Johannes Berg [Thu, 23 May 2024 10:09:45 +0000 (12:09 +0200)]
wifi: cfg80211: move enum ieee80211_ap_reg_power to cfg80211
This really shouldn't have been in ieee80211.h, since it
doesn't directly represent the spec. Move it to cfg80211
rather than mac80211 since upcoming changes will use it
there.
Breno Leitao [Fri, 7 Jun 2024 10:20:44 +0000 (03:20 -0700)]
wifi: mac80211: Remove generic .ndo_get_stats64
Commit 3e2f544dd8a33 ("net: get stats64 if device if driver is
configured") moved the callback to dev_get_tstats64() to net core, so,
unless the driver is doing some custom stats collection, it does not
need to set .ndo_get_stats64.
Since this driver is now relying in NETDEV_PCPU_STAT_TSTATS, then, it
doesn't need to set the dev_get_tstats64() generic .ndo_get_stats64
function pointer.
In this driver specifically, .ndo_get_stats64 basically points to
dev_fetch_sw_netstats(). Now it will point to dev_get_tstats64(), which
calls netdev_stats_to_stats64() and dev_fetch_sw_netstats().
netdev_stats_to_stats64() seems irrelevant for this driver.
Breno Leitao [Fri, 7 Jun 2024 10:20:43 +0000 (03:20 -0700)]
wifi: mac80211: Move stats allocation to core
With commit 34d21de99cea9 ("net: Move {l,t,d}stats allocation to core and
convert veth & vrf"), stats allocation could be done on net core instead
of this driver.
With this new approach, the driver doesn't have to bother with error
handling (allocation failure checking, making sure free happens in the
right spot, etc). This is core responsibility now.
Move mac80211 driver to leverage the core allocation.
Johannes Berg [Fri, 7 Jun 2024 18:17:17 +0000 (20:17 +0200)]
wifi: cfg80211: make hash table duplicates more survivable
Jiazi Li reported that they occasionally see hash table duplicates
as evidenced by the WARN_ON() in rb_insert_bss() in this code. It
isn't clear how that happens, nor have I been able to reproduce it,
but if it does happen, the kernel crashes later, when it tries to
unhash the entry that's now not hashed.
Try to make this situation more survivable by removing the BSS from
the list(s) as well, that way it's fully leaked here (as had been
the intent in the hash insert error path), and no longer reachable
through the list(s) so it shouldn't be unhashed again later.
Johannes Berg [Wed, 12 Jun 2024 08:05:33 +0000 (10:05 +0200)]
wifi: mac80211: correct EHT EIRP TPE parsing
For the EHT EIRP transmit power envelope, the 320 MHz is in
the last octet, but if we've copied 4 octets (count == 3),
the next one is at index 4 not 5 (count + 2). Fix this, and
just hardcode the offset since count is always 3 here.
====================
Intel Wired LAN Driver Updates 2024-06-03
This series includes miscellaneous improvements for the ice as well as a
cleanup to the Makefiles for all Intel net drivers.
Andy fixes all of the Intel net driver Makefiles to use the documented
'*-y' syntax for specifying object files to link into kernel driver
modules, rather than the '*-objs' syntax which works but is documented as
reserved for user-space host programs.
Jacob has a cleanup to refactor rounding logic in the ice driver into a
common roundup_u64 helper function.
Michal Schmidt replaces irq_set_affinity_hint() to use
irq_update_affinity_hint() which behaves better with user-applied affinity
settings.
Michal Schmidt [Fri, 7 Jun 2024 21:22:34 +0000 (14:22 -0700)]
ice: use irq_update_affinity_hint()
irq_set_affinity_hint() is deprecated. Use irq_update_affinity_hint()
instead. This removes the side-effect of actually applying the affinity.
The driver does not really need to worry about spreading its IRQs across
CPUs. The core code already takes care of that.
On the contrary, when the driver applies affinities by itself, it breaks
the users' expectations:
1. The user configures irqbalance with IRQBALANCE_BANNED_CPULIST in
order to prevent IRQs from being moved to certain CPUs that run a
real-time workload.
2. ice reconfigures VSIs at runtime due to a MIB change
(ice_dcb_process_lldp_set_mib_change). Reopening a VSI resets the
affinity in ice_vsi_req_irq_msix().
3. ice has no idea about irqbalance's config, so it may move an IRQ to
a banned CPU. The real-time workload suffers unacceptable latency.
I am not sure if updating the affinity hints is at all useful, because
irqbalance ignores them since 2016 ([1]), but at least it's harmless.
This ice change is similar to i40e commit d34c54d1739c ("i40e: Use
irq_update_affinity_hint()").
Jacob Keller [Fri, 7 Jun 2024 21:22:33 +0000 (14:22 -0700)]
ice: add and use roundup_u64 instead of open coding equivalent
In ice_ptp_cfg_clkout(), the ice driver needs to calculate the nearest next
second of a current time value specified in nanoseconds. It implements this
using div64_u64, because the time value is a u64. It could use div_u64
since NSEC_PER_SEC is smaller than 32-bits.
Ideally this would be implemented directly with roundup(), but that can't
work on all platforms due to a division which requires using the specific
macros and functions due to platform restrictions, and to ensure that the
most appropriate and fast instructions are used.
The kernel doesn't currently provide any 64-bit equivalents for doing
roundup. Attempting to use roundup() on a 32-bit platform will result in a
link failure due to not having a direct 64-bit division.
The closest equivalent for this is DIV64_U64_ROUND_UP, which does a
division always rounding up. However, this only computes the division, and
forces use of the div64_u64 in cases where the divisor is a 32bit value and
could make use of div_u64.
Introduce DIV_U64_ROUND_UP based on div_u64, and then use it to implement
roundup_u64 which takes a u64 input value and a u32 rounding value.
The name roundup_u64 matches the naming scheme of div_u64, and future
patches could implement roundup64_u64 if they need to round by a multiple
that is greater than 32-bits.
Replace the logic in ice_ptp.c which does this equivalent with the newly
added roundup_u64.
Andy Shevchenko [Fri, 7 Jun 2024 21:22:32 +0000 (14:22 -0700)]
net: intel: Use *-y instead of *-objs in Makefile
*-objs suffix is reserved rather for (user-space) host programs while
usually *-y suffix is used for kernel drivers (although *-objs works
for that purpose for now).
Let's correct the old usages of *-objs in Makefiles.
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Pucha Himasekhar Reddy <himasekharx.reddy.pucha@intel.com> Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://lore.kernel.org/r/20240607-next-2024-06-03-intel-next-batch-v3-1-d1470cee3347@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
We've added 54 non-merge commits during the last 10 day(s) which contain
a total of 50 files changed, 1887 insertions(+), 527 deletions(-).
The main changes are:
1) Add a user space notification mechanism via epoll when a struct_ops
object is getting detached/unregistered, from Kui-Feng Lee.
