- Kernel - mailweb.openeuler.org

[PATCH OLK-6.6] tcp: challenge ACK for non-exact RST in SYN-RECEIVED
by JiangJieHua 28 Aug '26

28 Aug '26

From: Yuxiang Yang <yangyx22(a)mails.tsinghua.edu.cn> stable inclusion from stable-v6.6.153 commit 8b0a3a094f4cae2fb92e4d08d4eef7246a9d9c49 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/16927 CVE: CVE-2026-68118 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit a28c4fcbf774e23b4779cae468e3497a5ad1f4a1 ] The SYN-RECEIVED request-socket path in tcp_check_req() accepts an in-window RST without requiring SEG.SEQ to exactly match RCV.NXT. A non-exact RST therefore removes the request instead of eliciting a challenge ACK. RFC 9293 section 3.10.7.4 applies the RFC 5961 reset check in SYN-RECEIVED: an exact RST resets the connection, while a non-exact in-window RST must trigger a challenge ACK and be dropped. Apply that check before the ACK-field validation, following the RFC sequence-number, RST, then ACK processing order. Factor the per-netns challenge ACK quota out of tcp_send_challenge_ack() so request sockets can share it. Use the request socket's send_ack() callback and its own out-of-window ACK timestamp to send and rate-limit the response. Reported-by: Yuxiang Yang <yangyx22(a)mails.tsinghua.edu.cn> Reported-by: Yizhou Zhao <zhaoyz24(a)mails.tsinghua.edu.cn> Reported-by: Ao Wang <wangao(a)seu.edu.cn> Reported-by: Xuewei Feng <fengxw06(a)126.com> Reported-by: Qi Li <qli01(a)tsinghua.edu.cn> Reported-by: Ke Xu <xuke(a)tsinghua.edu.cn> Fixes: 282f23c6ee34 ("tcp: implement RFC 5961 3.2") Cc: stable(a)vger.kernel.org Signed-off-by: Yuxiang Yang <yangyx22(a)mails.tsinghua.edu.cn> Reviewed-by: Eric Dumazet <edumazet(a)google.com> Link: https://patch.msgid.link/20260717081443.809393-2-yangyx22@mails.tsinghua.ed… Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Signed-off-by: JiangJieHua <jiangjiehua1(a)huawei.com> --- include/net/tcp.h | 2 ++ net/ipv4/tcp_input.c | 56 ++++++++++++++++++++++++++++++---------- net/ipv4/tcp_minisocks.c | 12 ++++++++- 3 files changed, 56 insertions(+), 14 deletions(-) diff --git a/include/net/tcp.h b/include/net/tcp.h index 8674996bd39c..a40864c5f280 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -1685,6 +1685,8 @@ static inline bool tcp_paws_reject(const struct tcp_options_received *rx_opt, bool tcp_oow_rate_limited(struct net *net, const struct sk_buff *skb, int mib_idx, u32 *last_oow_ack_time); +void tcp_reqsk_send_challenge_ack(struct sock *sk, struct sk_buff *skb, + struct request_sock *req); static inline void tcp_mib_init(struct net *net) { diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index e6499904caf5..6af54c18f55c 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -3744,24 +3744,17 @@ bool tcp_oow_rate_limited(struct net *net, const struct sk_buff *skb, return __tcp_oow_rate_limited(net, mib_idx, last_oow_ack_time); } -/* RFC 5961 7 [ACK Throttling] */ -static void tcp_send_challenge_ack(struct sock *sk) +/* Consume one slot from the per-netns RFC 5961 challenge ACK quota. + * Returns true if a