__hci_cmd_sync_status shall only be used if hci_req_sync_lock is _not_
required which is not the case of hci_dev_cmd so it needs to use
hci_cmd_sync_status which uses hci_req_sync_lock internally.
Fixes: f1a8f402f13f ("Bluetooth: L2CAP: Fix deadlock") Reported-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
If a port is blocking in the common instance but forwarding in an MST
instance, traffic egressing the bridge will be dropped because the
state of the common instance is overriding that of the MST instance.
Fix this by skipping the port state check in MST mode to allow
checking the vlan state via br_allowed_egress(). This is similar to
what happens in br_handle_frame_finish() when checking ingress
traffic, which was introduced in the change below.
Fixes: ec7328b59176 ("net: bridge: mst: Multiple Spanning Tree (MST) mode") Signed-off-by: Elliot Ayrey <elliot.ayrey@alliedtelesis.co.nz> Acked-by: Nikolay Aleksandrov <razor@blackwall.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
If the driver uses a page pool, it creates a page pool with
page_pool_create().
The reference count of page pool is 1 as default.
A page pool will be destroyed only when a reference count reaches 0.
page_pool_destroy() is used to destroy page pool, it decreases a
reference count.
When a page pool is destroyed, ->disconnect() is called, which is
mem_allocator_disconnect().
This function internally acquires mutex_lock().
If the driver uses XDP, it registers a memory model with
xdp_rxq_info_reg_mem_model().
The xdp_rxq_info_reg_mem_model() internally increases a page pool
reference count if a memory model is a page pool.
Now the reference count is 2.
To destroy a page pool, the driver should call both page_pool_destroy()
and xdp_unreg_mem_model().
The xdp_unreg_mem_model() internally calls page_pool_destroy().
Only page_pool_destroy() decreases a reference count.
If a driver calls page_pool_destroy() then xdp_unreg_mem_model(), we
will face an invalid wait context warning.
Because xdp_unreg_mem_model() calls page_pool_destroy() with
rcu_read_lock().
The page_pool_destroy() internally acquires mutex_lock().
To fix this problem, it uses rhashtable_lookup_fast() instead of
rhashtable_lookup() with rcu_read_lock().
Using xa without rcu_read_lock() here is safe.
xa is freed by __xdp_mem_allocator_rcu_free() and this is called by
call_rcu() of mem_xa_remove().
The mem_xa_remove() is called by page_pool_destroy() if a reference
count reaches 0.
The xa is already protected by the reference count mechanism well in the
control plane.
So removing rcu_read_lock() for page_pool_destroy() is safe.
Fixes: c3f812cea0d7 ("page_pool: do not release pool until inflight == 0.") Signed-off-by: Taehee Yoo <ap420073@gmail.com> Reviewed-by: Jakub Kicinski <kuba@kernel.org> Link: https://patch.msgid.link/20240712095116.3801586-1-ap420073@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The issue arises because virtio is assuming it's running in NAPI context
even when it's not, such as in the netpoll case.
To resolve this, modify virtnet_poll_tx() to only set NAPI when budget
is available. Same for virtnet_poll_cleantx(), which always assumed that
it was in a NAPI context.
Fixes: df133f3f9625 ("virtio_net: bulk free tx skbs") Suggested-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Breno Leitao <leitao@debian.org> Reviewed-by: Jakub Kicinski <kuba@kernel.org> Acked-by: Michael S. Tsirkin <mst@redhat.com> Acked-by: Jason Wang <jasowang@redhat.com> Reviewed-by: Heng Qi <hengqi@linux.alibaba.com> Link: https://patch.msgid.link/20240712115325.54175-1-leitao@debian.org Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
After commit 78db544b5d27 ("Bluetooth: hci_core: Remove le_restart_scan
work"), 'scan_start' and 'scan_duration' of 'struct discovery_state'
are still initialized but actually unused. So remove the aforementioned
fields and adjust 'hci_discovery_filter_clear()' and 'le_scan_disable()'
accordingly. Compile tested only.
On a Broadcast Sink, after synchronizing to the PA transimitted by a
Broadcast Source, the BIGInfo advertising reports emitted by the
Controller hold the encryption field, which indicates whether the
Broadcast Source is transmitting encrypted streams.
This updates the PA sync hcon QoS with the encryption value reported
in the BIGInfo report, so that this information is accurate if the
userspace tries to access the QoS struct via getsockopt.
Fixes: 1d11d70d1f6b ("Bluetooth: ISO: Pass BIG encryption info through QoS") Signed-off-by: Iulia Tanasescu <iulia.tanasescu@nxp.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
This handles the timeout error codes sent by the chip as part of the
bootloader signatures during firmware download process.
When the bootloader does not receive a response packet from the host
within a specific time, it adds an error code to the bootloader
signature while requesting for the FW chunk from the same offset.
The host is expected to clear this error code with a NAK, and reply to
only those bootloader signatures which have error code 0.
However, the driver was ignoring this error code and replying with the
firmware chunks instead, which is apparently ignored by the chip and the
chip resends the same bootloader signature with the error code again. This
happens in a loop until the error code self clears and firmware download
proceeds ahead, adding a couple of milliseconds to the total firmware
download time.
Commit 689ca16e5232 was an initial implementation which simply printed
the following line during driver debug:
- FW Download received err 0x04 from chip
This commit adds the expected handling to the error codes.
This error handling is valid for data_req bootloader signatures for V3
and future bootloader versions.
Signed-off-by: Neeraj Sanjay Kale <neeraj.sanjaykale@nxp.com> Fixes: 689ca16e5232 ("Bluetooth: NXP: Add protocol support for NXP Bluetooth chipsets") Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Fixes: c2b636b3f788 ("Bluetooth: btintel_pcie: Add support for PCIe transport") Signed-off-by: Kiran K <kiran.k@intel.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Current flow iterates the ACPI table associated with Bluetooth
controller looking for PPAG method. Method name can be directly passed
to acpi_evaluate_object function instead of iterating the table.
Fixes: c585a92b2f9c ("Bluetooth: btintel: Set Per Platform Antenna Gain(PPAG)") Signed-off-by: Kiran K <kiran.k@intel.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
BCM4377_TIMEOUT is always used to wait for completitions and their API
expects a timeout in jiffies instead of msecs.
Fixes: 8a06127602de ("Bluetooth: hci_bcm4377: Add new driver for BCM4377 PCIe boards") Signed-off-by: Sven Peter <sven@svenpeter.dev> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Lately, an additional locking was added by commit c0a40097f0bc
("drivers: core: synchronize really_probe() and dev_uevent()"). The
locking protects dev_uevent() calling. This function is used to send
messages from the kernel to user space. Uevent messages notify user space
about changes in device states, such as when a device is added, removed,
or changed. These messages are used by udev (or other similar user-space
tools) to apply device-specific rules.
After reloading devlink instance, udev events should be processed. This
locking causes a short delay of udev events handling.
One example for useful udev rule is renaming ports. 'forwading.config'
can be configured to use names after udev rules are applied. Some tests run
devlink_reload() and immediately use the updated names. This worked before
the above mentioned commit was pushed, but now the delay of uevent messages
causes that devlink_reload() returns before udev events are handled and
tests fail.
