If PHY is not available on DSA port (described at devicetree but absent or
failed to detect) then kernel prints warning after 3700 secs:
[ 3707.948771] ------------[ cut here ]------------
[ 3707.948784] Type was not set for devlink port.
[ 3707.948894] WARNING: CPU: 1 PID: 17 at net/core/devlink.c:8097 0xc083f9d8
We should unregister the devlink port as a user port and
re-register it as an unused port before executing "continue" in case of
dsa_port_setup error.
Fixes: 86f8b1c01a ("net: dsa: Do not make user port errors fatal")
Signed-off-by: Maxim Kochetkov <fido_max@inbox.ru>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Handle return error code of eth_mac_addr();
Fixes: 3d23a05c75 ("gianfar: Enable changing mac addr when if up")
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In myri10ge_sw_tso, the skb_list_walk_safe macro will set
(curr) = (segs) and (next) = (curr)->next. If status!=0 is true,
the memory pointed by curr and segs will be free by dev_kfree_skb_any(curr).
But later, the segs is used by segs = segs->next and causes a uaf.
As (next) = (curr)->next, my patch replaces seg->next to next.
Fixes: 536577f36f ("net: myri10ge: use skb_list_walk_safe helper for gso segments")
Signed-off-by: Lv Yunlong <lyl2019@mail.ustc.edu.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
Marc Kleine-Budde says:
====================
pull-request: can 2021-03-29
this is a pull request of 3 patches for net/master.
The two patch are by Oliver Hartkopp. He fixes length check in the
proto_ops::getname callback for the CAN RAW, BCM and ISOTP protocols,
which were broken by the introduction of the J1939 protocol.
The last patch is by me and fixes the a BUILD_BUG_ON() check which
triggers on ARCH=arm with CONFIG_AEABI unset.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Ido Schimmel says:
====================
mlxsw: spectrum: Fix ECN marking in tunnel decapsulation
Patch #1 fixes a discrepancy between the software and hardware data
paths with regards to ECN marking after decapsulation. See the changelog
for a detailed description.
Patch #2 extends the ECN decap test to cover all possible combinations
of inner and outer ECN markings. The test passes over both data paths.
v2:
* Only set ECT(1) if inner is ECT(0)
* Introduce a new helper to determine inner ECN. Share it between NVE
and IP-in-IP tunnels
* Extend the selftest
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Test that all possible combinations of inner and outer ECN bits result
in the correct inner ECN marking according to RFC 6040 4.2.
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cited commit changed the behavior of the software data path with regards
to the ECN marking of decapsulated packets. However, the commit did not
change other callers of __INET_ECN_decapsulate(), namely mlxsw. The
driver is using the function in order to ensure that the hardware and
software data paths act the same with regards to the ECN marking of
decapsulated packets.
The discrepancy was uncovered by commit 5aa3c334a4 ("selftests:
forwarding: vxlan_bridge_1d: Fix vxlan ecn decapsulate value") that
aligned the selftest to the new behavior. Without this patch the
selftest passes when used with veth pairs, but fails when used with
mlxsw netdevs.
Fix this by instructing the device to propagate the ECT(1) mark from the
outer header to the inner header when the inner header is ECT(0), for
both NVE and IP-in-IP tunnels.
A helper is added in order not to duplicate the code between both tunnel
types.
Fixes: b723748750 ("tunnel: Propagate ECT(1) when decapsulating as recommended by RFC6040")
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit ea7800565a ("can: add optional DLC element to Classical
CAN frame structure") the struct can_frame::can_dlc was put into an
anonymous union with another u8 variable.
For various reasons some members in struct can_frame and canfd_frame
including the first 8 byes of data are expected to have the same
memory layout. This is enforced by a BUILD_BUG_ON check in af_can.c.
