Commmit f5ea16137a3f ("NFSv4: Retry LOCK on OLD_STATEID during delegation
return") attempted to solve this problem by using nfs4's generic async error
handling, but introduced a regression where v4.0 lock recovery would hang.
The additional complexity introduced by overloading that error handling is
not necessary for this case. This patch expects that commit to be
reverted.
The problem as originally explained in the above commit is:
There's a small window where a LOCK sent during a delegation return can
race with another OPEN on client, but the open stateid has not yet been
updated. In this case, the client doesn't handle the OLD_STATEID error
from the server and will lose this lock, emitting:
"NFS: nfs4_handle_delegation_recall_error: unhandled error -10024".
Fix this by using the old_stateid refresh helpers if the server replies
with OLD_STATEID.
The Lenovo Thinkbook 14s Yoga ITL has 4 new symbols/shortcuts on their
F9-F11 and PrtSc keys:
F9: Has a symbol of a head with a headset, the manual says "Service key"
F10: Has a symbol of a telephone horn which has been picked up from the
receiver, the manual says: "Answer incoming calls"
F11: Has a symbol of a telephone horn which is resting on the receiver,
the manual says: "Reject incoming calls"
PrtSc: Has a symbol of a siccor and a dashed ellipse, the manual says:
"Open the Windows 'Snipping' Tool app"
This commit adds support for these 4 new hkey events.
The commit 06470f7468c8 ("mac80211: add API to allow filtering frames in BA sessions")
added reorder_buf_filtered to mark frames filtered by firmware, and it
can only work correctly if hw.max_rx_aggregation_subframes <= 64 since
it stores the bitmap in a u64 variable.
However, new HE or EHT devices can support BlockAck number up to 256 or
1024, and then using a higher subframe index leads UBSAN warning:
UBSAN: shift-out-of-bounds in net/mac80211/rx.c:1129:39
shift exponent 215 is too large for 64-bit type 'long long unsigned int'
Call Trace:
<IRQ>
dump_stack_lvl+0x48/0x70
dump_stack+0x10/0x20
__ubsan_handle_shift_out_of_bounds+0x1ac/0x360
ieee80211_release_reorder_frame.constprop.0.cold+0x64/0x69 [mac80211]
ieee80211_sta_reorder_release+0x9c/0x400 [mac80211]
ieee80211_prepare_and_rx_handle+0x1234/0x1420 [mac80211]
ieee80211_rx_list+0xaef/0xf60 [mac80211]
ieee80211_rx_napi+0x53/0xd0 [mac80211]
Since only old hardware that supports <=64 BlockAck uses
ieee80211_mark_rx_ba_filtered_frames(), limit the use as it is, so add a
WARN_ONCE() and comment to note to avoid using this function if hardware
capability is not suitable.
When building for power4, newer binutils don't recognise the "dcbfl"
extended mnemonic.
dcbfl RA, RB is equivalent to dcbf RA, RB, 1.
Switch to "dcbf" to avoid the build error.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current analog gain TLV seems to have completely incorrect values in
it. The gain starts at 0.5dB, proceeds in 1dB steps, and has no mute
value, correct the control to match.
If the target ring is configured with IOPOLL, then we always need to hold
the target ring uring_lock before posting CQEs. We could just grab it
unconditionally, but since we don't expect many target rings to be of this
type, make grabbing the uring_lock conditional on the ring type.
We don't need to take both uring_locks at once, msg_ring can be split in
two parts, first getting a file from the filetable of the first ring and
then installing it into the second one.
KVM: x86/mmu: Fix an sign-extension bug with mmu_seq that hangs vCPUs
Upstream commit ba6e3fe25543 ("KVM: x86/mmu: Grab mmu_invalidate_seq in
kvm_faultin_pfn()") unknowingly fixed the bug in v6.3 when refactoring
how KVM tracks the sequence counter snapshot.
Take the vCPU's mmu_seq snapshot as an "unsigned long" instead of an "int"
when checking to see if a page fault is stale, as the sequence count is
stored as an "unsigned long" everywhere else in KVM. This fixes a bug
where KVM will effectively hang vCPUs due to always thinking page faults
are stale, which results in KVM refusing to "fix" faults.
mmu_invalidate_seq (née mmu_notifier_seq) is a sequence counter used when
KVM is handling page faults to detect if userspace mappings relevant to
the guest were invalidated between snapshotting the counter and acquiring
mmu_lock, i.e. to ensure that the userspace mapping KVM is using to
resolve the page fault is fresh. If KVM sees that the counter has
changed, KVM simply resumes the guest without fixing the fault.
What _should_ happen is that the source of the mmu_notifier invalidations
eventually goes away, mmu_invalidate_seq becomes stable, and KVM can once
again fix guest page fault(s).
But for a long-lived VM and/or a VM that the host just doesn't particularly
like, it's possible for a VM to be on the receiving end of 2 billion (with
a B) mmu_notifier invalidations. When that happens, bit 31 will be set in
mmu_invalidate_seq. This causes the value to be turned into a 32-bit
negative value when implicitly cast to an "int" by is_page_fault_stale(),
and then sign-extended into a 64-bit unsigned when the signed "int" is
implicitly cast back to an "unsigned long" on the call to
mmu_invalidate_retry_hva().
As a result of the casting and sign-extension, given a sequence counter of
e.g. 0x8002dc25, mmu_invalidate_retry_hva() ends up doing
if (0x8002dc25 != 0xffffffff8002dc25)
and signals that the page fault is stale and needs to be retried even
though the sequence counter is stable, and KVM effectively hangs any vCPU
that takes a page fault (EPT violation or #NPF when TDP is enabled).
Reported-by: Brian Rak <brak@vultr.com> Reported-by: Amaan Cheval <amaan.cheval@gmail.com> Reported-by: Eric Wheeler <kvm@lists.ewheeler.net> Closes: https://lore.kernel.org/all/f023d927-52aa-7e08-2ee5-59a2fbc65953@gameservers.com Fixes: a955cad84cda ("KVM: x86/mmu: Retry page fault if root is invalidated by memslot update") Signed-off-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Preserve TDP MMU roots until they are explicitly invalidated by gifting
the TDP MMU itself a reference to a root when it is allocated. Keeping a
reference in the TDP MMU fixes a flaw where the TDP MMU exhibits terrible
performance, and can potentially even soft-hang a vCPU, if a vCPU
frequently unloads its roots, e.g. when KVM is emulating SMI+RSM.
When KVM emulates something that invalidates _all_ TLB entries, e.g. SMI
and RSM, KVM unloads all of the vCPUs roots (KVM keeps a small per-vCPU
cache of previous roots). Unloading roots is a simple way to ensure KVM
flushes and synchronizes all roots for the vCPU, as KVM flushes and syncs
when allocating a "new" root (from the vCPU's perspective).
In the shadow MMU, KVM keeps track of all shadow pages, roots included, in
a per-VM hash table. Unloading a shadow MMU root just wipes it from the
per-vCPU cache; the root is still tracked in the per-VM hash table. When
KVM loads a "new" root for the vCPU, KVM will find the old, unloaded root
in the per-VM hash table.
Unlike the shadow MMU, the TDP MMU doesn't track "inactive" roots in a
per-VM structure, where "active" in this case means a root is either
in-use or cached as a previous root by at least one vCPU. When a TDP MMU
root becomes inactive, i.e. the last vCPU reference to the root is put,
KVM immediately frees the root (asterisk on "immediately" as the actual
freeing may be done by a worker, but for all intents and purposes the root
is gone).
The TDP MMU behavior is especially problematic for 1-vCPU setups, as
unloading all roots effectively frees all roots. The issue is mitigated
to some degree in multi-vCPU setups as a different vCPU usually holds a
reference to an unloaded root and thus keeps the root alive, allowing the
vCPU to reuse its old root after unloading (with a flush+sync).