challenge ACK may be sent. + */ +static bool tcp_challenge_ack_allowed(struct net *net) { - struct tcp_sock *tp = tcp_sk(sk); - struct net *net = sock_net(sk); u32 count, now, ack_limit; - /* First check our per-socket dupack rate limit. */ - if (__tcp_oow_rate_limited(net, - LINUX_MIB_TCPACKSKIPPEDCHALLENGE, - &tp->last_oow_ack_time)) - return; - ack_limit = READ_ONCE(net->ipv4.sysctl_tcp_challenge_ack_limit); if (ack_limit == INT_MAX) - goto send_ack; + return true; - /* Then check host-wide RFC 5961 rate limit. */ now = jiffies / HZ; if (now != READ_ONCE(net->ipv4.tcp_challenge_timestamp)) { u32 half = (ack_limit + 1) >> 1; @@ -3773,12 +3766,49 @@ static void tcp_send_challenge_ack(struct sock *sk) count = READ_ONCE(net->ipv4.tcp_challenge_count); if (count > 0) { WRITE_ONCE(net->ipv4.tcp_challenge_count, count - 1); -send_ack: + return true; + } + return false; +} + +/* RFC 5961 7 [ACK Throttling] */ +static void tcp_send_challenge_ack(struct sock *sk) +{ + struct tcp_sock *tp = tcp_sk(sk); + struct net *net = sock_net(sk); + + /* First check our per-socket dupack rate limit. */ + if (__tcp_oow_rate_limited(net, + LINUX_MIB_TCPACKSKIPPEDCHALLENGE, + &tp->last_oow_ack_time)) + return; + + /* Then check the per-netns RFC 5961 rate limit. */ + if (tcp_challenge_ack_allowed(net)) { NET_INC_STATS(net, LINUX_MIB_TCPCHALLENGEACK); tcp_send_ack(sk); } } +/* Send a challenge ACK from a SYN-RECEIVED request socket. Uses + * __tcp_oow_rate_limited() directly so that an RST carrying payload + * cannot bypass the per-request rate limit. + */ +void tcp_reqsk_send_challenge_ack(struct sock *sk, struct sk_buff *skb, + struct request_sock *req) +{ + struct net *net = sock_net(sk); + + if (__tcp_oow_rate_limited(net, LINUX_MIB_TCPACKSKIPPEDCHALLENGE, + &tcp_rsk(req)->last_oow_ack_time)) + return; + + if (tcp_challenge_ack_allowed(net)) { + NET_INC_STATS(net, LINUX_MIB_TCPCHALLENGEACK); + req->rsk_ops->send_ack(sk, skb, req); + } +} + static void tcp_store_ts_recent(struct tcp_sock *tp) { tp->rx_opt.ts_recent = tp->rx_opt.rcv_tsval; diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index 8099563937e3..6138f546e90d 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -763,7 +763,7 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb, * elsewhere and is checked directly against the child socket rather * than req because user data may have been sent out. */ - if ((flg & TCP_FLAG_ACK) && !fastopen && + if ((flg & TCP_FLAG_ACK) && !(flg & TCP_FLAG_RST) && !fastopen && (TCP_SKB_CB(skb)->ack_seq != tcp_rsk(req)->snt_isn + 1)) return sk; @@ -802,6 +802,16 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb, flg &= ~TCP_FLAG_SYN; } + /* RFC 5961 section 3.2, as clarified by RFC 9293 section + * 3.10.7.4, requires a challenge ACK for a non-exact + * in-window RST in SYN-RECEIVED. + */ + if ((flg & TCP_FLAG_RST) && + TCP_SKB_CB(skb)->seq != tcp_rsk(req)->rcv_nxt) { + tcp_reqsk_send_challenge_ack(sk, skb, req); + return NULL; + } + /* RFC793: "second check the RST bit" and * "fourth, check the SYN bit" */ -- 2.33.8