Since the above mentioned commit this check fails on ARM kernels
compiled with the ARM OABI (which means CONFIG_AEABI not set). In this
case -mabi=apcs-gnu is passed to the compiler, which leads to a
structure size boundary of 32, instead of 8 compared to CONFIG_AEABI
enabled. This means the the union in struct can_frame takes 4 bytes
instead of the expected 1.
Rong Chen illustrates the problem with pahole in the ARM OABI case:
| struct can_frame {
| canid_t can_id; /* 0 4 */
| union {
| __u8 len; /* 4 1 */
| __u8 can_dlc; /* 4 1 */
| }; /* 4 4 */
| __u8 __pad; /* 8 1 */
| __u8 __res0; /* 9 1 */
| __u8 len8_dlc; /* 10 1 */
|
| /* XXX 5 bytes hole, try to pack */
|
| __u8 data[8]
| __attribute__((__aligned__(8))); /* 16 8 */
|
| /* size: 24, cachelines: 1, members: 6 */
| /* sum members: 19, holes: 1, sum holes: 5 */
| /* forced alignments: 1, forced holes: 1, sum forced holes: 5 */
| /* last cacheline: 24 bytes */
| } __attribute__((__aligned__(8)));
Marking the anonymous union as __attribute__((packed)) fixes the
BUILD_BUG_ON problem on these compilers.
Fixes: ea7800565a ("can: add optional DLC element to Classical CAN frame structure")
Reported-by: kernel test robot <lkp@intel.com>
Suggested-by: Rong Chen <rong.a.chen@intel.com>
Link: https://lore.kernel.org/linux-can/2c82ec23-3551-61b5-1bd8-178c3407ee83@hartkopp.net/
Link: https://lore.kernel.org/r/20210325125850.1620-3-socketcan@hartkopp.net
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
In pvc_xmit, if __skb_pad(skb, pad, false) failed, it will free
the skb in the first time and goto drop. But the same skb is freed
by kfree_skb(skb) in the second time in drop.
Maintaining the original function unchanged, my patch adds a new
label out to avoid the double free if __skb_pad() failed.
Fixes: f5083d0cee ("drivers/net/wan/hdlc_fr: Improvements to the code of pvc_xmit")
Signed-off-by: Lv Yunlong <lyl2019@mail.ustc.edu.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2021-03-25
This series contains updates to virtchnl header file and i40e driver.
Norbert removes added padding from virtchnl RSS structures as this
causes issues when iterating over the arrays.
Mateusz adds Asym_Pause as supported to allow these settings to be set
as the hardware supports it.
Eryk fixes an issue where encountering a VF reset alongside releasing
VFs could cause a call trace.
Arkadiusz moves TC setup before resource setup as previously it was
possible to enter with a null q_vector causing a kernel oops.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Antoine Tenart says:
====================
net: do not modify the shared tunnel info when PMTU triggers an ICMP reply
The series fixes an issue were a shared ip_tunnel_info is modified when
PMTU triggers an ICMP reply in vxlan and geneve, making following
packets in that flow to have a wrong destination address if the flow
isn't updated. A detailled information is given in each of the two
commits.
This was tested manually with OVS and I ran the PTMU selftests with
kmemleak enabled (all OK, none was skipped).
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
When the interface is part of a bridge or an Open vSwitch port and a
packet exceed a PMTU estimate, an ICMP reply is sent to the sender. When
using the external mode (collect metadata) the source and destination
addresses are reversed, so that Open vSwitch can match the packet
against an existing (reverse) flow.
But inverting the source and destination addresses in the shared
ip_tunnel_info will make following packets of the flow to use a wrong
destination address (packets will be tunnelled to itself), if the flow
isn't updated. Which happens with Open vSwitch, until the flow times
out.
Fixes this by uncloning the skb's ip_tunnel_info before inverting its
source and destination addresses, so that the modification will only be
made for the PTMU packet, not the following ones.
