all but the
trailing edge of the stale fragment before we test for sum. By doing this
we can guarantee that we have all cases covered, including the case of a
fragment that spans multiple descriptors. We don't need to worry about
checking the inner portions of this since 12K is the maximum aligned DMA
size and that is larger than any MSS will ever be since the MTU limit for
jumbos is something on the order of 9K.
Fugang Duan [Wed, 3 Jan 2018 02:39:30 +0000 (10:39 +0800)]
net: fec: defer probe if regulator is not ready
Defer probe if regulator is not ready. E.g. some regulator is fixed
regulator controlled by i2c expander gpio, the i2c device may be probed
after the driver, then it should handle the case of defer probe error.
Amritha Nambiar [Fri, 17 Nov 2017 23:51:47 +0000 (15:51 -0800)]
i40e: Remove UDP support for big buffer
Since UDP based filters are not supported via big buffer cloud
filters, remove UDP support. Also change a few return types to
indicate unsupported vs invalid configuration.
syzbot noticed a NULL pointer dereference panic in sctp_stream_free()
which was caused by an incomplete error handling in sctp_stream_init().
By not clearing stream->outcnt, it made a for() in sctp_stream_free()
think that it had elements to free, but not, leading to the panic.
As suggested by Xin Long, this patch also simplifies the error path by
moving it to the only if() that uses it.
See-also: https://www.spinics.net/lists/netdev/msg473756.html
See-also: https://www.spinics.net/lists/netdev/msg465024.html Reported-by: syzbot <[email protected]> Fixes: f952be79cebd ("sctp: introduce struct sctp_stream_out_ext") Signed-off-by: Marcelo Ricardo Leitner <[email protected]> Reviewed-by: Xin Long <[email protected]> Acked-by: Neil Horman <[email protected]> Signed-off-by: David S. Miller <[email protected]>
David S. Miller [Wed, 3 Jan 2018 16:25:45 +0000 (11:25 -0500)]
Merge branch '1GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-queue
Jeff Kirsher says:
====================
Intel Wired LAN Driver Updates 2018-01-02
This series contains fixes for e1000 and e1000e.
Tushar Dave adds a check to the driver so that it won't attempt to disable a
device that is already disabled for e1000.
Benjamin Poirier provides a fix to e1000e, where a previous commit that
Benjamin submitted changed the meaning of the return value for
"check_for_link" for copper media and not all the instances were properly
updated. Benjamin fixes the remaining instances that needed the change.
====================
Mohamed Ghannam [Tue, 2 Jan 2018 19:44:34 +0000 (19:44 +0000)]
RDS: Heap OOB write in rds_message_alloc_sgs()
When args->nr_local is 0, nr_pages gets also 0 due some size
calculation via rds_rm_size(), which is later used to allocate
pages for DMA, this bug produces a heap Out-Of-Bound write access
to a specific memory region.
Felix Janda [Mon, 1 Jan 2018 18:33:20 +0000 (19:33 +0100)]
uapi libc compat: add fallback for unsupported libcs
libc-compat.h aims to prevent symbol collisions between uapi and libc
headers for each supported libc. This requires continuous coordination
between them.
The goal of this commit is to improve the situation for libcs (such as
musl) which are not yet supported and/or do not wish to be explicitly
supported, while not affecting supported libcs. More precisely, with
this commit, unsupported libcs can request the suppression of any
specific uapi definition by defining the correspondings _UAPI_DEF_*
macro as 0. This can fix symbol collisions for them, as long as the
libc headers are included before the uapi headers. Inclusion in the
other order is outside the scope of this commit.
All infrastructure in order to enable this fallback for unsupported
libcs is already in place, except that libc-compat.h unconditionally
defines all _UAPI_DEF_* macros to 1 for all unsupported libcs so that
any previous definitions are ignored. In order to fix this, this commit
merely makes these definitions conditional.
Josh Poimboeuf [Sun, 31 Dec 2017 16:18:06 +0000 (10:18 -0600)]
x86/dumpstack: Fix partial register dumps
The show_regs_safe() logic is wrong. When there's an iret stack frame,
it prints the entire pt_regs -- most of which is random stack data --
instead of just the five registers at the end.
show_regs_safe() is also poorly named: the on_stack() checks aren't for
safety. Rename the function to show_regs_if_on_stack() and add a
comment to explain why the checks are needed.
These issues were introduced with the "partial register dump" feature of
the following commit:
b02fcf9ba121 ("x86/unwinder: Handle stack overflows more gracefully")
That patch had gone through a few iterations of development, and the
above issues were artifacts from a previous iteration of the patch where
'regs' pointed directly to the iret frame rather than to the (partially
empty) pt_regs.
Tom Lendacky [Wed, 27 Dec 2017 05:43:54 +0000 (23:43 -0600)]
x86/cpu, x86/pti: Do not enable PTI on AMD processors
AMD processors are not subject to the types of attacks that the kernel
page table isolation feature protects against. The AMD microarchitecture
does not allow memory references, including speculative references, that
access higher privileged data when running in a lesser privileged mode
when that access would result in a page fault.
