| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Network driver documentation indicates we should be avoiding returning
NETDEV_TX_BUSY from ndo_start_xmit in normal cases, since it requires
the packets to be requeued. Instead the queue should be stopped after
a packet is added to the TX ring when there may not be enough room for an
additional one. Also, when TX ring entries are completed, we should only
wake the queue if we know there is room for another full maximally
fragmented packet.
Print a warning if there is insufficient space at the start of start_xmit,
since this should no longer happen.
Combined with increasing the default TX ring size (in a subsequent
patch), this appears to recover the TX performance lost by previous changes
to actually manage the TX ring state properly.
Fixes: 8a3b7a252dca9 ("drivers/net/ethernet/xilinx: added Xilinx AXI Ethernet driver")
Signed-off-by: Robert Hancock <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The check for the number of available TX ring slots was off by 1 since a
slot is required for the skb header as well as each fragment. This could
result in overwriting a TX ring slot that was still in use.
Fixes: 8a3b7a252dca9 ("drivers/net/ethernet/xilinx: added Xilinx AXI Ethernet driver")
Signed-off-by: Robert Hancock <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The check for whether a TX ring slot was available was incorrect,
since a slot which had been loaded with transmit data but the device had
not started transmitting would be treated as available, potentially
causing non-transmitted slots to be overwritten. The control field in
the descriptor should be checked, rather than the status field (which may
only be updated when the device completes the entry).
Fixes: 8a3b7a252dca9 ("drivers/net/ethernet/xilinx: added Xilinx AXI Ethernet driver")
Signed-off-by: Robert Hancock <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The driver will not work properly if the TX ring size is set to below
MAX_SKB_FRAGS + 1 since it needs to hold at least one full maximally
fragmented packet in the TX ring. Limit setting the ring size to below
this value.
Fixes: 8b09ca823ffb4 ("net: axienet: Make RX/TX ring sizes configurable")
Signed-off-by: Robert Hancock <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
This driver was missing some required memory barriers:
Use dma_rmb to ensure we see all updates to the descriptor after we see
that an entry has been completed.
Use wmb and rmb to avoid stale descriptor status between the TX path and
TX complete IRQ path.
Fixes: 8a3b7a252dca9 ("drivers/net/ethernet/xilinx: added Xilinx AXI Ethernet driver")
Signed-off-by: Robert Hancock <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
In some cases where the Xilinx Ethernet core was used in 1000Base-X or
SGMII modes, which use the internal PCS/PMA PHY, and the MGT
transceiver clock source for the PCS was not running at the time the
FPGA logic was loaded, the core would come up in a state where the
PCS could not be found on the MDIO bus. To fix this, the Ethernet core
(including the PCS) should be reset after enabling the clocks, prior to
attempting to access the PCS using of_mdio_find_device.
Fixes: 1a02556086fc (net: axienet: Properly handle PCS/PMA PHY for 1000BaseX mode)
Signed-off-by: Robert Hancock <[email protected]>
Reviewed-by: Andrew Lunn <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
When resetting the device, wait for the PhyRstCmplt bit to be set
in the interrupt status register before continuing initialization, to
ensure that the core is actually ready. When using an external PHY, this
also ensures we do not start trying to access the PHY while it is still
in reset. The PHY reset is initiated by the core reset which is
triggered just above, but remains asserted for 5ms after the core is
reset according to the documentation.
The MgtRdy bit could also be waited for, but unfortunately when using
7-series devices, the bit does not appear to work as documented (it
seems to behave as some sort of link state indication and not just an
indication the transceiver is ready) so it can't really be relied on for
this purpose.
Fixes: 8a3b7a252dca9 ("drivers/net/ethernet/xilinx: added Xilinx AXI Ethernet driver")
Signed-off-by: Robert Hancock <[email protected]>
Reviewed-by: Andrew Lunn <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The previous timeout of 1ms was too short to handle some cases where the
core is reset just after the input clocks were started, which will
be introduced in an upcoming patch. Increase the timeout to 50ms. Also
simplify the reset timeout checking to use read_poll_timeout.
Fixes: 8a3b7a252dca9 ("drivers/net/ethernet/xilinx: added Xilinx AXI Ethernet driver")
Signed-off-by: Robert Hancock <[email protected]>
Reviewed-by: Andrew Lunn <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Daniel Borkmann says:
====================
pull-request: bpf 2022-01-19
We've added 12 non-merge commits during the last 8 day(s) which contain
a total of 12 files changed, 262 insertions(+), 64 deletions(-).
The main changes are:
1) Various verifier fixes mainly around register offset handling when
passed to helper functions, from Daniel Borkmann.
2) Fix XDP BPF link handling to assert program type,
from Toke Høiland-Jørgensen.
3) Fix regression in mount parameter handling for BPF fs,
from Yafang Shao.
4) Fix incorrect integer literal when marking scratched stack slots
in verifier, from Christy Lee.
* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
bpf, selftests: Add ringbuf memory type confusion test
bpf, selftests: Add various ringbuf tests with invalid offset
bpf: Fix ringbuf memory type confusion when passing to helpers
bpf: Fix out of bounds access for ringbuf helpers
bpf: Generally fix helper register offset check
bpf: Mark PTR_TO_FUNC register initially with zero offset
bpf: Generalize check_ctx_reg for reuse with other types
bpf: Fix incorrect integer literal used for marking scratched stack.
bpf/selftests: Add check for updating XDP bpf_link with wrong program type
bpf/selftests: convert xdp_link test to ASSERT_* macros
xdp: check prog type before updating BPF link
bpf: Fix mount source show for bpffs
====================
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>
|
|
Add two tests, one which asserts that ring buffer memory can be passed to
other helpers for populating its entry area, and another one where verifier
rejects different type of memory passed to bpf_ringbuf_submit().
Signed-off-by: Daniel Borkmann <[email protected]>
Acked-by: John Fastabend <[email protected]>
Acked-by: Alexei Starovoitov <[email protected]>
|
|
Assert that the verifier is rejecting invalid offsets on the ringbuf entries:
# ./test_verifier | grep ring
#947/u ringbuf: invalid reservation offset 1 OK
#947/p ringbuf: invalid reservation offset 1 OK
#948/u ringbuf: invalid reservation offset 2 OK
#948/p ringbuf: invalid reservation offset 2 OK
Signed-off-by: Daniel Borkmann <[email protected]>
Acked-by: John Fastabend <[email protected]>
Acked-by: Alexei Starovoitov <[email protected]>
|
|
The bpf_ringbuf_submit() and bpf_ringbuf_discard() have ARG_PTR_TO_ALLOC_MEM
in their bpf_func_proto definition as their first argument, and thus both expect
the result from a prior bpf_ringbuf_reserve() call which has a return type of
RET_PTR_TO_ALLOC_MEM_OR_NULL.