2) Big batch of BPF selftest refactoring for sockmap and BPF congctl
tests, from Geliang Tang.
3) Add BTF field (type and string fields, right now) iterator support
to libbpf instead of using existing callback-based approaches,
from Andrii Nakryiko.
4) Extend BPF selftests for the latter with a new btf_field_iter
selftest, from Alan Maguire.
5) Add new kfuncs for a generic, open-coded bits iterator,
from Yafang Shao.
6) Fix BPF selftests' kallsyms_find() helper under kernels configured
with CONFIG_LTO_CLANG_THIN, from Yonghong Song.
7) Remove a bunch of unused structs in BPF selftests,
from David Alan Gilbert.
8) Convert test_sockmap section names into names understood by libbpf
so it can deduce program type and attach type, from Jakub Sitnicki.
9) Extend libbpf with the ability to configure log verbosity
via LIBBPF_LOG_LEVEL environment variable, from Mykyta Yatsenko.
10) Fix BPF selftests with regards to bpf_cookie and find_vma flakiness
in nested VMs, from Song Liu.
11) Extend riscv32/64 JITs to introduce shift/add helpers to generate Zba
optimization, from Xiao Wang.
12) Enable BPF programs to declare arrays and struct fields with kptr,
bpf_rb_root, and bpf_list_head, from Kui-Feng Lee.
* tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next: (54 commits)
selftests/bpf: Drop useless arguments of do_test in bpf_tcp_ca
selftests/bpf: Use start_test in test_dctcp in bpf_tcp_ca
selftests/bpf: Use start_test in test_dctcp_fallback in bpf_tcp_ca
selftests/bpf: Add start_test helper in bpf_tcp_ca
selftests/bpf: Use connect_to_fd_opts in do_test in bpf_tcp_ca
libbpf: Auto-attach struct_ops BPF maps in BPF skeleton
selftests/bpf: Add btf_field_iter selftests
selftests/bpf: Fix send_signal test with nested CONFIG_PARAVIRT
libbpf: Remove callback-based type/string BTF field visitor helpers
bpftool: Use BTF field iterator in btfgen
libbpf: Make use of BTF field iterator in BTF handling code
libbpf: Make use of BTF field iterator in BPF linker code
libbpf: Add BTF field iterator
selftests/bpf: Ignore .llvm.<hash> suffix in kallsyms_find()
selftests/bpf: Fix bpf_cookie and find_vma in nested VM
selftests/bpf: Test global bpf_list_head arrays.
selftests/bpf: Test global bpf_rb_root arrays and fields in nested struct types.
selftests/bpf: Test kptr arrays and kptrs in nested struct fields.
bpf: limit the number of levels of a nested struct type.
bpf: look into the types of the fields of a struct type recursively.
...
====================
Jakub Kicinski [Tue, 11 Jun 2024 00:40:25 +0000 (17:40 -0700)]
Merge tag 'wireless-next-2024-06-07' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless-next
Kalle Valo says:
====================
wireless-next patches for v6.11
The first "new features" pull request for v6.11 with changes both in
stack and in drivers. Nothing out of ordinary, except that we have
two conflicts this time:
cfg80211/mac80211
* parse Transmit Power Envelope (TPE) data in mac80211 instead of in drivers
wilc1000
* read MAC address during probe to make it visible to user space
iwlwifi
* bump FW API to 91 for BZ/SC devices
* report 64-bit radiotap timestamp
* enable P2P low latency by default
* handle Transmit Power Envelope (TPE) advertised by AP
* start using guard()
rtlwifi
* RTL8192DU support
ath12k
* remove unsupported tx monitor handling
* channel 2 in 6 GHz band support
* Spatial Multiplexing Power Save (SMPS) in 6 GHz band support
* multiple BSSID (MBSSID) and Enhanced Multi-BSSID Advertisements (EMA)
support
* dynamic VLAN support
* add panic handler for resetting the firmware state
ath10k
* add qcom,no-msa-ready-indicator Device Tree property
* LED support for various chipsets
* tag 'wireless-next-2024-06-07' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless-next: (194 commits)
wifi: ath12k: add hw_link_id in ath12k_pdev
wifi: ath12k: add panic handler
wifi: rtw89: chan: Use swap() in rtw89_swap_sub_entity()
wifi: brcm80211: remove unused structs
wifi: brcm80211: use sizeof(*pointer) instead of sizeof(type)
wifi: ath12k: do not process consecutive RDDM event
dt-bindings: net: wireless: ath11k: Drop "qcom,ipq8074-wcss-pil" from example
wifi: ath12k: fix memory leak in ath12k_dp_rx_peer_frag_setup()
wifi: rtlwifi: handle return value of usb init TX/RX
wifi: rtlwifi: Enable the new rtl8192du driver
wifi: rtlwifi: Add rtl8192du/sw.c
wifi: rtlwifi: Constify rtl_hal_cfg.{ops,usb_interface_cfg} and rtl_priv.cfg
wifi: rtlwifi: Add rtl8192du/dm.{c,h}
wifi: rtlwifi: Add rtl8192du/fw.{c,h} and rtl8192du/led.{c,h}
wifi: rtlwifi: Add rtl8192du/rf.{c,h}
wifi: rtlwifi: Add rtl8192du/trx.{c,h}
wifi: rtlwifi: Add rtl8192du/phy.{c,h}
wifi: rtlwifi: Add rtl8192du/hw.{c,h}
wifi: rtlwifi: Add new members to struct rtl_priv for RTL8192DU
wifi: rtlwifi: Add rtl8192du/table.{c,h}
...
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
====================
David S. Miller [Mon, 10 Jun 2024 12:48:06 +0000 (13:48 +0100)]
Merge branch 'fix-changing-dsa-conduit'
Marek Behún says:
====================
Fix changing DSA conduit
This series fixes an issue in the DSA code related to host interface UC
address installed into port FDB and port conduit address database when
live-changing port conduit.
The first patch refactores/deduplicates the installation/uninstallation
of the interface's MAC address and the second patch fixes the issue.
Cover letter for v1 and v2:
https://patchwork.kernel.org/project/netdevbpf/cover/20240429163627.16031-1-kabel@kernel.org/
https://patchwork.kernel.org/project/netdevbpf/cover/20240502122922.28139-1-kabel@kernel.org/
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Marek Behún [Wed, 5 Jun 2024 13:33:29 +0000 (15:33 +0200)]
net: dsa: update the unicast MAC address when changing conduit
When changing DSA user interface conduit while the user interface is up,
DSA exhibits different behavior in comparison to when the interface is
down. This different behavior concerns the primary unicast MAC address
stored in the port standalone FDB and in the conduit device UC database.