2 1

[PATCH OLK-6.6] nvmet-tcp: fix page fragment cache leak in error path
by Lai Zewei 28 Aug '26

28 Aug '26

mainline inclusion from mainline-v7.2-rc1 commit 4dae393956093c807212918fd91a8fc70df15338 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/17264 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- In nvmet_tcp_alloc_queue(), when a connection is closed during the allocation process (e.g., nvmet_tcp_set_queue_sock() returns -ENOTCONN), the error handling jumps to out_destroy_sq and then to out_ida_remove without draining the page fragment cache. Although nvmet_tcp_free_cmd() is called in some error paths to release individual page fragments, the underlying page cache reference held by queue->pf_cache is never released. The first allocation using pf_cache is the call to nvmet_tcp_alloc_cmd() for queue->connect, which happens after ida_alloc() returns successfully. This results in a page leak each time a connection fails during allocation, which could lead to memory exhaustion over time if connections are repeatedly opened and closed. Fix this by calling page_frag_cache_drain() before freeing the queue structure in the out_ida_remove label. Fixes: 872d26a391da ("nvmet-tcp: add NVMe over TCP target driver") Reviewed-by: Christoph Hellwig <hch(a)lst.de> Signed-off-by: Geliang Tang <tanggeliang(a)kylinos.cn> Signed-off-by: Keith Busch <kbusch(a)kernel.org> Conflicts: drivers/nvme/target/tcp.c [ctx conflicts] Signed-off-by: Lai Zewei <laizewei3(a)huawei.com> --- drivers/nvme/target/tcp.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/drivers/nvme/target/tcp.c b/drivers/nvme/target/tcp.c index 6c6197a53fa3..fb6af1be694f 100644 --- a/drivers/nvme/target/tcp.c +++ b/drivers/nvme/target/tcp.c @@ -1754,6 +1754,14 @@ static int nvmet_tcp_alloc_queue(struct nvmet_tcp_port *port, nvmet_tcp_free_cmd(&queue->connect); out_ida_remove: ida_free(&nvmet_tcp_queue_ida, queue->idx); + /* + * Drain the page fragment cache if any allocations were done. + * The first allocation using pf_cache is nvmet_tcp_alloc_cmd() + * for queue->connect after ida_alloc(). + */ + if (queue->pf_cache.va) + __page_frag_cache_drain(virt_to_head_page(queue->pf_cache.va), + queue->pf_cache.pagecnt_bias); out_free_queue: kfree(queue); return ret; -- 2.52.0

2 1

[PATCH openEuler-1.0-LTS] RDMA/mlx5: Release the HW‑provided UAR index rather than the SW one
by Chen Jinghuang 28 Aug '26

28 Aug '26

From: Leon Romanovsky <leonro(a)nvidia.com> mainline inclusion from mainline-v7.2-rc1 commit 449ae7927152e46acbe5f19f97eafdae6d3a96b1 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/18162 CVE: CVE-2026-74296 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- Free the UAR index returned by the hardware. Fixes: 4ed131d0bb15 ("IB/mlx5: Expose dynamic mmap allocation") Link: https://patch.msgid.link/r/20260611-fix-uar-release-v1-1-f5464d845dbf@nvidi… Signed-off-by: Leon Romanovsky <leonro(a)nvidia.com> Signed-off-by: Jason Gunthorpe <jgg(a)nvidia.com> Conflicts: drivers/infiniband/hw/mlx5/main.c [ Context conflict ] Signed-off-by: Chen jinghuang <chenjinghuang2(a)huawei.com> --- drivers/infiniband/hw/mlx5/main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index 4f340d6db582..051de9f3ed82 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -2148,7 +2148,7 @@ static int uar_mmap(struct mlx5_ib_dev *dev, enum mlx5_ib_mmap_cmd cmd, if (!dyn_uar) return err; - mlx5_cmd_free_uar(dev->mdev, idx); + mlx5_cmd_free_uar(dev->mdev, uar_index); free_bfreg: mlx5_ib_free_bfreg(dev, bfregi, bfreg_dyn_idx); -- 2.34.1

2 1

[PATCH OLK-5.10] iommu/amd: Wait for completion instead of returning early in iommu_completion_wait()
by Zhang Yuwei 28 Aug '26