While the non-NULL memory from bpf_ringbuf_reserve() can be passed to other
helpers, the two sinks (bpf_ringbuf_submit(), bpf_ringbuf_discard()) right now
only enforce a register type of PTR_TO_MEM.
This can lead to potential type confusion since it would allow other PTR_TO_MEM
memory to be passed into the two sinks which did not come from bpf_ringbuf_reserve().
Add a new MEM_ALLOC composable type attribute for PTR_TO_MEM, and enforce that:
- bpf_ringbuf_reserve() returns NULL or PTR_TO_MEM | MEM_ALLOC
- bpf_ringbuf_submit() and bpf_ringbuf_discard() only take PTR_TO_MEM | MEM_ALLOC
but not plain PTR_TO_MEM arguments via ARG_PTR_TO_ALLOC_MEM
- however, other helpers might treat PTR_TO_MEM | MEM_ALLOC as plain PTR_TO_MEM
to populate the memory area when they use ARG_PTR_TO_{UNINIT_,}MEM in their
func proto description
Fixes: 457f44363a88 ("bpf: Implement BPF ring buffer and verifier support for it")
Reported-by: Alexei Starovoitov <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Acked-by: John Fastabend <[email protected]>
Acked-by: Alexei Starovoitov <[email protected]>
|
|
Both bpf_ringbuf_submit() and bpf_ringbuf_discard() have ARG_PTR_TO_ALLOC_MEM
in their bpf_func_proto definition as their first argument. They both expect
the result from a prior bpf_ringbuf_reserve() call which has a return type of
RET_PTR_TO_ALLOC_MEM_OR_NULL.
Meaning, after a NULL check in the code, the verifier will promote the register
type in the non-NULL branch to a PTR_TO_MEM and in the NULL branch to a known
zero scalar. Generally, pointer arithmetic on PTR_TO_MEM is allowed, so the
latter could have an offset.
The ARG_PTR_TO_ALLOC_MEM expects a PTR_TO_MEM register type. However, the non-
zero result from bpf_ringbuf_reserve() must be fed into either bpf_ringbuf_submit()
or bpf_ringbuf_discard() but with the original offset given it will then read
out the struct bpf_ringbuf_hdr mapping.
The verifier missed to enforce a zero offset, so that out of bounds access
can be triggered which could be used to escalate privileges if unprivileged
BPF was enabled (disabled by default in kernel).
Fixes: 457f44363a88 ("bpf: Implement BPF ring buffer and verifier support for it")
Reported-by: <[email protected]> (SecCoder Security Lab)
Signed-off-by: Daniel Borkmann <[email protected]>
Acked-by: John Fastabend <[email protected]>
Acked-by: Alexei Starovoitov <[email protected]>
|
|
Right now the assertion on check_ptr_off_reg() is only enforced for register
types PTR_TO_CTX (and open coded also for PTR_TO_BTF_ID), however, this is
insufficient since many other PTR_TO_* register types such as PTR_TO_FUNC do
not handle/expect register offsets when passed to helper functions.
Given this can slip-through easily when adding new types, make this an explicit
allow-list and reject all other current and future types by default if this is
encountered.
Also, extend check_ptr_off_reg() to handle PTR_TO_BTF_ID as well instead of
duplicating it. For PTR_TO_BTF_ID, reg->off is used for BTF to match expected
BTF ids if struct offset is used. This part still needs to be allowed, but the
dynamic off from the tnum must be rejected.
Fixes: 69c087ba6225 ("bpf: Add bpf_for_each_map_elem() helper")
Fixes: eaa6bcb71ef6 ("bpf: Introduce bpf_per_cpu_ptr()")
Signed-off-by: Daniel Borkmann <[email protected]>
Acked-by: John Fastabend <[email protected]>
Acked-by: Alexei Starovoitov <[email protected]>
|
|
Similar as with other pointer types where we use ldimm64, clear the register
content to zero first, and then populate the PTR_TO_FUNC type and subprogno
number. Currently this is not done, and leads to reuse of stale register
tracking data.
Given for special ldimm64 cases we always clear the register offset, make it
common for all cases, so it won't be forgotten in future.
Fixes: 69c087ba6225 ("bpf: Add bpf_for_each_map_elem() helper")
Signed-off-by: Daniel Borkmann <[email protected]>
Acked-by: John Fastabend <[email protected]>
Acked-by: Alexei Starovoitov <[email protected]>
|
|
Generalize the check_ctx_reg() helper function into a more generic named one
so that it can be reused for other register types as well to check whether
their offset is non-zero. No functional change.
Signed-off-by: Daniel Borkmann <[email protected]>
Acked-by: John Fastabend <[email protected]>
Acked-by: Alexei Starovoitov <[email protected]>
|
|
When under stress, cleanup_net() can have to dismantle
netns in big numbers. ops_exit_list() currently calls
many helpers [1] that have no schedule point, and we can
end up with soft lockups, particularly on hosts
with many cpus.
Even for moderate amount of netns processed by cleanup_net()
this patch avoids latency spikes.
[1] Some of these helpers like fib_sync_up() and fib_sync_down_dev()
are very slow because net/ipv4/fib_semantics.c uses host-wide hash tables,
and ifindex is used as the only input of two hash functions.
ifindexes tend to be the same for all netns (lo.ifindex==1 per instance)
This will be fixed in a separate patch.
Fixes: 72ad937abd0a ("net: Add support for batching network namespace cleanups")
Signed-off-by: Eric Dumazet <[email protected]>
Cc: Eric W. Biederman <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Commit 7cfa9c92d0a3 ("net: sfp: avoid power switch on address-change
modules") unintetionally changed the semantics for high power modules
without the digital diagnostics monitoring. We repeatedly attempt to
read the power status from the non-existing 0xa2 address in a futile
hope this failure is temporary:
[ 8.856051] sfp sfp-eth3: module NTT 0000000000000000 rev 0000 sn 0000000000000000 dc 160408
[ 8.865843] mvpp2 f4000000.ethernet eth3: switched to inband/1000base-x link mode
[ 8.873469] sfp sfp-eth3: Failed to read EEPROM: -5
[ 8.983251] sfp sfp-eth3: Failed to read EEPROM: -5
[ 9.103250] sfp sfp-eth3: Failed to read EEPROM: -5
We previosuly assumed such modules were powered up in the correct mode,
continuing without further configuration as long as the required power
class was supported by the host.
