- 12 Nov, 2022 24 commits
-
-
Ofer Levi authored
CQE compression feature improves performance by reducing PCI bandwidth bottleneck on CQEs write. Enhanced CQE compression introduced in ConnectX-6 and it aims to reduce CPU utilization of SW side packets decompression by eliminating the need to rewrite ownership bit, which is likely to cost a cache-miss, is replaced by validity byte handled solely by HW. Another advantage of the enhanced feature is that session packets are available to SW as soon as a single CQE slot is filled, instead of waiting for session to close, this improves packet latency from NIC to host. Performance: Following are tested scenarios and reults comparing basic and enahnced CQE compression. setup: IXIA 100GbE connected directly to port 0 and port 1 of ConnectX-6 Dx 100GbE dual port. Case #1 RX only, single flow goes to single queue: IRQ rate reduced by ~ 30%, CPU utilization improved by 2%. Case #2 IP forwarding from port 1 to port 0 single flow goes to single queue: Avg latency improved from 60us to 21us, frame loss improved from 0.5% to 0.0%. Case #3 IP forwarding from port 1 to port 0 Max Throughput IXIA sends 100%, 8192 UDP flows, goes to 24 queues: Enhanced is equal or slightly better than basic. Testing the basic compression feature with this patch shows there is no perfrormance degradation of the basic compression feature. Signed-off-by:
Ofer Levi <oferle@nvidia.com> Reviewed-by:
Tariq Toukan <tariqt@nvidia.com> Signed-off-by:
Saeed Mahameed <saeedm@nvidia.com>
-
Gal Pressman authored
Replace the min/max operations with a single clamp. Signed-off-by:
Gal Pressman <gal@nvidia.com> Reviewed-by:
-
Anisse Astier authored
This is never used, and probably something that was intended to be used before per-protocol hash tables were chosen instead. Signed-off-by:
Anisse Astier <anisse@astier.eu> Reviewed-by:
Leon Romanovsky <leonro@nvidia.com> Signed-off-by:
-
Eli Cohen authored
hca_id is an identifier of an mlx5_core instance within the hardware. This identifier may be required for troubleshooting. Expose it to debugfs. Example: $ cat /sys/kernel/debug/mlx5/mlx5_core.sf.2/vhca_id 0x12 Signed-off-by:
Eli Cohen <elic@nvidia.com> Reviewed-by:
Roi Dayan <roid@nvidia.com> Signed-off-by:
-
Moshe Shemesh authored
Before this patch, devlink traps are registered only on full driver probe and unregistered on driver removal. As devlink traps are not usable once driver functionality is unloaded, it should be unrgeistered also on flows that unload the driver and then registered when loaded back, e.g. devlink reload flow. Signed-off-by:
Moshe Shemesh <moshe@nvidia.com> Reviewed-by:
Aya Levin <ayal@nvidia.com> Signed-off-by:
-
Colin Ian King authored
There is a spelling mistake in an error message. Fix it. Signed-off-by:
Colin Ian King <colin.i.king@gmail.com> Reviewed-by:
-
Roi Dayan authored
There is no need for the warn if entry already removed. Use debug print like in the update flow. Also update the messages so user can identify if the it's from the update flow or remove flow. Signed-off-by:
Vlad Buslov <vladbu@nvidia.com> Signed-off-by:
-
Eric Dumazet authored
Use try_cmpxchg() (instead of cmpxchg()) in a more readable way. oval = smp_load_acquire(&sk->sk_tsq_flags); do { ... } while (!try_cmpxchg(&sk->sk_tsq_flags, &oval, nval)); Reduce indentation level. Signed-off-by:Eric Dumazet <edumazet@google.com> Link: https://lore.kernel.org/r/20221110190239.3531280-1-edumazet@google.comSigned-off-by:
Jakub Kicinski <kuba@kernel.org>
-
Eric Dumazet authored
try_cmpxchg() is slighly more efficient (at least on x86), and smp_load_acquire(&sk->sk_tsq_flags) could avoid a KCSAN report. Signed-off-by:
-
Ido Schimmel authored
No changes in generated code. Reported-by:
Petr Machata <petrm@nvidia.com> Reviewed-by:
Ido Schimmel <idosch@nvidia.com> Link: https://lore.kernel.org/r/20221110085422.521059-1-idosch@nvidia.comSigned-off-by:
-
Jakub Kicinski authored
Alex Elder says: ==================== dt-bindings: net: qcom,ipa: relax some restrictions The first patch in this series simply removes an unnecessary requirement in the IPA binding. Previously, if the modem was doing GSI firmware loading, the firmware name property was required to *not* be present. There is no harm in having the firmware name be specified, so this restriction isn't needed. The second patch restates a requirement on the "memory-region" property more accurately. These binding changes have no impact on existing code or DTS files. These aren't really bug fixes, so no need to back-port. ==================== Link: https://lore.kernel.org/r/20221110195619.1276302-1-elder@linaro.orgSigned-off-by:
-
Alex Elder authored
Either the AP or modem loads GSI firmware. If the modem-init property is present, the modem loads it. Otherwise, the AP loads it, and in that case the memory-region property must be defined. Currently this requirement is expressed as one or the other of the modem-init or the memory-region property being required. But it's harmless for the memory-region to be present if the modem is loading firmware (it'll just be ignored). Restate the requirement so that the memory-region property is required only if modem-init is not present. Signed-off-by:
Alex Elder <elder@linaro.org> Reviewed-by:
Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Signed-off-by:
-
Alex Elder authored
Commit d8604b20 ("dt-bindings: net: qcom,ipa: add firmware-name property") added a requirement for a "firmware-name" property that is more restrictive than necessary. If the AP loads GSI firmware, the name of the firmware file to use may optionally be provided via a "firmware-name" property. If the *modem* loads GSI firmware, "firmware-name" doesn't need to be supplied--but it's harmless to do so (it will simply be ignored). Remove the unnecessary restriction, and allow "firware-name" to be supplied even if it's not needed. Signed-off-by:
-
Angelo Dureghello authored
Enabling set_policy capability for mv88e6321. Signed-off-by:
Angelo Dureghello <angelo.dureghello@timesys.com> Reviewed-by:
Andrew Lunn <andrew@lunn.ch> Link: https://lore.kernel.org/r/20221110091027.998073-1-angelo.dureghello@timesys.comSigned-off-by:
-
Jakub Kicinski authored
Mat Martineau says: ==================== mptcp: Miscellaneous refactoring and small fixes Patches 1-3 do some refactoring to more consistently handle sock casts, and to remove some duplicate code. No functional changes. Patch 4 corrects a variable name in a self test, but does not change functionality since the same value gets used due to bash's scoping rules. Patch 5 rewords a comment. ==================== Link: https://lore.kernel.org/r/20221110232322.125068-1-mathew.j.martineau@linux.intel.comSigned-off-by:
-
Mat Martineau authored
We kept getting initial patches from new contributors to remove a duplicate 'the' (since grammar checking scripts flag it), but submitters never followed up after code review. Reviewed-by:
Matthieu Baerts <matthieu.baerts@tessares.net> Signed-off-by:
Mat Martineau <mathew.j.martineau@linux.intel.com> Signed-off-by:
-
Geliang Tang authored
'time' is the local variable of run_test() function, while 'max_time' is the local variable of do_transfer() function. So in do_transfer(), $max_time should be used, not $time. Please note that here $time == $max_time so the behaviour is not changed but the right variable is used. Reviewed-by:
Geliang Tang <geliang.tang@suse.com> Signed-off-by:
-
Geliang Tang authored
Use '(struct sock *)msk' to get 'sk' from 'msk' in a more direct way instead of using '&msk->sk.icsk_inet.sk'. Reviewed-by:
-
Geliang Tang authored