Adjust devlink_reload() to not assume that udev events are already
processed when devlink reload is done, instead, wait for udev events to
ensure they are processed before returning from the function.
Without this patch:
TESTS='rif_mac_profile' ./resource_scale.sh
TEST: 'rif_mac_profile' 4 [ OK ]
sysctl: cannot stat /proc/sys/net/ipv6/conf/swp1/disable_ipv6: No such file or directory
sysctl: cannot stat /proc/sys/net/ipv6/conf/swp1/disable_ipv6: No such file or directory
sysctl: cannot stat /proc/sys/net/ipv6/conf/swp2/disable_ipv6: No such file or directory
sysctl: cannot stat /proc/sys/net/ipv6/conf/swp2/disable_ipv6: No such file or directory
Cannot find device "swp1"
Cannot find device "swp2"
TEST: setup_wait_dev (: Interface swp1 does not come up.) [FAIL]
With this patch:
$ TESTS='rif_mac_profile' ./resource_scale.sh
TEST: 'rif_mac_profile' 4 [ OK ]
TEST: 'rif_mac_profile' overflow 5 [ OK ]
This is relevant not only for this test.
Fixes: bc7cbb1e9f4c ("selftests: forwarding: Add devlink_lib.sh") Signed-off-by: Amit Cohen <amcohen@nvidia.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> Signed-off-by: Petr Machata <petrm@nvidia.com> Link: https://patch.msgid.link/89367666e04b38a8993027f1526801ca327ab96a.1720709333.git.petrm@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Fix a possible null dereference when a PSE supports both c33 and PoDL, but
only one of the netlink attributes is specified. The c33 or PoDL PSE
capabilities are already validated in the ethnl_set_pse_validate() call.
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com> Reported-by: Jakub Kicinski <kuba@kernel.org> Closes: https://lore.kernel.org/netdev/20240705184116.13d8235a@kernel.org/ Fixes: 4d18e3ddf427 ("net: ethtool: pse-pd: Expand pse commands with the PSE PoE interface") Link: https://patch.msgid.link/20240711-fix_pse_pd_deref-v3-2-edd78fc4fe42@bootlin.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
For a PSE supporting both c33 and PoDL, setting config for one type of PoE
leaves the other type's config null. Currently, this case returns
EOPNOTSUPP, which is incorrect. Instead, we should do nothing if the
configuration is empty.
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com> Fixes: d83e13761d5b ("net: pse-pd: Use regulator framework within PSE framework") Link: https://patch.msgid.link/20240711-fix_pse_pd_deref-v3-1-edd78fc4fe42@bootlin.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When loading a EXT program without specifying `attr->attach_prog_fd`,
the `prog->aux->dst_prog` will be null. At this time, calling
resolve_prog_type() anywhere will result in a null pointer dereference.
One way to fix it is by forcing `attach_prog_fd` non-empty when
bpf_prog_load(). But this will lead to `libbpf_probe_bpf_prog_type`
API broken which use verifier log to probe prog type and will log
nothing if we reject invalid EXT prog before bpf_check().
Another way is by adding null check in resolve_prog_type().
The issue was introduced by commit 4a9c7bbe2ed4 ("bpf: Resolve to
prog->aux->dst_prog->type only for BPF_PROG_TYPE_EXT") which wanted
to correct type resolution for BPF_PROG_TYPE_TRACING programs. Before
that, the type resolution of BPF_PROG_TYPE_EXT prog actually follows
the logic below:
It implies that when EXT program is not yet attached to `dst_prog`,
the prog type should be EXT itself. This code worked fine in the past.
So just keep using it.
Fix this by returning `prog->type` for BPF_PROG_TYPE_EXT if `dst_prog`
is not present in resolve_prog_type().
Fixes: 4a9c7bbe2ed4 ("bpf: Resolve to prog->aux->dst_prog->type only for BPF_PROG_TYPE_EXT") Signed-off-by: Tengda Wu <wutengda@huaweicloud.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Daniel Borkmann <daniel@iogearbox.net> Cc: Martin KaFai Lau <kafai@fb.com> Link: https://lore.kernel.org/bpf/20240711145819.254178-2-wutengda@huaweicloud.com Signed-off-by: Sasha Levin <sashal@kernel.org>
"data" actually refers to a file_lease and not a file_lock. Both structs
have their file_lock_core as the first field though, so this bug should
be harmless without struct randomization in play.
Reported-by: Florian Evers <florian-evers@gmx.de> Closes: https://bugzilla.kernel.org/show_bug.cgi?id=219008 Fixes: 05580bbfc6bc ("nfsd: adapt to breakup of struct file_lock") Signed-off-by: Jeff Layton <jlayton@kernel.org> Tested-by: Florian Evers <florian-evers@gmx.de> Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
adjust_jmp_off() incorrectly used the insn->imm field for all overflow check,
which is incorrect as that should only be done or the BPF_JMP32 | BPF_JA case,
not the general jump instruction case. Fix it by using insn->off for overflow
check in the general case.
Fixes: 5337ac4c9b80 ("bpf: Fix the corner case with may_goto and jump to the 1st insn.") Signed-off-by: Shung-Hsi Yu <shung-hsi.yu@suse.com> Link: https://lore.kernel.org/r/20240712080127.136608-2-shung-hsi.yu@suse.com Signed-off-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
As reported by Mirsad [1] we still see format warnings in kernel/bpf/btf.o
at W=1 warning level:
CC kernel/bpf/btf.o
./kernel/bpf/btf.c: In function ‘btf_type_seq_show_flags’:
./kernel/bpf/btf.c:7553:21: warning: assignment left-hand side might be a candidate for a format attribute [-Wsuggest-attribute=format]
7553 | sseq.showfn = btf_seq_show;
| ^
./kernel/bpf/btf.c: In function ‘btf_type_snprintf_show’:
./kernel/bpf/btf.c:7604:31: warning: assignment left-hand side might be a candidate for a format attribute [-Wsuggest-attribute=format]
7604 | ssnprintf.show.showfn = btf_snprintf_show;
| ^
Combined with CONFIG_WERROR=y these can halt the build.
The fix (annotating the structure field with __printf())
suggested by Mirsad resolves these. Apologies I missed this last time.
No other W=1 warnings were observed in kernel/bpf after this fix.
To have enough space to write all possible sprintf() args. Currently
'name' size is 16, but the first '%s' specifier may already need at
least 16 characters, since 'bnad->netdev->name' is used there.
For '%d' specifiers, assume that they require:
* 1 char for 'tx_id + tx_info->tcb[i]->id' sum, BNAD_MAX_TXQ_PER_TX is 8
* 2 chars for 'rx_id + rx_info->rx_ctrl[i].ccb->id', BNAD_MAX_RXP_PER_RX
is 16
And replace sprintf with snprintf.
Detected using the static analysis tool - Svace.
Fixes: 8b230ed8ec96 ("bna: Brocade 10Gb Ethernet device driver") Signed-off-by: Alexey Kodanev <aleksei.kodanev@bell-sw.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
-Werror=suggest-attribute=format warns about two functions
in kernel/bpf/btf.c [1]; add __printf() annotations to silence
these warnings since for CONFIG_WERROR=y they will trigger
build failures.