The TDP MMU flaw has been known for some time, as until very recently,
KVM's handling of CR0.WP also triggered unloading of all roots. The
CR0.WP toggling scenario was eventually addressed by not unloading roots
when _only_ CR0.WP is toggled, but such an approach doesn't Just Work
for emulating SMM as KVM must emulate a full TLB flush on entry and exit
to/from SMM. Given that the shadow MMU plays nice with unloading roots
at will, teaching the TDP MMU to do the same is far less complex than
modifying KVM to track which roots need to be flushed before reuse.
Note, preserving all possible TDP MMU roots is not a concern with respect
to memory consumption. Now that the role for direct MMUs doesn't include
information about the guest, e.g. CR0.PG, CR0.WP, CR4.SMEP, etc., there
are _at most_ six possible roots (where "guest_mode" here means L2):
And because each vCPU can track 4 valid roots, a VM can already have all
6 root combinations live at any given time. Not to mention that, in
practice, no sane VMM will advertise different guest.MAXPHYADDR values
across vCPUs, i.e. KVM won't ever use both 4-level and 5-level roots for
a single VM. Furthermore, the vast majority of modern hypervisors will
utilize EPT/NPT when available, thus the guest_mode=%true cases are also
unlikely to be utilized.
The commit 14af9963ba1e ("bonding: Support macvlans on top of tlb/rlb mode
bonds") aims to enable the use of macvlans on top of rlb bond mode. However,
the current rlb bond mode only handles ARP packets to update remote neighbor
entries. This causes an issue when a macvlan is on top of the bond, and
remote devices send packets to the macvlan using the bond's MAC address
as the destination. After delivering the packets to the macvlan, the macvlan
will rejects them as the MAC address is incorrect. Consequently, this commit
makes macvlan over bond non-functional.
To address this problem, one potential solution is to check for the presence
of a macvlan port on the bond device using netif_is_macvlan_port(bond->dev)
and return NULL in the rlb_arp_xmit() function. However, this approach
doesn't fully resolve the situation when a VLAN exists between the bond and
macvlan.
So let's just do a partial revert for commit 14af9963ba1e in rlb_arp_xmit().
As the comment said, Don't modify or load balance ARPs that do not originate
locally.
Fixes: 14af9963ba1e ("bonding: Support macvlans on top of tlb/rlb mode bonds") Reported-by: susan.zheng@veritas.com Closes: https://bugzilla.redhat.com/show_bug.cgi?id=2117816 Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Negative ifindexes are illegal, but the kernel does not validate the
ifindex in the ancillary header of RTM_NEWLINK messages, resulting in
the kernel generating a warning [1] when such an ifindex is specified.
Destroy work waits for the RCU grace period then it releases the objects
with no mutex held. All releases objects follow this path for
transactions, therefore, order is guaranteed and references to top-level
objects in the hierarchy remain valid.
However, netlink notifier might interfer with pending destroy work.
rcu_barrier() is not correct because objects are not release via RCU
callback. Flush destroy work before releasing objects from netlink
notifier path.
Add check for pf->vf not being NULL before dereferencing
pf->vf[vsi->vf_id] in updating VSI filter sync.
Add a similar check before dereferencing !pf->vf[vsi->vf_id].trusted
in the condition for clearing promisc mode bit.
Fixes: c87c938f62d8 ("i40e: Add VF VLAN pruning") Signed-off-by: Andrii Staikov <andrii.staikov@intel.com> Signed-off-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Tested-by: Rafal Romanowski <rafal.romanowski@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
When replacing an existing root qdisc, with one that is of the same kind, the
request boils down to essentially a parameterization change i.e not one that
requires allocation and grafting of a new qdisc. syzbot was able to create a
scenario which resulted in a taprio qdisc replacing an existing taprio qdisc
with a combination of NLM_F_CREATE, NLM_F_REPLACE and NLM_F_EXCL leading to
create and graft scenario.
The fix ensures that only when the qdisc kinds are different that we should
allow a create and graft, otherwise it goes into the "change" codepath.
While at it, fix the code and comments to improve readability.
While syzbot was able to create the issue, it did not zone on the root cause.
Analysis from Vladimir Oltean <vladimir.oltean@nxp.com> helped narrow it down.
v1->V2 changes:
- remove "inline" function definition (Vladmir)
- remove extrenous braces in branches (Vladmir)
- change inline function names (Pedro)
- Run tdc tests (Victor)
v2->v3 changes:
- dont break else/if (Simon)
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: syzbot+a3618a167af2021433cd@syzkaller.appspotmail.com Closes: https://lore.kernel.org/netdev/20230816225759.g25x76kmgzya2gei@skbuf/T/ Tested-by: Vladimir Oltean <vladimir.oltean@nxp.com> Tested-by: Victor Nogueira <victor@mojatatu.com> Reviewed-by: Pedro Tammela <pctammela@mojatatu.com> Reviewed-by: Victor Nogueira <victor@mojatatu.com> Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The IGC_PTM_CTRL_SHRT_CYC defines the time between two consecutive PTM
requests. The bit resolution of this field is six bits. That bit five was
missing in the mask. This patch comes to correct the typo in the
IGC_PTM_CTRL_SHRT_CYC macro.
Fixes: a90ec8483732 ("igc: Add support for PTP getcrosststamp()") Signed-off-by: Sasha Neftin <sasha.neftin@intel.com> Tested-by: Naama Meir <naamax.meir@linux.intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Reviewed-by: Simon Horman <horms@kernel.org> Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Link: https://lore.kernel.org/r/20230821171721.2203572-1-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The original implementation had a very simple handling for single frame
transmissions as it just sent the single frame without a timeout handling.
With the new echo frame handling the echo frame was also introduced for
single frames but the former exception ('simple without timers') has been
maintained by accident. This leads to a 1 second timeout when closing the
socket and to an -ECOMM error when CAN_ISOTP_WAIT_TX_DONE is selected.
As the echo handling is always active (also for single frames) remove the
wrong extra condition for single frames.
Fixes: 9f39d36530e5 ("can: isotp: add support for transmission without flow control") Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Link: https://lore.kernel.org/r/20230821144547.6658-2-socketcan@hartkopp.net Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Before adding a port to bond, it need to be set down first. In the
lacpdu test the author set the port down specifically. But commit a4abfa627c38 ("net: rtnetlink: Enslave device before bringing it up")
changed the operation order, the kernel will set the port down _after_
adding to bond. So all the ports will be down at last and the test failed.
In fact, the veth interfaces are already inactive when added. This
means there's no need to set them down again before adding to the bond.
Let's just remove the link down operation.
Fixes: a4abfa627c38 ("net: rtnetlink: Enslave device before bringing it up") Reported-by: Zhengchao Shao <shaozhengchao@huawei.com> Closes: https://lore.kernel.org/netdev/a0ef07c7-91b0-94bd-240d-944a330fcabd@huawei.com/ Signed-off-by: Hangbin Liu <liuhangbin@gmail.com> Link: https://lore.kernel.org/r/20230817082459.1685972-1-liuhangbin@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
During stress test with attaching and detaching VF from KVM and
simultaneously changing VFs spoofcheck and trust there was a
NULL pointer dereference in ice_reset_vf that VF's VSI is null.
More than one instance of ice_reset_vf() can be running at a given
time. When we rebuild the VSI in ice_reset_vf, another reset can be
triaged from ice_service_task. In this case we can access the currently
uninitialized VSI and cause panic. The window for this racing condition
has been around for a long time but it's much worse after commit 227bf4500aaa ("ice: move VSI delete outside deconfig") because
the reset runs faster. ice_reset_vf() using vf->cfg_lock and when
we move this lock before accessing to the VF VSI, we can fix
BUG for all cases.