If we put a switch port down while changing the conduit with
ip link set sw0p0 down
ip link set sw0p0 type dsa conduit conduit1
ip link set sw0p0 up
we delete the address in dsa_user_close() and install the (possibly
different) address in dsa_user_open().
But when changing the conduit on the fly, the old address is not
deleted and the new one is not installed.
Since we explicitly want to support live-changing the conduit, uninstall
the old address before calling dsa_port_assign_conduit() and install the
(possibly different) new address after the call.
Because conduit change might also trigger address change (the user
interface is supposed to inherit the conduit interface MAC address if no
address is defined in hardware (dp->mac is a zero address)), move the
eth_hw_addr_inherit() call from dsa_user_change_conduit() to
dsa_port_change_conduit(), just before installing the new address.
Although this is in theory a flaw in DSA core, it needs not be
backported, since there is currently no DSA driver that can be affected
by this. The only DSA driver that supports changing conduit is felix,
and, as explained by Vladimir Oltean [1]:
There are 2 reasons why with felix the bug does not manifest itself.
First is because both the 'ocelot' and the alternate 'ocelot-8021q'
tagging protocols have the 'promisc_on_conduit = true' flag. So the
unicast address doesn't have to be in the conduit's RX filter -
neither the old or the new conduit.
Second, dsa_user_host_uc_install() theoretically leaves behind host
FDB entries installed towards the wrong (old) CPU port. But in
felix_fdb_add(), we treat any FDB entry requested towards any CPU port
as if it was a multicast FDB entry programmed towards _all_ CPU ports.
For that reason, it is installed towards the port mask of the PGID_CPU
port group ID:
if (dsa_port_is_cpu(dp))
port = PGID_CPU;
Therefore no Fixes tag for this change.
[1] https://lore.kernel.org/netdev/20240507201827.47suw4fwcjrbungy@skbuf/ Signed-off-by: Marek Behún <kabel@kernel.org> Reviewed-by: Vladimir Oltean <olteanv@gmail.com> Tested-by: Vladimir Oltean <olteanv@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Marek Behún [Wed, 5 Jun 2024 13:33:28 +0000 (15:33 +0200)]
net: dsa: deduplicate code adding / deleting the port address to fdb
The sequence
if (dsa_switch_supports_uc_filtering(ds))
dsa_port_standalone_host_fdb_add(dp, addr, 0);
if (!ether_addr_equal(addr, conduit->dev_addr))
dev_uc_add(conduit, addr);
is executed both in dsa_user_open() and dsa_user_set_mac_addr().
Its reverse is executed both in dsa_user_close() and
dsa_user_set_mac_addr().
Refactor these sequences into new functions dsa_user_host_uc_install()
and dsa_user_host_uc_uninstall().
Signed-off-by: Marek Behún <kabel@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Mon, 10 Jun 2024 12:15:40 +0000 (13:15 +0100)]
Merge branch 'rtnetlink-rtnl_lock'
Jakub Kicinski says:
====================
rtnetlink: move rtnl_lock handling out of af_netlink
With the changes done in commit 5b4b62a169e1 ("rtnetlink: make
the "split" NLM_DONE handling generic") we can also move the
rtnl locking out of af_netlink.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Jakub Kicinski [Thu, 6 Jun 2024 19:29:06 +0000 (12:29 -0700)]
net: netlink: remove the cb_mutex "injection" from netlink core
Back in 2007, in commit af65bdfce98d ("[NETLINK]: Switch cb_lock spinlock
to mutex and allow to override it") netlink core was extended to allow
subsystems to replace the dump mutex lock with its own lock.
The mechanism was used by rtnetlink to take rtnl_lock but it isn't
sufficiently flexible for other users. Over the 17 years since
it was added no other user appeared. Since rtnetlink needs conditional
locking now, and doesn't use it either, axe this feature complete.
Signed-off-by: Jakub Kicinski <kuba@kernel.org> Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Jakub Kicinski [Thu, 6 Jun 2024 19:29:05 +0000 (12:29 -0700)]
rtnetlink: move rtnl_lock handling out of af_netlink
Now that we have an intermediate layer of code for handling
rtnl-level netlink dump quirks, we can move the rtnl_lock
taking there.
For dump handlers with RTNL_FLAG_DUMP_SPLIT_NLM_DONE we can
avoid taking rtnl_lock just to generate NLM_DONE, once again.
Signed-off-by: Jakub Kicinski <kuba@kernel.org> Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Andy Shevchenko [Thu, 6 Jun 2024 16:15:49 +0000 (19:15 +0300)]
net: dsa: hellcreek: Replace kernel.h with what is used
kernel.h is included solely for some other existing headers.
Include them directly and get rid of kernel.h.
While at it, sort headers alphabetically for easier maintenance.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de> Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Mon, 10 Jun 2024 10:54:19 +0000 (11:54 +0100)]
Merge branch 'tcp-up-pin-tw-timer'
Florian Westphal says:
====================
net: tcp: un-pin tw timer
Changes since previous iteration:
- Patch 1: update a comment, I copied Erics v7 RvB tag.
- Patch 2: move bh off/on into hashdance_schedule and get rid of
comment mentioning pinned tw timer.
I did not copy Erics RvB tag over from v7 because of the change.
- Patch 3 is unchanged, so I kept Erics RvB tag.
This is v8 of the series where the tw_timer is un-pinned to get rid of
interferences in isolated CPUs setups.
First patch makes necessary preparations, existing code relies on
TIMER_PINNED to avoid races.
Second patch un-pins the TW timer. Could be folded into the first one,
but it might help wrt. bisection.
Third patch is a minor cleanup to move a helper from .h to the only
remaining compilation unit.
Tested with iperf3 and stress-ng socket mode.
====================
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Florian Westphal [Thu, 6 Jun 2024 15:11:38 +0000 (17:11 +0200)]
net: tcp: un-pin the tw_timer
After previous patch, even if timer fires immediately on another CPU,
context that schedules the timer now holds the ehash spinlock, so timer
cannot reap tw socket until ehash lock is released.
BH disable is moved into hashdance_schedule.
Signed-off-by: Florian Westphal <fw@strlen.de> Reviewed-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Valentin Schneider [Thu, 6 Jun 2024 15:11:37 +0000 (17:11 +0200)]
net: tcp/dccp: prepare for tw_timer un-pinning
The TCP timewait timer is proving to be problematic for setups where
scheduler CPU isolation is achieved at runtime via cpusets (as opposed to
statically via isolcpus=domains).
What happens there is a CPU goes through tcp_time_wait(), arming the
time_wait timer, then gets isolated. TCP_TIMEWAIT_LEN later, the timer
fires, causing interference for the now-isolated CPU. This is conceptually
similar to the issue described in commit e02b93124855 ("workqueue: Unbind
kworkers before sending them to exit()")
Move inet_twsk_schedule() to within inet_twsk_hashdance(), with the ehash
lock held. Expand the lock's critical section from inet_twsk_kill() to
inet_twsk_deschedule_put(), serializing the scheduling vs descheduling of
the timer. IOW, this prevents the following race:
Thanks to Paolo Abeni for suggesting to leverage the ehash lock.