The imc perf fd close() calls are missing from all error paths. In
addition, get_mem_bw_imc() handles fds in a for loop but close() is
based on two fixed indexes READ and WRITE.
Open code inner for loops to READ+WRITE entries for clarity and add a
function to close() IMC fds properly in all cases.
When BPF_TRAMP_F_CALL_ORIG is set, the trampoline calls
__bpf_tramp_enter() and __bpf_tramp_exit() functions, passing them
the struct bpf_tramp_image *im pointer as an argument in R0.
The trampoline generation code uses emit_addr_mov_i64() to emit
instructions for moving the bpf_tramp_image address into R0, but
emit_addr_mov_i64() assumes the address to be in the vmalloc() space
and uses only 48 bits. Because bpf_tramp_image is allocated using
kzalloc(), its address can use more than 48-bits, in this case the
trampoline will pass an invalid address to __bpf_tramp_enter/exit()
causing a kernel crash.
Fix this by using emit_a64_mov_i64() in place of emit_addr_mov_i64()
as it can work with addresses that are greater than 48-bits.
If bpf_object__load() fails in test_xdp_adjust_frags_tail_grow(), "obj"
opened before this should be closed. So use "goto out" to close it instead
of using "return" here.
This is because BPF trampoline is not implemented on Loongarch yet,
"link" returned by bpf_map__attach_struct_ops() is NULL. test_progs
crashs when this NULL link passes to bpf_link__update_map(). This
patch adds NULL checks for all links in bpf_tcp_ca to fix these errors.
If "link" is NULL, goto the newly added label "out" to destroy the skel.
v2:
- use "goto out" instead of "return" as Eduard suggested.
In the error path when update_lookup_map() fails in drop_on_reuseport in
prog_tests/sk_lookup.c, "server1", the fd of server 1, should be closed.
This patch fixes this by using "goto close_srv1" lable instead of "detach"
to close "server1" in this case.
Apparently despite it being marked inline, the compiler
may not inline __down_write_common() which makes it difficult
to identify the cause of lock contention, as the wchan of the
blocked function will always be listed as __down_write_common().
So add __always_inline annotation to the common function (as
well as the inlined helper callers) to force it to be inlined
so a more useful blocking function will be listed (via wchan).
This mirrors commit 92cc5d00a431 ("locking/rwsem: Add
__always_inline annotation to __down_read_common() and inlined
callers") which did the same for __down_read_common.
I sort of worry that I'm playing wack-a-mole here, and talking
with compiler people, they tell me inline means nothing, which
makes me want to cry a little. So I'm wondering if we need to
replace all the inlines with __always_inline, or remove them
because either we mean something by it, or not.
Fixes: c995e638ccbb ("locking/rwsem: Fold __down_{read,write}*()") Reported-by: Tim Murray <timmurray@google.com> Signed-off-by: John Stultz <jstultz@google.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Waiman Long <longman@redhat.com> Link: https://lkml.kernel.org/r/20240709060831.495366-1-jstultz@google.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Looks like not all compilers allow strlen(constant) as
a constant, so don't do that. Instead, revert back to
defining the length as the first submission had it.
Fixes: b5d14b0c6716 ("wifi: virt_wifi: avoid reporting connection success with wrong SSID") Reported-by: kernel test robot <lkp@intel.com> Closes: https://lore.kernel.org/oe-kbuild-all/202407090934.NnR1TUbW-lkp@intel.com/ Closes: https://lore.kernel.org/oe-kbuild-all/202407090944.mpwLHGt9-lkp@intel.com/ Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
If we fail to call crypto_sync_skcipher_setkey, we should free the
memory allocation for cipher, replace err_return with err_free_cipher
to free the memory of cipher.
Fixes: 4891f2d008e4 ("gss_krb5: import functionality to derive keys into the kernel") Signed-off-by: Gaosheng Cui <cuigaosheng1@huawei.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Since CONFIG_NFSD_LEGACY_CLIENT_TRACKING is a new config option, its
initial default setting should have been Y (if we are to follow the
common practice of "default Y, wait, default N, wait, remove code").
Paul also suggested adding a clearer remedy action to the warning
message.
Reported-by: Paul Menzel <pmenzel@molgen.mpg.de>
Message-Id: <d2ab4ee7-ba0f-44ac-b921-90c8fa5a04d2@molgen.mpg.de> Fixes: 74fd48739d04 ("nfsd: new Kconfig option for legacy client tracking") Reviewed-by: Jeff Layton <jlayton@kernel.org> Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
I realized this while having a map containing both a struct bpf_timer and
a struct bpf_wq: the third argument provided to the bpf_wq callback is
not the struct bpf_wq pointer itself, but the pointer to the value in
the map.
Which means that the users need to double cast the provided "value" as
this is not a struct bpf_wq *.
This is a change of API, but there doesn't seem to be much users of bpf_wq
right now, so we should be able to go with this right now.
When user issues a connection with a different SSID than the one
virt_wifi has advertised, the __cfg80211_connect_result() will
trigger the warning: WARN_ON(bss_not_found).
The issue is because the connection code in virt_wifi does not
check the SSID from user space (it only checks the BSSID), and
virt_wifi will call cfg80211_connect_result() with WLAN_STATUS_SUCCESS
even if the SSID is different from the one virt_wifi has advertised.
Eventually cfg80211 won't be able to find the cfg80211_bss and generate
the warning.
Fixed it by checking the SSID (from user space) in the connection code.
xfrm_policy_kill() is called at different places to delete xfrm
policy. It will call xfrm_pol_put(). But xfrm_dev_policy_delete() is
not called to free the policy offloaded to hardware.
The three commits cited here are to handle this issue by calling
xfrm_dev_policy_delete() outside xfrm_get_policy(). But they didn't
cover all the cases. An example, which is not handled for now, is
xfrm_policy_insert(). It is called when XFRM_MSG_UPDPOLICY request is
received. Old policy is replaced by new one, but the offloaded policy
is not deleted, so driver doesn't have the chance to release hardware
resources.
To resolve this issue for all cases, move xfrm_dev_policy_delete()
into xfrm_policy_kill(), so the offloaded policy can be deleted from
hardware when it is called, which avoids hardware resources leakage.
Fixes: 919e43fad516 ("xfrm: add an interface to offload policy") Fixes: bf06fcf4be0f ("xfrm: add missed call to delete offloaded policies") Fixes: 982c3aca8bac ("xfrm: delete offloaded policy") Signed-off-by: Jianbo Liu <jianbol@nvidia.com> Reviewed-by: Cosmin Ratiu <cratiu@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In cited commit, netdev_tracker_alloc() is called for the newly
allocated xfrm state, but dev_hold() is missed, which causes netdev
reference count imbalance, because netdev_put() is called when the
state is freed in xfrm_dev_state_free(). Fix the issue by replacing
netdev_tracker_alloc() with netdev_hold().
Fixes: f8a70afafc17 ("xfrm: add TX datapath support for IPsec packet offload mode") Signed-off-by: Jianbo Liu <jianbol@nvidia.com> Reviewed-by: Cosmin Ratiu <cratiu@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In rtw89_sta_info_get_iter() 'status->he_gi' is compared to array size.
But then 'rate->he_gi' is used as array index instead of 'status->he_gi'.