Panic occurs sometimes in ice_vsi_is_rx_queue_active() and sometimes
in ice_vsi_stop_all_rx_rings()
With our reproducer, we can hit BUG:
~8h before commit 227bf4500aaa ("ice: move VSI delete outside deconfig").
~20m after commit 227bf4500aaa ("ice: move VSI delete outside deconfig").
After this fix we are not able to reproduce it after ~48h
There was commit cf90b74341ee ("ice: Fix call trace with null VSI during
VF reset") which also tried to fix this issue, but it was only
partially resolved and the bug still exists.
After this commit we are not able to attach VF to VM:
virsh attach-interface v0 hostdev --managed 0000:41:01.0 --mac 52:52:52:52:52:52
error: Failed to attach interface
error: Cannot set interface MAC to 52:52:52:52:52:52 for ifname enp65s0f0np0 vf 0: Resource temporarily unavailable
ice_check_vf_ready_for_cfg() already contain waiting for reset.
New condition in ice_check_vf_ready_for_reset() causing only problems.
Fixes: 7255355a0636 ("ice: Fix ice VF reset during iavf initialization") Signed-off-by: Petr Oros <poros@redhat.com> Reviewed-by: Simon Horman <horms@kernel.org> Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Rafal Romanowski <rafal.romanowski@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The driver is misconfiguring the hardware for some values of MTU such that
it could use multiple descriptors to receive a packet when it could have
simply used one.
Change the driver to use a round-up instead of the result of a shift, as
the shift can truncate the lower bits of the size, and result in the
problem noted above. It also aligns this driver with similar code in i40e.
The insidiousness of this problem is that everything works with the wrong
size, it's just not working as well as it could, as some MTU sizes end up
using two or more descriptors, and there is no way to tell that is
happening without looking at ice_trace or a bus analyzer.
Fixes: efc2214b6047 ("ice: Add support for XDP") Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com> Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com> Reviewed-by: Leon Romanovsky <leonro@nvidia.com> Tested-by: Pucha Himasekhar Reddy <himasekharx.reddy.pucha@intel.com> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 7804 Comm: syz-executor.1 Not tainted 6.5.0-rc6-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 07/26/2023
==================================================================
Fixes: 23f57406b82d ("ipv4: avoid using shared IP generator for connected sockets") Reported-by: syzbot <syzkaller@googlegroups.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Reviewed-by: David Ahern <dsahern@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
veth and vxcan need to make sure the ifindexes of the peer
are not negative, core does not validate this.
Using iproute2 with user-space-level checking removed:
Before:
# ./ip link add index 10 type veth peer index -1
# ip link show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: enp1s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP mode DEFAULT group default qlen 1000
link/ether 52:54:00:74:b2:03 brd ff:ff:ff:ff:ff:ff
10: veth1@veth0: <BROADCAST,MULTICAST,M-DOWN> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
link/ether 8a:90:ff:57:6d:5d brd ff:ff:ff:ff:ff:ff
-1: veth0@veth1: <BROADCAST,MULTICAST,M-DOWN> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
link/ether ae:ed:18:e6:fa:7f brd ff:ff:ff:ff:ff:ff
Now:
$ ./ip link add index 10 type veth peer index -1
Error: ifindex can't be negative.
This problem surfaced in net-next because an explicit WARN()
was added, the root cause is older.
Fixes: e6f8f1a739b6 ("veth: Allow to create peer link with given ifindex") Fixes: a8f820a380a2 ("can: add Virtual CAN Tunnel driver (vxcan)") Reported-by: syzbot+5ba06978f34abb058571@syzkaller.appspotmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Reviewed-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The fixed_phy_register() function returns error pointers and never
returns NULL. Update the checks accordingly.
Fixes: b0ba512e25d7 ("net: bcmgenet: enable driver to work without a device tree") Signed-off-by: Ruan Jinjie <ruanjinjie@huawei.com> Reviewed-by: Leon Romanovsky <leonro@nvidia.com> Acked-by: Doug Berger <opendmb@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The fixed_phy_register() function returns error pointers and never
returns NULL. Update the checks accordingly.
Fixes: c25b23b8a387 ("bgmac: register fixed PHY for ARM BCM470X / BCM5301X chipsets") Signed-off-by: Ruan Jinjie <ruanjinjie@huawei.com> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Reviewed-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
802.1X PAE frames are link-local frames, therefore they must be trapped to
the CPU port. Currently, the MT753X switches treat 802.1X PAE frames as
regular multicast frames, therefore flooding them to user ports. To fix
this, set 802.1X PAE frames to be trapped to the CPU port(s).
Fixes: b8f126a8d543 ("net-next: dsa: add dsa support for Mediatek MT7530 switch") Signed-off-by: Arınç ÜNAL <arinc.unal@arinc9.com> Reviewed-by: Vladimir Oltean <olteanv@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Remove assumptions about shared buffer cell size and instead query the
cell size from devlink. Adjust the test to send small packets that fit
inside a single cell.
Tested on Spectrum-{1,2,3,4}.
Fixes: 4735402173e6 ("mlxsw: spectrum: Extend to support Spectrum-4 ASIC") Signed-off-by: Ido Schimmel <idosch@nvidia.com> Reviewed-by: Petr Machata <petrm@nvidia.com> Signed-off-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/r/f7dfbf3c4d1cb23838d9eb99bab09afaa320c4ca.1692268427.git.petrm@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The field 'virtual router' was extended to 12 bits in Spectrum-4.
Therefore, the element 'MLXSW_AFK_ELEMENT_VIRT_ROUTER_MSB' needs 3 bits for
Spectrum < 4 and 4 bits for Spectrum >= 4.
The elements are stored in an internal storage scratchpad. Currently, the
MSB is defined there as 3 bits. It means that for Spectrum-4, only 2K VRFs
can be used for multicast routing, as the highest bit is not really used by
the driver. Fix the definition of 'VIRT_ROUTER_MSB' to use 4 bits. Adjust
the definitions of 'virtual router' field in the blocks accordingly - use
'_avoid_size_check' for Spectrum-2 instead of for Spectrum-4. Fix the mask
in parse function to use 4 bits.
Fixes: 6d5d8ebb881c ("mlxsw: Rename virtual router flex key element") Signed-off-by: Amit Cohen <amcohen@nvidia.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> Signed-off-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/r/79bed2b70f6b9ed58d4df02e9798a23da648015b.1692268427.git.petrm@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The two most significant bits of the "local_port" field in the SSPR
register are always cleared since they are overwritten by the deprecated
and overlapping "sub_port" field.
On systems with more than 255 local ports (e.g., Spectrum-4), this
results in the firmware maintaining invalid mappings between system port
and local port. Specifically, two different systems ports (0x1 and
0x101) point to the same local port (0x1), which eventually leads to
firmware errors.
Fix by removing the deprecated "sub_port" field.
Fixes: fd24b29a1b74 ("mlxsw: reg: Align existing registers to use extended local_port field") Signed-off-by: Ido Schimmel <idosch@nvidia.com> Signed-off-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/r/9b909a3033c8d3d6f67f237306bef4411c5e6ae4.1692268427.git.petrm@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently, in Spectrum-2 and above, time stamps are extracted from the CQE
into the time stamp fields in 'struct mlxsw_skb_cb', only when the CQE
time stamp type is UTC. The time stamps are read directly from the CQE and
software can get the time stamp in UTC format using CQEv2.
From Spectrum-4, the time stamps that are read from the CQE are allowed
to be also from MIRROR_UTC type.
Therefore, we get a warning [1] from the driver that the time stamp fields
were not set, when LLDP control packet is sent.
Allow the time stamp type to be MIRROR_UTC and set the time stamp in this
case as well.