Restore this behaviour, while preserving the intent of subsequent
patches to avoid the "Address Change Sequence not supported" warning
if we are not going to be accessing the DDM address.
Fixes: 7cfa9c92d0a3 ("net: sfp: avoid power switch on address-change modules")
Reported-by: 照山周一郎 <[email protected]>
Tested-by: 照山周一郎 <[email protected]>
Signed-off-by: Russell King (Oracle) <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Gal Pressman says:
====================
net: Couple of skb memory leak fixes
As discussed in:
https://lore.kernel.org/netdev/[email protected]/
These are the two followup suggestions from Eric and Jakub.
Patch #1 adds a sk_defer_free_flush() call to the kTLS splice_read
handler.
Patch #2 verifies the defer list is empty on socket destroy, and calls a
defer free flush as well.
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
The cited Fixes patch moved to a deferred skb approach where the skbs
are not freed immediately under the socket lock. Add a WARN_ON_ONCE()
to verify the deferred list is empty on socket destroy, and empty it to
prevent potential memory leaks.
Fixes: f35f821935d8 ("tcp: defer skb freeing after socket lock is released")
Signed-off-by: Gal Pressman <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
This patch is a followup to
commit ffef737fd037 ("net/tls: Fix skb memory leak when running kTLS traffic")
Which was missing another sk_defer_free_flush() call in
tls_sw_splice_read().
Fixes: f35f821935d8 ("tcp: defer skb freeing after socket lock is released")
Signed-off-by: Gal Pressman <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
In the blamed commit, the call to the function
switchdev_bridge_port_offload was passing the wrong argument for
atomic_nb. It was ocelot_netdevice_nb instead of ocelot_swtchdev_nb.
This patch fixes this issue.
Fixes: 4e51bf44a03af6 ("net: bridge: move the switchdev object replay helpers to "push" mode")
Signed-off-by: Horatiu Vultur <[email protected]>
Reviewed-by: Vladimir Oltean <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Wrong hash sends single stream to multiple output interfaces.
The offset calculation was relative to skb->head, fix it to be relative
to skb->data.
Fixes: a815bde56b15 ("net, bonding: Refactor bond_xmit_hash for use with
xdp_buff")
Reviewed-by: Jussi Maki <[email protected]>
Reviewed-by: Saeed Mahameed <[email protected]>
Reviewed-by: Gal Pressman <[email protected]>
Signed-off-by: Moshe Tal <[email protected]>
Acked-by: Jay Vosburgh <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
A hung_task is observed when removing SMC-R devices. Suppose that
a link group has two active links(lnk_A, lnk_B) associated with two
different SMC-R devices(dev_A, dev_B). When dev_A is removed, the
link group will be removed from smc_lgr_list and added into
lgr_linkdown_list. lnk_A will be cleared and smcibdev(A)->lnk_cnt
will reach to zero. However, when dev_B is removed then, the link
group can't be found in smc_lgr_list and lnk_B won't be cleared,
making smcibdev->lnk_cnt never reaches zero, which causes a hung_task.
This patch fixes this issue by restoring the implementation of
smc_smcr_terminate_all() to what it was before commit 349d43127dac
("net/smc: fix kernel panic caused by race of smc_sock"). The original
implementation also satisfies the intention that make sure QP destroy
earlier than CQ destroy because we will always wait for smcibdev->lnk_cnt
reaches zero, which guarantees QP has been destroyed.
Fixes: 349d43127dac ("net/smc: fix kernel panic caused by race of smc_sock")
Signed-off-by: Wen Gu <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
In the past, free_fib_info() was supposed to be called
under RTNL protection.
This eventually was no longer the case.
Instead of enforcing RTNL it seems we simply can
move fib_info_cnt changes to occur when fib_info_lock
is held.
v2: David Laight suggested to update fib_info_cnt
only when an entry is added/deleted to/from the hash table,
as fib_info_cnt is used to make sure hash table size
is optimal.
BUG: KCSAN: data-race in fib_create_info / free_fib_info
write to 0xffffffff86e243a0 of 4 bytes by task 26429 on cpu 0:
fib_create_info+0xe78/0x3440 net/ipv4/fib_semantics.c:1428
fib_table_insert+0x148/0x10c0 net/ipv4/fib_trie.c:1224
fib_magic+0x195/0x1e0 net/ipv4/fib_frontend.c:1087
fib_add_ifaddr+0xd0/0x2e0 net/ipv4/fib_frontend.c:1109
fib_netdev_event+0x178/0x510 net/ipv4/fib_frontend.c:1466
notifier_call_chain kernel/notifier.c:83 [inline]
raw_notifier_call_chain+0x53/0xb0 kernel/notifier.c:391
__dev_notify_flags+0x1d3/0x3b0
dev_change_flags+0xa2/0xc0 net/core/dev.c:8872
do_setlink+0x810/0x2410 net/core/rtnetlink.c:2719
rtnl_group_changelink net/core/rtnetlink.c:3242 [inline]
__rtnl_newlink net/core/rtnetlink.c:3396 [inline]
rtnl_newlink+0xb10/0x13b0 net/core/rtnetlink.c:3506
rtnetlink_rcv_msg+0x745/0x7e0 net/core/rtnetlink.c:5571
netlink_rcv_skb+0x14e/0x250 net/netlink/af_netlink.c:2496
rtnetlink_rcv+0x18/0x20 net/core/rtnetlink.c:5589
netlink_unicast_kernel net/netlink/af_netlink.c:1319 [inline]
netlink_unicast+0x5fc/0x6c0 net/netlink/af_netlink.c:1345
netlink_sendmsg+0x726/0x840 net/netlink/af_netlink.c:1921
sock_sendmsg_nosec net/socket.c:704 [inline]
sock_sendmsg net/socket.c:724 [inline]
____sys_sendmsg+0x39a/0x510 net/socket.c:2409
___sys_sendmsg net/socket.c:2463 [inline]
__sys_sendmsg+0x195/0x230 net/socket.c:2492
__do_sys_sendmsg net/socket.c:2501 [inline]
__se_sys_sendmsg net/socket.c:2499 [inline]
__x64_sys_sendmsg+0x42/0x50 net/socket.c:2499
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x44/0xd0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