28 Aug '26

From: Guanghui Feng <guanghuifeng(a)linux.alibaba.com> stable inclusion from stable-v6.6.148 commit ab7faf5a172ebfdc423ebb3eea4d472740de82f9 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/18421 CVE: CVE-2026-68329 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit 1e75a8255f11c81fb07e81e5029cfd75804350a0 ] need_sync is a per-IOMMU flag shared by all domains and devices behind that IOMMU. It is set whenever a command is queued with sync == true and cleared when a completion-wait (CWAIT) command is queued. However, a cleared need_sync only means that a covering CWAIT has been queued, not that all previously queued commands have actually completed in hardware. iommu_completion_wait() read need_sync locklessly and returned early when it was false. This breaks the "block until all previously queued commands have completed" contract in a multi-CPU scenario: CPU2: queue inv-B => need_sync = true CPU1: queue CWAIT(N); need_sync = false; then wait_on_sem(N) CPU2: read need_sync == false => return 0 (no wait!) CPU2 returns without waiting for any sequence number even though its inv-B may not have completed yet (CWAIT(N), queued after inv-B, has not been signaled). CPU2 then proceeds to, for example, free page-table pages while the IOMMU can still walk stale translations, opening a use-after-free window. This is a logical race in the meaning of the flag, not a memory-visibility issue, so barriers alone do not help. Fix it without losing the optimization of avoiding redundant CWAIT commands: take iommu->lock before testing need_sync, and when it is false do not return early but wait for the last allocated sequence number (cmd_sem_val). Since need_sync == false implies no sync command was queued after the last CWAIT, that CWAIT is FIFO-ordered after every not-yet-completed command, so waiting for its sequence number guarantees all prior commands (possibly queued by another CPU) have completed. The common path with pending work is unchanged and no extra hardware command is issued. Signed-off-by: Guanghui Feng <guanghuifeng(a)linux.alibaba.com> Fixes: 815b33fdc279 ("x86/amd-iommu: Cleanup completion-wait handling") Reviewed-by: Vasant Hegde <vasant.hegde(a)amd.com> Signed-off-by: Will Deacon <will(a)kernel.org> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Hulk Robot <hulkrobot(a)huawei.com> Conflicts: drivers/iommu/amd/iommu.c [context conflict] Signed-off-by: Zhang Yuwei <zhangyuwei20(a)huawei.com> --- drivers/iommu/amd/iommu.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c index ba17dd2b8e6f..4a8dc1326006 100644 --- a/drivers/iommu/amd/iommu.c +++ b/drivers/iommu/amd/iommu.c @@ -1140,11 +1140,23 @@ static int iommu_completion_wait(struct amd_iommu *iommu) int ret; u64 data; - if (!iommu->need_sync) - return 0; - raw_spin_lock_irqsave(&iommu->lock, flags); + if (!iommu->need_sync) { + /* + * No command has been queued since the last completion-wait. + * A concurrent CPU may have already queued that CWAIT and + * cleared need_sync; need_sync == false only means a covering + * CWAIT is queued, not that all prior commands have completed. + * Wait for the last allocated sequence number so that any + * command queued before this call (possibly on another CPU) + * is guaranteed to have completed before returning. + */ + data = iommu->cmd_sem_val; + raw_spin_unlock_irqrestore(&iommu->lock, flags); + return wait_on_sem(iommu, data); + } + data = ++iommu->cmd_sem_val; build_completion_wait(&cmd, iommu, data); @@ -1154,9 +1166,7 @@ static int iommu_completion_wait(struct amd_iommu *iommu) if (ret) return ret; - ret = wait_on_sem(iommu, data); - - return ret; + return wait_on_sem(iommu, data); } static int iommu_flush_dte(struct amd_iommu *iommu, u16 devid) -- 2.22.0

2 1

[PATCH OLK-5.10] ceph: fix hanging __ceph_get_caps() with stale mds_wanted
by Tengda Wu 28 Aug '26