Disable page table isolation by default on AMD processors by not setting
the X86_BUG_CPU_INSECURE feature, which controls whether X86_FEATURE_PTI
is set.
82c3768b8d68 ("efi/capsule-loader: Use a cached copy of the capsule header")
... refactored the capsule loading code that maps the capsule header,
to avoid having to map it several times.
However, as it turns out, the vmap() call we ended up removing did not
just map the header, but the entire capsule image, and dropping this
virtual mapping breaks capsules that are processed by the firmware
immediately (i.e., without a reboot).
Unfortunately, that change was part of a larger refactor that allowed
a quirk to be implemented for Quark, which has a non-standard memory
layout for capsules, and we have slightly painted ourselves into a
corner by allowing quirk code to mangle the capsule header and memory
layout.
So we need to fix this without breaking Quark. Fortunately, Quark does
not appear to care about the virtual mapping, and so we can simply
do a partial revert of commit:
2a457fb31df6 ("efi/capsule-loader: Use page addresses rather than struct page pointers")
... and create a vmap() mapping of the entire capsule (including header)
based on the reinstated struct page array, unless running on Quark, in
which case we pass the capsule header copy as before.
'add_efi_memmap' is an early param, but do_add_efi_memmap() has no
chance to run because the code path is before parse_early_param().
I believe it worked when the param was introduced but probably later
some other changes caused the wrong order and nobody noticed it.
Move efi_memblock_x86_reserve_range() after parse_early_param()
to fix it.
Jon Maloy [Fri, 29 Dec 2017 18:48:02 +0000 (19:48 +0100)]
tipc: fix problems with multipoint-to-point flow control
In commit 04d7b574b245 ("tipc: add multipoint-to-point flow control") we
introduced a protocol for preventing buffer overflow when many group
members try to simultaneously send messages to the same receiving member.
Stress test of this mechanism has revealed a couple of related bugs:
- When the receiving member receives an advertisement REMIT message from
one of the senders, it will sometimes prematurely activate a pending
member and send it the remitted advertisement, although the upper
limit for active senders has been reached. This leads to accumulation
of illegal advertisements, and eventually to messages being dropped
because of receive buffer overflow.
- When the receiving member leaves REMITTED state while a received
message is being read, we miss to look at the pending queue, to
activate the oldest pending peer. This leads to some pending senders
being starved out, and never getting the opportunity to profit from
the remitted advertisement.
We fix the former in the function tipc_group_proto_rcv() by returning
directly from the function once it becomes clear that the remitting
peer cannot leave REMITTED state at that point.
We fix the latter in the function tipc_group_update_rcv_win() by looking
up and activate the longest pending peer when it becomes clear that the
remitting peer now can leave REMITTED state.
ethtool: do not print warning for applications using legacy API
In kernel log ths message appears on every boot:
"warning: `NetworkChangeNo' uses legacy ethtool link settings API,
link modes are only partially reported"
When ethtool link settings API changed, it started complaining about
usages of old API. Ironically, the original patch was from google but
the application using the legacy API is chrome.
Linux ABI is fixed as much as possible. The kernel must not break it
and should not complain about applications using legacy API's.
This patch just removes the warning since using legacy API's
in Linux is perfectly acceptable.
Fixes: 3f1ac7a700d0 ("net: ethtool: add new ETHTOOL_xLINKSETTINGS API") Signed-off-by: Stephen Hemminger <[email protected]> Signed-off-by: David Decotigny <[email protected]> Signed-off-by: David S. Miller <[email protected]>
e1000e_check_for_copper_link() and e1000_check_for_copper_link_ich8lan()
are the two functions that may be assigned to mac.ops.check_for_link when
phy.media_type == e1000_media_type_copper. Commit 19110cfbb34d ("e1000e:
Separate signaling for link check/link up") changed the meaning of the
return value of check_for_link for copper media but only adjusted the first
function. This patch adjusts the second function likewise.
Russell King [Tue, 26 Dec 2017 23:15:17 +0000 (23:15 +0000)]
sfp: fix sfp-bus oops when removing socket/upstream
When we remove a socket or upstream, and the other side isn't
registered, we dereference a NULL pointer, causing a kernel oops.
Fix this.
Fixes: ce0aa27ff3f6 ("sfp: add sfp-bus to bridge between network devices and sfp cages") Signed-off-by: Russell King <[email protected]> Reviewed-by: Florian Fainelli <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Russell King [Tue, 26 Dec 2017 23:15:12 +0000 (23:15 +0000)]
phylink: ensure we report link down when LOS asserted
Although we disable the netdev carrier, we fail to report in the kernel
log that the link went down. Fix this.