This also restores a comment from commit ec94c2696f0b ("tcp/dccp: avoid
one atomic operation for timewait hashdance") as inet_twsk_hashdance() had
a "Step 1" and "Step 3" comment, but the "Step 2" had gone missing.
inet_twsk_deschedule_put() now acquires the ehash spinlock to synchronize
with inet_twsk_hashdance_schedule().
To ease possible regression search, actual un-pin is done in next patch.
Link: https://lore.kernel.org/all/ZPhpfMjSiHVjQkTk@localhost.localdomain/ Reviewed-by: Eric Dumazet <edumazet@google.com> Signed-off-by: Valentin Schneider <vschneid@redhat.com> Co-developed-by: Florian Westphal <fw@strlen.de> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net>
Ido Schimmel [Thu, 6 Jun 2024 14:49:43 +0000 (16:49 +0200)]
mlxsw: spectrum_acl: Fix ACL scale regression and firmware errors
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>
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>
Ido Schimmel [Thu, 6 Jun 2024 14:49:41 +0000 (16:49 +0200)]
lib: objagg: Fix general protection fault
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>
Ido Schimmel [Thu, 6 Jun 2024 14:49:40 +0000 (16:49 +0200)]
mlxsw: spectrum_acl_atcam: Fix wrong comment
The key is encoded, not encrypted.
Fixes: c22291f7cf45 ("mlxsw: spectrum: acl: Implement delta for ERP") 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>
Ido Schimmel [Thu, 6 Jun 2024 14:49:39 +0000 (16:49 +0200)]
lib: test_objagg: Fix spelling
Fixes: 0a020d416d0a ("lib: introduce initial implementation of object aggregation manager") 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>
Ido Schimmel [Thu, 6 Jun 2024 14:49:38 +0000 (16:49 +0200)]
lib: objagg: Fix spelling
Fixes: 0a020d416d0a ("lib: introduce initial implementation of object aggregation manager") 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>
Dan Carpenter [Thu, 6 Jun 2024 14:23:44 +0000 (17:23 +0300)]
dmaengine: ti: k3-udma-glue: clean up return in k3_udma_glue_rx_get_irq()
Currently the k3_udma_glue_rx_get_irq() function returns either negative
error codes or zero on error. Generally, in the kernel, zero means
success so this be confusing and has caused bugs in the past. Also the
"tx" version of this function only returns negative error codes. Let's
clean this "rx" function so both functions match.
This patch has no effect on runtime.
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org> Acked-by: Peter Ujfalusi <peter.ujfalusi@gmail.com> Acked-by: Vinod Koul <vkoul@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Jakub Kicinski [Wed, 5 Jun 2024 17:16:44 +0000 (10:16 -0700)]
tools: ynl: make user space policies const
Dan, who's working on C++ YNL, pointed out that the C code
does not make policies const. Sprinkle some 'const's around.
Reported-by: Dan Melnic <dmm@meta.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org> Reviewed-by: Donald Hunter <donald.hunter@gmail.com> Reviewed-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David Wei [Wed, 5 Jun 2024 16:19:24 +0000 (09:19 -0700)]
page_pool: remove WARN_ON() with OR
Having an OR in WARN_ON() makes me sad because it's impossible to tell
which condition is true when triggered.
Split a WARN_ON() with an OR in page_pool_disable_direct_recycling().
Signed-off-by: David Wei <dw@davidwei.uk> Reviewed-by: Mina Almasry <almasrymina@google.com> Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com> Acked-by: Jesper Dangaard Brouer <hawk@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
net: ti: icssg-prueth: Add multicast filtering support
Add multicast filtering support for ICSSG Driver. Multicast addresses will
be updated by __dev_mc_sync() API. icssg_prueth_add_macst () and
icssg_prueth_del_mcast() will be sync and unsync APIs for the driver
respectively.
To add a mac_address for a port, driver needs to call icssg_fdb_add_del()
and pass the mac_address and BIT(port_id) to the API. The ICSSG firmware
will then configure the rules and allow filtering.
If a mac_address is added to port0 and the same mac_address needs to be
added for port1, driver needs to pass BIT(port0) | BIT(port1) to the
icssg_fdb_add_del() API. If driver just pass BIT(port1) then the entry for
port0 will be overwritten / lost. This is a design constraint on the
firmware side.
To overcome this in the driver, to add any mac_address for let's say portX
driver first checks if the same mac_address is already added for any other
port. If yes driver calls icssg_fdb_add_del() with BIT(portX) |
BIT(other_existing_port). If not, driver calls icssg_fdb_add_del() with
BIT(portX).
The same thing is applicable for deleting mac_addresses as well. This
logic is in icssg_prueth_add_mcast / icssg_prueth_del_mcast APIs.
Signed-off-by: MD Danish Anwar <danishanwar@ti.com> Reviewed-by: Wojciech Drewek <wojciech.drewek@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Geliang Tang [Thu, 30 May 2024 07:41:12 +0000 (15:41 +0800)]
selftests/bpf: Drop useless arguments of do_test in bpf_tcp_ca
bpf_map_lookup_elem() has been removed from do_test(), it makes the
sk_stg_map argument of do_test() useless. In addition, two exactly the
same opts are passed in all the places where do_test() is invoked, so
cli_opts argument can be dropped too.
This patch drops these two useless arguments of do_test() in bpf_tcp_ca.c.
Geliang Tang [Thu, 30 May 2024 07:41:11 +0000 (15:41 +0800)]
selftests/bpf: Use start_test in test_dctcp in bpf_tcp_ca
The "if (sk_stg_map)" block in do_test() is only used by test_dctcp(),
it makes sense to move it from do_test() into test_dctcp(). Then
do_test() can be used by other tests except test_dctcp().
Geliang Tang [Thu, 30 May 2024 07:41:10 +0000 (15:41 +0800)]
selftests/bpf: Use start_test in test_dctcp_fallback in bpf_tcp_ca
The newly added helper start_test() can be used in test_dctcp_fallback()
too, to replace start_server_str() and connect_to_fd_opts(). In that
way, two network_helper_opts srv_opts and cli_opts are used instead of
the previously shared opts.
Geliang Tang [Thu, 30 May 2024 07:41:09 +0000 (15:41 +0800)]
selftests/bpf: Add start_test helper in bpf_tcp_ca
For moving the "if (sk_stg_map)" block out of do_test(), extract the
code before this block as a new function start_test(). It creates
server-side and client-side sockets and returns them to the caller.