This can lead to go beyond array boundaries in case of 'rate->he_gi' is
not equal to 'status->he_gi' and is bigger than array size. Looks like
"copy-paste" mistake.
Fix this mistake by replacing 'rate->he_gi' with 'status->he_gi'.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
For uncore PMUs, a single context is shared across all CPUs in a domain.
The domain can be a CCX, like in the case of the L3 PMU, or a socket,
like in the case of DF and UMC PMUs. This information is available via
the PMU's cpumask.
For contexts shared across a socket, the domain is currently determined
from topology_die_id() which is incorrect after the introduction of
commit 63edbaa48a57 ("x86/cpu/topology: Add support for the AMD
0x80000026 leaf") as it now returns a CCX identifier on Zen 4 and later
systems which support CPUID leaf 0x80000026.
Use topology_logical_package_id() instead as it always returns a socket
identifier irrespective of the availability of CPUID leaf 0x80000026.
Fixes: 63edbaa48a57 ("x86/cpu/topology: Add support for the AMD 0x80000026 leaf") Signed-off-by: Sandipan Das <sandipan.das@amd.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lkml.kernel.org/r/20240626074942.1044818-1-sandipan.das@amd.com Signed-off-by: Sasha Levin <sashal@kernel.org>
X86_FEATURE_PERFCTR_NB and X86_FEATURE_PERFCTR_LLC are derived from
CPUID leaf 0x80000001 ECX bits 24 and 28 respectively and denote the
availability of DF and L3 counters. When these bits are not set, the
corresponding PMUs have no counters and hence, should not be registered.
Fixes: 07888daa056e ("perf/x86/amd/uncore: Move discovery and registration") Signed-off-by: Sandipan Das <sandipan.das@amd.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lkml.kernel.org/r/20240626074404.1044230-1-sandipan.das@amd.com Signed-off-by: Sasha Levin <sashal@kernel.org>
The default aux_watermark is half the AUX area buffer size. In general,
on a 64-bit architecture, the AUX area buffer size could be a bigger than
fits in a 32-bit type, but the calculation does not allow for that
possibility.
However the aux_watermark value is recorded in a u32, so should not be
more than U32_MAX either.
Fix by doing the calculation in a correctly sized type, and limiting the
result to U32_MAX.
Fixes: d68e6799a5c8 ("perf: Cap allocation order at aux_watermark") Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lore.kernel.org/r/20240624201101.60186-7-adrian.hunter@intel.com Signed-off-by: Sasha Levin <sashal@kernel.org>
nr_pages is unsigned long but gets passed to rb_alloc_aux() as an int,
and is stored as an int.
Only power-of-2 values are accepted, so if nr_pages is a 64_bit value, it
will be passed to rb_alloc_aux() as zero.
That is not ideal because:
1. the value is incorrect
2. rb_alloc_aux() is at risk of misbehaving, although it manages to
return -ENOMEM in that case, it is a result of passing zero to get_order()
even though the get_order() result is documented to be undefined in that
case.
Fix by simply validating the maximum supported value in the first place.
Use -ENOMEM error code for consistency with the current error code that
is returned in that case.
Fixes: 45bfb2e50471 ("perf: Add AUX area to ring buffer for raw data streams") Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lore.kernel.org/r/20240624201101.60186-6-adrian.hunter@intel.com Signed-off-by: Sasha Levin <sashal@kernel.org>
perf_buffer->aux_nr_pages uses a 32-bit type, so a cast is needed to
calculate a 64-bit size.
Fixes: 45bfb2e50471 ("perf: Add AUX area to ring buffer for raw data streams") Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lore.kernel.org/r/20240624201101.60186-5-adrian.hunter@intel.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently, perf allocates an array of page pointers which is limited in
size by MAX_PAGE_ORDER. That in turn limits the maximum Intel PT buffer
size to 2GiB. Should that limitation be lifted, the Intel PT driver can
support larger sizes, except for one calculation in
pt_topa_entry_for_page(), which is limited to 32-bits.
Fix pt_topa_entry_for_page() address calculation by adding a cast.
Fixes: 39152ee51b77 ("perf/x86/intel/pt: Get rid of reverse lookup table for ToPA") Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lore.kernel.org/r/20240624201101.60186-4-adrian.hunter@intel.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Zero-extending results of atomic probe operations fails with:
verifier bug. zext_dst is set, but no reg is defined
The problem is that insn_def_regno() handles BPF_ATOMICs, but not
BPF_PROBE_ATOMICs. Fix by adding the missing condition.
Fixes: d503a04f8bc0 ("bpf: Add support for certain atomics in bpf_arena to x86 JIT") Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Daniel Borkmann <daniel@iogearbox.net> Link: https://lore.kernel.org/bpf/20240701234304.14336-2-iii@linux.ibm.com Signed-off-by: Sasha Levin <sashal@kernel.org>
We get the size of the trampoline image during the dry run phase and
allocate memory based on that size. The allocated image will then be
populated with instructions during the real patch phase. But after
commit 26ef208c209a ("bpf: Use arch_bpf_trampoline_size"), the `im`
argument is inconsistent in the dry run and real patch phase. This may
cause emit_imm in RV64 to generate a different number of instructions
when generating the 'im' address, potentially causing out-of-bounds
issues. Let's emit the maximum number of instructions for the "im"
address during dry run to fix this problem.
TX power index is read from the efuse on init, the values get written to
the TX power registers when the channel gets switched.
When the chip has not yet been calibrated, the efuse values are 0xFF,
which on some boards leads to USB timeouts for reading/writing registers
after the first frames have been sent.
The vendor driver (v5.11.5-1) checks for these invalid values and sets
default values instead. Implement something similar in
rtl8188fu_parse_efuse().
Fixes: c888183b21f3 ("wifi: rtl8xxxu: Support new chip RTL8188FU") Signed-off-by: Martin Kaistra <martin.kaistra@linutronix.de> Reviewed-by: Ping-Ke Shih <pkshih@realtek.com> Signed-off-by: Ping-Ke Shih <pkshih@realtek.com> Link: https://patch.msgid.link/20240624140037.231657-1-martin.kaistra@linutronix.de Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently, the Rx data path only supports parsing peer metadata of version
zero. However, the QCN9274 platform configures the peer metadata version
as V1B. When V1B peer metadata is parsed using the version zero logic,
invalid data is populated, causing valid packets to be dropped. To address
this issue, refactor the peer metadata version and add the version based
parsing to populate the data from peer metadata correctly.
Fixes: 287033810990 ("wifi: ath12k: add support for peer meta data version") Signed-off-by: Karthikeyan Periyasamy <quic_periyasa@quicinc.com> Acked-by: Jeff Johnson <quic_jjohnson@quicinc.com> Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com> Link: https://patch.msgid.link/20240624145418.2043461-1-quic_periyasa@quicinc.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Advertise the driver support for multiple BSSID (MBSSID) and
enhanced multi-BSSID advertisements (EMA) by setting extended
capabilities.
Configure mbssid_max_interfaces and ema_max_profile_periodicity
fields in structure wiphy which are used to advertise maximum number
of interfaces and profile periodicity supported by the driver.
Add new WMI fields to configure maximum vdev count supported for
MBSSID and profile periodicity in case of EMA.