Fixes: 9525ae83959b ("phylink: add phylink infrastructure") Signed-off-by: Russell King <[email protected]> Reviewed-by: Florian Fainelli <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Gao Feng [Tue, 26 Dec 2017 13:44:32 +0000 (21:44 +0800)]
macvlan: Fix one possible double free
Because the macvlan_uninit would free the macvlan port, so there is one
double free case in macvlan_common_newlink. When the macvlan port is just
created, then register_netdevice or netdev_upper_dev_link failed and they
would invoke macvlan_uninit. Then it would reach the macvlan_port_destroy
which triggers the double free.
Roi Dayan [Tue, 26 Dec 2017 05:48:51 +0000 (07:48 +0200)]
net/sched: Fix update of lastuse in act modules implementing stats_update
We need to update lastuse to to the most updated value between what
is already set and the new value.
If HW matching fails, i.e. because of an issue, the stats are not updated
but it could be that software did match and updated lastuse.
Fixes: 5712bf9c5c30 ("net/sched: act_mirred: Use passed lastuse argument") Fixes: 9fea47d93bcc ("net/sched: act_gact: Update statistics when offloaded to hardware") Signed-off-by: Roi Dayan <[email protected]> Reviewed-by: Paul Blakey <[email protected]> Acked-by: Jiri Pirko <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Erez Shitrit [Sun, 31 Dec 2017 13:33:15 +0000 (15:33 +0200)]
IB/ipoib: Fix race condition in neigh creation
When using enhanced mode for IPoIB, two threads may execute xmit in
parallel to two different TX queues while the target is the same.
In this case, both of them will add the same neighbor to the path's
neigh link list and we might see the following message:
Analysis:
Two SKB are scheduled to be transmitted from two cores.
In ipoib_start_xmit, both gets NULL when calling ipoib_neigh_get.
Two calls to neigh_add_path are made. One thread takes the spin-lock
and calls ipoib_neigh_alloc which creates the neigh structure,
then (after the __path_find) the neigh is added to the path's neigh
link list. When the second thread enters the critical section it also
calls ipoib_neigh_alloc but in this case it gets the already allocated
ipoib_neigh structure, which is already linked to the path's neigh
link list and adds it again to the list. Which beside of triggering
the list, it creates a loop in the linked list. This loop leads to
endless loop inside path_rec_completion.
Solution:
Check list_empty(&neigh->list) before adding to the list.
Add a similar fix in "ipoib_multicast.c::ipoib_mcast_send"
Fixes: b63b70d87741 ('IPoIB: Use a private hash table for path lookup in xmit path') Signed-off-by: Erez Shitrit <[email protected]> Reviewed-by: Alex Vesker <[email protected]> Signed-off-by: Leon Romanovsky <[email protected]> Signed-off-by: Jason Gunthorpe <[email protected]>
Leon Romanovsky [Sun, 31 Dec 2017 13:33:14 +0000 (15:33 +0200)]
IB/mlx4: Fix mlx4_ib_alloc_mr error flow
ibmr.device is being set only after ib_alloc_mr() is successfully complete.
Therefore, in case imlx4_mr_enable() returns with error, the error flow
unwinder calls to mlx4_free_priv_pages(), which uses ibmr.device.
Such usage causes to NULL dereference oops and to fix it, the IB device
should be set in the mr struct earlier stage (e.g. prior to calling
mlx4_free_priv_pages()).
Fixes: 1b2cd0fc673c ("IB/mlx4: Support the new memory registration API") Signed-off-by: Nitzan Carmi <[email protected]> Signed-off-by: Leon Romanovsky <[email protected]> Signed-off-by: Jason Gunthorpe <[email protected]>
Input: xen-kbdfront - do not advertise multi-touch pressure support
Some user-space applications expect multi-touch pressure
on contact to be reported if it is advertised in device
properties. Otherwise, such applications may treat reports
not as actual touches, but hovering. Currently this is
only advertised, but not reported.
Fix this by not advertising that ABS_MT_PRESSURE is supported.
Ido Schimmel [Mon, 25 Dec 2017 08:05:33 +0000 (09:05 +0100)]
mlxsw: spectrum: Relax sanity checks during enslavement
Since commit 25cc72a33835 ("mlxsw: spectrum: Forbid linking to devices that
have uppers") the driver forbids enslavement to netdevs that already
have uppers of their own, as this can result in various ordering
problems.
This requirement proved to be too strict for some users who need to be
able to enslave ports to a bridge that already has uppers. In this case,
we can allow the enslavement if the bridge is already known to us, as
any configuration performed on top of the bridge was already reflected
to the device.
Ido Schimmel [Mon, 25 Dec 2017 07:57:35 +0000 (08:57 +0100)]
mlxsw: spectrum_router: Fix NULL pointer deref
When we remove the neighbour associated with a nexthop we should always
refuse to write the nexthop to the adjacency table. Regardless if it is
already present in the table or not.
Otherwise, we risk dereferencing the NULL pointer that was set instead
of the neighbour.