There are two network devices(veth1 and veth3) in ns1, and ipvlan1 with
L3S mode and ipvlan2 with L2 mode are created based on them as
figure (1). In this case, ipvlan_register_nf_hook() will be called to
register nf hook which is needed by ipvlans in L3S mode in ns1 and value
of ipvl_nf_hook_refcnt is set to 1.
When veth3 migrates from ns1 to ns2 as figure (2), veth3 will register in
ns2 and calls call_netdevice_notifiers with NETDEV_REGISTER event:
dev_change_net_namespace
call_netdevice_notifiers
ipvlan_device_event
ipvlan_migrate_l3s_hook
ipvlan_register_nf_hook(newnet) (I)
ipvlan_unregister_nf_hook(oldnet) (II)
In function ipvlan_migrate_l3s_hook(), ipvl_nf_hook_refcnt in ns1 is not 0
since veth1 with ipvlan1 still in ns1, (I) and (II) will be called to
register nf_hook in ns2 and unregister nf_hook in ns1. As a result,
ipvl_nf_hook_refcnt in ns1 is decreased incorrectly and this in ns2
is increased incorrectly. When the second net namespace is removed, a
reference count leak warning in ipvlan_ns_exit() will be triggered.
This patch add a check before ipvlan_migrate_l3s_hook() is called. The
warning can be triggered as follows:
$ ip netns add ns1
$ ip netns add ns2
$ ip netns exec ns1 ip link add veth1 type veth peer name veth2
$ ip netns exec ns1 ip link add veth3 type veth peer name veth4
$ ip netns exec ns1 ip link add ipv1 link veth1 type ipvlan mode l3s
$ ip netns exec ns1 ip link add ipv2 link veth3 type ipvlan mode l2
$ ip netns exec ns1 ip link set veth3 netns ns2
$ ip net del ns2
Fixes: 3133822f5ac1 ("ipvlan: use pernet operations and restrict l3s hooks to master netns") Signed-off-by: Lu Wei <luwei32@huawei.com> Reviewed-by: Florian Westphal <fw@strlen.de> Link: https://lore.kernel.org/r/20230817145449.141827-1-luwei32@huawei.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The blamed commit resolved a bug where frames would still get stuck at
egress, even though they're smaller than the maxSDU[tc], because the
driver did not take into account the extra 33 ns that the queue system
needs for scheduling the frame.
It now takes that into account, but the arithmetic that we perform in
vsc9959_tas_remaining_gate_len_ps() is buggy, because we operate on
64-bit unsigned integers, so gate_len_ns - VSC9959_TAS_MIN_GATE_LEN_NS
may become a very large integer if gate_len_ns < 33 ns.
In practice, this means that we've introduced a regression where all
traffic class gates which are permanently closed will not get detected
by the driver, and we won't enable oversize frame dropping for them.
Before:
mscc_felix 0000:00:00.5: port 0: max frame size 1526 needs 12400000 ps, 1152000 ps for mPackets at speed 1000
mscc_felix 0000:00:00.5: port 0 tc 0 min gate len 1000000, sending all frames
mscc_felix 0000:00:00.5: port 0 tc 1 min gate len 0, sending all frames
mscc_felix 0000:00:00.5: port 0 tc 2 min gate len 0, sending all frames
mscc_felix 0000:00:00.5: port 0 tc 3 min gate len 0, sending all frames
mscc_felix 0000:00:00.5: port 0 tc 4 min gate len 0, sending all frames
mscc_felix 0000:00:00.5: port 0 tc 5 min gate len 0, sending all frames
mscc_felix 0000:00:00.5: port 0 tc 6 min gate len 0, sending all frames
mscc_felix 0000:00:00.5: port 0 tc 7 min gate length 5120 ns not enough for max frame size 1526 at 1000 Mbps, dropping frames over 615 octets including FCS
After:
mscc_felix 0000:00:00.5: port 0: max frame size 1526 needs 12400000 ps, 1152000 ps for mPackets at speed 1000
mscc_felix 0000:00:00.5: port 0 tc 0 min gate len 1000000, sending all frames
mscc_felix 0000:00:00.5: port 0 tc 1 min gate length 0 ns not enough for max frame size 1526 at 1000 Mbps, dropping frames over 1 octets including FCS
mscc_felix 0000:00:00.5: port 0 tc 2 min gate length 0 ns not enough for max frame size 1526 at 1000 Mbps, dropping frames over 1 octets including FCS
mscc_felix 0000:00:00.5: port 0 tc 3 min gate length 0 ns not enough for max frame size 1526 at 1000 Mbps, dropping frames over 1 octets including FCS
mscc_felix 0000:00:00.5: port 0 tc 4 min gate length 0 ns not enough for max frame size 1526 at 1000 Mbps, dropping frames over 1 octets including FCS
mscc_felix 0000:00:00.5: port 0 tc 5 min gate length 0 ns not enough for max frame size 1526 at 1000 Mbps, dropping frames over 1 octets including FCS
mscc_felix 0000:00:00.5: port 0 tc 6 min gate length 0 ns not enough for max frame size 1526 at 1000 Mbps, dropping frames over 1 octets including FCS
mscc_felix 0000:00:00.5: port 0 tc 7 min gate length 5120 ns not enough for max frame size 1526 at 1000 Mbps, dropping frames over 615 octets including FCS
Fixes: 11afdc6526de ("net: dsa: felix: tc-taprio intervals smaller than MTU should send at least one packet") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/r/20230817120111.3522827-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Cited fixes commit introduced linecard notifications for register,
however it didn't add them for unregister. Fix that by adding them.
Fixes: c246f9b5fd61 ("devlink: add support to create line card and expose to user") Signed-off-by: Jiri Pirko <jiri@nvidia.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/r/20230817125240.2144794-1-jiri@resnulli.us Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The devlink code is hard to navigate with 13kLoC in one file.
I really like the way Michal split the ethtool into per-command
files and core. It'd probably be too much to split it all up,
but we can at least separate the core parts out of the per-cmd
implementations and put it in a directory so that new commands
can be separate files.
Move the code, subsequent commit will do a partial split.
On SDP interfaces, frame oversize and undersize errors are
observed as driver is not considering packet sizes of all
subscribers of the link before updating the link config.
This patch fixes the same.
Fixes: 9b7dd87ac071 ("octeontx2-af: Support to modify min/max allowed packet lengths") Signed-off-by: Hariprasad Kelam <hkelam@marvell.com> Signed-off-by: Sunil Goutham <sgoutham@marvell.com> Reviewed-by: Leon Romanovsky <leonro@nvidia.com> Link: https://lore.kernel.org/r/20230817063006.10366-1-hkelam@marvell.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
__tracing_open() { // 1. File 'trace' is being opened;
...
*iter->trace = *tr->current_trace; // 2. Tracer 'function_graph' is
// currently set;
...
iter->trace->open(iter); // 3. Call graph_trace_open() here,
// and memory are allocated in it;
...
}
s_start() { // 4. The opened file is being read;
...
*iter->trace = *tr->current_trace; // 5. If tracer is switched to
// 'nop' or others, then memory
// in step 3 are leaked!!!
...
}
To fix it, in s_start(), close tracer before switching then reopen the
new tracer after switching. And some tracers like 'wakeup' may not update
'iter->private' in some cases when reopen, then it should be cleared
to avoid being mistakenly closed again.
The root cause is that:
1. After `echo 0 > tracing_cpumask`, 'record_disabled' of cpu buffers
in 'tr->array_buffer.buffer' became 1 (see tracing_set_cpumask());
2. After `echo 1 > snapshot`, 'tr->array_buffer.buffer' is swapped
with 'tr->max_buffer.buffer', then the 'record_disabled' became 0
(see update_max_tr());
3. After `echo fff > tracing_cpumask`, the 'record_disabled' become -1;
Then array_buffer and max_buffer are both unavailable due to value of
'record_disabled' is not 0.