28 Aug '26

From: Max Kellermann <max.kellermann(a)ionos.com> stable inclusion from stable-v5.10.266 commit b5661524c5a45085a866864ca9b8ae2513dfd67a category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/18404 CVE: CVE-2026-80527 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit 50958bb928bad3bdba9e5d1b7ff4bbadcf6951e6 ] A reader can hang forever in __ceph_get_caps() when the client no longer holds `FILE_RD`, but local cap state still says that the capability is already wanted (via `mds_wanted`). One way to trigger this is through MDS cap revocation. If another client performs a conflicting operation, the MDS can revoke `FILE_RD` from the reader; the next read then has to reacquire `FILE_RD`. If the cap update that should request `FILE_RD` never reaches the MDS after `cap->mds_wanted` was raised, the reader is left holding only non-file caps while local `mds_wanted` still includes the file read caps. In that state, try_get_cap_refs() sees `need <= mds_wanted` and returns 0, so __ceph_get_caps() just waits on `i_cap_wq`. If the cap update that was supposed to request `FILE_RD never reaches the MDS after `cap->mds_wanted was` raised, no further request is sent and the waiter can sleep indefinitely until unrelated cap traffic happens to wake it up. The ordering issue is that `cap->mds_wanted` is updated in __prep_cap() before the `CEPH_MSG_CLIENT_CAPS message` is actually queued for send. That makes one field serve two different meanings at once: what this client wants, and what the client believes the MDS already knows it wants. A proper fix would be to split those states and track whether a cap update is actually in flight or has been observed by the MDS. However, simply moving the `cap->mds_wanted assignment` later would not be sufficient: queueing the message in the messenger does not guarantee that the MDS processed that specific wanted set, and reconnect or message loss can still invalidate that assumption. Fixing that properly would require a larger rework of the cap state machine. To allow simpler backports to stable kernels, this patch implements a simpler workaround: - stop waiting forever in __ceph_get_caps(); after a bounded wait, fall back to the renew path - make ceph_renew_caps() issue a synchronous `OPEN` request whenever the inode still does not actually hold the wanted caps, instead of only calling ceph_check_caps() The extra issued-vs-wanted check in ceph_renew_caps() is necessary because the previous test only checked whether the inode still had any real caps at all. That is not enough after revocation: the client can still hold something like `pLs` and yet be missing `FILE_RD` completely. In that case, falling back to ceph_check_caps() is not sufficient, because it still trusts `cap->mds_wanted` and may resend nothing. By requiring `(issued & wanted) == wanted` before taking the asynchronous path, the code only uses ceph_check_caps() when the `wanted caps` are already actually issued. Otherwise, it sends the synchronous `OPEN` renew. This preserves the existing asynchronous fast path when the wanted caps are already issued, avoids changing cap-state semantics, and fixes the hang by guaranteeing that a stalled waiter eventually retries through a path that does not rely on the stale `mds_wanted` state. [ idryomov: move CEPH_GET_CAPS_WAIT_TIMEOUT from libceph.h to mds_client.h, formatting ] Cc: stable(a)vger.kernel.org Fixes: 0a454bdd501a ("ceph: reorganize __send_cap for less spinlock abuse") Signed-off-by: Max Kellermann <max.kellermann(a)ionos.com> Reviewed-by: Alex Markuze <amarkuze(a)redhat.com> Signed-off-by: Ilya Dryomov <idryomov(a)gmail.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Signed-off-by: Tengda Wu <wutengda2(a)huawei.com> --- fs/ceph/caps.c | 17 +++++++++++++++-- fs/ceph/file.c | 9 +++++---- fs/ceph/mds_client.h | 2 ++ 3 files changed, 22 insertions(+), 6 deletions(-) diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 31e5d87ec945..9bebdafd5ec9 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -2995,7 +2995,19 @@ int ceph_get_caps(struct file *filp, int need, int want, ret = -ERESTARTSYS; break; } - wait_woken(&wait, TASK_INTERRUPTIBLE, MAX_SCHEDULE_TIMEOUT); + + /* + * If a cap update is lost after + * mds_wanted was raised, waiting + * forever will never make progress. + * Retry the renew path periodically + * so we can resend synchronously. + */ + if (!wait_woken(&wait, TASK_INTERRUPTIBLE, + CEPH_GET_CAPS_WAIT_TIMEOUT)) { + ret = -ESTALE; + break; + } } remove_wait_queue(&ci->i_cap_wq, &wait); @@ -3027,7 +3039,8 @@ int ceph_get_caps(struct file *filp, int need, int want, continue; } if (ret == -ESTALE) { - /* session was killed, try renew caps */ + /* session was killed or a waited cap + * request needs a retry */ ret = ceph_renew_caps(inode, flags); if (ret == 0) continue; diff --git a/fs/ceph/file.c b/fs/ceph/file.c index 215bd9a3c535..f20d5f708009 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -277,21 +277,22 @@ static int ceph_init_file(struct inode *inode, struct file *file, int fmode) } /* - * try renew caps after session gets killed. + * Retry cap acquisition after a stale session or a lost cap update. */ int ceph_renew_caps(struct inode *inode, int fmode) { struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb); struct ceph_inode_info *ci = ceph_inode(inode); struct ceph_mds_request *req; - int err, flags, wanted; + int err, flags, wanted, issued; spin_lock(&ci->i_ceph_lock); __ceph_touch_fmode(ci, mdsc, fmode); wanted = __ceph_caps_file_wanted(ci); + issued = __ceph_caps_issued(ci, NULL); if (__ceph_is_any_real_caps(ci) && - (!(wanted & CEPH_CAP_ANY_WR) || ci->i_auth_cap)) { - int issued = __ceph_caps_issued(ci, NULL); + (!(wanted & CEPH_CAP_ANY_WR) || ci->i_auth_cap) && + (issued & wanted) == wanted) { spin_unlock(&ci->i_ceph_lock); dout("renew caps %p want %s issued %s updating mds_wanted\n", inode, ceph_cap_string(wanted), ceph_cap_string(issued)); diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h index 1c958510f00f..b1939f916228 100644 --- a/fs/ceph/mds_client.h +++ b/fs/ceph/mds_client.h @@ -61,6 +61,8 @@ enum ceph_feature_type { struct ceph_fs_client; struct ceph_cap; +#define CEPH_GET_CAPS_WAIT_TIMEOUT (5 * HZ) + /* * parsed info about a single inode. pointers are into the encoded * on-wire structures within the mds reply message payload. -- 2.34.1