Fixes: a7ff87acd995 ("mlxsw: spectrum_router: Implement next-hop routing") Signed-off-by: Ido Schimmel <[email protected]> Reported-by: Alexander Petrovskiy <[email protected]> Signed-off-by: Jiri Pirko <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Xin Long [Mon, 25 Dec 2017 06:45:12 +0000 (14:45 +0800)]
ip6_tunnel: allow ip6gre dev mtu to be set below 1280
Commit 582442d6d5bc ("ipv6: Allow the MTU of ipip6 tunnel to be set
below 1280") fixed a mtu setting issue. It works for ipip6 tunnel.
But ip6gre dev updates the mtu also with ip6_tnl_change_mtu. Since
the inner packet over ip6gre can be ipv4 and it's mtu should also
be allowed to set below 1280, the same issue also exists on ip6gre.
This patch is to fix it by simply changing to check if parms.proto
is IPPROTO_IPV6 in ip6_tnl_change_mtu instead, to make ip6gre to
go to 'else' branch.
Xin Long [Mon, 25 Dec 2017 06:43:58 +0000 (14:43 +0800)]
geneve: update skb dst pmtu on tx path
Commit a93bf0ff4490 ("vxlan: update skb dst pmtu on tx path") has fixed
a performance issue caused by the change of lower dev's mtu for vxlan.
The same thing needs to be done for geneve as well.
Note that geneve cannot adjust it's mtu according to lower dev's mtu
when creating it. The performance is very low later when netperfing
over it without fixing the mtu manually. This patch could also avoid
this issue.
Eli Cooper [Mon, 25 Dec 2017 02:43:49 +0000 (10:43 +0800)]
ip6_tunnel: disable dst caching if tunnel is dual-stack
When an ip6_tunnel is in mode 'any', where the transport layer
protocol can be either 4 or 41, dst_cache must be disabled.
This is because xfrm policies might apply to only one of the two
protocols. Caching dst would cause xfrm policies for one protocol
incorrectly used for the other.
Chris Mason [Fri, 15 Dec 2017 19:58:27 +0000 (11:58 -0800)]
btrfs: fix refcount_t usage when deleting btrfs_delayed_nodes
refcounts have a generic implementation and an asm optimized one. The
generic version has extra debugging to make sure that once a refcount
goes to zero, refcount_inc won't increase it.
The btrfs delayed inode code wasn't expecting this, and we're tripping
over the warnings when the generic refcounts are used. We ended up with
this race:
Process A Process B
btrfs_get_delayed_node()
spin_lock(root->inode_lock)
radix_tree_lookup()
__btrfs_release_delayed_node()
refcount_dec_and_test(&delayed_node->refs)
our refcount is now zero
refcount_add(2) <---
warning here, refcount
unchanged
spin_lock(root->inode_lock)
radix_tree_delete()
With the generic refcounts, we actually warn again when process B above
tries to release his refcount because refcount_add() turned into a
no-op.
We saw this in production on older kernels without the asm optimized
refcounts.
The fix used here is to use refcount_inc_not_zero() to detect when the
object is in the middle of being freed and return NULL. This is almost
always the right answer anyway, since we usually end up pitching the
delayed_node if it didn't have fresh data in it.
This also changes __btrfs_release_delayed_node() to remove the extra
check for zero refcounts before radix tree deletion.
btrfs_get_delayed_node() was the only path that was allowing refcounts
to go from zero to one.
Nikolay Borisov [Wed, 13 Dec 2017 11:50:07 +0000 (13:50 +0200)]
btrfs: Fix flush bio leak
Commit e0ae99941423 ("btrfs: preallocate device flush bio") reworked
the way the flush bio is allocated and used. Concretely it allocates
the bio in __alloc_device and then re-uses it multiple times with a
very simple endio routine that just calls complete() without consuming
a reference. Allocated bios by default come with a ref count of 1,
which is then consumed by the endio routine (or not, in which case they
should be bio_put by the caller). The way the impleementation works now
is that the flush bio has a refcount of 2 and we only ever bio_put it
once, leaving it to hang indefinitely. Fix this by removing the extra
bio_get in __alloc_device.
Changing the error return code in some situations turns out to
be harmful in practice. In particular Michael Ellerman reports
that DHCP fails on his powerpc machines, and this revert gets
things working again.
Johannes Berg agrees that this revert is the best course of
action for now.
Fixes: 029b6d140550 ("Revert "net: core: maybe return -EEXIST in __dev_alloc_name"") Reported-by: Michael Ellerman <[email protected]> Signed-off-by: David S. Miller <[email protected]>
Robin Murphy [Tue, 2 Jan 2018 12:33:14 +0000 (12:33 +0000)]
iommu/arm-smmu-v3: Cope with duplicated Stream IDs
For PCI devices behind an aliasing PCIe-to-PCI/X bridge, the bridge
alias to DevFn 0.0 on the subordinate bus may match the original RID of
the device, resulting in the same SID being present in the device's
fwspec twice. This causes trouble later in arm_smmu_write_strtab_ent()
when we wind up visiting the STE a second time and find it already live.
Avoid the issue by giving arm_smmu_install_ste_for_dev() the cleverness
to skip over duplicates. It seems mildly counterintuitive compared to
preventing the duplicates from existing in the first place, but since
the DT and ACPI probe paths build their fwspecs differently, this is
actually the cleanest and most self-contained way to deal with it.
