- 02 Aug, 2024 28 commits
-
-
Jakub Kicinski authored
Lorenzo Bianconi says: ==================== Add second QDMA support for EN7581 eth controller EN7581 SoC supports two independent QDMA controllers to connect the Ethernet Frame Engine (FE) to the CPU. Introduce support for the second QDMA controller. This is a preliminary series to support multiple FE ports (e.g. connected to a second PHY controller). ==================== Link: https://patch.msgid.link/cover.1722522582.git.lorenzo@kernel.orgSigned-off-by:
Jakub Kicinski <kuba@kernel.org>
-
Lorenzo Bianconi authored
Link the running gdm port to the qdma controller used to connect with the CPU. Moreover, load all QDMA controllers available on EN7581 SoC. Signed-off-by:
Lorenzo Bianconi <lorenzo@kernel.org> Link: https://patch.msgid.link/95b515df34ba4727f7ae5b14a1d0462cceec84ff.1722522582.git.lorenzo@kernel.orgSigned-off-by:
-
Lorenzo Bianconi authored
This is a preliminary patch to support multi-QDMA controllers. Signed-off-by:
-
Lorenzo Bianconi authored
Map MMIO regions of both qdma controllers available on EN7581 SoC. Run airoha_hw_cleanup routine for both QDMA controllers available on EN7581 SoC removing airoha_eth module or in airoha_probe error path. This is a preliminary patch to support multi-QDMA controllers. Signed-off-by:
-
Lorenzo Bianconi authored
This is a preliminary patch to support multi-QDMA controllers. Signed-off-by:
-
Lorenzo Bianconi authored
Move airoha_eth pointer in airoha_qdma structure from airoha_tx_irq_queue/airoha_queue ones. This is a preliminary patch to introduce support for multi-QDMA controllers available on EN7581. Signed-off-by:
-
Lorenzo Bianconi authored
QDMA controllers have independent irq lines, so move irqmask in airoha_qdma structure. This is a preliminary patch to support multiple QDMA controllers. Signed-off-by:
-
Lorenzo Bianconi authored
QDMA controllers available in EN7581 SoC have independent tx/rx hw queues so move them in airoha_queues structure. Signed-off-by:
-
Lorenzo Bianconi authored
Introduce airoha_qdma struct and move qdma IO register mapping in airoha_qdma. This is a preliminary patch to enable both QDMA controllers available on EN7581 SoC. Signed-off-by:
-
Simon Horman authored
Build bot reports undefined references to devlink functions. And local testing revealed undefined references to page_pool functions. Based on a patch by Jakub Kicinski <kuba@kernel.org> Fixes: 1a9d4889 ("eth: fbnic: Allocate core device specific structures and devlink interface") Reported-by:
kernel test robot <lkp@intel.com> Closes: https://lore.kernel.org/oe-kbuild-all/202408011219.hiPmwwAs-lkp@intel.com/Signed-off-by:
Simon Horman <horms@kernel.org> Link: https://patch.msgid.link/20240802-fbnic-select-v2-1-41f82a3e0178@kernel.orgSigned-off-by:
-
Jakub Kicinski authored
Print more stack frames and the failing line when check fails. This helps when tests use helpers to do the checks. Before: # At ./ksft/drivers/net/hw/rss_ctx.py line 92: # Check failed 1037698 >= 396893.0 traffic on other queues:[344612, 462380, 233020, 449174, 342298] not ok 8 rss_ctx.test_rss_context_queue_reconfigure After: # Check| At ./ksft/drivers/net/hw/rss_ctx.py, line 387, in test_rss_context_queue_reconfigure: # Check| test_rss_queue_reconfigure(cfg, main_ctx=False) # Check| At ./ksft/drivers/net/hw/rss_ctx.py, line 230, in test_rss_queue_reconfigure: # Check| _send_traffic_check(cfg, port, ctx_ref, { 'target': (0, 3), # Check| At ./ksft/drivers/net/hw/rss_ctx.py, line 92, in _send_traffic_check: # Check| ksft_lt(sum(cnts[i] for i in params['noise']), directed / 2, # Check failed 1045235 >= 405823.5 traffic on other queues (context 1)':[460068, 351995, 565970, 351579, 127270] not ok 8 rss_ctx.test_rss_context_queue_reconfigure Link: https://patch.msgid.link/20240801232317.545577-1-kuba@kernel.orgSigned-off-by: -
Stanislav Fomichev authored
Petr suggested to use errno.EOPNOTSUPP instead of hard-coded 95 in the new test case. Adjust existing ones to match this style. Signed-off-by:
Stanislav Fomichev <sdf@fomichev.me> Link: https://patch.msgid.link/20240802000309.2368-3-sdf@fomichev.meSigned-off-by:
-
Stanislav Fomichev authored
Add new @ksft_disruptive decorator to mark the tests that might be disruptive to the system. Depending on how well the previous test works in the CI we might want to disable disruptive tests by default and only let the developers run them manually. KSFT framework runs disruptive tests by default. DISRUPTIVE=False environment (or config file) can be used to disable these tests. ksft_setup should be called by the test cases that want to use new decorator (ksft_setup is only called via NetDrvEnv/NetDrvEpEnv for now). In the future we can add similar decorators to, for example, avoid running slow tests all the time. And/or have some option to run only 'fast' tests for some sort of smoke test scenario. $ DISRUPTIVE=False ./stats.py KTAP version 1 1..5 ok 1 stats.check_pause ok 2 stats.check_fec ok 3 stats.pkt_byte_sum ok 4 stats.qstat_by_ifindex ok 5 stats.check_down # SKIP marked as disruptive # Totals: pass:4 fail:0 xfail:0 xpass:0 skip:1 error:0 v3: - parse yes and properly treat non-zero nums as true (Petr) v2: - convert from cli argument to env variable (Jakub) Signed-off-by:
-
Stanislav Fomichev authored
Verify that total device stats don't decrease after it has been turned down. Also make sure the device doesn't crash when we access per-queue stats when it's down (in case it tries to access some pointers that are NULL). KTAP version 1 1..5 ok 1 stats.check_pause ok 2 stats.check_fec ok 3 stats.pkt_byte_sum ok 4 stats.qstat_by_ifindex ok 5 stats.check_down # Totals: pass:5 fail:0 xfail:0 xpass:0 skip:0 error:0 v3: - use errno.EOPNOTSUPP (Petr) - move qstat[0] under try (Petr) v2: - KTAP output formatting (Jakub) - defer instead of try/finally (Jakub) - disappearing stats is an error (Jakub) - ksft_ge instead of open coding (Jakub) Signed-off-by:
-
Jakub Kicinski authored
We re-define values of enum netdev_priv_flags as preprocessor macros with the same name. I guess this was done to avoid breaking out of tree modules which may use #ifdef X for kernel compatibility? Commit 7aa98047 ("net: move net_device priv_flags out from UAPI") which added the enum doesn't say. In any case, the flags with defines are quite old now, and defines for new flags don't get added. OOT drivers have to resort to code greps for compat detection, anyway. Let's delete these defines, save LoC, help LXR link to the right place. Reviewed-by:
Jiri Pirko <jiri@nvidia.com> Link: https://patch.msgid.link/20240801163401.378723-1-kuba@kernel.orgSigned-off-by:
-
David S. Miller authored
Radhey Shyam Pandey says: ==================== net: axienet: Fix coding style issues This patchset replace all occurences of (1<<x) by BIT(x) to get rid of checkpatch.pl "CHECK" output "Prefer using the BIT macro". It also removes unnecessary ftrace-like logging, add missing blank line after declaration and remove unnecessary parentheses around 'ndev->mtu <= XAE_JUMBO_MTU' and 'ndev->mtu > XAE_MTU'. Changes for v2: - Split each coding style change into separate patch. ==================== Signed-off-by:David S. Miller <davem@davemloft.net>
-
Radhey Shyam Pandey authored
Remove unnecessary parentheses around 'ndev->mtu <= XAE_JUMBO_MTU' and 'ndev->mtu > XAE_MTU'. Reported by checkpatch. CHECK: Unnecessary parentheses around 'ndev->mtu > XAE_MTU' + if ((ndev->mtu > XAE_MTU) && + (ndev->mtu <= XAE_JUMBO_MTU)) { CHECK: Unnecessary parentheses around 'ndev->mtu <= XAE_JUMBO_MTU' + if ((ndev->mtu > XAE_MTU) && + (ndev->mtu <= XAE_JUMBO_MTU)) { Signed-off-by:Radhey Shyam Pandey <radhey.shyam.pandey@amd.com> Reviewed-by:
Andrew Lunn <andrew@lunn.ch> Signed-off-by:
-
Radhey Shyam Pandey authored
remove unnecessary ftrace-like logging. Also fixes below checkpatch WARNING. WARNING: Unnecessary ftrace-like logging - prefer using ftrace + dev_dbg(&ndev->dev, "%s\n", __func__); Signed-off-by:
-
Radhey Shyam Pandey authored
Add missing blank line after declaration. Fixes below checkpatch warnings. WARNING: Missing a blank line after declarations + struct sockaddr *addr = p; + axienet_set_mac_address(ndev, addr->sa_data); WARNING: Missing a blank line after declarations + struct axienet_local *lp = netdev_priv(ndev); + disable_irq(lp->tx_irq); Signed-off-by:
-
Appana Durga Kedareswara Rao authored
Replace all occurences of (1<<x) by BIT(x) to get rid of checkpatch.pl "CHECK" output "Prefer using the BIT macro". Signed-off-by:
Appana Durga Kedareswara Rao <appana.durga.rao@xilinx.com> Signed-off-by:
-
David S. Miller authored
Luigi Leonardi says: ==================== vsock: avoid queuing on intermediate queue if possible This series introduces an optimization for vsock/virtio to reduce latency and increase the throughput: When the guest sends a packet to the host, and the intermediate queue (send_pkt_queue) is empty, if there is enough space, the packet is put directly in the virtqueue. v3->v4 While running experiments on fio with 64B payload, I realized that there was a mistake in my fio configuration, so I re-ran all the experiments and now the latency numbers are indeed lower with the patch applied. I also noticed that I was kicking the host without the lock. - Fixed a configuration mistake on fio and re-ran all experiments. - Fio latency measurement using 64B payload. - virtio_transport_send_skb_fast_path sends kick with the tx_lock acquired - Addressed all minor style changes requested by maintainer. - Rebased on latest net-next - Link to v3: https://lore.kernel.org/r/20240711-pinna-v3-0-697d4164fe80@outlook.com v2->v3 - Performed more experiments using iperf3 using multiple streams - Handling of reply packets removed from virtio_transport_send_skb, as is needed just by the worker. - Removed atomic_inc/atomic_sub when queuing directly to the vq. - Introduced virtio_transport_send_skb_fast_path that handles the steps for sending on the vq. - Fixed a missing mutex_unlock in error path. - Changed authorship of the second commit - Rebased on latest net-next v1->v2 In this v2 I replaced a mutex_lock with a mutex_trylock because it was insidea RCU critical section. I also added a check on tx_run, so if the module is being removed the packet is not queued. I'd like to thank Stefano for reporting the tx_run issue. Applied all Stefano's suggestions: - Minor code style changes - Minor commit text rewrite Performed more experiments: - Check if all the packets go directly to the vq (Matias' suggestion) - Used iperf3 to see if there is any improvement in overall throughput from guest to host - Pinned the vhost process to a pCPU. - Run fio using 512B payload Rebased on latest net-next ==================== Signed-off-by:
-
Luigi Leonardi authored
Introduce two tests, one for SOCK_STREAM and one for SOCK_SEQPACKET, which use SIOCOUTQ ioctl to check that the number of unsent bytes is zero after delivering a packet. vsock_connect and vsock_accept are no longer static: this is to create more generic tests, allowing code to be reused for SEQPACKET and STREAM. Signed-off-by:
Luigi Leonardi <luigi.leonardi@outlook.com> Reviewed-by:
Stefano Garzarella <sgarzare@redhat.com> Signed-off-by:
-
Luigi Leonardi authored
Introduce support for virtio_transport_unsent_bytes ioctl for virtio_transport, vhost_vsock and vsock_loopback. For all transports the unsent bytes counter is incremented in virtio_transport_get_credit. In virtio_transport (G2H) and in vhost-vsock (H2G) the counter is decremented when the skbuff is consumed. In vsock_loopback the same skbuff is passed from the transmitter to the receiver, so the counter is decremented before queuing the skbuff to the receiver. Signed-off-by:
-
Luigi Leonardi authored
Add support for ioctl(s) in AF_VSOCK. The only ioctl available is SIOCOUTQ/TIOCOUTQ, which returns the number of unsent bytes in the socket. This information is transport-specific and is delegated to them using a callback. Suggested-by:
Daan De Meyer <daan.j.demeyer@gmail.com> Signed-off-by:
-
Rob Herring (Arm) authored
Use of_property_read_bool() to read boolean properties rather than of_find_property(). This is part of a larger effort to remove callers of of_find_property() and similar functions. of_find_property() leaks the DT struct property and data pointers which is a problem for dynamically allocated nodes which may be freed. Signed-off-by:
Rob Herring (Arm) <robh@kernel.org> Reviewed-by:
MD Danish Anwar <danishanwar@ti.com> Link: https://patch.msgid.link/20240731191601.1714639-2-robh@kernel.orgSigned-off-by:
-
Rob Herring (Arm) authored
Replace of_get_property() with the type specific of_property_count_u32_elems() to get the property length. This is part of a larger effort to remove callers of of_get_property() and similar functions. of_get_property() leaks the DT property data pointer which is a problem for dynamically allocated nodes which may be freed. Signed-off-by:
-
Rob Herring (Arm) authored
The check for "leds" and "gpio-controller" both being present is never true because "leds" is a node, not a property. This could be fixed with a check for child node, but there's really no need for the kernel to validate a DT. Just continue ignoring the LEDs if GPIOs are present. Signed-off-by:
-
John Wang authored
When executing ioctl to allocate tags, if the peer address is 0, mctp_alloc_local_tag now replaces it with 0xff. However, during tag dropping, this replacement is not performed, potentially causing the key not to be dropped as expected. Signed-off-by:
John Wang <wangzhiqiang02@ieisystem.com> Reviewed-by:
Jeremy Kerr <jk@codeconstruct.com.au> Link: https://patch.msgid.link/20240730084636.184140-1-wangzhiqiang02@ieisystem.comSigned-off-by:
-
- 01 Aug, 2024 12 commits
-
-
git://git.kernel.org/pub/scm/linux/kernel/git/netdev/netJakub Kicinski authored
Cross-merge networking fixes after downstream PR. No conflicts or adjacent changes. Link: https://patch.msgid.link/20240801131917.34494-1-pabeni@redhat.comSigned-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/netdev/netLinus Torvalds authored
Pull networking fixes from Paolo Abeni: "Including fixes from wireless, bleutooth, BPF and netfilter. Current release - regressions: - core: drop bad gso csum_start and offset in virtio_net_hdr - wifi: mt76: fix null pointer access in mt792x_mac_link_bss_remove - eth: tun: add missing bpf_net_ctx_clear() in do_xdp_generic() - phy: aquantia: only poll GLOBAL_CFG regs on aqr113, aqr113c and aqr115c Current release - new code bugs: - smc: prevent UAF in inet_create() - bluetooth: btmtk: fix kernel crash when entering btmtk_usb_suspend - eth: bnxt: reject unsupported hash functions Previous releases - regressions: - sched: act_ct: take care of padding in struct zones_ht_key - netfilter: fix null-ptr-deref in iptable_nat_table_init(). - tcp: adjust clamping window for applications specifying SO_RCVBUF Previous releases - always broken: - ethtool: rss: small fixes to spec and GET - mptcp: - fix signal endpoint re-add - pm: fix backup support in signal endpoints - wifi: ath12k: fix soft lockup on suspend - eth: bnxt_en: fix RSS logic in __bnxt_reserve_rings() - eth: ice: fix AF_XDP ZC timeout and concurrency issues - eth: mlx5: - fix missing lock on sync reset reload - fix error handling in irq_pool_request_irq" * tag 'net-6.11-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (76 commits) mptcp: fix duplicate data handling mptcp: fix bad RCVPRUNED mib accounting ipv6: fix ndisc_is_useropt() handling for PIO igc: Fix double reset adapter triggered from a single taprio cmd net: MAINTAINERS: Demote Qualcomm IPA to "maintained" net: wan: fsl_qmc_hdlc: Discard received CRC net: wan: fsl_qmc_hdlc: Convert carrier_lock spinlock to a mutex net/mlx5e: Add a check for the return value from mlx5_port_set_eth_ptys net/mlx5e: Fix CT entry update leaks of modify header context net/mlx5e: Require mlx5 tc classifier action support for IPsec prio capability net/mlx5: Fix missing lock on sync reset reload net/mlx5: Lag, don't use the hardcoded value of the first port net/mlx5: DR, Fix 'stack guard page was hit' error in dr_rule net/mlx5: Fix error handling in irq_pool_request_irq net/mlx5: Always drain health in shutdown callback net: Add skbuff.h to MAINTAINERS r8169: don't increment tx_dropped in case of NETDEV_TX_BUSY netfilter: iptables: Fix potential null-ptr-deref in ip6table_nat_table_init(). netfilter: iptables: Fix null-ptr-deref in iptable_nat_table_init(). net: drop bad gso csum_start and offset in virtio_net_hdr ... -