The function mptcp_subflow_process_delegated() uses the input ssk first, while __mptcp_check_push() invokes the packet scheduler first. So this patch adds a new parameter named 'first' for the function __mptcp_subflow_push_pending() to deal with these two cases separately. With this change, the code that invokes the packet scheduler in the function __mptcp_check_push() can be removed, and replaced by invoking __mptcp_subflow_push_pending() directly. Reviewed-by:
-
Geliang Tang authored
Use msk instead of mptcp_sk(sk) in the functions where the variable "msk = mptcp_sk(sk)" has been defined. Reviewed-by:
-
https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-nextJakub Kicinski authored
Andrii Nakryiko says: ==================== bpf-next 2022-11-11 We've added 49 non-merge commits during the last 9 day(s) which contain a total of 68 files changed, 3592 insertions(+), 1371 deletions(-). The main changes are: 1) Veristat tool improvements to support custom filtering, sorting, and replay of results, from Andrii Nakryiko. 2) BPF verifier precision tracking fixes and improvements, from Andrii Nakryiko. 3) Lots of new BPF documentation for various BPF maps, from Dave Tucker, Donald Hunter, Maryam Tahhan, Bagas Sanjaya. 4) BTF dedup improvements and libbpf's hashmap interface clean ups, from Eduard Zingerman. 5) Fix veth driver panic if XDP program is attached before veth_open, from John Fastabend. 6) BPF verifier clean ups and fixes in preparation for follow up features, from Kumar Kartikeya Dwivedi. 7) Add access to hwtstamp field from BPF sockops programs, from Martin KaFai Lau. 8) Various fixes for BPF selftests and samples, from Artem Savkov, Domenico Cerasuolo, Kang Minchul, Rong Tao, Yang Jihong. 9) Fix redirection to tunneling device logic, preventing skb->len == 0, from Stanislav Fomichev. * tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next: (49 commits) selftests/bpf: fix veristat's singular file-or-prog filter selftests/bpf: Test skops->skb_hwtstamp selftests/bpf: Fix incorrect ASSERT in the tcp_hdr_options test bpf: Add hwtstamp field for the sockops prog selftests/bpf: Fix xdp_synproxy compilation failure in 32-bit arch bpf, docs: Document BPF_MAP_TYPE_ARRAY docs/bpf: Document BPF map types QUEUE and STACK docs/bpf: Document BPF ARRAY_OF_MAPS and HASH_OF_MAPS docs/bpf: Document BPF_MAP_TYPE_CPUMAP map docs/bpf: Document BPF_MAP_TYPE_LPM_TRIE map libbpf: Hashmap.h update to fix build issues using LLVM14 bpf: veth driver panics when xdp prog attached before veth_open selftests: Fix test group SKIPPED result selftests/bpf: Tests for btf_dedup_resolve_fwds libbpf: Resolve unambigous forward declarations libbpf: Hashmap interface update to allow both long and void* keys/values samples/bpf: Fix sockex3 error: Missing BPF prog type selftests/bpf: Fix u32 variable compared with less than zero Documentation: bpf: Escape underscore in BPF type name prefix selftests/bpf: Use consistent build-id type for liburandom_read.so ... ==================== Link: https://lore.kernel.org/r/20221111233733.1088228-1-andrii@kernel.orgSigned-off-by:
-
Jakub Kicinski authored
Eric Dumazet says: ==================== net: vlan: claim one bit from sk_buff First patch claims skb->vlan_present. This means some bpf changes, eg for sparc32 that I could not test. Second patch removes one conditional test in gro_list_prepare(). ==================== Link: https://lore.kernel.org/r/20221109095759.1874969-1-edumazet@google.comSigned-off-by:
-
Eric Dumazet authored
We can remove a conditional test in gro_list_prepare() by comparing vlan_all fields of the two skbs. Notes: While comparing the vlan_proto is not strictly needed, because part of the following compare_ether_header() call, using 32bit word is actually faster than using 16bit values. napi_reuse_skb() makes sure to clear skb->vlan_all, as it already calls __vlan_hwaccel_clear_tag() Signed-off-by:
Martin KaFai Lau <martin.lau@kernel.org> Signed-off-by:
-
Eric Dumazet authored