Set WMI_RSRC_CFG_FLAGS2_CALC_NEXT_DTIM_COUNT_SET flag to allow
firmware to track and update the DTIM counts for each nontransmitted
profile.
When updating a channel context, the code can apply wider
bandwidth TDLS STA channel definitions to each and every
channel context used by the device, an approach that will
surely lead to problems if there is ever more than one.
Restrict the wider BW TDLS STA consideration to only TDLS
STAs that are actually related to links using the channel
context being updated.
When offloading xfrm states to hardware, the offloading
device is attached to the skbs secpath. If a skb is free
is deferred, an unregister netdevice hangs because the
netdevice is still refcounted.
Fix this by removing the netdevice from the xfrm states
when the netdevice is unregistered. To find all xfrm states
that need to be cleared we add another list where skbs
linked to that are unlinked from the lists (deleted)
but not yet freed.
When upgrading to libbpf 1.3 we noticed a big performance hit while
loading programs using CORE on non base-BTF symbols. This was tracked
down to the new BTF sanity check logic. The issue is the base BTF
definitions are checked first for the base BTF and then again for every
module BTF.
Loading 5 dummy programs (using libbpf-rs) that are using CORE on a
non-base BTF symbol on my system:
- Before this fix: 3s.
- With this fix: 0.1s.
Fix this by only checking the types starting at the BTF start id. This
should ensure the base BTF is still checked as expected but only once
(btf->start_id == 1 when creating the base BTF), and then only
additional types are checked for each module BTF.
$ make test_progs V=1
[...]
./tools/sbin/bpftool gen object ./ip_check_defrag.bpf.linked2.o ./ip_check_defrag.bpf.linked1.o
libbpf: failed to find symbol for variable 'bpf_dynptr_slice' in section '.ksyms'
Error: failed to link './ip_check_defrag.bpf.linked1.o': No such file or directory (2)
[...]
Upon investigation, I discovered that the btf_types referenced in the '.ksyms'
section had a kind of BTF_KIND_FUNC instead of BTF_KIND_VAR:
The KSZ9477 allows HSR in-HW offloading for any of two selected ports.
This patch adds check if one tries to use more than two ports with
HSR offloading enabled.
The problem is with RedBox configuration (HSR-SAN) - when configuring:
ip link add name hsr0 type hsr slave1 lan1 slave2 lan2 interlink lan3 \
supervision 45 version 1
The lan1 (port0) and lan2 (port1) are correctly configured as ports, which
can use HSR offloading on ksz9477.
However, when we do already have two bits set in hsr_ports, we need to
return (-ENOTSUPP), so the interlink port (lan3) would be used with
SW based HSR RedBox support.
Otherwise, I do see some strange network behavior, as some HSR frames are
visible on non-HSR network and vice versa.
This causes the switch connected to interlink port (lan3) to drop frames
and no communication is possible.
Moreover, conceptually - the interlink (i.e. HSR-SAN port - lan3/port2)
shall be only supported in software as it is also possible to use ksz9477
with only SW based HSR (i.e. port0/1 -> hsr0 with offloading, port2 ->
HSR-SAN/interlink, port4/5 -> hsr1 with SW based HSR).
Fixes: 5055cccfc2d1 ("net: hsr: Provide RedBox support (HSR-SAN)") Signed-off-by: Lukasz Majewski <lukma@denx.de> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Reviewed-by: Dan Carpenter <dan.carpenter@linaro.org> Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
FEC_ECR_EN1588 bit gets cleared after MAC reset in `fec_stop()`, which
makes all 1588 functionality shut down, and all the extended registers
disappear, on link-down, making the adapter fall back to compatibility
"dumb mode". However, some functionality needs to be retained (e.g. PPS)
even without link.
Fixes: 6605b730c061 ("FEC: Add time stamping code and a PTP hardware clock") Cc: Richard Cochran <richardcochran@gmail.com> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Link: https://lore.kernel.org/netdev/5fa9fadc-a89d-467a-aae9-c65469ff5fe1@lunn.ch/ Signed-off-by: Csókás, Bence <csokas.bence@prolan.hu> Reviewed-by: Wei Fang <wei.fang@nxp.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Syzbot reports uninitialized memory access in udf_rename() when updating
checksum of '..' directory entry of a moved directory. This is indeed
true as we pass on-stack diriter.fi to the udf_update_tag() and because
that has only struct fileIdentDesc included in it and not the impUse or
name fields, the checksumming function is going to checksum random stack
contents beyond the end of the structure. This is actually harmless
because the following udf_fiiter_write_fi() will recompute the checksum
from on-disk buffers where everything is properly included. So all that
is needed is just removing the bogus calculation.
Previously, the offload data path decrypted the packet before checking
the direction, leading to error logging and packet dropping. However,
dropped packets wouldn't be visible in tcpdump or audit log.
With this fix, the offload path, upon noticing SA direction mismatch,
will pass the packet to the stack without decrypting it. The L3 layer
will then log the error, audit, and drop ESP without decrypting or
decapsulating it.
This also ensures that the slow path records the error and audit log,
making dropped packets visible in tcpdump.
Fixes: 304b44f0d5a4 ("xfrm: Add dir validation to "in" data path lookup") Signed-off-by: Antony Antony <antony.antony@secunet.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
We will soon be generating kfunc prototypes from BTF. As part of that,
we need to align the manual signatures in bpf_kfuncs.h with the actual
kfunc definitions. There is currently a conflicting signature for
bpf_session_cookie() w.r.t. return type.
The original intent was to return long * and not __u64 *. You can see
evidence of that intent in a3a5113393cc ("selftests/bpf: Add kprobe
session cookie test").
If non-BSS and remain-on-channel (ROC) blocking were to occur
simultaneously, they'd step on each other's toes, unblocking
when not yet supported. Disentangle these bits, and ROC doesn't
need to use the non_bss_link() function then.
Fixes: a1efeb823084 ("wifi: iwlwifi: mvm: Block EMLSR when a p2p/softAP vif is active") Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com> Link: https://msgid.link/20240605140556.461fcf7b95bb.Id0d21dcb739d426ff15ec068b5df8abaab58884d@changeid Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When EMLSR gets unblocked, the current code checks if the last exit was
due to an EXIT reason (as opposed to a BLOCKING one), and if so, it
does nothing, as in this case a MLO scan was scheduled to run in 30
seconds.
But the code doesn't consider the time that passed from the last exit,
so if immediately after the exit a blocker occurred (e.g. non-BSS
interface), and lasts for more than 30 seconds, then the MLO scan and the
following link selection will decide not to enter EMLSR, and when the
unblocking event finally happens, the reason is still set to the EXIT one,
so it will do nothing, and we will not have the chance to re-enable EMLSR.
Fix this by checking also the time that has passed since the last exit,
only if it is less than 30 seconds, we can count on the scheduled MLO
scan.
Note that clearing the reason itself can't be done since it is needed
for the EMLSR prevention mechanism.
Fixes: 2f33561ea8f9 ("wifi: iwlwifi: mvm: trigger link selection after exiting EMLSR") Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com> Reviewed-by: Johannes Berg <johannes.berg@intel.com> Link: https://msgid.link/20240605140327.58556fc4cfa9.I4c55b3cd9f20b21b37f28258d0fb6842ba413966@changeid Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Function ath12k_debugfs_register() is called once inside the function
ath12k_mac_hw_register(). However, with single wiphy model, there could
be multiple pdevs registered under single hardware. Hence, need to register
debugfs for each one of them.