Simon Horman authored
Since commit f946270d ("ethtool: netlink: always pass genl_info to .prepare_data") the info argument of prepare_data callbacks is never NULL. Remove checks present in callback implementations. Link: https://lore.kernel.org/netdev/20240703121237.3f8b9125@kernel.org/Signed-off-by:
-
Yue Haibing authored
Commit f4f943c9 ("RDS: IB: ack more receive completions to improve performance") removed rds_ib_recv_tasklet_fn() implementation but not the declaration. And commit ec16227e ("RDS/IB: Infiniband transport") declared but never implemented other functions. Signed-off-by:
Yue Haibing <yuehaibing@huawei.com> Reviewed-by:
Allison Henderson <allison.henderson@oracle.com> Link: https://patch.msgid.link/20240731063630.3592046-1-yuehaibing@huawei.comSigned-off-by:
-
Daniel Golle authored
Clocks for SerDes and PHY are going to be handled by standalone drivers for each of those hardware components. Drop them from the Ethernet driver. The clocks which are being removed for this patch are responsible for the for the SerDes PCS and PHYs used for the 2nd and 3rd MAC which are anyway not yet supported. Hence backwards compatibility is not an issue. Signed-off-by:
Daniel Golle <daniel@makrotopia.org> Reviewed-by:
AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://patch.msgid.link/b5faaf69b5c6e3e155c64af03706c3c423c6a1c9.1722335682.git.daniel@makrotopia.orgSigned-off-by:
-
Anand Khoje authored
In non FLR context, at times CX-5 requests release of ~8 million FW pages. This needs humongous number of cmd mailboxes, which to be released once the pages are reclaimed. Release of humongous number of cmd mailboxes is consuming cpu time running into many seconds. Which with non preemptible kernels is leading to critical process starving on that cpu’s RQ. On top of it, the FW does not use all the mailbox messages as it has a limit of releasing 50K pages at once per MLX5_CMD_OP_MANAGE_PAGES + MLX5_PAGES_TAKE device command. Hence, the allocation of these many mailboxes is extra and adds unnecessary overhead. To alleviate this, this change restricts the total number of pages a worker will try to reclaim to maximum 50K pages in one go. Our tests have shown significant benefit of this change in terms of time consumed by dma_pool_free(). During a test where an event was raised by HCA to release 1.3 Million pages, following observations were made: - Without this change: Number of mailbox messages allocated was around 20K, to accommodate the DMA addresses of 1.3 million pages. The average time spent by dma_pool_free() to free the DMA pool is between 16 usec to 32 usec. value ------------- Distribution ------------- count 256 | 0 512 |@ 287 1024 |@@@ 1332 2048 |@ 656 4096 |@@@@@ 2599 8192 |@@@@@@@@@@ 4755 16384 |@@@@@@@@@@@@@@@ 7545 32768 |@@@@@ 2501 65536 | 0 - With this change: Number of mailbox messages allocated was around 800; this was to accommodate DMA addresses of only 50K pages. The average time spent by dma_pool_free() to free the DMA pool in this case lies between 1 usec to 2 usec. value ------------- Distribution ------------- count 256 | 0 512 |@@@@@@@@@@@@@@@@@@ 346 1024 |@@@@@@@@@@@@@@@@@@@@@@ 435 2048 | 0 4096 | 0 8192 | 1 16384 | 0 Signed-off-by:Anand Khoje <anand.a.khoje@oracle.com> Reviewed-by:
Leon Romanovsky <leonro@nvidia.com> Reviewed-by:
Zhu Yanjun <yanjun.zhu@linux.dev> Acked-by:
Saeed Mahameed <saeedm@nvidia.com> Link: https://patch.msgid.link/20240730073634.114407-1-anand.a.khoje@oracle.comSigned-off-by:
-
Paolo Abeni authored
Matthieu Baerts says: ==================== mptcp: fix duplicate data handling In some cases, the subflow-level's copied_seq counter was incorrectly increased, leading to an unexpected subflow reset. Patch 1/2 fixes the RCVPRUNED MIB counter that was attached to the wrong event since its introduction in v5.14, backported to v5.11. Patch 2/2 fixes the copied_seq counter issues, is present since v5.10. Signed-off-by:
Matthieu Baerts (NGI0) <matttbe@kernel.org> ==================== Link: https://patch.msgid.link/20240731-upstream-net-20240731-mptcp-dup-data-v1-0-bde833fa628a@kernel.orgSigned-off-by:
Paolo Abeni <pabeni@redhat.com>
-