skb->vlan_present seems redundant. We can instead derive it from this boolean expression: vlan_present = skb->vlan_proto != 0 || skb->vlan_tci != 0 Add a new union, to access both fields in a single load/store when possible. union { u32 vlan_all; struct { __be16 vlan_proto; __u16 vlan_tci; }; }; This allows following patch to remove a conditional test in GRO stack. Note: We move remcsum_offload to keep TC_AT_INGRESS_MASK and SKB_MONO_DELIVERY_TIME_MASK unchanged. Signed-off-by:Yonghong Song <yhs@fb.com> Acked-by:
-
- 11 Nov, 2022 16 commits
-
-
Andrii Nakryiko authored
Fix the bug of filtering out filename too early, before we know the program name, if using unified file-or-prog filter (i.e., -f <any-glob>). Because we try to filter BPF object file early without opening and parsing it, if any_glob (file-or-prog) filter is used we have to accept any filename just to get program name, which might match any_glob. Fixes: 10b1b3f3 ("selftests/bpf: consolidate and improve file/prog filtering in veristat") Signed-off-by:
Andrii Nakryiko <andrii@kernel.org> Acked-by:
Stanislav Fomichev <sdf@google.com> Link: https://lore.kernel.org/r/20221111181242.2101192-1-andrii@kernel.orgSigned-off-by:
-
Andrii Nakryiko authored
Martin KaFai Lau says: ==================== From: Martin KaFai Lau <martin.lau@kernel.org> The bpf-tc prog has already been able to access the skb_hwtstamps(skb)->hwtstamp. This set extends the same hwtstamp access to the sockops prog. v2: - Fixed the btf_dump selftest which depends on the last member of 'struct bpf_sock_ops'. ==================== Signed-off-by: -
Martin KaFai Lau authored
This patch tests reading the skops->skb_hwtstamp field. A local test was also done such that the shinfo hwtstamp was temporary set to a non zero value in the kernel bpf_skops_parse_hdr() and the same value can be read by the skops test. An adjustment is needed to the btf_dump selftest because the changes in the 'struct bpf_sock_ops'. Signed-off-by:
-
Martin KaFai Lau authored
This patch fixes the incorrect ASSERT test in tcp_hdr_options during the CHECK to ASSERT macro cleanup. Fixes: 3082f8cd ("selftests/bpf: Convert tcp_hdr_options test to ASSERT_* macros") Signed-off-by:
Wang Yufen <wangyufen@huawei.com> Acked-by:
-
Martin KaFai Lau authored
The bpf-tc prog has already been able to access the skb_hwtstamps(skb)->hwtstamp. This patch extends the same hwtstamp access to the sockops prog. In sockops, the skb is also available to the bpf prog during the BPF_SOCK_OPS_PARSE_HDR_OPT_CB event. There is a use case that the hwtstamp will be useful to the sockops prog to better measure the one-way-delay when the sender has put the tx timestamp in the tcp header option. Signed-off-by:
-
Yang Jihong authored
xdp_synproxy fails to be compiled in the 32-bit arch, log is as follows: xdp_synproxy.c: In function 'parse_options': xdp_synproxy.c:175:36: error: left shift count >= width of type [-Werror=shift-count-overflow] 175 | *tcpipopts = (mss6 << 32) | (ttl << 24) | (wscale << 16) | mss4; | ^~ xdp_synproxy.c: In function 'syncookie_open_bpf_maps': xdp_synproxy.c:289:28: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast] 289 | .map_ids = (__u64)map_ids, | ^ Fix it. Fixes: fb5cd0ce ("selftests/bpf: Add selftests for raw syncookie helpers") Signed-off-by:Yang Jihong <yangjihong1@huawei.com> Signed-off-by:
-
Dave Tucker authored
Add documentation for the BPF_MAP_TYPE_ARRAY including kernel version introduced, usage and examples. Also document BPF_MAP_TYPE_PERCPU_ARRAY which is similar. Co-developed-by:
Donald Hunter <donald.hunter@gmail.com> Signed-off-by:
Dave Tucker <dave@dtucker.co.uk> Signed-off-by:
Maryam Tahhan <mtahhan@redhat.com> Reviewed-by:
Bagas Sanjaya <bagasdotme@gmail.com> Link: https://lore.kernel.org/bpf/20221109174604.31673-2-donald.hunter@gmail.com
-
Donald Hunter authored
