skb_ensure_writable should succeed when it's trying to write to the
header of the unreadable skbs, so it doesn't need an unconditional
skb_frags_readable check. The preceding pskb_may_pull() call will
succeed if write_len is within the head and fail if we're trying to
write to the unreadable payload, so we don't need an additional check.
Removing this check restores DSCP functionality with unreadable skbs as
it's called from dscp_tg.
Cc: willemb@google.com
Cc: asml.silence@gmail.com
Fixes: 65249feb6b3d ("net: add support for skbs with unreadable frags")
Signed-off-by: Mina Almasry <almasrymina@google.com>
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://patch.msgid.link/20250615200733.520113-1-almasrymina@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Now that the only remaining caller of __skb_checksum() is skb_checksum(), fold __skb_checksum() into skb_checksum(). This makes struct skb_checksum_ops unnecessary, so remove that too and simply do the "regular" net checksum. It also makes the wrapper functions csum_partial_ext() and csum_block_add_ext() unnecessary, so remove those too and just use the underlying functions. Signed-off-by: Eric Biggers <ebiggers@google.com> Reviewed-by: Hannes Reinecke <hare@suse.de> Link: https://patch.msgid.link/20250519175012.36581-7-ebiggers@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add skb_crc32c(), which calculates the CRC32C of a sk_buff. It will
replace __skb_checksum(), which unnecessarily supports arbitrary
checksums. Compared to __skb_checksum(), skb_crc32c():
- Uses the correct type for CRC32C values (u32, not __wsum).
- Does not require the caller to provide a skb_checksum_ops struct.
- Is faster because it does not use indirect calls and does not use
the very slow crc32c_combine().
According to commit 2817a336d4d5 ("net: skb_checksum: allow custom
update/combine for walking skb") which added __skb_checksum(), the
original motivation for the abstraction layer was to avoid code
duplication for CRC32C and other checksums in the future. However:
- No additional checksums showed up after CRC32C. __skb_checksum()
is only used with the "regular" net checksum and CRC32C.
- Indirect calls are expensive. Commit 2544af0344ba ("net: avoid
indirect calls in L4 checksum calculation") worked around this
using the INDIRECT_CALL_1 macro. But that only avoided the indirect
call for the net checksum, and at the cost of an extra branch.
- The checksums use different types (__wsum and u32), causing casts
to be needed.
- It made the checksums of fragments be combined (rather than
chained) for both checksums, despite this being highly
counterproductive for CRC32C due to how slow crc32c_combine() is.
This can clearly be seen in commit 4c2f24549644 ("sctp: linearize
early if it's not GSO") which tried to work around this performance
bug. With a dedicated function for each checksum, we can instead
just use the proper strategy for each checksum.
As shown by the following tables, the new function skb_crc32c() is
faster than __skb_checksum(), with the improvement varying greatly from
5% to 2500% depending on the case. The largest improvements come from
fragmented packets, mainly due to eliminating the inefficient
crc32c_combine(). But linear packets are improved too, especially
shorter ones, mainly due to eliminating indirect calls. These
benchmarks were done on AMD Zen 5. On that CPU, Linux uses IBRS instead
of retpoline; an even greater improvement might be seen with retpoline:
Linear sk_buffs
Length in bytes __skb_checksum cycles skb_crc32c cycles
=============== ===================== =================
64 43 18
256 94 77
1420 204 161
16384 1735 1642
Nonlinear sk_buffs (even split between head and one fragment)
Length in bytes __skb_checksum cycles skb_crc32c cycles
=============== ===================== =================
64 579 22
256 829 77
1420 1506 194
16384 4365 1682
Signed-off-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Link: https://patch.msgid.link/20250519175012.36581-3-ebiggers@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Augment dmabuf binding to be able to handle TX. Additional to all the RX binding, we also create tx_vec needed for the TX path. Provide API for sendmsg to be able to send dmabufs bound to this device: - Provide a new dmabuf_tx_cmsg which includes the dmabuf to send from. - MSG_ZEROCOPY with SCM_DEVMEM_DMABUF cmsg indicates send from dma-buf. Devmem is uncopyable, so piggyback off the existing MSG_ZEROCOPY implementation, while disabling instances where MSG_ZEROCOPY falls back to copying. We additionally pipe the binding down to the new zerocopy_fill_skb_from_devmem which fills a TX skb with net_iov netmems instead of the traditional page netmems. We also special case skb_frag_dma_map to return the dma-address of these dmabuf net_iovs instead of attempting to map pages. The TX path may release the dmabuf in a context where we cannot wait. This happens when the user unbinds a TX dmabuf while there are still references to its netmems in the TX path. In that case, the netmems will be put_netmem'd from a context where we can't unmap the dmabuf, Resolve this by making __net_devmem_dmabuf_binding_free schedule_work'd. Based on work by Stanislav Fomichev <sdf@fomichev.me>. A lot of the meat of the implementation came from devmem TCP RFC v1[1], which included the TX path, but Stan did all the rebasing on top of netmem/net_iov. Cc: Stanislav Fomichev <sdf@fomichev.me> Signed-off-by: Kaiyuan Zhang <kaiyuanz@google.com> Signed-off-by: Mina Almasry <almasrymina@google.com> Acked-by: Stanislav Fomichev <sdf@fomichev.me> Link: https://patch.msgid.link/20250508004830.4100853-5-almasrymina@google.