Fixes: c1a800e88d ("geneve: Support for PMTU discovery on directly bridged links")
Tested-by: Eelco Chaudron <echaudro@redhat.com>
Reviewed-by: Eelco Chaudron <echaudro@redhat.com>
Signed-off-by: Antoine Tenart <atenart@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When the interface is part of a bridge or an Open vSwitch port and a
packet exceed a PMTU estimate, an ICMP reply is sent to the sender. When
using the external mode (collect metadata) the source and destination
addresses are reversed, so that Open vSwitch can match the packet
against an existing (reverse) flow.
But inverting the source and destination addresses in the shared
ip_tunnel_info will make following packets of the flow to use a wrong
destination address (packets will be tunnelled to itself), if the flow
isn't updated. Which happens with Open vSwitch, until the flow times
out.
Fixes this by uncloning the skb's ip_tunnel_info before inverting its
source and destination addresses, so that the modification will only be
made for the PTMU packet, not the following ones.
Fixes: fc68c99577 ("vxlan: Support for PMTU discovery on directly bridged links")
Tested-by: Eelco Chaudron <echaudro@redhat.com>
Reviewed-by: Eelco Chaudron <echaudro@redhat.com>
Signed-off-by: Antoine Tenart <atenart@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Xiaoming Ni says:
====================
nfc: fix Resource leakage and endless loop
fix Resource leakage and endless loop in net/nfc/llcp_sock.c,
reported by "kiyin(尹亮)".
Link: https://www.openwall.com/lists/oss-security/2020/11/01/1
====================
math: Export mul_u64_u64_div_u64
Fixes: f51d7bf1db ("ptp_qoriq: fix overflow in ptp_qoriq_adjfine() u64 calcalation")
Signed-off-by: David S. Miller <davem@davemloft.net>
When sock_wait_state() returns -EINPROGRESS, "sk->sk_state" is
LLCP_CONNECTING. In this case, llcp_sock_connect() is repeatedly invoked,
nfc_llcp_sock_link() will add sk to local->connecting_sockets twice.
sk->sk_node->next will point to itself, that will make an endless loop
and hang-up the system.
To fix it, check whether sk->sk_state is LLCP_CONNECTING in
llcp_sock_connect() to avoid repeated invoking.
Fixes: b4011239a0 ("NFC: llcp: Fix non blocking sockets connections")
Reported-by: "kiyin(尹亮)" <kiyin@tencent.com>
Link: https://www.openwall.com/lists/oss-security/2020/11/01/1
Cc: <stable@vger.kernel.org> #v3.11
Signed-off-by: Xiaoming Ni <nixiaoming@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Based on the IOMMU configuration, the current cache control settings can
result in possible coherency issues. The hardware team has recommended
new settings for the PCI device path to eliminate the issue.
Fixes: 6f595959c0 ("amd-xgbe: Adjust register settings to improve performance")
Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
Acked-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The xMII interface clock depends on the PHY interface (MII, RMII, RGMII)
as well as the current link speed. Explicitly configure the GSWIP to
automatically select the appropriate xMII interface clock.
This fixes an issue seen by some users where ports using an external
RMII or RGMII PHY were deaf (no RX or TX traffic could be seen). Most
likely this is due to an "invalid" xMII clock being selected either by
the bootloader or hardware-defaults.
Fixes: 14fceff477 ("net: dsa: Add Lantiq / Intel DSA driver for vrx200")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Correct the Micrel phy documentation for the ksz9021 and ksz9031 phys
for how the phy skews are set.
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
In setups with fixed-link settings there is no mdio node in DTS.
axienet_probe() already handles that gracefully but lp->mii_bus is
then NULL.
Fix code that tries to blindly grab the MDIO lock by introducing two helper
functions that make the locking conditional.