Add documentation for BPF_MAP_TYPE_QUEUE and BPF_MAP_TYPE_STACK, including usage and examples. Signed-off-by:
-
Donald Hunter authored
Add documentation for the ARRAY_OF_MAPS and HASH_OF_MAPS map types, including usage and examples. Reported-by:
kernel test robot <lkp@intel.com> Signed-off-by:
-
Maryam Tahhan authored
Add documentation for BPF_MAP_TYPE_CPUMAP including kernel version introduced, usage and examples. Co-developed-by:
Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by:
-
Donald Hunter authored
Add documentation for BPF_MAP_TYPE_LPM_TRIE including kernel BPF helper usage, userspace usage and examples. Signed-off-by:
-
Eduard Zingerman authored
A fix for the LLVM compilation error while building bpftool. Replaces the expression: _Static_assert((p) == NULL || ...) by expression: _Static_assert((__builtin_constant_p((p)) ? (p) == NULL : 0) || ...) When "p" is not a constant the former is not considered to be a constant expression by LLVM 14. The error was introduced in the following patch-set: [1]. The error was reported here: [2]. [1] https://lore.kernel.org/bpf/20221109142611.879983-1-eddyz87@gmail.com/ [2] https://lore.kernel.org/all/202211110355.BcGcbZxP-lkp@intel.com/Reported-by:
Eduard Zingerman <eddyz87@gmail.com> Signed-off-by:
-
David S. Miller authored
Jacob Keller says: ==================== ptp: convert remaining users of .adjfreq A handful of drivers remain which still use the .adjfreq interface instead of the newer .adjfine interface. The new interface is preferred as it has a more precise adjustment using scaled parts per million. A handful of the remaining drivers are implemented with a common pattern that can be refactored to use the adjust_by_scaled_ppm and diff_by_scaled_ppm helper functions. These include the ptp_phc, ptp_ixp64x, tg3, hclge, stmac, cpts and bnxt drivers. These are each refactored in a separate change. The remaining drivers, bnx2x, liquidio, cxgb4, fec, and qede implement .adjfreq in a way different from the normal pattern expected by adjust_by_scaled_ppm. Fixing these drivers to properly use .adjfine requires specific knowledge of the hardware implementation. Instead I simply refactor them to use .adjfine and convert scaled_ppm into ppb using the scaled_ppm_to_ppb function. Finally, the .adjfreq implementation interface is removed entirely. This simplifies the interface and ensures that new drivers must implement the new interface as they no longer have an alternative. This still leaves parts per billion used as part of the max_adj interface, and the core PTP stack still converts scaled_ppm to ppb to check this. I plan to investigate fixing this in the future. ==================== Signed-off-by:David S. Miller <davem@davemloft.net>
-
Jacob Keller authored
Now that all drivers have been converted to .adjfine, we can remove the .adjfreq from the interface structure. Signed-off-by:
Jacob Keller <jacob.e.keller@intel.com> Cc: Richard Cochran <richardcochran@gmail.com> Signed-off-by:
-
Jacob Keller authored
Convert all remaining drivers that still use .adjfreq to the newer .adjfine implementation. These drivers are not straightforward, as they use non-standard methods of programming their hardware. They are all converted to use scaled_ppm_to_ppb to get the parts per billion value that their logic depends on. Signed-off-by:
-
Jacob Keller authored
When the BNXT_FW_CAP_PTP_RTC flag is not set, the bnxt driver implements .adjfreq on a cyclecounter in terms of the straightforward "base * ppb / 1 billion" calculation. When BNXT_FW_CAP_PTP_RTC is set, the driver forwards the ppb value to firmware for configuration. Convert the driver to the newer .adjfine interface, updating the cyclecounter calculation to use adjust_by_scaled_ppm to perform the calculation. Use scaled_ppm_to_ppb when forwarding the correction to firmware. Signed-off-by:
Pavan Chebbi <pavan.chebbi@broadcom.com> Signed-off-by:
-