To fix it, enable or disable both array_buffer and max_buffer at the same
time in tracing_set_cpumask().
Link: https://lkml.kernel.org/r/20230805033816.3284594-2-zhengyejian1@huawei.com Cc: <mhiramat@kernel.org> Cc: <vnagarnaik@google.com> Cc: <shuah@kernel.org> Fixes: 71babb2705e2 ("tracing: change CPU ring buffer state from tracing_cpumask") Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
We always assumed that a device might either have AUX or FLAT
CCS, but this is an approximation that is not always true, e.g.
PVC represents an exception.
Set the basis for future finer selection by implementing a
boolean gen12_needs_ccs_aux_inv() function that tells whether aux
invalidation is needed or not.
Currently PVC is the only exception to the above mentioned rule.
Requires: 059ae7ae2a1c ("drm/i915/gt: Cleanup aux invalidation registers") Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com> Cc: Matt Roper <matthew.d.roper@intel.com> Cc: Jonathan Cavitt <jonathan.cavitt@intel.com> Cc: <stable@vger.kernel.org> # v5.8+ Reviewed-by: Matt Roper <matthew.d.roper@intel.com> Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com> Reviewed-by: Nirmoy Das <nirmoy.das@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20230725001950.1014671-3-andi.shyti@linux.intel.com
(cherry picked from commit c827655b87ad201ebe36f2e28d16b5491c8f7801) Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The length information for available buffer space for CCA
replies is covered with two fields in the T6 header prepended
on each CCA reply: fromcardlen1 and fromcardlen2. The sum of
these both values must not exceed the AP bus limit for this
card (24KB for CEX8, 12KB CEX7 and older) minus the always
present headers.
The current code adjusted the fromcardlen2 value in case
of exceeding the AP bus limit when there was a non-zero
value given from userspace. Some tests now showed that this
was the wrong assumption. Instead the userspace value given for
this field should always be trusted and if the sum of the
two fields exceeds the AP bus limit for this card the first
field fromcardlen1 should be adjusted instead.
So now the calculation is done with this new insight in mind.
Also some additional checks for overflow have been introduced
and some comments to provide some documentation for future
maintainers of this complicated calculation code.
Furthermore the 128 bytes of fix overhead which is used
in the current code is not correct. Investigations showed
that for a reply always the same two header structs are
prepended before a possible payload. So this is also fixed
with this patch.
It's a bug in the concurrent scenario of unregister_netdevice_many()
and raw_release() as following:
cpu0 cpu1
unregister_netdevice_many(can_dev)
unlist_netdevice(can_dev) // dev_get_by_index() return NULL after this
net_set_todo(can_dev)
raw_release(can_socket)
dev = dev_get_by_index(, ro->ifindex); // dev == NULL
if (dev) { // receivers in dev_rcv_lists not free because dev is NULL
raw_disable_allfilters(, dev, );
dev_put(dev);
}
...
ro->bound = 0;
...
call_netdevice_notifiers(NETDEV_UNREGISTER, )
raw_notify(, NETDEV_UNREGISTER, )
if (ro->bound) // invalid because ro->bound has been set 0
raw_disable_allfilters(, dev, ); // receivers in dev_rcv_lists will never be freed
Add a net_device pointer member in struct raw_sock to record bound
can_dev, and use rtnl_lock to serialize raw_socket members between
raw_bind(), raw_release(), raw_setsockopt() and raw_notify(). Use
ro->dev to decide whether to free receivers in dev_rcv_lists.
Fixes: 8d0caedb7596 ("can: bcm/raw/isotp: use per module netdevice notifier") Reviewed-by: Oliver Hartkopp <socketcan@hartkopp.net> Acked-by: Oliver Hartkopp <socketcan@hartkopp.net> Signed-off-by: Ziyang Xuan <william.xuanziyang@huawei.com> Link: https://lore.kernel.org/all/20230711011737.1969582-1-william.xuanziyang@huawei.com Cc: stable@vger.kernel.org Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Before removing checkpoint buffer from the t_checkpoint_list, we have to
check both BH_Dirty and BH_Lock bits together to distinguish buffers
have not been or were being written back. But __cp_buffer_busy() checks
them separately, it first check lock state and then check dirty, the
window between these two checks could be raced by writing back
procedure, which locks buffer and clears buffer dirty before I/O
completes. So it cannot guarantee checkpointing buffers been written
back to disk if some error happens later. Finally, it may clean
checkpoint transactions and lead to inconsistent filesystem.
jbd2_journal_forget() and __journal_try_to_free_buffer() also have the
same problem (journal_unmap_buffer() escape from this issue since it's
running under the buffer lock), so fix them through introducing a new
helper to try holding the buffer lock and remove really clean buffer.
journal_clean_one_cp_list() and journal_shrink_one_cp_list() are almost
the same, so merge them into journal_shrink_one_cp_list(), remove the
nr_to_scan parameter, always scan and try to free the whole checkpoint
list.
cpu_has_octeon_cache was tied to 0 for generic cpu-features,
whith this generic kernel built for octeon CPU won't boot.
Just enable this flag by cpu_type. It won't hurt orther platforms
because compiler will eliminate the code path on other processors.
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com> Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Stable-dep-of: 5487a7b60695 ("MIPS: cpu-features: Use boot_cpu_type for CPU type based features") Signed-off-by: Sasha Levin <sashal@kernel.org>
wait till linux guest boots, then hotplug device:
(qemu) device_add qxl,bus=rp1
hotplug on guest side fails with:
pci 0000:01:00.0: [1b36:0100] type 00 class 0x038000
pci 0000:01:00.0: reg 0x10: [mem 0x00000000-0x03ffffff]
pci 0000:01:00.0: reg 0x14: [mem 0x00000000-0x03ffffff]
pci 0000:01:00.0: reg 0x18: [mem 0x00000000-0x00001fff]
pci 0000:01:00.0: reg 0x1c: [io 0x0000-0x001f]
pci 0000:01:00.0: BAR 0: no space for [mem size 0x04000000]
pci 0000:01:00.0: BAR 0: failed to assign [mem size 0x04000000]
pci 0000:01:00.0: BAR 1: no space for [mem size 0x04000000]
pci 0000:01:00.0: BAR 1: failed to assign [mem size 0x04000000]
pci 0000:01:00.0: BAR 2: assigned [mem 0xfe800000-0xfe801fff]
pci 0000:01:00.0: BAR 3: assigned [io 0x1000-0x101f]
qxl 0000:01:00.0: enabling device (0000 -> 0003)
Unable to create vram_mapping
qxl: probe of 0000:01:00.0 failed with error -12
However when using native PCIe hotplug
'-global ICH9-LPC.acpi-pci-hotplug-with-bridge-support=off'
it works fine, since kernel attempts to reassign unused resources.
Use the same machinery as native PCIe hotplug to (re)assign resources.
Link: https://lore.kernel.org/r/20230424191557.2464760-1-imammedo@redhat.com Signed-off-by: Igor Mammedov <imammedo@redhat.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: Michael S. Tsirkin <mst@redhat.com> Acked-by: Rafael J. Wysocki <rafael@kernel.org> Cc: stable@vger.kernel.org Signed-off-by: Sasha Levin <sashal@kernel.org>
- It's really the only one where this matters. I tried looking around,
and I didn't find any non-pci vga-compatible controllers for x86
(since that's the only platform where we had this until a few
patches ago), where a driver participating in the aperture claim
dance would interfere.
- I also don't expect that any future bus anytime soon will
not just look like pci towards the OS, that's been the case for like
25+ years by now for practically everything (even non non-x86).