Move the caller inside the loop which iterates over all underlying pdevs.
Currently for CCMP256, GCMP128 and GCMP256 ciphers, in ath11k_install_key()
IEEE80211_KEY_FLAG_GENERATE_IV_MGMT is not set. And in ath11k_mac_mgmt_tx_wmi()
a length of IEEE80211_CCMP_MIC_LEN is reserved for all ciphers.
This results in unexpected management frame drop in case either of above 3 ciphers
is used. The reason is, without IEEE80211_KEY_FLAG_GENERATE_IV_MGMT set, mac80211
will not generate CCMP/GCMP headers in frame for ath11k. Also MIC length reserved
is wrong. Such frame is dropped later by hardware:
ath11k_pci 0000:5a:00.0: mac tx mgmt frame, buf id 0
ath11k_pci 0000:5a:00.0: mgmt tx compl ev pdev_id 1, desc_id 0, status 1
From user point of view, we have observed very low throughput due to this issue:
action frames are all dropped so ADDBA response from DUT never reaches AP. AP
can not use aggregation thus throughput is low.
Fix this by setting IEEE80211_KEY_FLAG_GENERATE_IV_MGMT flag and by reserving proper
MIC length for those ciphers.
Currently ACPI notification handler is installed when driver loads and only
gets removed when driver unloads. During resume after firmware is reloaded,
ath12k tries to install it by default. Since it is installed already, ACPI
subsystem rejects it and returns an error:
Fix it by removing that handler when going to suspend. This also avoid any
possible ACPI call to firmware before firmware is reloaded/reinitialized.
Note ab->acpi also needs to be cleared in ath12k_acpi_stop() such that we
are in a clean state when ACPI structures are reinitialized in
ath12k_acpi_start().
Fixes: 576771c9fa21 ("wifi: ath12k: ACPI TAS support") Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com> Acked-by: Jeff Johnson <quic_jjohnson@quicinc.com> Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com> Link: https://msgid.link/20240531024000.9291-1-quic_bqiang@quicinc.com Signed-off-by: Sasha Levin <sashal@kernel.org>
The commit which tried to fix the concurrency issues of concurrent
static_key_slow_inc() failed to fix the equivalent issues
vs. static_key_slow_dec():
CPU0 CPU1
static_key_slow_dec()
static_key_slow_try_dec()
key->enabled == 1
val = atomic_fetch_add_unless(&key->enabled, -1, 1);
if (val == 1)
return false;
jump_label_lock();
if (atomic_dec_and_test(&key->enabled)) {
--> key->enabled == 0
__jump_label_update()
static_key_slow_dec()
static_key_slow_try_dec()
key->enabled == 0
val = atomic_fetch_add_unless(&key->enabled, -1, 1);
--> key->enabled == -1 <- FAIL
There is another bug in that code, when there is a concurrent
static_key_slow_inc() which enables the key as that sets key->enabled to -1
so on the other CPU
val = atomic_fetch_add_unless(&key->enabled, -1, 1);
will succeed and decrement to -2, which is invalid.
Cure all of this by replacing the atomic_fetch_add_unless() with a
atomic_try_cmpxchg() loop similar to static_key_fast_inc_not_disabled().
[peterz: add WARN_ON_ONCE for the -1 race] Fixes: 4c5ea0a9cd02 ("locking/static_key: Fix concurrent static_key_slow_inc()") Reported-by: Yue Sun <samsun1006219@gmail.com> Reported-by: Xingwei Lee <xrivendell7@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lkml.kernel.org/r/20240610124406.422897838@linutronix.de Signed-off-by: Sasha Levin <sashal@kernel.org>
Yue and Xingwei reported a jump label failure. It's caused by the lack of
serialization in set_attr_rdpmc():
CPU0 CPU1
Assume: x86_pmu.attr_rdpmc == 0
if (val != x86_pmu.attr_rdpmc) {
if (val == 0)
...
else if (x86_pmu.attr_rdpmc == 0)
static_branch_dec(&rdpmc_never_available_key);
if (val != x86_pmu.attr_rdpmc) {
if (val == 0)
...
else if (x86_pmu.attr_rdpmc == 0)
FAIL, due to imbalance ---> static_branch_dec(&rdpmc_never_available_key);
The reported BUG() is a consequence of the above and of another bug in the
jump label core code. The core code needs a separate fix, but that cannot
prevent the imbalance problem caused by set_attr_rdpmc().
Prevent this by serializing set_attr_rdpmc() locally.
Fixes: a66734297f78 ("perf/x86: Add /sys/devices/cpu/rdpmc=2 to allow rdpmc for all tasks") Closes: https://lore.kernel.org/r/CAEkJfYNzfW1vG=ZTMdz_Weoo=RXY1NDunbxnDaLyj8R4kEoE_w@mail.gmail.com Reported-by: Yue Sun <samsun1006219@gmail.com> Reported-by: Xingwei Lee <xrivendell7@gmail.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lkml.kernel.org/r/20240610124406.359476013@linutronix.de Signed-off-by: Sasha Levin <sashal@kernel.org>
ACLs that reside in the algorithmic TCAM (A-TCAM) in Spectrum-2 and
newer ASICs can share the same mask if their masks only differ in up to
8 consecutive bits. For example, consider the following filters:
# tc filter add dev swp1 ingress pref 1 proto ip flower dst_ip 192.0.2.0/24 action drop
# tc filter add dev swp1 ingress pref 1 proto ip flower dst_ip 198.51.100.128/25 action drop
The second filter can use the same mask as the first (dst_ip/24) with a
delta of 1 bit.
However, the above only works because the two filters have different
values in the common unmasked part (dst_ip/24). When entries have the
same value in the common unmasked part they create undesired collisions
in the device since many entries now have the same key. This leads to
firmware errors such as [1] and to a reduced scale.
Fix by adjusting the hash table key to only include the value in the
common unmasked part. That is, without including the delta bits. That
way the driver will detect the collision during filter insertion and
spill the filter into the circuit TCAM (C-TCAM).
Add a test case that fails without the fix and adjust existing cases
that check C-TCAM spillage according to the above limitation.
[1]
mlxsw_spectrum2 0000:06:00.0: EMAD reg access failed (tid=3379b18a00003394,reg_id=3027(ptce3),type=write,status=8(resource not available))
Fixes: c22291f7cf45 ("mlxsw: spectrum: acl: Implement delta for ERP") Reported-by: Alexander Zubkov <green@qrator.net> Signed-off-by: Ido Schimmel <idosch@nvidia.com> Reviewed-by: Amit Cohen <amcohen@nvidia.com> Tested-by: Alexander Zubkov <green@qrator.net> Signed-off-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
ACLs in Spectrum-2 and newer ASICs can reside in the algorithmic TCAM
(A-TCAM) or in the ordinary circuit TCAM (C-TCAM). The former can
contain more ACLs (i.e., tc filters), but the number of masks in each
region (i.e., tc chain) is limited.