Kasan reports a double free when finalise_stage_fn fails: the io_pgtable
ops are freed by arm_smmu_domain_finalise and then again by
arm_smmu_domain_free. Prevent this by leaving pgtbl_ops empty on failure.
Markus Heiser [Tue, 12 Dec 2017 11:38:37 +0000 (12:38 +0100)]
docs: fix, intel_guc_loader.c has been moved to intel_guc_fw.c
With commit d9e2e0143c the 'GuC-specific firmware loader' doc
section was removed from intel_guc_loader.c without a
replacement. So lets remove it from the Kernel-doc::
With commit e8668bbcb0 intel_guc_loader.c was renamed to to
intel_guc_fw.c and to name just one, intel_guc_init_hw() was
renamed to intel_guc_fw_upload(). Since we get errors in the
Sphinx build like:
- Error: Cannot open file ./drivers/gpu/drm/i915/intel_guc_loader.c
Change the kernel-doc directive from intel_guc_loader.c to
intel_guc_fw.c
s390/dasd: fix wrongly assigned configuration data
We store per path and per device configuration data to identify the
path or device correctly. The per path configuration data might get
mixed up if the original request gets into error recovery and is
started with a random path mask.
This would lead to a wrong identification of a path in case of a CUIR
event for example.
Fix by copying the path mask from the original request to the error
recovery request in case it is a path verification request.
Hans Verkuil [Mon, 4 Dec 2017 13:32:46 +0000 (14:32 +0100)]
omapdrm/dss/hdmi4_cec: fix interrupt handling
The omap4 CEC hardware cannot tell a Nack from a Low Drive from an
Arbitration Lost error, so just report a Nack, which is almost
certainly the reason for the error anyway.
This also simplifies the implementation. The only three interrupts
that need to be enabled are:
Transmit Buffer Full/Empty Change event: triggered when the
transmit finished successfully and cleared the buffer.
Receiver FIFO Not Empty event: triggered when a message was received.
Frame Retransmit Count Exceeded event: triggered when a transmit
failed repeatedly, usually due to the message being Nacked. Other
reasons are possible (Low Drive, Arbitration Lost) but there is no
way to know. If this happens the TX buffer needs to be cleared
manually.
While testing various error conditions I noticed that the hardware
can receive messages up to 18 bytes in total, which exceeds the legal
maximum of 16. This could cause a buffer overflow, so we check for
this and constrain the size to 16 bytes.
The old incorrect interrupt handler could cause the CEC framework to
enter into a bad state because it mis-detected the "Start Bit Irregularity
event" as an ARB_LOST transmit error when it actually is a receive error
which should be ignored.
Ville Syrjälä [Mon, 13 Nov 2017 13:36:22 +0000 (15:36 +0200)]
drm/i915: Put all non-blocking modesets onto an ordered wq
We have plenty of global registers and whatnot programmed without
any further locking by the modeset code. Currently non-bocking
modesets are allowed to execute in parallel which could corrupt
said registers.
To avoid the problem let's run all non-blocking modesets on an
ordered workqueue. We still put page flips etc. to system_unbound_wq
allowing page flips on one pipe to execute in parallel with page flips
or a modeset on a another pipe (assuming no known state is shared
between them, at which point they would have been added to the same
atomic commit and serialized that way).
Blocking modesets are already serialized with each other by
connection_mutex, and thus are safe. To serialize them with
non-blocking modesets we just flush the workqueue before executing
blocking modesets.
Commit 77affa31722b ("drm/i915/psr: Fix compiler warnings for
hsw_psr_disable()") swapped status and control registers while fixing
indentation. The _ctl at the end of the status register name must have to
led to this.
John Sperbeck [Mon, 1 Jan 2018 05:24:58 +0000 (21:24 -0800)]
powerpc/mm: Fix SEGV on mapped region to return SEGV_ACCERR
The recent refactoring of the powerpc page fault handler in commit c3350602e876 ("powerpc/mm: Make bad_area* helper functions") caused
access to protected memory regions to indicate SEGV_MAPERR instead of
the traditional SEGV_ACCERR in the si_code field of a user-space
signal handler. This can confuse debug libraries that temporarily
change the protection of memory regions, and expect to use SEGV_ACCERR
as an indication to restore access to a region.
This commit restores the previous behavior. The following program
exhibits the issue:
The problem seems to be that it's not properly decrementing the nlink count
so that the inode can be scrapped.
Note that this doesn't fix local creation followed by remote deletion.
That's harder to handle and will require a separate patch as we're not told
that the file has been deleted - only that the directory has changed.
Dan Carpenter [Tue, 2 Jan 2018 10:02:19 +0000 (10:02 +0000)]
afs: Potential uninitialized variable in afs_extract_data()
Smatch warns that:
fs/afs/rxrpc.c:922 afs_extract_data()
error: uninitialized symbol 'remote_abort'.