read to 0xffffffff86e243a0 of 4 bytes by task 31505 on cpu 1:
free_fib_info+0x35/0x80 net/ipv4/fib_semantics.c:252
fib_info_put include/net/ip_fib.h:575 [inline]
nsim_fib4_rt_destroy drivers/net/netdevsim/fib.c:294 [inline]
nsim_fib4_rt_replace drivers/net/netdevsim/fib.c:403 [inline]
nsim_fib4_rt_insert drivers/net/netdevsim/fib.c:431 [inline]
nsim_fib4_event drivers/net/netdevsim/fib.c:461 [inline]
nsim_fib_event drivers/net/netdevsim/fib.c:881 [inline]
nsim_fib_event_work+0x15ca/0x2cf0 drivers/net/netdevsim/fib.c:1477
process_one_work+0x3fc/0x980 kernel/workqueue.c:2298
process_scheduled_works kernel/workqueue.c:2361 [inline]
worker_thread+0x7df/0xa70 kernel/workqueue.c:2447
kthread+0x2c7/0x2e0 kernel/kthread.c:327
ret_from_fork+0x1f/0x30
value changed: 0x00000d2d -> 0x00000d2e
Reported by Kernel Concurrency Sanitizer on:
CPU: 1 PID: 31505 Comm: kworker/1:21 Not tainted 5.16.0-rc6-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Workqueue: events nsim_fib_event_work
Fixes: 48bb9eb47b27 ("netdevsim: fib: Add dummy implementation for FIB offload")
Signed-off-by: Eric Dumazet <[email protected]>
Reported-by: syzbot <[email protected]>
Cc: David Laight <[email protected]>
Cc: Ido Schimmel <[email protected]>
Cc: Jiri Pirko <[email protected]>
Reviewed-by: Ido Schimmel <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The declaration of smc_wr_tx_dismiss_slots() is unused.
So remove it.
Fixes: 349d43127dac ("net/smc: fix kernel panic caused by race of smc_sock")
Signed-off-by: Wen Gu <[email protected]>
Reviewed-by: Dust Li <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
In pci_generic.c there is a 'mru_default' in struct mhi_pci_dev_info.
This value shall be used for whole mhi if it's given a value for a specific product.
But in function mhi_net_rx_refill_work(), it's still using hard code value MHI_DEFAULT_MRU.
'mru_default' shall have higher priority than MHI_DEFAULT_MRU.
And after checking, this change could help fix a data connection lost issue.
Fixes: 5c2c85315948 ("bus: mhi: pci-generic: configurable network interface MRU")
Signed-off-by: Shujun Wang <[email protected]>
Signed-off-by: Slark Xiao <[email protected]>
Reviewed-by: Loic Poulain <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Existing genphy_loopback() is not applicable for Marvell PHY. Besides
configuring bit-6 and bit-13 in Page 0 Register 0 (Copper Control
Register), it is also required to configure same bits in Page 2
Register 21 (MAC Specific Control Register 2) according to speed of
the loopback is operating.
Tested working on Marvell88E1510 PHY for all speeds (1000/100/10Mbps).
FIXME: Based on trial and error test, it seem 1G need to have delay between
soft reset and loopback enablement.
Fixes: 014068dcb5b1 ("net: phy: genphy_loopback: add link speed configuration")
Cc: <[email protected]> # 5.15.x
Signed-off-by: Mohammad Athari Bin Ismail <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
A dma_request_chan() call is hidden in emac_configure_dma().
It must be released in the probe if an error occurs, as already done in
the remove function.
Add the corresponding dma_release_channel() call.
Fixes: 47869e82c8b8 ("sun4i-emac.c: add dma support")
Signed-off-by: Christophe JAILLET <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Clang static analysis reports this problem
mtk_eth_soc.c:394:7: warning: Branch condition evaluates
to a garbage value
if (err)
^~~
err is not initialized and only conditionally set.
So intitialize err.
Fixes: 7e538372694b ("net: ethernet: mediatek: Re-add support SGMII")
Signed-off-by: Tom Rix <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The following command sequence:
tc qdisc del dev swp0 clsact
tc qdisc add dev swp0 ingress_block 1 clsact
tc qdisc add dev swp1 ingress_block 1 clsact
tc filter add block 1 flower action drop
tc qdisc del dev swp0 clsact
produces the following NPD:
Unable to handle kernel NULL pointer dereference at virtual address 0000000000000014
pc : vcap_entry_set+0x14/0x70
lr : ocelot_vcap_filter_del+0x198/0x234
Call trace:
vcap_entry_set+0x14/0x70
ocelot_vcap_filter_del+0x198/0x234
ocelot_cls_flower_destroy+0x94/0xe4
felix_cls_flower_del+0x70/0x84
dsa_slave_setup_tc_block_cb+0x13c/0x60c
dsa_slave_setup_tc_block_cb_ig+0x20/0x30
tc_setup_cb_reoffload+0x44/0x120
fl_reoffload+0x280/0x320
tcf_block_playback_offloads+0x6c/0x184
tcf_block_unbind+0x80/0xe0
tcf_block_setup+0x174/0x214
tcf_block_offload_cmd.isra.0+0x100/0x13c
tcf_block_offload_unbind+0x5c/0xa0
__tcf_block_put+0x54/0x174
tcf_block_put_ext+0x5c/0x74
clsact_destroy+0x40/0x60
qdisc_destroy+0x4c/0x150
qdisc_put+0x70/0x90
qdisc_graft+0x3f0/0x4c0
tc_get_qdisc+0x1cc/0x364
rtnetlink_rcv_msg+0x124/0x340
The reason is that the driver isn't prepared to receive two tc filters
with the same cookie. It unconditionally creates a new struct
ocelot_vcap_filter for each tc filter, and it adds all filters with the
same identifier (cookie) to the ocelot_vcap_block.
The problem is here, in ocelot_vcap_filter_del():
/* Gets index of the filter */
index = ocelot_vcap_block_get_filter_index(block, filter);
if (index < 0)
return index;
/* Delete filter */
ocelot_vcap_block_remove_filter(ocelot, block, filter);
/* Move up all the blocks over the deleted filter */
for (i = index; i < block->count; i++) {
struct ocelot_vcap_filter *tmp;
tmp = ocelot_vcap_block_find_filter_by_index(block, i);
vcap_entry_set(ocelot, i, tmp);
}
what will happen is ocelot_vcap_block_get_filter_index() will return the
index (@index) of the first filter found with that cookie. This is _not_
the index of _this_ filter, but the other one with the same cookie,
because ocelot_vcap_filter_equal() gets fooled.