- Also it's a bit funny if we have one part of the vga removal in the
pci function, and the other in the generic one.
v2: Rebase.
v4:
- fix Daniel's S-o-b address
v5:
- add back an S-o-b tag with Daniel's Intel address
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Cc: Thomas Zimmermann <tzimmermann@suse.de> Cc: Javier Martinez Canillas <javierm@redhat.com> Cc: Helge Deller <deller@gmx.de> Cc: linux-fbdev@vger.kernel.org Reviewed-by: Javier Martinez Canillas <javierm@redhat.com> Link: https://patchwork.freedesktop.org/patch/msgid/20230406132109.32050-6-tzimmermann@suse.de
Stable-dep-of: 5ae3716cfdcd ("video/aperture: Only remove sysfb on the default vga pci device") Signed-off-by: Sasha Levin <sashal@kernel.org>
Otherwise it's a bit silly, and we might throw out the driver for the
screen the user is actually looking at. I haven't found a bug report
for this case yet, but we did get bug reports for the analog case
where we're throwing out the efifb driver.
v2: Flip the check around to make it clear it's a special case for
kicking out the vgacon driver only (Thomas)
Only really pci devices have a business setting this - it's for
figuring out whether the legacy vga stuff should be nuked too. And
with the preceding two patches those are all using the pci version of
this.
Which means for all other callers primary == false and we can remove
it now.
v2:
- Reorder to avoid compile fail (Thomas)
- Include gma500, which retained it's called to the non-pci version.
v4:
- fix Daniel's S-o-b address
v5:
- add back an S-o-b tag with Daniel's Intel address
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Cc: Thomas Zimmermann <tzimmermann@suse.de> Cc: Javier Martinez Canillas <javierm@redhat.com> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Maxime Ripard <mripard@kernel.org> Cc: Deepak Rawat <drawat.floss@gmail.com> Cc: Neil Armstrong <neil.armstrong@linaro.org> Cc: Kevin Hilman <khilman@baylibre.com> Cc: Jerome Brunet <jbrunet@baylibre.com> Cc: Martin Blumenstingl <martin.blumenstingl@googlemail.com> Cc: Thierry Reding <thierry.reding@gmail.com> Cc: Jonathan Hunter <jonathanh@nvidia.com> Cc: Emma Anholt <emma@anholt.net> Cc: Helge Deller <deller@gmx.de> Cc: David Airlie <airlied@gmail.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: linux-hyperv@vger.kernel.org Cc: linux-amlogic@lists.infradead.org Cc: linux-arm-kernel@lists.infradead.org Cc: linux-tegra@vger.kernel.org Cc: linux-fbdev@vger.kernel.org Acked-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com> Acked-by: Thierry Reding <treding@nvidia.com> Reviewed-by: Javier Martinez Canillas <javierm@redhat.com> Link: https://patchwork.freedesktop.org/patch/msgid/20230406132109.32050-4-tzimmermann@suse.de
Stable-dep-of: 5ae3716cfdcd ("video/aperture: Only remove sysfb on the default vga pci device") Signed-off-by: Sasha Levin <sashal@kernel.org>
This one nukes all framebuffers, which is a bit much. In reality
gma500 is igpu and never shipped with anything discrete, so there should
not be any difference.
v2: Unfortunately the framebuffer sits outside of the pci bars for
gma500, and so only using the pci helpers won't be enough. Otoh if we
only use non-pci helper, then we don't get the vga handling, and
subsequent refactoring to untangle these special cases won't work.
It's not pretty, but the simplest fix (since gma500 really is the only
quirky pci driver like this we have) is to just have both calls.
v4:
- fix Daniel's S-o-b address
v5:
- add back an S-o-b tag with Daniel's Intel address
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Cc: Patrik Jakobsson <patrik.r.jakobsson@gmail.com> Cc: Thomas Zimmermann <tzimmermann@suse.de> Cc: Javier Martinez Canillas <javierm@redhat.com> Reviewed-by: Javier Martinez Canillas <javierm@redhat.com> Link: https://patchwork.freedesktop.org/patch/msgid/20230406132109.32050-2-tzimmermann@suse.de
Stable-dep-of: 5ae3716cfdcd ("video/aperture: Only remove sysfb on the default vga pci device") Signed-off-by: Sasha Levin <sashal@kernel.org>
On server-initiated disconnect, rpcrdma_xprt_disconnect() was DMA-
unmapping the Receive buffers, but rpcrdma_post_recvs() neglected
to remap them after a new connection had been established. The
result was immediate failure of the new connection with the Receives
flushing with LOCAL_PROT_ERR.
Fixes: 671c450b6fe0 ("xprtrdma: Fix oops in Receive handler after device removal") Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Another highly rare error case when a page allocating loop (inside
__nfs4_get_acl_uncached, this time) is not properly unwound on error.
Since pages array is allocated being uninitialized, need to free only
lower array indices. NULL checks were useful before commit 62a1573fcf84
("NFSv4 fix acl retrieval over krb5i/krb5p mounts") when the array had
been initialized to zero on stack.
Found by Linux Verification Center (linuxtesting.org).
There is a slight issue with error handling code inside
nfs42_proc_getxattr(). If page allocating loop fails then we free the
failing page array element which is NULL but __free_page() can't deal with
NULL args.
Found by Linux Verification Center (linuxtesting.org).
Something is currently broken in the f2fs code, Guenter has reported
boot problems with it for a few releases now, so revert the most recent
f2fs changes in the hope to get this back to a working filesystem.
Something is currently broken in the f2fs code, Guenter has reported
boot problems with it for a few releases now, so revert the most recent
f2fs changes in the hope to get this back to a working filesystem.
Something is currently broken in the f2fs code, Guenter has reported
boot problems with it for a few releases now, so revert the most recent
f2fs changes in the hope to get this back to a working filesystem.
- it collects all (tail) call's of __x86_return_thunk and places them
into .return_sites. These are typically compiler generated, but
RET also emits this same.
- it fudges the validation of the __x86_return_thunk symbol; because
this symbol is inside another instruction, it can't actually find
the instruction pointed to by the symbol offset and gets upset.
Because these two things pertained to the same symbol, there was no
pressing need to separate these two separate things.
However, alas, along comes SRSO and more crazy things to deal with
appeared.
The SRSO patch itself added the following symbol names to identify as
rethunk:
'srso_untrain_ret', 'srso_safe_ret' and '__ret'
Where '__ret' is the old retbleed return thunk, 'srso_safe_ret' is a
new similarly embedded return thunk, and 'srso_untrain_ret' is
completely unrelated to anything the above does (and was only included
because of that INT3 vs UD2 issue fixed previous).
Clear things up by adding a second category for the embedded instruction
thing.
For stack-validation of a frame-pointer build, objtool validates that
every CALL instruction is preceded by a frame-setup. The new SRSO
return thunks violate this with their RSB stuffing trickery.
Extend the __fentry__ exception to also cover the embedded_insn case
used for this. This cures:
vmlinux.o: warning: objtool: srso_untrain_ret+0xd: call without frame pointer save/setup
Macro TEXT_TEXT references TEXT_MAIN which normally expands to only
".text". However, with CONFIG_LTO_CLANG, TEXT_MAIN becomes
".text .text.[0-9a-zA-Z_]*" which wrongly matches also the thunk
sections. The output layout is then different than expected. For
instance, the currently defined range [__indirect_thunk_start,
__indirect_thunk_end] becomes empty.
Prevent the problem by using ".." as the first separator, for example,
".text..__x86.indirect_thunk". This pattern is utilized by other
explicit section names which start with one of the standard prefixes,
such as ".text" or ".data", and that need to be individually selected in
the linker script.