Smatch is right that "remote_abort" might be uninitialized when we pass
it to afs_set_call_complete(). I don't know if that function uses the
uninitialized variable. Anyway, the comment for rxrpc_kernel_recv_data(),
says that "*_abort should also be initialised to 0." and this patch does
that.
David Howells [Tue, 2 Jan 2018 10:02:19 +0000 (10:02 +0000)]
fscache: Fix the default for fscache_maybe_release_page()
Fix the default for fscache_maybe_release_page() for when the cookie isn't
valid or the page isn't cached. It mustn't return false as that indicates
the page cannot yet be freed.
The problem with the default is that if, say, there's no cache, but a
network filesystem's pages are using up almost all the available memory, a
system can OOM because the filesystem ->releasepage() op will not allow
them to be released as fscache_maybe_release_page() incorrectly prevents
it.
This can be tested by writing a sequence of 512MiB files to an AFS mount.
It does not affect NFS or CIFS because both of those wrap the call in a
check of PG_fscache and it shouldn't bother Ceph as that only has
PG_private set whilst writeback is in progress. This might be an issue for
9P, however.
Note that the pages aren't entirely stuck. Removing a file or unmounting
will clear things because that uses ->invalidatepage() instead.
Eric Biggers [Mon, 1 Jan 2018 15:28:31 +0000 (09:28 -0600)]
capabilities: fix buffer overread on very short xattr
If userspace attempted to set a "security.capability" xattr shorter than
4 bytes (e.g. 'setfattr -n security.capability -v x file'), then
cap_convert_nscap() read past the end of the buffer containing the xattr
value because it accessed the ->magic_etc field without verifying that
the xattr value is long enough to contain that field.
Fix it by validating the xattr value size first.
This bug was found using syzkaller with KASAN. The KASAN report was as
follows (cleaned up slightly):
BUG: KASAN: slab-out-of-bounds in cap_convert_nscap+0x514/0x630 security/commoncap.c:498
Read of size 4 at addr ffff88002d8741c0 by task syz-executor1/2852
Linus Torvalds [Sun, 31 Dec 2017 21:13:56 +0000 (13:13 -0800)]
Merge branch 'x86/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 fixes from Thomas Gleixner:
"A couple of fixlets for x86:
- Fix the ESPFIX double fault handling for 5-level pagetables
- Fix the commandline parsing for 'apic=' on 32bit systems and update
documentation
- Make zombie stack traces reliable
- Fix kexec with stack canary
- Fix the delivery mode for APICs which was missed when the x86
vector management was converted to single target delivery. Caused a
regression due to the broken hardware which ignores affinity
settings in lowest prio delivery mode.
- Unbreak modules when AMD memory encryption is enabled
- Remove an unused parameter of prepare_switch_to"
* 'x86/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/apic: Switch all APICs to Fixed delivery mode
x86/apic: Update the 'apic=' description of setting APIC driver
x86/apic: Avoid wrong warning when parsing 'apic=' in X86-32 case
x86-32: Fix kexec with stack canary (CONFIG_CC_STACKPROTECTOR)
x86: Remove unused parameter of prepare_switch_to
x86/stacktrace: Make zombie stack traces reliable
x86/mm: Unbreak modules that use the DMA API
x86/build: Make isoimage work on Debian
x86/espfix/64: Fix espfix double-fault handling on 5-level systems
Linus Torvalds [Sun, 31 Dec 2017 20:30:34 +0000 (12:30 -0800)]
Merge branch 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull timer fixes from Thomas Gleixner:
"A pile of fixes for long standing issues with the timer wheel and the
NOHZ code:
- Prevent timer base confusion accross the nohz switch, which can
cause unlocked access and data corruption
- Reinitialize the stale base clock on cpu hotplug to prevent subtle
side effects including rollovers on 32bit
- Prevent an interrupt storm when the timer softirq is already
pending caused by tick_nohz_stop_sched_tick()
- Move the timer start tracepoint to a place where it actually makes
sense
- Add documentation to timerqueue functions as they caused confusion
several times now"
* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
timerqueue: Document return values of timerqueue_add/del()
timers: Invoke timer_start_debug() where it makes sense
nohz: Prevent a timer interrupt storm in tick_nohz_stop_sched_tick()
timers: Reinitialize per cpu bases on hotplug
timers: Use deferrable base independent of base::nohz_active
Linus Torvalds [Sun, 31 Dec 2017 20:27:19 +0000 (12:27 -0800)]
Merge branch 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull scheduler fixes from Thomas Gleixner:
"Three patches addressing the fallout of the CPU_ISOLATION changes
especially with NO_HZ_FULL plus documentation of boot parameter
dependency"
* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
sched/isolation: Document boot parameters dependency on CONFIG_CPU_ISOLATION=y
sched/isolation: Enable CONFIG_CPU_ISOLATION=y by default
sched/isolation: Make CONFIG_NO_HZ_FULL select CONFIG_CPU_ISOLATION
Linus Torvalds [Sun, 31 Dec 2017 19:47:24 +0000 (11:47 -0800)]
Merge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull perf fixes from Thomas Gleixner:
- plug a memory leak in the intel pmu init code
- clang fixes
- tooling fix to avoid including kernel headers
- a fix for jvmti to generate correct debug information for inlined
code
- replace backtick with a regular shell function
- fix the build in hardened environments
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf/x86/intel: Plug memory leak in intel_pmu_init()
x86/asm: Allow again using asm.h when building for the 'bpf' clang target
tools arch s390: Do not include header files from the kernel sources
perf jvmti: Generate correct debug information for inlined code
perf tools: Fix up build in hardened environments
perf tools: Use shell function for perl cflags retrieval
Linus Torvalds [Sun, 31 Dec 2017 19:23:11 +0000 (11:23 -0800)]
Merge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull irq fixes from Thomas Gleixner:
"A rather large update after the kaisered maintainer finally found time
to handle regression reports.