Then later, ocelot_vcap_block_remove_filter() is coded to remove all
filters that are ocelot_vcap_filter_equal() with the passed @filter.
So unexpectedly, both filters get deleted from the list.
Then ocelot_vcap_filter_del() will attempt to move all the other filters
up, again finding them by index (@i). The block count is 2, @index was 0,
so it will attempt to move up filter @i=0 and @i=1. It assigns tmp =
ocelot_vcap_block_find_filter_by_index(block, i), which is now a NULL
pointer because ocelot_vcap_block_remove_filter() has removed more than
one filter.
As far as I can see, this problem has been there since the introduction
of tc offload support, however I cannot test beyond the blamed commit
due to hardware availability. In any case, any fix cannot be backported
that far, due to lots of changes to the code base.
Therefore, let's go for the correct solution, which is to not call
ocelot_vcap_filter_add() and ocelot_vcap_filter_del(), unless the filter
is actually unique and not shared. For the shared filters, we should
just modify the ingress port mask and call ocelot_vcap_filter_replace(),
a function introduced by commit 95706be13b9f ("net: mscc: ocelot: create
a function that replaces an existing VCAP filter"). This way,
block->rules will only contain filters with unique cookies, by design.
Fixes: 07d985eef073 ("net: dsa: felix: Wire up the ocelot cls_flower methods")
Signed-off-by: Vladimir Oltean <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
wait_for_unix_gc() reads unix_tot_inflight & gc_in_progress
without synchronization.
Adds READ_ONCE()/WRITE_ONCE() and their associated comments
to better document the intent.
BUG: KCSAN: data-race in unix_inflight / wait_for_unix_gc
write to 0xffffffff86e2b7c0 of 4 bytes by task 9380 on cpu 0:
unix_inflight+0x1e8/0x260 net/unix/scm.c:63
unix_attach_fds+0x10c/0x1e0 net/unix/scm.c:121
unix_scm_to_skb net/unix/af_unix.c:1674 [inline]
unix_dgram_sendmsg+0x679/0x16b0 net/unix/af_unix.c:1817
unix_seqpacket_sendmsg+0xcc/0x110 net/unix/af_unix.c:2258
sock_sendmsg_nosec net/socket.c:704 [inline]
sock_sendmsg net/socket.c:724 [inline]
____sys_sendmsg+0x39a/0x510 net/socket.c:2409
___sys_sendmsg net/socket.c:2463 [inline]
__sys_sendmmsg+0x267/0x4c0 net/socket.c:2549
__do_sys_sendmmsg net/socket.c:2578 [inline]
__se_sys_sendmmsg net/socket.c:2575 [inline]
__x64_sys_sendmmsg+0x53/0x60 net/socket.c:2575
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x44/0xd0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
read to 0xffffffff86e2b7c0 of 4 bytes by task 9375 on cpu 1:
wait_for_unix_gc+0x24/0x160 net/unix/garbage.c:196
unix_dgram_sendmsg+0x8e/0x16b0 net/unix/af_unix.c:1772
unix_seqpacket_sendmsg+0xcc/0x110 net/unix/af_unix.c:2258
sock_sendmsg_nosec net/socket.c:704 [inline]
sock_sendmsg net/socket.c:724 [inline]
____sys_sendmsg+0x39a/0x510 net/socket.c:2409
___sys_sendmsg net/socket.c:2463 [inline]
__sys_sendmmsg+0x267/0x4c0 net/socket.c:2549
__do_sys_sendmmsg net/socket.c:2578 [inline]
__se_sys_sendmmsg net/socket.c:2575 [inline]
__x64_sys_sendmmsg+0x53/0x60 net/socket.c:2575
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x44/0xd0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
value changed: 0x00000002 -> 0x00000004
Reported by Kernel Concurrency Sanitizer on:
CPU: 1 PID: 9375 Comm: syz-executor.1 Not tainted 5.16.0-rc7-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Fixes: 9915672d4127 ("af_unix: limit unix_tot_inflight")
Signed-off-by: Eric Dumazet <[email protected]>
Reported-by: syzbot <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>
|
|
If the device is PCI based like intel-eth-pci, pm_runtime_enable() is
already called by pci_pm_init().
So only pm_runtime_enable() when it's not already enabled.
Signed-off-by: Kai-Heng Feng <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
platform_get_irq() returns negative error number instead 0 on failure.
And the doc of platform_get_irq() provides a usage example:
int irq = platform_get_irq(pdev, 0);
if (irq < 0)
return irq;
Fix the check of return value to catch errors correctly.
Fixes: 115978859272 ("i825xx: Move the Intel 82586/82593/82596 based drivers")
Signed-off-by: Miaoqian Lin <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Since commit adeef3e32146 ("net: constify netdev->dev_addr") the bmac
driver no longer builds with the following errors (pmac32_defconfig):
linux/drivers/net/ethernet/apple/bmac.c: In function ‘bmac_probe’:
linux/drivers/net/ethernet/apple/bmac.c:1287:20: error: assignment of read-only location ‘*(dev->dev_addr + (sizetype)j)’
1287 | dev->dev_addr[j] = rev ? bitrev8(addr[j]): addr[j];
| ^
Fix it by making the modifications to a local macaddr variable and then
passing that to eth_hw_addr_set().
We don't use the existing addr variable because the bitrev8() would
mutate it, but it is already used unreversed later in the function.