In order to mitigate the effects of the above limitation, the device
allows filters to share a single mask if their masks only differ in up
to 8 consecutive bits. For example, dst_ip/25 can be represented using
dst_ip/24 with a delta of 1 bit. The C-TCAM does not have a limit on the
number of masks being used (and therefore does not support mask
aggregation), but can contain a limited number of filters.
The driver uses the "objagg" library to perform the mask aggregation by
passing it objects that consist of the filter's mask and whether the
filter is to be inserted into the A-TCAM or the C-TCAM since filters in
different TCAMs cannot share a mask.
The set of created objects is dependent on the insertion order of the
filters and is not necessarily optimal. Therefore, the driver will
periodically ask the library to compute a more optimal set ("hints") by
looking at all the existing objects.
When the library asks the driver whether two objects can be aggregated
the driver only compares the provided masks and ignores the A-TCAM /
C-TCAM indication. This is the right thing to do since the goal is to
move as many filters as possible to the A-TCAM. The driver also forbids
two identical masks from being aggregated since this can only happen if
one was intentionally put in the C-TCAM to avoid a conflict in the
A-TCAM.
The above can result in the following set of hints:
H1: {mask X, A-TCAM} -> H2: {mask Y, A-TCAM} // X is Y + delta
H3: {mask Y, C-TCAM} -> H4: {mask Z, A-TCAM} // Y is Z + delta
After getting the hints from the library the driver will start migrating
filters from one region to another while consulting the computed hints
and instructing the device to perform a lookup in both regions during
the transition.
Assuming a filter with mask X is being migrated into the A-TCAM in the
new region, the hints lookup will return H1. Since H2 is the parent of
H1, the library will try to find the object associated with it and
create it if necessary in which case another hints lookup (recursive)
will be performed. This hints lookup for {mask Y, A-TCAM} will either
return H2 or H3 since the driver passes the library an object comparison
function that ignores the A-TCAM / C-TCAM indication.
This can eventually lead to nested objects which are not supported by
the library [1].
Fix by removing the object comparison function from both the driver and
the library as the driver was the only user. That way the lookup will
only return exact matches.
I do not have a reliable reproducer that can reproduce the issue in a
timely manner, but before the fix the issue would reproduce in several
minutes and with the fix it does not reproduce in over an hour.
Note that the current usefulness of the hints is limited because they
include the C-TCAM indication and represent aggregation that cannot
actually happen. This will be addressed in net-next.
Fixes: 9069a3817d82 ("lib: objagg: implement optimization hints assembly and use hints for object creation") Signed-off-by: Ido Schimmel <idosch@nvidia.com> Reviewed-by: Amit Cohen <amcohen@nvidia.com> Tested-by: Alexander Zubkov <green@qrator.net> Signed-off-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The library supports aggregation of objects into other objects only if
the parent object does not have a parent itself. That is, nesting is not
supported.
Aggregation happens in two cases: Without and with hints, where hints
are a pre-computed recommendation on how to aggregate the provided
objects.
Nesting is not possible in the first case due to a check that prevents
it, but in the second case there is no check because the assumption is
that nesting cannot happen when creating objects based on hints. The
violation of this assumption leads to various warnings and eventually to
a general protection fault [1].
Before fixing the root cause, error out when nesting happens and warn.
Fixes: 9069a3817d82 ("lib: objagg: implement optimization hints assembly and use hints for object creation") Reported-by: Alexander Zubkov <green@qrator.net> Signed-off-by: Ido Schimmel <idosch@nvidia.com> Reviewed-by: Amit Cohen <amcohen@nvidia.com> Tested-by: Alexander Zubkov <green@qrator.net> Signed-off-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Use preempt_model_preemptible() to detect a preemptible kernel when
deciding whether or not to reschedule in order to drop a contended
spinlock or rwlock. Because PREEMPT_DYNAMIC selects PREEMPTION, kernels
built with PREEMPT_DYNAMIC=y will yield contended locks even if the live
preemption model is "none" or "voluntary". In short, make kernels with
dynamically selected models behave the same as kernels with statically
selected models.
Somewhat counter-intuitively, NOT yielding a lock can provide better
latency for the relevant tasks/processes. E.g. KVM x86's mmu_lock, a
rwlock, is often contended between an invalidation event (takes mmu_lock
for write) and a vCPU servicing a guest page fault (takes mmu_lock for
read). For _some_ setups, letting the invalidation task complete even
if there is mmu_lock contention provides lower latency for *all* tasks,
i.e. the invalidation completes sooner *and* the vCPU services the guest
page fault sooner.
But even KVM's mmu_lock behavior isn't uniform, e.g. the "best" behavior
can vary depending on the host VMM, the guest workload, the number of
vCPUs, the number of pCPUs in the host, why there is lock contention, etc.
In other words, simply deleting the CONFIG_PREEMPTION guard (or doing the
opposite and removing contention yielding entirely) needs to come with a
big pile of data proving that changing the status quo is a net positive.
Opportunistically document this side effect of preempt=full, as yielding
contended spinlocks can have significant, user-visible impact.
Fixes: c597bfddc9e9 ("sched: Provide Kconfig support for default dynamic preempt mode") Signed-off-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Ankur Arora <ankur.a.arora@oracle.com> Reviewed-by: Chen Yu <yu.c.chen@intel.com> Link: https://lore.kernel.org/kvm/ef81ff36-64bb-4cfe-ae9b-e3acf47bff24@proxmox.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Move the declarations and inlined implementations of the preempt_model_*()
helpers to preempt.h so that they can be referenced in spinlock.h without
creating a potential circular dependency between spinlock.h and sched.h.
No functional change intended.
Signed-off-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Ankur Arora <ankur.a.arora@oracle.com> Link: https://lkml.kernel.org/r/20240528003521.979836-2-ankur.a.arora@oracle.com
Stable-dep-of: c793a62823d1 ("sched/core: Drop spinlocks on contention iff kernel is preemptible") Signed-off-by: Sasha Levin <sashal@kernel.org>
udf_evict_inode() calls udf_setsize() to truncate deleted inode.
However inode deletion through udf_evict_inode() can happen from inode
reclaim context and udf_setsize() grabs mapping->invalidate_lock which
isn't generally safe to acquire from fs reclaim context since we
allocate pages under mapping->invalidate_lock for example in a page
fault path. This is however not a real deadlock possibility as by the
time udf_evict_inode() is called, nobody can be accessing the inode,
even less work with its page cache. So this is just a lockdep triggering
false positive. Fix the problem by moving mapping->invalidate_lock
locking outsize of udf_setsize() into udf_setattr() as grabbing
mapping->invalidate_lock from udf_evict_inode() is pointless.
Reported-by: syzbot+0333a6f4b88bcd68a62f@syzkaller.appspotmail.com Fixes: b9a861fd527a ("udf: Protect truncate and file type conversion with invalidate_lock") Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Sasha Levin <sashal@kernel.org>
SG_MAX_SINGLE_ALLOC is used to limit maximum number of entries that
will be allocated in one piece of scatterlist. When the entries of
scatterlist exceeds SG_MAX_SINGLE_ALLOC, sg chain will be used. From
commit 7c703e54cc71 ("arch: switch the default on ARCH_HAS_SG_CHAIN"),
we can know that the macro CONFIG_ARCH_NO_SG_CHAIN is used to identify
whether sg chain is supported. So, SMC-R's rmb buffer should be limited
by SG_MAX_SINGLE_ALLOC only when the macro CONFIG_ARCH_NO_SG_CHAIN is
defined.