2 1

[PATCH OLK-6.6 0/5] CVE-2026-74314
by Pu Lehui 28 Aug '26

28 Aug '26

Benjamin Tissoires (1): bpf: Do not walk twice the hash map on free Hou Tao (1): bpf: Factor out htab_elem_value helper() Justin Suess (1): bpf: Cancel special fields on map value recycle Mykyta Yatsenko (2): bpf: htab: extract helper for freeing special structs bpf: Extract internal structs validation logic into helpers include/linux/bpf.h | 8 ++ kernel/bpf/arraymap.c | 12 +- kernel/bpf/hashtab.c | 111 +++++++++--------- kernel/bpf/helpers.c | 6 + kernel/bpf/syscall.c | 5 + .../selftests/bpf/prog_tests/linked_list.c | 33 +++--- .../selftests/bpf/prog_tests/map_kptr.c | 10 +- .../testing/selftests/bpf/progs/linked_list.c | 51 ++++++++ 8 files changed, 152 insertions(+), 84 deletions(-) -- 2.34.1

2 6

[PATCH OLK-5.10] iommu/amd: Wait for completion instead of returning early in iommu_completion_wait()
by Zhang Yuwei 28 Aug '26

28 Aug '26

From: Guanghui Feng <guanghuifeng(a)linux.alibaba.com> stable inclusion from stable-v6.6.148 commit ab7faf5a172ebfdc423ebb3eea4d472740de82f9 category: bugfix bugzilla: NA Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit 1e75a8255f11c81fb07e81e5029cfd75804350a0 ] need_sync is a per-IOMMU flag shared by all domains and devices behind that IOMMU. It is set whenever a command is queued with sync == true and cleared when a completion-wait (CWAIT) command is queued. However, a cleared need_sync only means that a covering CWAIT has been queued, not that all previously queued commands have actually completed in hardware. iommu_completion_wait() read need_sync locklessly and returned early when it was false. This breaks the "block until all previously queued commands have completed" contract in a multi-CPU scenario: CPU2: queue inv-B => need_sync = true CPU1: queue CWAIT(N); need_sync = false; then wait_on_sem(N) CPU2: read need_sync == false => return 0 (no wait!) CPU2 returns without waiting for any sequence number even though its inv-B may not have completed yet (CWAIT(N), queued after inv-B, has not been signaled). CPU2 then proceeds to, for example, free page-table pages while the IOMMU can still walk stale translations, opening a use-after-free window. This is a logical race in the meaning of the flag, not a memory-visibility issue, so barriers alone do not help. Fix it without losing the optimization of avoiding redundant CWAIT commands: take iommu->lock before testing need_sync, and when it is false do not return early but wait for the last allocated sequence number (cmd_sem_val). Since need_sync == false implies no sync command was queued after the last CWAIT, that CWAIT is FIFO-ordered after every not-yet-completed command, so waiting for its sequence number guarantees all prior commands (possibly queued by another CPU) have completed. The common path with pending work is unchanged and no extra hardware command is issued. Signed-off-by: Guanghui Feng <guanghuifeng(a)linux.alibaba.com> Fixes: 815b33fdc279 ("x86/amd-iommu: Cleanup completion-wait handling") Reviewed-by: Vasant Hegde <vasant.hegde(a)amd.com> Signed-off-by: Will Deacon <will(a)kernel.org> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Hulk Robot <hulkrobot(a)huawei.com> Conflicts: drivers/iommu/amd/iommu.c [context conflict] Signed-off-by: Zhang Yuwei <zhangyuwei20(a)huawei.com> --- drivers/iommu/amd/iommu.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c index ba17dd2b8e6f..4a8dc1326006 100644 --- a/drivers/iommu/amd/iommu.c +++ b/drivers/iommu/amd/iommu.c @@ -1140,11 +1140,23 @@ static int iommu_completion_wait(struct amd_iommu *iommu) int ret; u64 data; - if (!iommu->need_sync) - return 0; - raw_spin_lock_irqsave(&iommu->lock, flags); + if (!iommu->need_sync) { + /* + * No command has been queued since the last completion-wait. + * A concurrent CPU may have already queued that CWAIT and + * cleared need_sync; need_sync == false only means a covering + * CWAIT is queued, not that all prior commands have completed. + * Wait for the last allocated sequence number so that any + * command queued before this call (possibly on another CPU) + * is guaranteed to have completed before returning. + */ + data = iommu->cmd_sem_val; + raw_spin_unlock_irqrestore(&iommu->lock, flags); + return wait_on_sem(iommu, data); + } + data = ++iommu->cmd_sem_val; build_completion_wait(&cmd, iommu, data); @@ -1154,9 +1166,7 @@ static int iommu_completion_wait(struct amd_iommu *iommu) if (ret) return ret; - ret = wait_on_sem(iommu, data); - - return ret; + return wait_on_sem(iommu, data); } static int iommu_flush_dte(struct amd_iommu *iommu, u16 devid) -- 2.22.0