Signed-off-by: Michael Ellerman <[email protected]>
Reviewed-by: Jakub Kicinski <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Since commit adeef3e32146 ("net: constify netdev->dev_addr") the mace
driver no longer builds with various errors (pmac32_defconfig):
linux/drivers/net/ethernet/apple/mace.c: In function ‘mace_probe’:
linux/drivers/net/ethernet/apple/mace.c:170:20: error: assignment of read-only location ‘*(dev->dev_addr + (sizetype)j)’
170 | dev->dev_addr[j] = rev ? bitrev8(addr[j]): addr[j];
| ^
linux/drivers/net/ethernet/apple/mace.c: In function ‘mace_reset’:
linux/drivers/net/ethernet/apple/mace.c:349:32: warning: passing argument 2 of ‘__mace_set_address’ discards ‘const’ qualifier from pointer target type
349 | __mace_set_address(dev, dev->dev_addr);
| ~~~^~~~~~~~~~
linux/drivers/net/ethernet/apple/mace.c:93:62: note: expected ‘void *’ but argument is of type ‘const unsigned char *’
93 | static void __mace_set_address(struct net_device *dev, void *addr);
| ~~~~~~^~~~
linux/drivers/net/ethernet/apple/mace.c: In function ‘__mace_set_address’:
linux/drivers/net/ethernet/apple/mace.c:388:36: error: assignment of read-only location ‘*(dev->dev_addr + (sizetype)i)’
388 | out_8(&mb->padr, dev->dev_addr[i] = p[i]);
| ^
Fix it by making the modifications to a local macaddr variable and then
passing that to eth_hw_addr_set(), as well as adding some missing const
qualifiers.
Signed-off-by: Michael Ellerman <[email protected]>
Reviewed-by: Jakub Kicinski <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
So that users can run/query them easily.
$ ./fcnal-test.sh -h
usage: fcnal-test.sh OPTS
-4 IPv4 tests only
-6 IPv6 tests only
-t <test> Test name/set to run
-p Pause on fail
-P Pause after each test
-v Be verbose
Tests:
ipv4_ping ipv4_tcp ipv4_udp ipv4_bind ipv4_runtime ipv4_netfilter ipv6_ping ipv6_tcp ipv6_udp ipv6_bind ipv6_runtime ipv6_netfilter use_cases
Suggested-by: David Ahern <[email protected]>
Signed-off-by: Li Zhijian <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
On boards with LAN9514 and no preconfigured MAC address we don't get an
ip address from DHCP after commit a049a30fc27c ("net: usb: Correct PHY handling
of smsc95xx") anymore. Adding an explicit reset before starting the phy
fixes the issue.
[1]
https://lore.kernel.org/netdev/[email protected]/
From: Gabriel Hojda <[email protected]>
Fixes: a049a30fc27c ("net: usb: Correct PHY handling of smsc95xx")
Signed-off-by: Gabriel Hojda <[email protected]>
Signed-off-by: Markus Reichl <[email protected]>
Tested-by: Alexander Stein <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The driver neglects to check the result of platform_get_irq_optional()'s
call and blithely passes the negative error codes to devm_request_irq()
(which takes *unsigned* IRQ #), causing it to fail with -EINVAL.
Stop calling devm_request_irq() with the invalid IRQ #s.
Fixes: 8562056f267d ("net: bcmgenet: request Wake-on-LAN interrupt")
Signed-off-by: Sergey Shtylyov <[email protected]>
Acked-by: Florian Fainelli <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
commit 56b765b79e9a ("htb: improved accuracy at high rates") broke
"overhead X", "linklayer atm" and "mpu X" attributes.
"overhead X" and "linklayer atm" have already been fixed. This restores
the "mpu X" handling, as might be used by DOCSIS or Ethernet shaping:
tc class add ... htb rate X overhead 4 mpu 64
The code being fixed is used by htb, tbf and act_police. Cake has its
own mpu handling. qdisc_calculate_pkt_len still uses the size table
containing values adjusted for mpu by user space.
iproute2 tc has always passed mpu into the kernel via a tc_ratespec
structure, but the kernel never directly acted on it, merely stored it
so that it could be read back by `tc class show`.
Rather, tc would generate length-to-time tables that included the mpu
(and linklayer) in their construction, and the kernel used those tables.
Since v3.7, the tables were no longer used. Along with "mpu", this also
broke "overhead" and "linklayer" which were fixed in 01cb71d2d47b
("net_sched: restore "overhead xxx" handling", v3.10) and 8a8e3d84b171
("net_sched: restore "linklayer atm" handling", v3.11).
"overhead" was fixed by simply restoring use of tc_ratespec::overhead -
this had originally been used by the kernel but was initially omitted
from the new non-table-based calculations.
"linklayer" had been handled in the table like "mpu", but the mode was
not originally passed in tc_ratespec. The new implementation was made to
handle it by getting new versions of tc to pass the mode in an extended
tc_ratespec, and for older versions of tc the table contents were analysed
at load time to deduce linklayer.
As "mpu" has always been given to the kernel in tc_ratespec,
accompanying the mpu-based table, we can restore system functionality
with no userspace change by making the kernel act on the tc_ratespec
value.
Fixes: 56b765b79e9a ("htb: improved accuracy at high rates")
Signed-off-by: Kevin Bracey <[email protected]>
Cc: Eric Dumazet <[email protected]>
Cc: Jiri Pirko <[email protected]>
Cc: Vimalkumar <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>
|
|
The Hucom Wireless HM-211S/K is an LTE module based on Qualcomm MDM9207.
This module supports LTE Band 1, 3, 5, 7, 8 and WCDMA Band 1.
Manual testing showed that only interface
number two replies to QMI messages.
T: Bus=01 Lev=02 Prnt=02 Port=01 Cnt=01 Dev#= 3 Spd=480 MxCh= 0
D: Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs= 1
P: Vendor=22de ProdID=9051 Rev= 3.18
S: Manufacturer=Android
S: Product=Android
S: SerialNumber=0123456789ABCDEF
C:* #Ifs= 4 Cfg#= 1 Atr=80 MxPwr=500mA
I:* If#= 0 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
E: Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=01(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=(none)
E: Ad=83(I) Atr=03(Int.) MxPS= 10 Ivl=32ms
E: Ad=82(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=qmi_wwan
E: Ad=85(I) Atr=03(Int.) MxPS= 8 Ivl=32ms
E: Ad=84(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=03(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
I:* If#= 3 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=42 Prot=01 Driver=(none)
E: Ad=04(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E: Ad=86(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
Signed-off-by: Kyoungkyu Park <[email protected]>
Acked-by: Bjørn Mork <[email protected]>
Link: https://lore.kernel.org/r/Yd+nxAA6KorDpQFv@choryu-tfx5470h
Signed-off-by: Jakub Kicinski <[email protected]>
|
|
We encountered some crashes caused by the race between SMC-R
link access and link clear that triggered by abnormal link
group termination, such as port error.