Geliang Tang [Thu, 30 May 2024 07:41:08 +0000 (15:41 +0800)]
selftests/bpf: Use connect_to_fd_opts in do_test in bpf_tcp_ca
This patch uses connect_to_fd_opts() instead of using connect_fd_to_fd()
and settcpca() in do_test() in prog_tests/bpf_tcp_ca.c to accept a struct
network_helper_opts argument.
Then define a dctcp dedicated post_socket_cb callback stg_post_socket_cb(),
invoking both settcpca() and bpf_map_update_elem() in it, and set it in
test_dctcp(). For passing map_fd into stg_post_socket_cb() callback, a new
member map_fd is added in struct cb_opts.
Add another "const struct network_helper_opts *cli_opts" to do_test() to
separate it from the server "opts".
drivers/net/ethernet/pensando/ionic/ionic_txrx.c d9c04209990b ("ionic: Mark error paths in the data path as unlikely") 491aee894a08 ("ionic: fix kernel panic in XDP_TX action")
net/ipv6/ip6_fib.c b4cb4a1391dc ("net: use unrcu_pointer() helper") b01e1c030770 ("ipv6: fix possible race in __fib6_drop_pcpu_from()")
Mykyta Yatsenko [Wed, 5 Jun 2024 17:51:35 +0000 (18:51 +0100)]
libbpf: Auto-attach struct_ops BPF maps in BPF skeleton
Similarly to `bpf_program`, support `bpf_map` automatic attachment in
`bpf_object__attach_skeleton`. Currently only struct_ops maps could be
attached.
On bpftool side, code-generate links in skeleton struct for struct_ops maps.
Similarly to `bpf_program_skeleton`, set links in `bpf_map_skeleton`.
On libbpf side, extend `bpf_map` with new `autoattach` field to support
enabling or disabling autoattach functionality, introducing
getter/setter for this field.
`bpf_object__(attach|detach)_skeleton` is extended with
attaching/detaching struct_ops maps logic.
Linus Torvalds [Thu, 6 Jun 2024 16:55:27 +0000 (09:55 -0700)]
Merge tag 'net-6.10-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Pull networking fixes from Jakub Kicinski:
"Including fixes from BPF and big collection of fixes for WiFi core and
drivers.
Current release - regressions:
- vxlan: fix regression when dropping packets due to invalid src
addresses
- bpf: fix a potential use-after-free in bpf_link_free()
- xdp: revert support for redirect to any xsk socket bound to the
same UMEM as it can result in a corruption
- virtio_net:
- add missing lock protection when reading return code from
control_buf
- fix false-positive lockdep splat in DIM
- Revert "wifi: wilc1000: convert list management to RCU"
- wifi: ath11k: fix error path in ath11k_pcic_ext_irq_config
Previous releases - regressions:
- rtnetlink: make the "split" NLM_DONE handling generic, restore the
old behavior for two cases where we started coalescing those
messages with normal messages, breaking sloppily-coded userspace
- wifi:
- cfg80211: validate HE operation element parsing
- cfg80211: fix 6 GHz scan request building
- mt76: mt7615: add missing chanctx ops
- ath11k: move power type check to ASSOC stage, fix connecting to
6 GHz AP
- ath11k: fix WCN6750 firmware crash caused by 17 num_vdevs
- rtlwifi: ignore IEEE80211_CONF_CHANGE_RETRY_LIMITS
- iwlwifi: mvm: fix a crash on 7265
Previous releases - always broken:
- ncsi: prevent multi-threaded channel probing, a spec violation
- vmxnet3: disable rx data ring on dma allocation failure
- ethtool: init tsinfo stats if requested, prevent unintentionally
reporting all-zero stats on devices which don't implement any
- dst_cache: fix possible races in less common IPv6 features
- tcp: auth: don't consider TCP_CLOSE to be in TCP_AO_ESTABLISHED
- ax25: fix two refcounting bugs
- eth: ionic: fix kernel panic in XDP_TX action
Misc:
- tcp: count CLOSE-WAIT sockets for TCP_MIB_CURRESTAB"
* tag 'net-6.10-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (107 commits)
selftests: net: lib: set 'i' as local
selftests: net: lib: avoid error removing empty netns name
selftests: net: lib: support errexit with busywait
net: ethtool: fix the error condition in ethtool_get_phy_stats_ethtool()
ipv6: fix possible race in __fib6_drop_pcpu_from()
af_unix: Annotate data-race of sk->sk_shutdown in sk_diag_fill().
af_unix: Use skb_queue_len_lockless() in sk_diag_show_rqlen().
af_unix: Use skb_queue_empty_lockless() in unix_release_sock().
af_unix: Use unix_recvq_full_lockless() in unix_stream_connect().
af_unix: Annotate data-race of net->unx.sysctl_max_dgram_qlen.
af_unix: Annotate data-races around sk->sk_sndbuf.
af_unix: Annotate data-races around sk->sk_state in UNIX_DIAG.
af_unix: Annotate data-race of sk->sk_state in unix_stream_read_skb().
af_unix: Annotate data-races around sk->sk_state in sendmsg() and recvmsg().
af_unix: Annotate data-race of sk->sk_state in unix_accept().
af_unix: Annotate data-race of sk->sk_state in unix_stream_connect().
af_unix: Annotate data-races around sk->sk_state in unix_write_space() and poll().
af_unix: Annotate data-race of sk->sk_state in unix_inq_len().
af_unix: Annodate data-races around sk->sk_state for writers.
af_unix: Set sk->sk_state under unix_state_lock() for truly disconencted peer.
...
Jakub Kicinski [Thu, 6 Jun 2024 15:23:44 +0000 (08:23 -0700)]
Merge branch 'selftests-net-lib-small-fixes'
Matthieu Baerts says:
====================
selftests: net: lib: small fixes
While looking at using 'lib.sh' for the MPTCP selftests [1], we found
some small issues with 'lib.sh'. Here they are:
- Patch 1: fix 'errexit' (set -e) support with busywait. 'errexit' is
supported in some functions, not all. A fix for v6.8+.
- Patch 2: avoid confusing error messages linked to the cleaning part
when the netns setup fails. A fix for v6.8+.
- Patch 3: set a variable as local to avoid accidentally changing the
value of a another one with the same name on the caller side. A fix
for v6.10-rc1+.
Without this, the 'i' variable declared before could be overridden by
accident, e.g.
for i in "${@}"; do
__ksft_status_merge "${i}" ## 'i' has been modified
foo "${i}" ## using 'i' with an unexpected value
done
After a quick look, it looks like 'i' is currently not used after having
been modified in __ksft_status_merge(), but still, better be safe than
sorry. I saw this while modifying the same file, not because I suspected
an issue somewhere.
selftests: net: lib: avoid error removing empty netns name
If there is an error to create the first netns with 'setup_ns()',
'cleanup_ns()' will be called with an empty string as first parameter.
The consequences is that 'cleanup_ns()' will try to delete an invalid
netns, and wait 20 seconds if the netns list is empty.