[ nathan: Fix conflicts with SRSO and fold in fix issue brought up by
Andrew Cooper in post-review:
https://lore.kernel.org/20230803230323.1478869-1-andrew.cooper3@citrix.com ]
Fixes: dc5723b02e52 ("kbuild: add support for Clang LTO") Signed-off-by: Petr Pavlu <petr.pavlu@suse.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Nathan Chancellor <nathan@kernel.org> Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de> Link: https://lore.kernel.org/r/20230711091952.27944-2-petr.pavlu@suse.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Skip the srso cmd line parsing which is not needed on Zen1/2 with SMT
disabled and with the proper microcode applied (latter should be the
case anyway) as those are not affected.
Fixes: 5a15d8348881 ("x86/srso: Tie SBPB bit setting to microcode patch detection") Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de> Link: https://lore.kernel.org/r/20230813104517.3346-1-bp@alien8.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Initially, it was thought that doing an innocuous division in the #DE
handler would take care to prevent any leaking of old data from the
divider but by the time the fault is raised, the speculation has already
advanced too far and such data could already have been used by younger
operations.
Therefore, do the innocuous division on every exit to userspace so that
userspace doesn't see any potentially old data from integer divisions in
kernel space.
Do the same before VMRUN too, to protect host data from leaking into the
guest too.
Fixes: 77245f1c3c64 ("x86/CPU/AMD: Do not leak quotient data after a division by 0") Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de> Cc: <stable@kernel.org> Link: https://lore.kernel.org/r/20230811213824.10025-1-bp@alien8.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use LEA instead of ADD when adjusting %rsp in srso_safe_ret{,_alias}()
so as to avoid clobbering flags. Drop one of the INT3 instructions to
account for the LEA consuming one more byte than the ADD.
KVM's emulator makes indirect calls into a jump table of sorts, where
the destination of each call is a small blob of code that performs fast
emulation by executing the target instruction with fixed operands.
E.g. to emulate ADC, fastop() invokes adcb_al_dl():
A major motivation for doing fast emulation is to leverage the CPU to
handle consumption and manipulation of arithmetic flags, i.e. RFLAGS is
both an input and output to the target of the call. fastop() collects
the RFLAGS result by pushing RFLAGS onto the stack and popping them back
into a variable (held in %rdi in this case):
asm("push %[flags]; popf; " CALL_NOSPEC " ; pushf; pop %[flags]\n"
Christian reported spurious module load crashes after some of Song's
module memory layout patches.
Turns out that if the very last instruction on the very last page of the
module is a 'JMP __x86_return_thunk' then __static_call_fixup() will
trip a fault and die.
And while the module rework made this slightly more likely to happen,
it's always been possible.
Fixes: ee88d363d156 ("x86,static_call: Use alternative RET encoding") Reported-by: Christian Bricart <christian@bricart.de> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Josh Poimboeuf <jpoimboe@kernel.org> Link: https://lkml.kernel.org/r/20230816104419.GA982867@hirez.programming.kicks-ass.net Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since there can only be one active return_thunk, there only needs be
one (matching) untrain_ret. It fundamentally doesn't make sense to
allow multiple untrain_ret at the same time.
Fold all the 3 different untrain methods into a single (temporary)
helper stub.
srso_safe_ret: // embedded in movabs instruction
add $8,%rsp
ret
int3
srso_return_thunk:
call srso_safe_ret
ud2
While retbleed does:
zen_untrain_ret:
test $0xcc, %bl
lfence
jmp zen_return_thunk
int3
zen_return_thunk: // embedded in the test instruction
ret
int3
Where Zen1/2 flush the BTB entry using the instruction decoder trick
(test,movabs) Zen3/4 use BTB aliasing. SRSO adds a return sequence
(srso_safe_ret()) which forces the function return instruction to
speculate into a trap (UD2). This RET will then mispredict and
execution will continue at the return site read from the top of the
stack.
Pick one of three options at boot (evey function can only ever return
once).
[ bp: Fixup commit message uarch details and add them in a comment in
the code too. Add a comment about the srso_select_mitigation()
dependency on retbleed_select_mitigation(). Add moar ifdeffery for
32-bit builds. Add a dummy srso_untrain_ret_alias() definition for
32-bit alternatives needing the symbol. ]
There is infrastructure to rewrite return thunks to point to any
random thunk one desires, unwrap that from CALL_THUNKS, which up to
now was the sole user of that.
[ bp: Make the thunks visible on 32-bit and add ifdeffery for the
32-bit builds. ]
vmlinux.o: warning: objtool: srso_untrain_ret() falls through to next function __x86_return_skl()
vmlinux.o: warning: objtool: __x86_return_thunk() falls through to next function __x86_return_skl()
This is because these functions (can) end with CALL, which objtool
does not consider a terminating instruction. Therefore, replace the
INT3 instruction (which is a non-fatal trap) with UD2 (which is a
fatal-trap).
This indicates execution will not continue past this point.
In the real workload, I encountered an issue which could cause the RTO
timer to retransmit the skb per 1ms with linear option enabled. The amount
of lost-retransmitted skbs can go up to 1000+ instantly.
The root cause is that if the icsk_rto happens to be zero in the 6th round
(which is the TCP_THIN_LINEAR_RETRIES value), then it will always be zero
due to the changed calculation method in tcp_retransmit_timer() as follows:
Above line could be converted to
icsk->icsk_rto = min(0 << 1, TCP_RTO_MAX) = 0
Therefore, the timer expires so quickly without any doubt.
I read through the RFC 6298 and found that the RTO value can be rounded
up to a certain value, in Linux, say TCP_RTO_MIN as default, which is
regarded as the lower bound in this patch as suggested by Eric.
Fixes: 36e31b0af587 ("net: TCP thin linear timeouts") Suggested-by: Eric Dumazet <edumazet@google.com> Signed-off-by: Jason Xing <kernelxing@tencent.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We can't simply free the connector after calling drm_connector_init on it.
We need to clean up the drm side first.
It might not fix all regressions from commit 2b5d1c29f6c4
("drm/nouveau/disp: PIOR DP uses GPIO for HPD, not PMGR AUX interrupts"),
but at least it fixes a memory corruption in error handling related to
that commit.
af_unix: Fix null-ptr-deref in unix_stream_sendpage().
Bing-Jhong Billy Jheng reported null-ptr-deref in unix_stream_sendpage()
with detailed analysis and a nice repro.
unix_stream_sendpage() tries to add data to the last skb in the peer's
recv queue without locking the queue.
If the peer's FD is passed to another socket and the socket's FD is
passed to the peer, there is a loop between them. If we close both
sockets without receiving FD, the sockets will be cleaned up by garbage
collection.
The garbage collection iterates such sockets and unlinks skb with
FD from the socket's receive queue under the queue's lock.
So, there is a race where unix_stream_sendpage() could access an skb
locklessly that is being released by garbage collection, resulting in
use-after-free.
To avoid the issue, unix_stream_sendpage() must lock the peer's recv
queue.
Note the issue does not exist in 6.5+ thanks to the recent sendpage()
refactoring.
This patch is originally written by Linus Torvalds.
This can clean up all irq warnings because of unbalanced
amdgpu_irq_get/put when unplugging/unbinding device, and leave
irq count decrease in each ip fini function.
Signed-off-by: Guchun Chen <guchun.chen@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When we use NT_ARM_SSVE to either enable streaming mode or change the
vector length for a process we do not currently do anything to ensure that
there is storage allocated for the SME specific register state. If the
task had not previously used SME or we changed the vector length then
the task will not have had TIF_SME set or backing storage for ZA/ZT
allocated, resulting in inconsistent register sizes when saving state
and spurious traps which flush the newly set register state.
We should set TIF_SME to disable traps and ensure that storage is
allocated for ZA and ZT if it is not already allocated. This requires
modifying sme_alloc() to make the flush of any existing register state
optional so we don't disturb existing state for ZA and ZT.