Here is an example of this kind of crashes:
BUG: kernel NULL pointer dereference, address: 0000000000000000
Workqueue: smc_hs_wq smc_listen_work [smc]
RIP: 0010:smc_llc_flow_initiate+0x44/0x190 [smc]
Call Trace:
<TASK>
? __smc_buf_create+0x75a/0x950 [smc]
smcr_lgr_reg_rmbs+0x2a/0xbf [smc]
smc_listen_work+0xf72/0x1230 [smc]
? process_one_work+0x25c/0x600
process_one_work+0x25c/0x600
worker_thread+0x4f/0x3a0
? process_one_work+0x600/0x600
kthread+0x15d/0x1a0
? set_kthread_struct+0x40/0x40
ret_from_fork+0x1f/0x30
</TASK>
smc_listen_work() __smc_lgr_terminate()
---------------------------------------------------------------
| smc_lgr_free()
| |- smcr_link_clear()
| |- memset(lnk, 0)
smc_listen_rdma_reg() |
|- smcr_lgr_reg_rmbs() |
|- smc_llc_flow_initiate() |
|- access lnk->lgr (panic) |
These crashes are similarly caused by clearing SMC-R link
resources when some functions is still accessing to them.
This patch tries to fix the issue by introducing reference
count of SMC-R links and ensuring that the sensitive resources
of links won't be cleared until reference count reaches zero.
The operation to the SMC-R link reference count can be concluded
as follows:
object [hold or initialized as 1] [put]
--------------------------------------------------------------------
links smcr_link_init() smcr_link_clear()
connections smc_conn_create() smc_conn_free()
Through this way, the clear of SMC-R links is later than the
free of all the smc connections above it, thus avoiding the
unsafe reference to SMC-R links.
Signed-off-by: Wen Gu <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
It is no longer suitable to identify whether a smc connection
is registered in a link group through checking if conn->lgr
is NULL, because conn->lgr won't be reset even the connection
is unregistered from a link group.
So this patch introduces a new helper smc_conn_lgr_valid() and
replaces all the check of conn->lgr in original implementation
with the new helper to judge if conn->lgr is valid to use.
Signed-off-by: Wen Gu <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Both fields can be read/written without synchronization,
add proper accessors and documentation.
Fixes: d5dd88794a13 ("inet: fix various use-after-free in defrags units")
Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Wen Gu says:
====================
net/smc: Fixes for race in smc link group termination
We encountered some crashes recently and they are caused by the
race between the access and free of link/link group in abnormal
smc link group termination. The crashes can be reproduced in
frequent abnormal link group termination, like setting RNICs up/down.
This set of patches tries to fix this by extending the life cycle
of link/link group to ensure that they won't be referred to after
cleared or freed.
v1 -> v2:
- Improve some comments.
- Move codes of waking up lgrs_deleted wait queue from smc_lgr_free()
to __smc_lgr_free().
- Move codes of waking up links_deleted wait queue from smcr_link_clear()
to __smcr_link_clear().
- Move codes of smc_ibdev_cnt_dec() and put_device() from smcr_link_clear()
to __smcr_link_clear()
- Move smc_lgr_put() to the end of __smcr_link_clear().
- Call smc_lgr_put() after 'out' tag in smcr_link_init() when link
initialization fails.
- Modify the location where smc connection holds the lgr or link.
before:
* hold lgr in smc_lgr_register_conn().
* hold link in smcr_lgr_conn_assign_link().
after:
* hold both lgr and link in smc_conn_create().
Modify the location to symmetrical with the place where smc connections
put the lgr or link, which is smc_conn_free().
- Initialize conn->freed as zero in smc_conn_create().
====================
Signed-off-by: David S. Miller <[email protected]>
|
|
We encountered some crashes caused by the race between the access
and the termination of link groups.
Here are some of panic stacks we met:
1) Race between smc_clc_wait_msg() and __smc_lgr_terminate()
BUG: kernel NULL pointer dereference, address: 00000000000002f0
Workqueue: smc_hs_wq smc_listen_work [smc]
RIP: 0010:smc_clc_wait_msg+0x3eb/0x5c0 [smc]
Call Trace:
<TASK>
? smc_clc_send_accept+0x45/0xa0 [smc]
? smc_clc_send_accept+0x45/0xa0 [smc]
smc_listen_work+0x783/0x1220 [smc]
? finish_task_switch+0xc4/0x2e0
? process_one_work+0x1ad/0x3c0
process_one_work+0x1ad/0x3c0
worker_thread+0x4c/0x390
? rescuer_thread+0x320/0x320
kthread+0x149/0x190
? set_kthread_struct+0x40/0x40
ret_from_fork+0x1f/0x30
</TASK>
smc_listen_work() abnormal case like port error
---------------------------------------------------------------
| __smc_lgr_terminate()
| |- smc_conn_kill()
| |- smc_lgr_unregister_conn()
| |- set conn->lgr = NULL
smc_clc_wait_msg() |
|- access conn->lgr (panic) |
2) Race between smc_setsockopt() and __smc_lgr_terminate()
BUG: kernel NULL pointer dereference, address: 00000000000002e8
RIP: 0010:smc_setsockopt+0x17a/0x280 [smc]
Call Trace:
<TASK>
__sys_setsockopt+0xfc/0x190
__x64_sys_setsockopt+0x20/0x30
do_syscall_64+0x34/0x90
entry_SYSCALL_64_after_hwframe+0x44/0xae
</TASK>
smc_setsockopt() abnormal case like port error
--------------------------------------------------------------
| __smc_lgr_terminate()
| |- smc_conn_kill()
| |- smc_lgr_unregister_conn()
| |- set conn->lgr = NULL
mod_delayed_work() |
|- access conn->lgr (panic) |
There are some other panic places and they are caused by the
similar reason as described above, which is accessing link
group after termination, thus getting a NULL pointer or invalid
resource.
Currently, there seems to be no synchronization between the
link group access and a sudden termination of it. This patch
tries to fix this by introducing reference count of link group
and not freeing link group until reference count is zero.
Link group might be referred to by links or smc connections. So
the operation to the link group reference count can be concluded
as follows:
object [hold or initialized as 1] [put]
-------------------------------------------------------------------
link group smc_lgr_create() smc_lgr_free()
connections smc_conn_create() smc_conn_free()
links smcr_link_init() smcr_link_clear()
Througth this way, we extend the life cycle of link group and
ensure it is longer than the life cycle of connections and links
above it, so that avoid invalid access to link group after its
termination.
Signed-off-by: Wen Gu <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
timeout in settings is used by each case under the same directory, so
it should adapt to the maximum runtime.