- The larger part addresses a regression caused by the x86 vector
management rework.
The reservation based model does not work reliably for MSI
interrupts, if they cannot be masked (yes, yet another hw
engineering trainwreck). The reason is that the reservation mode
assigns a dummy vector when the interrupt is allocated and switches
to a real vector when the interrupt is requested.
If the MSI entry cannot be masked then the initialization might
raise an interrupt before the interrupt is requested, which ends up
as spurious interrupt and causes device malfunction and worse. The
fix is to exclude MSI interrupts which do not support masking from
reservation mode and assign a real vector right away.
- Extend the extra lockdep class setup for nested interrupts with a
class for the recently added irq_desc::request_mutex so lockdep can
differeniate and does not emit false positive warnings.
- A ratelimit guard for the bad irq printout so in case a bad irq
comes back immediately the system does not drown in dmesg spam"
* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
genirq/msi, x86/vector: Prevent reservation mode for non maskable MSI
genirq/irqdomain: Rename early argument of irq_domain_activate_irq()
x86/vector: Use IRQD_CAN_RESERVE flag
genirq: Introduce IRQD_CAN_RESERVE flag
genirq/msi: Handle reactivation only on success
gpio: brcmstb: Make really use of the new lockdep class
genirq: Guard handle_bad_irq log messages
kernel/irq: Extend lockdep class for request mutex
Linus Torvalds [Sun, 31 Dec 2017 18:57:10 +0000 (10:57 -0800)]
Merge branch 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull objtool fixes from Thomas Gleixner:
"Three fixlets for objtool:
- Address two segfaults related to missing parameter and clang
objects
- Make it compile clean with clang"
* 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
objtool: Fix seg fault with clang-compiled objects
objtool: Fix seg fault caused by missing parameter
objtool: Fix Clang enum conversion warning
Linus Torvalds [Sun, 31 Dec 2017 18:52:51 +0000 (10:52 -0800)]
Merge tag 'char-misc-4.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc
Pull char/misc fixes from Greg KH:
"Here are six small fixes of some of the char/misc drivers that have
been sent in to resolve reported issues.
Nothing major, a binder use-after-free fix, some thunderbolt bugfixes,
a hyper-v bugfix, and an nvmem driver fix. All of these have been in
linux-next with no reported issues for a while"
* tag 'char-misc-4.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
nvmem: meson-mx-efuse: fix reading from an offset other than 0
binder: fix proc->files use-after-free
vmbus: unregister device_obj->channels_kset
thunderbolt: Mask ring interrupt properly when polling starts
MAINTAINERS: Add thunderbolt.rst to the Thunderbolt driver entry
thunderbolt: Make pathname to force_power shorter
Linus Torvalds [Sun, 31 Dec 2017 18:50:05 +0000 (10:50 -0800)]
Merge tag 'driver-core-4.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core
Pull driver core fixes from Greg KH:
"Here are two driver core fixes for 4.15-rc6, resolving some reported
issues.
The first is a cacheinfo fix for DT based systems to resolve a
reported issue that has been around for a while, and the other is to
resolve a regression in the kobject uevent code that showed up in
4.15-rc1.
Both have been in linux-next for a while with no reported issues"
* tag 'driver-core-4.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core:
kobject: fix suppressing modalias in uevents delivered over netlink
drivers: base: cacheinfo: fix cache type for non-architected system cache
Linus Torvalds [Sun, 31 Dec 2017 18:44:00 +0000 (10:44 -0800)]
Merge tag 'usb-4.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Pull USB/PHY fixes from Greg KH:
"Here are a number of small USB and PHY driver fixes for 4.15-rc6.
Nothing major, but there are a number of regression fixes in here that
resolve issues that have been reported a bunch. There are also the
usual xhci fixes as well as a number of new usb serial device ids.