Instead of just checking if the name is not empty, convert the string
separated by spaces to an array. Manipulating the array is cleaner, and
calling 'cleanup_ns()' with an empty array will be a no-op.
Fixes: 25ae948b4478 ("selftests/net: add lib.sh") Cc: stable@vger.kernel.org Acked-by: Geliang Tang <geliang@kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Reviewed-by: Petr Machata <petrm@nvidia.com> Reviewed-by: Hangbin Liu <liuhangbin@gmail.com> Link: https://lore.kernel.org/r/20240605-upstream-net-20240605-selftests-net-lib-fixes-v1-2-b3afadd368c9@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Yonghong Song [Wed, 5 Jun 2024 20:12:03 +0000 (13:12 -0700)]
selftests/bpf: Fix send_signal test with nested CONFIG_PARAVIRT
Alexei reported that send_signal test may fail with nested CONFIG_PARAVIRT
configs. In this particular case, the base VM is AMD with 166 cpus, and I
run selftests with regular qemu on top of that and indeed send_signal test
failed. I also tried with an Intel box with 80 cpus and there is no issue.
In the test, two processes are communicating through pipe. Further debugging
with strace found that the above splat is triggered as read() syscall could
not receive the data even if the corresponding write() syscall in another
process successfully wrote data into the pipe.
The failed subtest is "send_signal_perf". The corresponding perf event has
sample_period 1 and config PERF_COUNT_SW_CPU_CLOCK. sample_period 1 means every
overflow event will trigger a call to the BPF program. So I suspect this may
overwhelm the system. So I increased the sample_period to 100,000 and the test
passed. The sample_period 10,000 still has the test failed.
In other parts of selftest, e.g., [1], sample_freq is used instead. So I
decided to use sample_freq = 1,000 since the test can pass as well.
Paolo Abeni [Thu, 6 Jun 2024 13:18:06 +0000 (15:18 +0200)]
Merge branch 'tcp-small-code-reorg'
Eric Dumazet says:
====================
tcp: small code reorg
Replace a WARN_ON_ONCE() that never triggered
to DEBUG_NET_WARN_ON_ONCE() in reqsk_free().
Move inet_reqsk_alloc() and reqsk_alloc()
to inet_connection_sock.c, to unclutter
net/ipv4/tcp_input.c and include/net/request_sock.h
====================
Davide Caratti [Wed, 5 Jun 2024 07:15:42 +0000 (09:15 +0200)]
mptcp: refer to 'MPTCP' socket in comments
We used to call it 'master' socket at the early stages of MPTCP
development, but the correct wording is 'MPTCP' socket opposed to 'TCP
subflows': convert the last 3 comments to use a more appropriate term.
Geliang Tang [Wed, 5 Jun 2024 07:15:41 +0000 (09:15 +0200)]
mptcp: add mptcp_space_from_win helper
As a wrapper of __tcp_space_from_win(), this patch adds a MPTCP dedicated
space_from_win helper mptcp_space_from_win() in protocol.h to paired with
mptcp_win_from_space().
Use it instead of __tcp_space_from_win() in both mptcp_rcv_space_adjust()
and mptcp_set_rcvlowat().
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn> Reviewed-by: Mat Martineau <martineau@kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Geliang Tang [Wed, 5 Jun 2024 07:15:40 +0000 (09:15 +0200)]
mptcp: use mptcp_win_from_space helper
The MPTCP dedicated win_from_space helper mptcp_win_from_space() is defined
in protocol.h, use it in mptcp_rcv_space_adjust() instead of using the TCP
one. Here scaling_ratio is the same as msk->scaling_ratio.
Signed-off-by: Geliang Tang <tanggeliang@kylinos.cn> Reviewed-by: Mat Martineau <martineau@kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Jason Xing [Wed, 5 Jun 2024 02:29:32 +0000 (10:29 +0800)]
net: allow rps/rfs related configs to be switched
After John Sperbeck reported a compile error if the CONFIG_RFS_ACCEL
is off, I found that I cannot easily enable/disable the config
because of lack of the prompt when using 'make menuconfig'. Therefore,
I decided to change rps/rfc related configs altogether.
Note that with this series there is skb_queue_empty() left in
unix_dgram_disconnected() that needs to be changed to lockless
version, and unix_peer(other) access there should be protected
by unix_state_lock().
This will require some refactoring, so another series will follow.
Kuniyuki Iwashima [Tue, 4 Jun 2024 16:52:39 +0000 (09:52 -0700)]
af_unix: Use skb_queue_empty_lockless() in unix_release_sock().
If the socket type is SOCK_STREAM or SOCK_SEQPACKET, unix_release_sock()
checks the length of the peer socket's recvq under unix_state_lock().
However, unix_stream_read_generic() calls skb_unlink() after releasing
the lock. Also, for SOCK_SEQPACKET, __skb_try_recv_datagram() unlinks
skb without unix_state_lock().
Thues, unix_state_lock() does not protect qlen.
Let's use skb_queue_empty_lockless() in unix_release_sock().
Kuniyuki Iwashima [Tue, 4 Jun 2024 16:52:38 +0000 (09:52 -0700)]
af_unix: Use unix_recvq_full_lockless() in unix_stream_connect().
Once sk->sk_state is changed to TCP_LISTEN, it never changes.
unix_accept() takes advantage of this characteristics; it does not
hold the listener's unix_state_lock() and only acquires recvq lock
to pop one skb.
It means unix_state_lock() does not prevent the queue length from
changing in unix_stream_connect().
Thus, we need to use unix_recvq_full_lockless() to avoid data-race.
Now we remove unix_recvq_full() as no one uses it.
Note that we can remove READ_ONCE() for sk->sk_max_ack_backlog in
unix_recvq_full_lockless() because of the following reasons:
(1) For SOCK_DGRAM, it is a written-once field in unix_create1()
(2) For SOCK_STREAM and SOCK_SEQPACKET, it is changed under the
listener's unix_state_lock() in unix_listen(), and we hold
the lock in unix_stream_connect()
Kuniyuki Iwashima [Tue, 4 Jun 2024 16:52:30 +0000 (09:52 -0700)]
af_unix: Annotate data-races around sk->sk_state in unix_write_space() and poll().
unix_poll() and unix_dgram_poll() read sk->sk_state locklessly and
calls unix_writable() which also reads sk->sk_state without holding
unix_state_lock().
Let's use READ_ONCE() in unix_poll() and unix_dgram_poll() and pass
it to unix_writable().
While at it, we remove TCP_SYN_SENT check in unix_dgram_poll() as
that state does not exist for AF_UNIX socket since the code was added.