Fixes: e12310a0d30f ("arm64/sme: Implement ptrace support for streaming mode SVE registers") Reported-by: David Spickett <David.Spickett@arm.com> Signed-off-by: Mark Brown <broonie@kernel.org> Cc: <stable@vger.kernel.org> # 5.19.x Link: https://lore.kernel.org/r/20230810-arm64-fix-ptrace-race-v1-1-a5361fad2bd6@kernel.org Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In SCTP protocol, it is using the same timer (T2 timer) for SHUTDOWN and
SHUTDOWN_ACK retransmission. However in sctp conntrack the default timeout
value for SCTP_CONNTRACK_SHUTDOWN_ACK_SENT state is 3 secs while it's 300
msecs for SCTP_CONNTRACK_SHUTDOWN_SEND/RECV state.
As Paolo Valerio noticed, this might cause unwanted expiration of the ct
entry. In my test, with 1s tc netem delay set on the NAT path, after the
SHUTDOWN is sent, the sctp ct entry enters SCTP_CONNTRACK_SHUTDOWN_SEND
state. However, due to 300ms (too short) delay, when the SHUTDOWN_ACK is
sent back from the peer, the sctp ct entry has expired and been deleted,
and then the SHUTDOWN_ACK has to be dropped.
Also, it is confusing these two sysctl options always show 0 due to all
timeout values using sec as unit:
This patch fixes it by also using 3 secs for sctp shutdown send and recv
state in sctp conntrack, which is also RTO.initial value in SCTP protocol.
Note that the very short time value for SCTP_CONNTRACK_SHUTDOWN_SEND/RECV
was probably used for a rare scenario where SHUTDOWN is sent on 1st path
but SHUTDOWN_ACK is replied on 2nd path, then a new connection started
immediately on 1st path. So this patch also moves from SHUTDOWN_SEND/RECV
to CLOSE when receiving INIT in the ORIGINAL direction.
Fixes: 9fb9cbb1082d ("[NETFILTER]: Add nf_conntrack subsystem.") Reported-by: Paolo Valerio <pvalerio@redhat.com> Signed-off-by: Xin Long <lucien.xin@gmail.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Patch series "Fix hugetlb free path race with memory errors".
In the discussion of Jiaqi Yan's series "Improve hugetlbfs read on
HWPOISON hugepages" the race window was discovered.
https://lore.kernel.org/linux-mm/20230616233447.GB7371@monkey/
Freeing a hugetlb page back to low level memory allocators is performed
in two steps.
1) Under hugetlb lock, remove page from hugetlb lists and clear destructor
2) Outside lock, allocate vmemmap if necessary and call low level free
Between these two steps, the hugetlb page will appear as a normal
compound page. However, vmemmap for tail pages could be missing.
If a memory error occurs at this time, we could try to update page
flags non-existant page structs.
A much more detailed description is in the first patch.
The first patch addresses the race window. However, it adds a
hugetlb_lock lock/unlock cycle to every vmemmap optimized hugetlb page
free operation. This could lead to slowdowns if one is freeing a large
number of hugetlb pages.
The second path optimizes the update_and_free_pages_bulk routine to only
take the lock once in bulk operations.
The second patch is technically not a bug fix, but includes a Fixes tag
and Cc stable to avoid a performance regression. It can be combined with
the first, but was done separately make reviewing easier.
This patch (of 2):
Freeing a hugetlb page and releasing base pages back to the underlying
allocator such as buddy or cma is performed in two steps:
- remove_hugetlb_folio() is called to remove the folio from hugetlb
lists, get a ref on the page and remove hugetlb destructor. This
all must be done under the hugetlb lock. After this call, the page
can be treated as a normal compound page or a collection of base
size pages.
- update_and_free_hugetlb_folio() is called to allocate vmemmap if
needed and the free routine of the underlying allocator is called
on the resulting page. We can not hold the hugetlb lock here.
One issue with this scheme is that a memory error could occur between
these two steps. In this case, the memory error handling code treats
the old hugetlb page as a normal compound page or collection of base
pages. It will then try to SetPageHWPoison(page) on the page with an
error. If the page with error is a tail page without vmemmap, a write
error will occur when trying to set the flag.
Address this issue by modifying remove_hugetlb_folio() and
update_and_free_hugetlb_folio() such that the hugetlb destructor is not
cleared until after allocating vmemmap. Since clearing the destructor
requires holding the hugetlb lock, the clearing is done in
remove_hugetlb_folio() if the vmemmap is present. This saves a
lock/unlock cycle. Otherwise, destructor is cleared in
update_and_free_hugetlb_folio() after allocating vmemmap.
Note that this will leave hugetlb pages in a state where they are marked
free (by hugetlb specific page flag) and have a ref count. This is not
a normal state. The only code that would notice is the memory error
code, and it is set up to retry in such a case.
A subsequent patch will create a routine to do bulk processing of
vmemmap allocation. This will eliminate a lock/unlock cycle for each
hugetlb page in the case where we are freeing a large number of pages.
Link: https://lkml.kernel.org/r/20230711220942.43706-1-mike.kravetz@oracle.com Link: https://lkml.kernel.org/r/20230711220942.43706-2-mike.kravetz@oracle.com Fixes: ad2fa3717b74 ("mm: hugetlb: alloc the vmemmap pages associated with each HugeTLB page") Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com> Reviewed-by: Muchun Song <songmuchun@bytedance.com> Tested-by: Naoya Horiguchi <naoya.horiguchi@nec.com> Cc: Axel Rasmussen <axelrasmussen@google.com> Cc: James Houghton <jthoughton@google.com> Cc: Jiaqi Yan <jiaqiyan@google.com> Cc: Miaohe Lin <linmiaohe@huawei.com> Cc: Michal Hocko <mhocko@suse.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[Why and How]
Current implementation requires FPGA builds to take a different
code path from DCN32 to write to OTG_PIXEL_RATE_DIV. Now that
we have a workaround to write to OTG_PIXEL_RATE_DIV register without
blanking display on hotplug on DCN32, we can allow the code paths for
FPGA to be exactly the same allowing for more consistent
testing.
Reviewed-by: Alvin Lee <Alvin.Lee2@amd.com> Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com> Signed-off-by: Saaem Rizvi <SyedSaaem.Rizvi@amd.com> Tested-by: Daniel Wheeler <daniel.wheeler@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: "Limonciello, Mario" <mario.limonciello@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Remove the capacity inversion detection which is now handled by
util_fits_cpu() returning -1 when we need to continue to look for a
potential CPU with better performance.
This ends up almost reverting patches below except for some comments:
commit da07d2f9c153 ("sched/fair: Fixes for capacity inversion detection")
commit aa69c36f31aa ("sched/fair: Consider capacity inversion in util_fits_cpu()")
commit 44c7b80bffc3 ("sched/fair: Detect capacity inversion")
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lore.kernel.org/r/20230201143628.270912-3-vincent.guittot@linaro.org Signed-off-by: Qais Yousef (Google) <qyousef@layalina.io> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
By taking into account uclamp_min, the 1:1 relation between task misfit
and cpu overutilized is no more true as a task with a small util_avg may
not fit a high capacity cpu because of uclamp_min constraint.
Add a new state in util_fits_cpu() to reflect the case that task would fit
a CPU except for the uclamp_min hint which is a performance requirement.
Use -1 to reflect that a CPU doesn't fit only because of uclamp_min so we
can use this new value to take additional action to select the best CPU
that doesn't match uclamp_min hint.
When util_fits_cpu() returns -1, we will continue to look for a possible
CPU with better performance, which replaces Capacity Inversion detection
with capacity_orig_of() - thermal_load_avg to detect a capacity inversion.
projects / kernel.git / log