Signed-off-by: Daniel Mack <daniel@zonque.org>
Reviewed-by: Radhey Shyam Pandey <radhey.shyam.pandey@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
DSA is aware of switches with global VLAN filtering since the blamed
commit, but it makes a bad decision when multiple bridges are spanning
the same switch:
ip link add br0 type bridge vlan_filtering 1
ip link add br1 type bridge vlan_filtering 1
ip link set swp2 master br0
ip link set swp3 master br0
ip link set swp4 master br1
ip link set swp5 master br1
ip link set swp5 nomaster
ip link set swp4 nomaster
[138665.939930] sja1105 spi0.1: port 3: dsa_core: VLAN filtering is a global setting
[138665.947514] DSA: failed to notify DSA_NOTIFIER_BRIDGE_LEAVE
When all ports leave br1, DSA blindly attempts to disable VLAN filtering
on the switch, ignoring the fact that br0 still exists and is VLAN-aware
too. It fails while doing that.
This patch checks whether any port exists at all and is under a
VLAN-aware bridge.
Fixes: d371b7c92d ("net: dsa: Unset vlan_filtering when ports leave the bridge")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Setup TC before the i40e_setup_pf_switch() call.
Memory must be initialized for all the queues
before using its resources.
Previously it could be possible that a call:
xdp_rxq_info_reg(&rx_ring->xdp_rxq, rx_ring->netdev,
rx_ring->queue_index, rx_ring->q_vector->napi.napi_id);
was made with q_vector being null.
Oops could show up with the following sequence:
- no driver loaded
- FW LLDP agent is on (flag disable-fw-lldp:off)
- link is up
- DCB configured with number of Traffic Classes that will not divide
completely the default number of queues (usually cpu cores)
- driver load
- set private flag: disable-fw-lldp:on
Fixes: 4b208eaa80 ("i40e: Add init and default config of software based DCB")
Fixes: b02e5a0ebb ("xsk: Propagate napi_id to XDP socket Rx path")
Signed-off-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Signed-off-by: Arkadiusz Kubalewski <arkadiusz.kubalewski@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Fix the reason of kernel oops when i40e driver removed VFs.
Added new __I40E_VFS_RELEASING state to signalize releasing
process by PF, that it makes possible to exit of reset VF procedure.
Without this patch, it is possible to suspend the VFs reset by
releasing VFs resources procedure. Retrying the reset after the
timeout works on the freed VF memory causing a kernel oops.
Fixes: d43d60e5eb ("i40e: ensure reset occurs when disabling VF")
Signed-off-by: Eryk Rybak <eryk.roch.rybak@intel.com>
Signed-off-by: Grzegorz Szczurek <grzegorzx.szczurek@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Merge misc fixes from Andrew Morton:
"14 patches.
Subsystems affected by this patch series: mm (hugetlb, kasan, gup,
selftests, z3fold, kfence, memblock, and highmem), squashfs, ia64,
gcov, and mailmap"
* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
mailmap: update Andrey Konovalov's email address
mm/highmem: fix CONFIG_DEBUG_KMAP_LOCAL_FORCE_MAP
mm: memblock: fix section mismatch warning again
kfence: make compatible with kmemleak
gcov: fix clang-11+ support
ia64: fix format strings for err_inject
ia64: mca: allocate early mca with GFP_ATOMIC
squashfs: fix xattr id and id lookup sanity checks
squashfs: fix inode lookup sanity checks
z3fold: prevent reclaim/free race for headless pages
selftests/vm: fix out-of-tree build
mm/mmu_notifiers: ensure range_end() is paired with range_start()
kasan: fix per-page tags for non-page_alloc pages
hugetlb_cgroup: fix imbalanced css_get and css_put pair for shared mappings
Pull rdma fixes from Jason Gunthorpe:
"Not much going on, just some small bug fixes:
- Typo causing a regression in mlx5 devx
- Regression in the recent hns rework causing the HW to get out of
sync
- Long-standing cxgb4 adaptor crash when destroying cm ids"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
RDMA/cxgb4: Fix adapter LE hash errors while destroying ipv6 listening server
RDMA/hns: Fix bug during CMDQ initialization
RDMA/mlx5: Fix typo in destroy_mkey inbox
Pull arm64 fixes from Will Deacon:
"Minor fixes all over, ranging from typos to tests to errata
workarounds:
- Fix possible memory hotplug failure with KASLR
- Fix FFR value in SVE kselftest
- Fix backtraces reported in /proc/$pid/stack
- Disable broken CnP implementation on NVIDIA Carmel
- Typo fixes and ACPI documentation clarification
- Fix some W=1 warnings"
* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
arm64: kernel: disable CNP on Carmel
arm64/process.c: fix Wmissing-prototypes build warnings
kselftest/arm64: sve: Do not use non-canonical FFR register value
arm64: mm: correct the inside linear map range during hotplug check
arm64: kdump: update ppos when reading elfcorehdr
arm64: cpuinfo: Fix a typo
Documentation: arm64/acpi : clarify arm64 support of IBFT
arm64: stacktrace: don't trace arch_stack_walk()
arm64: csum: cast to the proper type
Remove padding from RSS structures. Previous layout
could lead to unwanted compiler optimizations
in loops when iterating over key and lut arrays.
Fixes: 65ece6de01 ("virtchnl: Add missing explicit padding to structures")
Signed-off-by: Norbert Ciosek <norbertx.ciosek@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Commit 34dc2efb39 ("memblock: fix section mismatch warning") marked
memblock_bottom_up() and memblock_set_bottom_up() as __init, but they
could be referenced from non-init functions like
memblock_find_in_range_node() on architectures that enable
CONFIG_ARCH_KEEP_MEMBLOCK.
For such builds kernel test robot reports:
WARNING: modpost: vmlinux.o(.text+0x74fea4): Section mismatch in reference from the function memblock_find_in_range_node() to the function .init.text:memblock_bottom_up()
The function memblock_find_in_range_node() references the function __init memblock_bottom_up().
This is often because memblock_find_in_range_node lacks a __init annotation or the annotation of memblock_bottom_up is wrong.
Replace __init annotations with __init_memblock annotations so that the
appropriate section will be selected depending on
CONFIG_ARCH_KEEP_MEMBLOCK.
Link: https://lore.kernel.org/lkml/202103160133.UzhgY0wt-lkp@intel.com
Link: https://lkml.kernel.org/r/20210316171347.14084-1-rppt@kernel.org
Fixes: 34dc2efb39 ("memblock: fix section mismatch warning")
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Reported-by: kernel test robot <lkp@intel.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Because memblock allocations are registered with kmemleak, the KFENCE
pool was seen by kmemleak as one large object. Later allocations
through kfence_alloc() that were registered with kmemleak via
slab_post_alloc_hook() would then overlap and trigger a warning.
Therefore, once the pool is initialized, we can remove (free) it from
kmemleak again, since it should be treated as allocator-internal and be
seen as "free memory".
The second problem is that kmemleak is passed the rounded size, and not
the originally requested size, which is also the size of KFENCE objects.
To avoid kmemleak scanning past the end of an object and trigger a
KFENCE out-of-bounds error, fix the size if it is a KFENCE object.
For simplicity, to avoid a call to kfence_ksize() in
slab_post_alloc_hook() (and avoid new IS_ENABLED(CONFIG_DEBUG_KMEMLEAK)
guard), just call kfence_ksize() in mm/kmemleak.c:create_object().