Fixes: 1586a5877db9 ("af_unix: do not report POLLOUT on listeners") Fixes: 3c73419c09a5 ("af_unix: fix 'poll for write'/ connected DGRAM sockets") Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Kuniyuki Iwashima [Tue, 4 Jun 2024 16:52:27 +0000 (09:52 -0700)]
af_unix: Set sk->sk_state under unix_state_lock() for truly disconencted peer.
When a SOCK_DGRAM socket connect()s to another socket, the both sockets'
sk->sk_state are changed to TCP_ESTABLISHED so that we can register them
to BPF SOCKMAP.
When the socket disconnects from the peer by connect(AF_UNSPEC), the state
is set back to TCP_CLOSE.
Then, the peer's state is also set to TCP_CLOSE, but the update is done
locklessly and unconditionally.
Let's say socket A connect()ed to B, B connect()ed to C, and A disconnects
from B.
After the first two connect()s, all three sockets' sk->sk_state are
TCP_ESTABLISHED:
So, when a socket disconnects from the peer, we should not set TCP_CLOSE to
the peer if the peer is connected to yet another socket, and this must be
done under unix_state_lock().
Note that we use WRITE_ONCE() for sk->sk_state as there are many lockless
readers. These data-races will be fixed in the following patches.
Fixes: 83301b5367a9 ("af_unix: Set TCP_ESTABLISHED for datagram sockets too") Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Eric Dumazet [Tue, 4 Jun 2024 16:51:50 +0000 (16:51 +0000)]
inet: remove (struct uncached_list)->quarantine
This list is used to tranfert dst that are handled by
rt_flush_dev() and rt6_uncached_list_flush_dev() out
of the per-cpu lists.
But quarantine list is not used later.
If we simply use list_del_init(&rt->dst.rt_uncached),
this also removes the dst from per-cpu list.
This patch also makes the future calls to rt_del_uncached_list()
and rt6_uncached_list_del() faster, because no spinlock
acquisition is needed anymore.
Aleksandr Mishin [Tue, 4 Jun 2024 08:25:00 +0000 (11:25 +0300)]
net: wwan: iosm: Fix tainted pointer delete is case of region creation fail
In case of region creation fail in ipc_devlink_create_region(), previously
created regions delete process starts from tainted pointer which actually
holds error code value.
Fix this bug by decreasing region index before delete.
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Fixes: 4dcd183fbd67 ("net: wwan: iosm: devlink registration") Signed-off-by: Aleksandr Mishin <amishin@t-argos.ru> Acked-by: Sergey Ryazanov <ryazanov.s.a@gmail.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://lore.kernel.org/r/20240604082500.20769-1-amishin@t-argos.ru Signed-off-by: Paolo Abeni <pabeni@redhat.com>
====================
Improve GbEth performance on Renesas RZ/G2L and related SoCs
This series aims to improve performance of the GbEth IP in the Renesas
RZ/G2L SoC family and the RZ/G3S SoC, which use the ravb driver. Along
the way, we do some refactoring and ensure that napi_complete_done() is
used in accordance with the NAPI documentation for both GbEth and R-Car
code paths.
Much of the performance improvement comes from enabling SW IRQ
Coalescing for all SoCs using the GbEth IP, and NAPI Threaded mode for
single core SoCs using the GbEth IP. These can be enabled/disabled at
runtime via sysfs, but our goal is to set sensible defaults which get
good performance on the affected SoCs.
The rest of the performance improvement comes from using a page pool to
allocate RX buffers, and reducing the allocation size from >8kB to 2kB.
The overall performance impact of this patch series seen in testing with
iperf3 is as follows (see patches 5-7 for more detailed results):
* RZ/G2L:
* TCP TX: +1.8% bandwidth
* TCP RX: +1% bandwidth at 47% less CPU load
* UDP RX: +1% bandwidth at 26% less CPU load
There is no significant impact on bandwidth or CPU load in testing on
RZ/G2H or R-Car M3N.
Fixing the crash in UDP RX testing for RZ/Five is a cumulative effect of
patches 1, 2, 5 & 6 so this is very difficult to break out as a bugfix
for backporting.
Changes v4->v5:
* Added Sergey's Reviewed-by tags.
* Improved the commit message for patch 2/7.
* Re-wrapped to 80 cols, except where this would significantly impact
readability.
* Use lower case `skb` consistently in comments.
* Included <net/page_pool/types.h> in ravb.h.
* Moved rx_buffer_size so it is in the same place in ravb_hw_info as
rx_max_desc_use was previously.
* Used reverse xmas tree ordering in variable declarations.
* Split lines after binary operators, instead of before.
* Factor subtraction of sizeof(__sum16) out of the if condition in
ravb_rx_csum_gbeth().
* Add blank lines after variable declarations where needed.
* Used goto instead of break to handle napi_build_skb() failure in
ravb_rx_gbeth(). Break was incorrectly scoped to the surrounding
switch statement, when it's the outer loop we really want to break
out of.
* Used continue instead of break to handle NULL priv->rx_1st_skb in
ravb_rx_gbeth() as we may still be able to process further
descriptors.
* Unconditionally set priv->rx_1st_skb = NULL after processing a
packet in ravb_rx_gbeth(). We don't need to check die_dt as this
will be a no-op for single descriptor packets.
* Moved napi_build_skb() call after dma_sync_single_for_cpu() in
ravb_rx_rcar() to align the order of operations with ravb_rx_gbeth()
and ensure the data is sync'd before it is accessed.
* Moved zeroing of rx_buff->page to the end of packet processing in
ravb_rx_rcar() to align the order of operations with
ravb_rx_gbeth().
Changes v3->v4:
* Dependency patches have merged so this is no longer an RFC.
* Fixed update of stats->rx_packets.
* Simplified refactoring following feedback from Niklas and Sergey.
* Renamed needs_irq_coalesce -> coalesce_irqs.
* Used a separate page pool for each RX queue.
* Passed struct ravb_rx_desc to ravb_alloc_rx_buffer() so that we can
simplify the calling function.
* Explained the calculation of rx_desc->ds_cc.
* Added handling of nonlinear SKBs in ravb_rx_csum_gbeth().
* Used Niklas' suggested commit message for patch 2/7.
* Added Sergey's Reviewed-by tags to patches 5/7 and 6/7.
Changes v2->v3:
* Incorporated feedback on RFC v2 from Sergey.
* Split out bugfixes and rebased. This changed the order of what was
the first 5 patches of v2 and things look a little different so I've
not picked up Reviewed-by tags from v2.
* Further refactoring and tidy up of RX ring refill and
ravb_rx_gbeth().
* Switched to using a page pool to allocate RX buffers.
* Re-tested and provided updated performance figures.
Changes v1->v2:
* Marked as RFC as the series depends on unmerged patches.
* Refactored R-Car code paths as well as GbEth code paths.
* Updated references to the patches this series depends on.