A normally running net/fib_nexthops.sh may be killed by this unsuitable
timeout. Furthermore, since the defect[1] of kselftests framework,
net/fib_nexthops.sh which might take at least (300 * 4) seconds would
block the whole kselftests framework previously.
$ git grep -w 'sleep 300' tools/testing/selftests/net
tools/testing/selftests/net/fib_nexthops.sh: sleep 300
tools/testing/selftests/net/fib_nexthops.sh: sleep 300
tools/testing/selftests/net/fib_nexthops.sh: sleep 300
tools/testing/selftests/net/fib_nexthops.sh: sleep 300
Enlarge the timeout by plus 300 based on the obvious largest runtime
to avoid the blocking.
[1]: https://www.spinics.net/lists/kernel/msg4185370.html
Signed-off-by: Zhou Jie <[email protected]>
Signed-off-by: Li Zhijian <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
Since commit b39648079db4 ("net: mscc: ocelot: disable flow control on
NPI interface"), flow control should be disabled on the DSA CPU port
when used in NPI mode.
However, the commit blamed in the Fixes: tag below broke this, because
it allowed felix_phylink_mac_link_up() to overwrite SYS_PAUSE_CFG_PAUSE_ENA
for the DSA CPU port.
This issue became noticeable since the device tree update from commit
8fcea7be5736 ("arm64: dts: ls1028a: mark internal links between Felix
and ENETC as capable of flow control").
The solution is to check whether this is the currently configured NPI
port from ocelot_phylink_mac_link_up(), and to not modify the statically
disabled PAUSE frame transmission if it is.
When the port is configured for lossless mode as opposed to tail drop
mode, but the link partner (DSA master) doesn't observe the transmitted
PAUSE frames, the switch termination throughput is much worse, as can be
seen below.
Before:
root@debian:~# iperf3 -c 192.168.100.2
Connecting to host 192.168.100.2, port 5201
[ 5] local 192.168.100.1 port 37504 connected to 192.168.100.2 port 5201
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 28.4 MBytes 238 Mbits/sec 357 22.6 KBytes
[ 5] 1.00-2.00 sec 33.6 MBytes 282 Mbits/sec 426 19.8 KBytes
[ 5] 2.00-3.00 sec 34.0 MBytes 285 Mbits/sec 343 21.2 KBytes
[ 5] 3.00-4.00 sec 32.9 MBytes 276 Mbits/sec 354 22.6 KBytes
[ 5] 4.00-5.00 sec 32.3 MBytes 271 Mbits/sec 297 18.4 KBytes
^C[ 5] 5.00-5.06 sec 2.05 MBytes 270 Mbits/sec 45 19.8 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-5.06 sec 163 MBytes 271 Mbits/sec 1822 sender
[ 5] 0.00-5.06 sec 0.00 Bytes 0.00 bits/sec receiver
After:
root@debian:~# iperf3 -c 192.168.100.2
Connecting to host 192.168.100.2, port 5201
[ 5] local 192.168.100.1 port 49470 connected to 192.168.100.2 port 5201
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 112 MBytes 941 Mbits/sec 259 143 KBytes
[ 5] 1.00-2.00 sec 110 MBytes 920 Mbits/sec 329 144 KBytes
[ 5] 2.00-3.00 sec 112 MBytes 936 Mbits/sec 255 144 KBytes
[ 5] 3.00-4.00 sec 110 MBytes 927 Mbits/sec 355 105 KBytes
[ 5] 4.00-5.00 sec 110 MBytes 926 Mbits/sec 350 156 KBytes
[ 5] 5.00-6.00 sec 110 MBytes 925 Mbits/sec 305 148 KBytes
[ 5] 6.00-7.00 sec 110 MBytes 924 Mbits/sec 320 143 KBytes
[ 5] 7.00-8.00 sec 110 MBytes 925 Mbits/sec 273 97.6 KBytes
[ 5] 8.00-9.00 sec 109 MBytes 913 Mbits/sec 299 141 KBytes
[ 5] 9.00-10.00 sec 110 MBytes 922 Mbits/sec 287 146 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bitrate Retr
[ 5] 0.00-10.00 sec 1.08 GBytes 926 Mbits/sec 3032 sender
[ 5] 0.00-10.00 sec 1.08 GBytes 925 Mbits/sec receiver
Fixes: de274be32cb2 ("net: dsa: felix: set TX flow control according to the phylink_mac_link_up resolution")
Reported-by: Xiaoliang Yang <[email protected]>
Signed-off-by: Vladimir Oltean <[email protected]>
Reviewed-by: Florian Fainelli <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The pointer skb is redundant, it is assigned a value that is never
read and hence can be removed. Cleans up clang scan warning:
drivers/atm/iphase.c:205:18: warning: Although the value stored
to 'skb' is used in the enclosing expression, the value is never
actually read from 'skb' [deadcode.DeadStores]
Signed-off-by: Colin Ian King <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|
|
The attach callback of struct Qdisc_ops is used by only a few qdiscs:
mq, mqprio and htb. qdisc_graft() contains the following logic
(pseudocode):
if (!qdisc->ops->attach) {
if (ingress)
do ingress stuff;
else
do egress stuff;
}
if (!ingress) {
...
if (qdisc->ops->attach)
qdisc->ops->attach(qdisc);
} else {
...
}
As we see, the attach callback is not called if the qdisc is being
attached to ingress (TC_H_INGRESS). That wasn't a problem for mq and
mqprio, since they contain a check that they are attached to TC_H_ROOT,
and they can't be attached to TC_H_INGRESS anyway.
However, the commit cited below added the attach callback to htb. It is
needed for the hardware offload, but in the non-offload mode it
simulates the "do egress stuff" part of the pseudocode above. The
problem is that when htb is attached to ingress, neither "do ingress
stuff" nor attach() is called. It results in an inconsistency, and the
following message is printed to dmesg:
unregister_netdevice: waiting for lo to become free. Usage count = 2
This commit addresses the issue by running "do ingress stuff" in the
ingress flow even in the attach callback is present, which is fine,
because attach isn't going to be called afterwards.
The bug was found by syzbot and reported by Eric.
Fixes: d03b195b5aa0 ("sch_htb: Hierarchical QoS hardware offload")
Signed-off-by: Maxim Mikityanskiy <[email protected]>
Reported-by: Eric Dumazet <[email protected]>
Reviewed-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
|