Link: https://lkml.kernel.org/r/20210317084740.3099921-1-elver@google.com
Signed-off-by: Marco Elver <elver@google.com>
Reported-by: Luis Henriques <lhenriques@suse.de>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Tested-by: Luis Henriques <lhenriques@suse.de>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Konovalov <andreyknvl@google.com>
Cc: Jann Horn <jannh@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Fix warning with %lx / u64 mismatch:
arch/ia64/kernel/err_inject.c: In function 'show_resources':
arch/ia64/kernel/err_inject.c:62:22: warning:
format '%lx' expects argument of type 'long unsigned int',
but argument 3 has type 'u64' {aka 'long long unsigned int'}
62 | return sprintf(buf, "%lx", name[cpu]); \
| ^~~~~~~
Link: https://lkml.kernel.org/r/20210313104312.1548232-1-slyfox@gentoo.org
Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The sleep warning happens at early boot right at secondary CPU
activation bootup:
smp: Bringing up secondary CPUs ...
BUG: sleeping function called from invalid context at mm/page_alloc.c:4942
in_atomic(): 0, irqs_disabled(): 1, non_block: 0, pid: 0, name: swapper/1
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.12.0-rc2-00007-g79e228d0b611-dirty #99
..
Call Trace:
show_stack+0x90/0xc0
dump_stack+0x150/0x1c0
___might_sleep+0x1c0/0x2a0
__might_sleep+0xa0/0x160
__alloc_pages_nodemask+0x1a0/0x600
alloc_page_interleave+0x30/0x1c0
alloc_pages_current+0x2c0/0x340
__get_free_pages+0x30/0xa0
ia64_mca_cpu_init+0x2d0/0x3a0
cpu_init+0x8b0/0x1440
start_secondary+0x60/0x700
start_ap+0x750/0x780
Fixed BSP b0 value from CPU 1
As I understand interrupts are not enabled yet and system has a lot of
memory. There is little chance to sleep and switch to GFP_ATOMIC should
be a no-op.
Link: https://lkml.kernel.org/r/20210315085045.204414-1-slyfox@gentoo.org
Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
If one or more notifiers fails .invalidate_range_start(), invoke
.invalidate_range_end() for "all" notifiers. If there are multiple
notifiers, those that did not fail are expecting _start() and _end() to
be paired, e.g. KVM's mmu_notifier_count would become imbalanced.
Disallow notifiers that can fail _start() from implementing _end() so
that it's unnecessary to either track which notifiers rejected _start(),
or had already succeeded prior to a failed _start().
Note, the existing behavior of calling _start() on all notifiers even
after a previous notifier failed _start() was an unintented "feature".
Make it canon now that the behavior is depended on for correctness.
As of today, the bug is likely benign:
1. The only caller of the non-blocking notifier is OOM kill.
2. The only notifiers that can fail _start() are the i915 and Nouveau
drivers.
3. The only notifiers that utilize _end() are the SGI UV GRU driver
and KVM.
4. The GRU driver will never coincide with the i195/Nouveau drivers.
5. An imbalanced kvm->mmu_notifier_count only causes soft lockup in the
_guest_, and the guest is already doomed due to being an OOM victim.
Fix the bug now to play nice with future usage, e.g. KVM has a
potential use case for blocking memslot updates in KVM while an
invalidation is in-progress, and failure to unblock would result in said
updates being blocked indefinitely and hanging.
Found by inspection. Verified by adding a second notifier in KVM that
periodically returns -EAGAIN on non-blockable ranges, triggering OOM,
and observing that KVM exits with an elevated notifier count.
Link: https://lkml.kernel.org/r/20210311180057.1582638-1-seanjc@google.com
Fixes: 93065ac753 ("mm, oom: distinguish blockable mode for mmu notifiers")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Suggested-by: Jason Gunthorpe <jgg@ziepe.ca>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ben Gardon <bgardon@google.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: "Jérôme Glisse" <jglisse@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Dimitri Sivanich <dimitri.sivanich@hpe.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
To allow performing tag checks on page_alloc addresses obtained via
page_address(), tag-based KASAN modes store tags for page_alloc
allocations in page->flags.
Currently, the default tag value stored in page->flags is 0x00.
Therefore, page_address() returns a 0x00ffff... address for pages that
were not allocated via page_alloc.