====================
Paul Barker [Tue, 4 Jun 2024 07:28:25 +0000 (08:28 +0100)]
net: ravb: Allocate RX buffers via page pool
This patch makes multiple changes that can't be separated:
1) Allocate plain RX buffers via a page pool instead of allocating
SKBs, then use build_skb() when a packet is received.
2) For GbEth IP, reduce the RX buffer size to 2kB.
3) For GbEth IP, merge packets which span more than one RX descriptor
as SKB fragments instead of copying data.
Implementing (1) without (2) would require the use of an order-1 page
pool (instead of an order-0 page pool split into page fragments) for
GbEth.
Implementing (2) without (3) would leave us no space to re-assemble
packets which span more than one RX descriptor.
Implementing (3) without (1) would not be possible as the network stack
expects to use put_page() or page_pool_put_page() to free SKB fragments
after an SKB is consumed.
RX checksum offload support is adjusted to handle both linear and
nonlinear (fragmented) packets.
This patch gives the following improvements during testing with iperf3.
* RZ/G2L:
* TCP RX: same bandwidth at -43% CPU load (70% -> 40%)
* UDP RX: same bandwidth at -17% CPU load (88% -> 74%)
Paul Barker [Tue, 4 Jun 2024 07:28:24 +0000 (08:28 +0100)]
net: ravb: Use NAPI threaded mode on 1-core CPUs with GbEth IP
NAPI Threaded mode (along with the previously enabled SW IRQ Coalescing)
is required to improve network stack performance for single core SoCs
using the GbEth IP (currently the RZ/G2L SoC family and the RZ/G3S SoC).
This patch gives the following improvements during testing with iperf3.
These losses are considered acceptable given the benefits shown above.
If UDP TX bandwidth must be maximised for a particular use case, NAPI
threaded mode can be disabled at runtime via sysfs writes.
The improvement of UDP RX bandwidth for the single core SoCs (RZ/G2UL &
RZ/G3S) is particularly critical.
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com> Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Paul Barker [Tue, 4 Jun 2024 07:28:23 +0000 (08:28 +0100)]
net: ravb: Enable SW IRQ Coalescing for GbEth
Software IRQ Coalescing is required to improve network stack performance
in the RZ/G2L SoC family and the RZ/G3S SoC, i.e. the SoCs which use the
GbEth IP.
This patch gives the following improvements during testing with iperf3:
* RZ/G2L:
* TCP RX: same bandwidth with -6% CPU load (76% -> 71%)
* UDP RX: same bandwidth with -10% CPU load (99% -> 89%)
Paul Barker [Tue, 4 Jun 2024 07:28:21 +0000 (08:28 +0100)]
net: ravb: Refactor RX ring refill
To reduce code duplication, we add a new RX ring refill function which
can handle both the initial RX ring population (which was split between
ravb_ring_init() and ravb_ring_format()) and the RX ring refill after
polling (in ravb_rx()).
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com> Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Paul Barker [Tue, 4 Jun 2024 07:28:20 +0000 (08:28 +0100)]
net: ravb: Align poll function with NAPI docs
Align ravb_poll() with the documentation in
`Documentation/networking/kapi.rst` and
`Documentation/networking/napi.rst`.
The documentation says that we should prefer napi_complete_done() over
napi_complete(), and using the former allows us to properly support busy
polling. We should ensure that napi_complete_done() is only called if
the work budget has not been exhausted, and we should only re-arm
interrupts if it returns true.
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com> Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Paul Barker [Tue, 4 Jun 2024 07:28:19 +0000 (08:28 +0100)]
net: ravb: Simplify poll & receive functions
We don't need to pass the work budget to ravb_rx() by reference, it's
cleaner to pass this by value and return the amount of work done. This
allows us to simplify the ravb_poll() function and use the common
`work_done` variable name seen in other network drivers for consistency
and ease of understanding.
This is a pure refactor and should not affect behaviour.
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com> Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Benchmark details:
VM based setup
CPU: Intel(R) Xeon(R) Platinum 8380 CPU, 24 cores
NIC: ConnectX-7 100GbE
iperf3 and irq running on same CPU over a single receive queue
====================
Dragos Tatulea [Mon, 3 Jun 2024 21:22:19 +0000 (00:22 +0300)]
net/mlx5e: SHAMPO, Coalesce skb fragments to page size
When doing hardware GRO (SHAMPO), the driver puts each data payload of a
packet from the wire into one skb fragment. TCP Zero-Copy expects page
sized skb fragments to be able to do it's page-flipping magic. With the
current way of arranging fragments by the driver, only specific MTUs
(page sized multiple + header size) will yield such page sized fragments
in a high percentage.
This change improves payload arrangement in the skb for hardware GRO by
coalescing payloads into a single skb fragment when possible.
To demonstrate the fix, running tcp_mmap with a MTU of 1500 yields:
- Before: 0 % bytes mmap'ed
- After : 81 % bytes mmap'ed
More importantly, coalescing considerably improves the HW GRO performance.
Here are the results for a iperf3 bandwidth benchmark:
+---------+--------+--------+------------------------+-----------+
| streams | SW GRO | HW GRO | HW GRO with coalescing | Unit |
|---------+--------+--------+------------------------+-----------|
| 1 | 36 | 42 | 57 | Gbits/sec |
| 4 | 34 | 39 | 50 | Gbits/sec |
| 8 | 31 | 35 | 43 | Gbits/sec |
+---------+--------+--------+------------------------+-----------+
Benchmark details:
VM based setup
CPU: Intel(R) Xeon(R) Platinum 8380 CPU, 24 cores
NIC: ConnectX-7 100GbE
iperf3 and irq running on same CPU over a single receive queue
Yoray Zack [Mon, 3 Jun 2024 21:22:18 +0000 (00:22 +0300)]
net/mlx5e: SHAMPO, Re-enable HW-GRO
Add back HW-GRO to the reported features.
As the current implementation of HW-GRO uses KSMs with a
specific fixed buffer size (256B) to map its headers buffer,
we reported the feature only if the NIC is supporting KSM and
the minimum value for buffer size is below the requested one.
A downstream patch will add skb fragment coalescing which will improve
performance considerably.
Benchmark details:
VM based setup
CPU: Intel(R) Xeon(R) Platinum 8380 CPU, 24 cores
NIC: ConnectX-7 100GbE
iperf3 and irq running on same CPU over a single receive queue
Yoray Zack [Mon, 3 Jun 2024 21:22:17 +0000 (00:22 +0300)]
net/mlx5e: SHAMPO, Use KSMs instead of KLMs
KSM Mkey is KLM Mkey with a fixed buffer size. Due to this fact,
it is a faster mechanism than KLM.
SHAMPO feature used KLMs Mkeys for memory mappings of its headers buffer.
As it used KLMs with the same buffer size for each entry,
we can use KSMs instead.
This commit changes the Mkeys that map the SHAMPO headers buffer
from KLMs to KSMs.
projects / users / hch / configfs.git / log