Fixes: a3fe3d01bd0d ("net/smc: introduce sg-logic for RMBs") Signed-off-by: Guangguan Wang <guangguan.wang@linux.alibaba.com> Co-developed-by: Wen Gu <guwen@linux.alibaba.com> Signed-off-by: Wen Gu <guwen@linux.alibaba.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Each object file contains information about which module it gets linked
into, so linking the same file into multiple modules now causes a warning:
scripts/Makefile.build:254: drivers/net/ethernet/hisilicon/hns3/Makefile: hns3_common/hclge_comm_cmd.o is added to multiple modules: hclge hclgevf
scripts/Makefile.build:254: drivers/net/ethernet/hisilicon/hns3/Makefile: hns3_common/hclge_comm_rss.o is added to multiple modules: hclge hclgevf
scripts/Makefile.build:254: drivers/net/ethernet/hisilicon/hns3/Makefile: hns3_common/hclge_comm_tqp_stats.o is added to multiple modules: hclge hclgevf
Change the way that hns3 is built by moving the three common files into a
separate module with exported symbols instead.
Fixes: 5f20be4e90e6 ("net: hns3: refactor hns3 makefile to support hns3_common module") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://lore.kernel.org/r/20240528161603.2443125-1-arnd@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
I noticed flakes in a packetdrill test, expecting an epoll_wait()
to return EPOLLERR | EPOLLHUP on a failed connect() attempt,
after multiple SYN retransmits. It sometimes return EPOLLERR only.
The issue is that tcp_write_err():
1) writes an error in sk->sk_err,
2) calls sk_error_report(),
3) then calls tcp_done().
tcp_done() is writing SHUTDOWN_MASK into sk->sk_shutdown,
among other things.
Problem is that the awaken user thread (from 2) sk_error_report())
might call tcp_poll() before tcp_done() has written sk->sk_shutdown.
tcp_poll() only sees a non zero sk->sk_err and returns EPOLLERR.
This patch fixes the issue by making sure to call sk_error_report()
after tcp_done().
tcp_write_err() also lacks an smp_wmb().
We can reuse tcp_done_with_error() to factor out the details,
as Neal suggested.
An earlier commit deleted the TSO support in the Cortina Gemini
driver because the driver was confusing gso_size and MTU,
probably because what the Linux kernel calls "gso_size" was
called "MTU" in the datasheet.
Restore the functionality properly reading the gso_size from
the skbuff.
Tested with iperf3, running a server on a different machine
and client on the device with the cortina gemini ethernet:
Several such packets often follow after each other verifying
the segmentation into 0x05a8 (1448) byte packages also on the
reveiving end. As can be seen, the ethernet frames are
0x05ea (1514) in size.
Performance with iperf3 before this patch: ~15.5 Mbit/s
Performance with iperf3 after this patch: ~175 Mbit/s
This was running a 60 second test (twice) the best measurement
was 179 Mbit/s.
For comparison if I run iperf3 with UDP I get around 1.05 Mbit/s
both before and after this patch.
While this is a gigabit ethernet interface, the CPU is a cheap
D-Link DIR-685 router (based on the ARMv5 Faraday FA526 at
~50 MHz), and the software is not supposed to drive traffic,
as the device has a DSA chip, so this kind of numbers can be
expected.
Fixes: ac631873c9e7 ("net: ethernet: cortina: Drop TSO support") Reviewed-by: Eric Dumazet <edumazet@google.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Base address of CE ring is defined as u32, currently this works
because DMA mask configured as 32 bit:
#define ATH12K_PCI_DMA_MASK 32
However this mask could be changed once firmware bugs are fixed
to fully support 36 bit DMA addressing. So to protect against any
future changes to the DMA mask, change the type of the fields that
are dependent upon it.
This is found during code review. Compile tested only.
Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices") Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com> Acked-by: Jeff Johnson <quic_jjohnson@quicinc.com> Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com> Link: https://msgid.link/20240524024021.37711-1-quic_bqiang@quicinc.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Base address of CE ring is defined as u32, currently this works
because coherent DMA mask configured as 32 bit:
#define ATH11K_PCI_COHERENT_DMA_MASK 32
However this mask could be changed once firmware bugs are fixed
to fully support 36 bit DMA addressing. So to protect against any
future changes to the DMA mask, change the type of the fields that
are dependent upon it.
This is found during code review. Compile tested only.
Fixes: d5c65159f289 ("ath11k: driver for Qualcomm IEEE 802.11ax devices") Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com> Acked-by: Jeff Johnson <quic_jjohnson@quicinc.com> Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com> Link: https://msgid.link/20240524021558.34452-1-quic_bqiang@quicinc.com Signed-off-by: Sasha Levin <sashal@kernel.org>
PCI PHY EQ value is missing after card off/on, so update the value after
device resume. The original commit only updates once at probe stage, which
could lead problem after suspend/resume.
The logic should be read a value from one register and write to another
register with a mask to avoid affecting unrelated bits.
When handling fragmented packets, the ath12k driver reassembles each
fragment into a normal packet and then reinjects it into the HW ring.
However, a firmware crash occurs during this reinjection process.
The issue arises because the driver populates peer metadata in
reo_ent_ring->queue_addr_lo, while the firmware expects the physical
address obtained from the corresponding peer’s queue descriptor. Fix it
by filling peer's queue descriptor's physical address in queue_addr_lo.
Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices") Signed-off-by: P Praneesh <quic_ppranees@quicinc.com> Acked-by: Jeff Johnson <quic_jjohnson@quicinc.com> Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com> Link: https://msgid.link/20240520070045.631029-4-quic_ppranees@quicinc.com Signed-off-by: Sasha Levin <sashal@kernel.org>
The monitor ring and the reo reinject ring share the same ring mask index.
When the driver receives an interrupt for the reo reinject ring, the
monitor ring is also processed, leading to invalid memory access. Since
monitor support is not yet enabled in ath12k, the ring mask for the monitor
ring should be removed.
Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices") Signed-off-by: P Praneesh <quic_ppranees@quicinc.com> Acked-by: Jeff Johnson <quic_jjohnson@quicinc.com> Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com> Link: https://msgid.link/20240520070045.631029-3-quic_ppranees@quicinc.com Signed-off-by: Sasha Levin <sashal@kernel.org>
For fragmented packets, ath12k reassembles each fragment as a normal
packet and then reinjects it into HW ring. In this case, the DMA
direction should be DMA_TO_DEVICE, not DMA_FROM_DEVICE. Otherwise,
an invalid payload may be reinjected into the HW and
subsequently delivered to the host.
Given that arbitrary memory can be allocated to the skb buffer,
knowledge about the data contained in the reinjected buffer is lacking.
Consequently, there’s a risk of private information being leaked.
The reason is, during resume, firmware is reinitialized but host does
not send current country code to firmware. So default reg rules with
country code set to 'na' is uploaded to host, as shown above.
Fix it by restoring country code to firmware during resume.
projects / thirdparty / kernel / stable.git / log