Paolo Abeni authored
When a subflow receives and discards duplicate data, the mptcp stack assumes that the consumed offset inside the current skb is zero. With multiple subflows receiving data simultaneously such assertion does not held true. As a result the subflow-level copied_seq will be incorrectly increased and later on the same subflow will observe a bad mapping, leading to subflow reset. Address the issue taking into account the skb consumed offset in mptcp_subflow_discard_data(). Fixes: 04e4cd4f ("mptcp: cleanup mptcp_subflow_discard_data()") Cc: stable@vger.kernel.org Link: https://github.com/multipath-tcp/mptcp_net-next/issues/501Signed-off-by:
Mat Martineau <martineau@kernel.org> Signed-off-by:
-
Paolo Abeni authored
Since its introduction, the mentioned MIB accounted for the wrong event: wake-up being skipped as not-needed on some edge condition instead of incoming skb being dropped after landing in the (subflow) receive queue. Move the increment in the correct location. Fixes: ce599c51 ("mptcp: properly account bulk freed memory") Cc: stable@vger.kernel.org Signed-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nfPaolo Abeni authored
Pablo Neira Ayuso says: ==================== Netfilter fixes for net The following patchset contains Netfilter fixes for net: Fix a possible null-ptr-deref sometimes triggered by iptables-restore at boot time. Register iptables {ipv4,ipv6} nat table pernet in first place to fix this issue. Patch #1 and #2 from Kuniyuki Iwashima. netfilter pull request 24-07-31 * tag 'nf-24-07-31' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf: netfilter: iptables: Fix potential null-ptr-deref in ip6table_nat_table_init(). netfilter: iptables: Fix null-ptr-deref in iptable_nat_table_init(). ==================== Link: https://patch.msgid.link/20240731213046.6194-1-pablo@netfilter.orgSigned-off-by: -
Maciej Żenczykowski authored
The current logic only works if the PIO is between two other ND user options. This fixes it so that the PIO can also be either before or after other ND user options (for example the first or last option in the RA). side note: there's actually Android tests verifying a portion of the old broken behaviour, so: https://android-review.googlesource.com/c/kernel/tests/+/3196704 fixes those up. Cc: Jen Linkova <furry@google.com> Cc: Lorenzo Colitti <lorenzo@google.com> Cc: Patrick Rohr <prohr@google.com> Cc: David Ahern <dsahern@kernel.org> Cc: YOSHIFUJI Hideaki / 吉藤英明 <yoshfuji@linux-ipv6.org> Cc: Jakub Kicinski <kuba@kernel.org> Signed-off-by:
Maciej Żenczykowski <maze@google.com> Fixes: 048c796b ("ipv6: adjust ndisc_is_useropt() to also return true for PIO") Link: https://patch.msgid.link/20240730001748.147636-1-maze@google.comSigned-off-by:
-
Breno Leitao authored
This patch modifies the skb_unref function to skip the early return optimization when CONFIG_DEBUG_NET is enabled. The change ensures that the reference count decrement always occurs in debug builds, allowing for more thorough checking of SKB reference counting. Previously, when the SKB's reference count was 1 and CONFIG_DEBUG_NET was not set, the function would return early after a memory barrier (smp_rmb()) without decrementing the reference count. This optimization assumes it's safe to proceed with freeing the SKB without the overhead of an atomic decrement from 1 to 0. With this change: - In non-debug builds (CONFIG_DEBUG_NET not set), behavior remains unchanged, preserving the performance optimization. - In debug builds (CONFIG_DEBUG_NET set), the reference count is always decremented, even when it's 1, allowing for consistent behavior and potentially catching subtle SKB management bugs. This modification enhances debugging capabilities for networking code without impacting performance in production kernels. It helps kernel developers identify and diagnose issues related to SKB management and reference counting in the network stack. Cc: Chris Mason <clm@fb.com> Suggested-by:
Breno Leitao <leitao@debian.org> Reviewed-by:
Eric Dumazet <edumazet@google.com> Link: https://patch.msgid.link/20240729104741.370327-1-leitao@debian.orgSigned-off-by:
-