All of these have been in linux-next for a while with no reported
issues"
* tag 'usb-4.15-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
usb: xhci: Add XHCI_TRUST_TX_LENGTH for Renesas uPD720201
xhci: Fix use-after-free in xhci debugfs
xhci: Fix xhci debugfs NULL pointer dereference in resume from hibernate
USB: serial: ftdi_sio: add id for Airbus DS P8GR
usb: Add device quirk for Logitech HD Pro Webcam C925e
usb: add RESET_RESUME for ELSA MicroLink 56K
usbip: fix usbip bind writing random string after command in match_busid
usbip: stub_rx: fix static checker warning on unnecessary checks
usbip: prevent leaking socket pointer address in messages
usbip: stub: stop printing kernel pointer addresses in messages
usbip: vhci: stop printing kernel pointer addresses in messages
USB: Fix off by one in type-specific length check of BOS SSP capability
USB: serial: option: adding support for YUGA CLM920-NC5
phy: rcar-gen3-usb2: select USB_COMMON
phy: rockchip-typec: add pm_runtime_disable in err case
phy: cpcap-usb: Fix platform_get_irq_byname's error checking.
phy: tegra: fix device-tree node lookups
USB: serial: qcserial: add Sierra Wireless EM7565
USB: serial: option: add support for Telit ME910 PID 0x1101
USB: chipidea: msm: fix ulpi-node lookup
Adam Borowski [Mon, 25 Dec 2017 15:38:58 +0000 (16:38 +0100)]
MAINTAINERS: mark arch/blackfin/ and its gubbins as orphaned
The blackfin architecture has seen no maintainer action of any kind since
April 2015. No new code, no pull requests, no acks to patches, no response
to mails, nothing.
The web site has an expired certificate (expiration Sep 2017, issued in
2013), the mailing list sees no answers either, with one exception:
https://sourceforge.net/p/adi-buildroot/mailman/adi-buildroot-devel/
>
> Steven is no longer working on this for ADI. Acked by me if this works. Thanks.
>
> Best regards,
> Aaron Wu
> Analog Devices Inc.
But, Aaron doesn't seem to respond to queries either.
Thomas Gleixner [Sun, 31 Dec 2017 15:52:15 +0000 (16:52 +0100)]
x86/ldt: Make LDT pgtable free conditional
Andy prefers to be paranoid about the pagetable free in the error path of
write_ldt(). Make it conditional and warn whenever the installment of a
secondary LDT fails.
Thomas Gleixner [Sun, 31 Dec 2017 10:24:34 +0000 (11:24 +0100)]
x86/ldt: Plug memory leak in error path
The error path in write_ldt() tries to free 'old_ldt' instead of the newly
allocated 'new_ldt', resulting in a memory leak. It also misses to clean up a
half populated LDT pagetable, which is not a leak as it gets cleaned up
when the process exits.
Free both the potentially half populated LDT pagetable and the newly
allocated LDT struct. This can be done unconditionally because once an LDT
is mapped subsequent maps will succeed, because the PTE page is already
populated and the two LDTs fit into that single page.
Thomas Gleixner [Sat, 30 Dec 2017 21:13:54 +0000 (22:13 +0100)]
x86/mm: Remove preempt_disable/enable() from __native_flush_tlb()
The preempt_disable/enable() pair in __native_flush_tlb() was added in
commit:
5cf0791da5c1 ("x86/mm: Disable preemption during CR3 read+write")
... to protect the UP variant of flush_tlb_mm_range().
That preempt_disable/enable() pair should have been added to the UP variant
of flush_tlb_mm_range() instead.
The UP variant was removed with commit:
ce4a4e565f52 ("x86/mm: Remove the UP asm/tlbflush.h code, always use the (formerly) SMP code")
... but the preempt_disable/enable() pair stayed around.
The latest change to __native_flush_tlb() in commit:
6fd166aae78c ("x86/mm: Use/Fix PCID to optimize user/kernel switches")
... added an access to a per CPU variable outside the preempt disabled
regions, which makes no sense at all. __native_flush_tlb() must always
be called with at least preemption disabled.
Remove the preempt_disable/enable() pair and add a WARN_ON_ONCE() to catch
bad callers independent of the smp_processor_id() debugging.
Thomas Gleixner [Sat, 30 Dec 2017 21:13:53 +0000 (22:13 +0100)]
x86/smpboot: Remove stale TLB flush invocations
smpboot_setup_warm_reset_vector() and smpboot_restore_warm_reset_vector()
invoke local_flush_tlb() for no obvious reason.
Digging in history revealed that the original code in the 2.1 era added
those because the code manipulated a swapper_pg_dir pagetable entry. The
pagetable manipulation was removed long ago in the 2.3 timeframe, but the
TLB flush invocations stayed around forever.
Remove them along with the pointless pr_debug()s which come from the same 2.1
change.
Linus Torvalds [Sat, 30 Dec 2017 22:31:30 +0000 (14:31 -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, both of which cause I/O hangs.
The storvsc one is from the hyper-v which can hang under certain hot
add/remove conditions and the other is generally, where removing a
target and a device in close proximity can result in the release
method being executed twice (and subsequent list and other corruption
and an eventual panic)"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
scsi: storvsc: Fix scsi_cmd error assignments in storvsc_handle_error
scsi: core: check for device state in __scsi_remove_target()
projects / linux.git / log