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Currently net_iovs support only pp ref counts, and do not support a page ref equivalent. This is fine for the RX path as net_iovs are used exclusively with the pp and only pp refcounting is needed there. The TX path however does not use pp ref counts, thus, support for get_page/put_page equivalent is needed for netmem. Support get_netmem/put_netmem. Check the type of the netmem before passing it to page or net_iov specific code to obtain a page ref equivalent. For dmabuf net_iovs, we obtain a ref on the underlying binding. This ensures the entire binding doesn't disappear until all the net_iovs have been put_netmem'ed. We do not need to track the refcount of individual dmabuf net_iovs as we don't allocate/free them from a pool similar to what the buddy allocator does for pages. This code is written to be extensible by other net_iov implementers. get_netmem/put_netmem will check the type of the netmem and route it to the correct helper: pages -> [get|put]_page() dmabuf net_iovs -> net_devmem_[get|put]_net_iov() new net_iovs -> new helpers Signed-off-by: Mina Almasry <almasrymina@google.com> Acked-by: Stanislav Fomichev <sdf@fomichev.me> Link: https://patch.msgid.link/20250508004830.4100853-3-almasrymina@google.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
When sending an skb over a socket using skb_send_sock_locked(), it is currently not possible to specify any flag to be set in msghdr->msg_flags. However, we may want to pass flags the user may have specified, like MSG_NOSIGNAL. Extend __skb_send_sock() with a new argument 'flags' and add a new interface named skb_send_sock_locked_with_flags(). Cc: Eric Dumazet <edumazet@google.com> Cc: Jakub Kicinski <kuba@kernel.org> Cc: Paolo Abeni <pabeni@redhat.com> Cc: Simon Horman <horms@kernel.org> Signed-off-by: Antonio Quartulli <antonio@openvpn.net> Link: https://patch.msgid.link/20250415-b4-ovpn-v26-12-577f6097b964@openvpn.net Reviewed-by: Sabrina Dubroca <sd@queasysnail.net> Tested-by: Oleksandr Natalenko <oleksandr@natalenko.name> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Since we are about to stash some more information into the pp_magic field, let's move the magic signature checks into a pair of helper functions so it can be changed in one place. Reviewed-by: Mina Almasry <almasrymina@google.com> Tested-by: Yonglong Liu <liuyonglong@huawei.com> Acked-by: Jesper Dangaard Brouer <hawk@kernel.org> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com> Link: https://patch.msgid.link/20250409-page-pool-track-dma-v9-1-6a9ef2e0cba8@redhat.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add SOF_TIMESTAMPING_TX_COMPLETION, for requesting a software timestamp when hardware reports a packet completed. Completion tstamp is useful for Bluetooth, as hardware timestamps do not exist in the HCI specification except for ISO packets, and the hardware has a queue where packets may wait. In this case the software SND timestamp only reflects the kernel-side part of the total latency (usually small) and queue length (usually 0 unless HW buffers congested), whereas the completion report time is more informative of the true latency. It may also be useful in other cases where HW TX timestamps cannot be obtained and user wants to estimate an upper bound to when the TX probably happened. Signed-off-by: Pauli Virtanen <pav@iki.fi> Reviewed-by: Willem de Bruijn <willemb@google.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Cross-merge networking fixes after downstream PR (net-6.14-rc5).
Conflicts:
drivers/net/ethernet/cadence/macb_main.c
fa52f15c745c ("net: cadence: macb: Synchronize stats calculations")
75696dd0fd72 ("net: cadence: macb: Convert to get_stats64")
https://lore.kernel.org/20250224125848.68ee63e5@canb.auug.org.au
Adjacent changes:
drivers/net/ethernet/intel/ice/ice_sriov.c
79990cf5e7ad ("ice: Fix deinitializing VF in error path")
a203163274a4 ("ice: simplify VF MSI-X managing")
net/ipv4/tcp.c
18912c520674 ("tcp: devmem: don't write truncated dmabuf CMSGs to userspace")
297d389e9e5b ("net: prefix devmem specific helpers")
net/mptcp/subflow.c
8668860b0ad3 ("mptcp: reset when MPTCP opts are dropped after join")
c3349a22c200 ("mptcp: consolidate subflow cleanup")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add a function to get an array of skbs from the NAPI percpu cache. It's supposed to be a drop-in replacement for kmem_cache_alloc_bulk(skbuff_head_cache, GFP_ATOMIC) and xdp_alloc_skb_bulk(GFP_ATOMIC). The difference (apart from the requirement to call it only from the BH) is that it tries to use as many NAPI cache entries for skbs as possible, and allocate new ones only if needed. The logic is as follows: * there is enough skbs in the cache: decache them and return to the caller; * not enough: try refilling the cache first. If there is now enough skbs, return; * still not enough: try allocating skbs directly to the output array with %GFP_ZERO, maybe we'll be able to get some. If there's now enough, return; * still not enough: return as many as we were able to obtain. Most of times, if called from the NAPI polling loop, the first one will be true, sometimes (rarely) the second one. The third and the fourth -- only under heavy memory pressure. It can save significant amounts of CPU cycles if there are GRO cycles and/or Tx completion cycles (anything that descends to napi_skb_cache_put()) happening on this CPU. Tested-by: Daniel Xu <dxu@dxuuu.xyz> Reviewed-by: Toke Høiland-Jørgensen <toke@redhat.com> Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com>