This might cause problems. A particular case we encountered is a
conflict with KFENCE. If a KFENCE-allocated slab object is being freed
via kfree(page_address(page) + offset), the address passed to kfree()
will get tagged with 0x00 (as slab pages keep the default per-page
tags). This leads to is_kfence_address() check failing, and a KFENCE
object ending up in normal slab freelist, which causes memory
corruptions.
This patch changes the way KASAN stores tag in page-flags: they are now
stored xor'ed with 0xff. This way, KASAN doesn't need to initialize
per-page flags for every created page, which might be slow.
With this change, page_address() returns natively-tagged (with 0xff)
pointers for pages that didn't have tags set explicitly.
This patch fixes the encountered conflict with KFENCE and prevents more
similar issues that can occur in the future.
Link: https://lkml.kernel.org/r/1a41abb11c51b264511d9e71c303bb16d5cb367b.1615475452.git.andreyknvl@google.com
Fixes: 2813b9c029 ("kasan, mm, arm64: tag non slab memory allocated via pagealloc")
Signed-off-by: Andrey Konovalov <andreyknvl@google.com>
Reviewed-by: Marco Elver <elver@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Peter Collingbourne <pcc@google.com>
Cc: Evgenii Stepanov <eugenis@google.com>
Cc: Branislav Rankov <Branislav.Rankov@arm.com>
Cc: Kevin Brodsky <kevin.brodsky@arm.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The current implementation of hugetlb_cgroup for shared mappings could
have different behavior. Consider the following two scenarios:
1.Assume initial css reference count of hugetlb_cgroup is 1:
1.1 Call hugetlb_reserve_pages with from = 1, to = 2. So css reference
count is 2 associated with 1 file_region.
1.2 Call hugetlb_reserve_pages with from = 2, to = 3. So css reference
count is 3 associated with 2 file_region.
1.3 coalesce_file_region will coalesce these two file_regions into
one. So css reference count is 3 associated with 1 file_region
now.
2.Assume initial css reference count of hugetlb_cgroup is 1 again:
2.1 Call hugetlb_reserve_pages with from = 1, to = 3. So css reference
count is 2 associated with 1 file_region.
Therefore, we might have one file_region while holding one or more css
reference counts. This inconsistency could lead to imbalanced css_get()
and css_put() pair. If we do css_put one by one (i.g. hole punch case),
scenario 2 would put one more css reference. If we do css_put all
together (i.g. truncate case), scenario 1 will leak one css reference.
The imbalanced css_get() and css_put() pair would result in a non-zero
reference when we try to destroy the hugetlb cgroup. The hugetlb cgroup
directory is removed __but__ associated resource is not freed. This
might result in OOM or can not create a new hugetlb cgroup in a busy
workload ultimately.
In order to fix this, we have to make sure that one file_region must
hold exactly one css reference. So in coalesce_file_region case, we
should release one css reference before coalescence. Also only put css
reference when the entire file_region is removed.
The last thing to note is that the caller of region_add() will only hold
one reference to h_cg->css for the whole contiguous reservation region.
But this area might be scattered when there are already some
file_regions reside in it. As a result, many file_regions may share only
one h_cg->css reference. In order to ensure that one file_region must
hold exactly one css reference, we should do css_get() for each
file_region and release the reference held by caller when they are done.
[linmiaohe@huawei.com: fix imbalanced css_get and css_put pair for shared mappings]
Link: https://lkml.kernel.org/r/20210316023002.53921-1-linmiaohe@huawei.com
Link: https://lkml.kernel.org/r/20210301120540.37076-1-linmiaohe@huawei.com
Fixes: 075a61d07a ("hugetlb_cgroup: add accounting for shared mappings")
Reported-by: kernel test robot <lkp@intel.com> (auto build test ERROR)
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Wanpeng Li <liwp.linux@gmail.com>
Cc: Mina Almasry <almasrymina@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