2 1

[PATCH OLK-5.10] watchdog: pretimeout: Fix UAF in watchdog_unregister_governor()
by Lin Ruifeng 28 Aug '26

28 Aug '26

From: Tzung-Bi Shih <tzungbi(a)kernel.org> stable inclusion from stable-v6.6.148 commit 2e47b91b9b4020fcc01def14d6b6556d66074cf4 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/17075 CVE: CVE-2026-68357 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- [ Upstream commit 7362ba0f9c96ac3ad6a2ca3995bd9fc9a28a8661 ] When a watchdog governor is unregistered, it updates existing watchdog devices that were using this governor by falling back to `default_gov`. If the governor being unregistered is currently set as `default_gov`, the `default_gov` is never cleared. This leads to 2 use-after-free issues: 1. New watchdog devices registered after this point will inherit the dangling `default_gov`. 2. Existing watchdog devices using the unregistered governor will have their `wdd->gov` reassigned to the dangling `default_gov`. Fix the UAF by clearing `default_gov` if it matches the governor being unregistered. Fixes: da0d12ff2b82 ("watchdog: pretimeout: add panic pretimeout governor") Signed-off-by: Tzung-Bi Shih <tzungbi(a)kernel.org> Link: https://lore.kernel.org/r/20260707101803.3598173-1-tzungbi@kernel.org Signed-off-by: Guenter Roeck <linux(a)roeck-us.net> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Lin Ruifeng <linruifeng4(a)huawei.com> --- drivers/watchdog/watchdog_pretimeout.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/watchdog/watchdog_pretimeout.c b/drivers/watchdog/watchdog_pretimeout.c index 01ca84be240f..fe4c8ace9a42 100644 --- a/drivers/watchdog/watchdog_pretimeout.c +++ b/drivers/watchdog/watchdog_pretimeout.c @@ -164,6 +164,8 @@ void watchdog_unregister_governor(struct watchdog_governor *gov) } spin_lock_irq(&pretimeout_lock); + if (default_gov == gov) + default_gov = NULL; list_for_each_entry(p, &pretimeout_list, entry) if (p->wdd->gov == gov) p->wdd->gov = default_gov; -- 2.34.1

2 1

[PATCH OLK-6.6] RDMA/nldev: Fix locking when accessing mr->pd
by Chen Jinghuang 27 Aug '26

27 Aug '26

From: Jason Gunthorpe <jgg(a)nvidia.com> mainline inclusion from mainline-v7.2-rc1 commit 50d5c02ab8e62325548bd3a6e6b758a9dcd6e7c3 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/17658 CVE: CVE-2026-74334 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- Sashiko points out that, due to rereg_mr, the PD is actually variable and all the touches in nldev are racy. Use mr->device instead of mr->pd->device. Getting the PD restrack ID is more tricky. To avoid disturbing all the happy paths, add an rdma_restrack_sync() operation which is sort of like flush_workqueue() or synchronize_irq(): after it returns, all the old nldev touches to the mr are gone and everything sees the new PD. This makes it safe to reach into the PD pointer. Fixes: da5c85078215 ("RDMA/nldev: add driver-specific resource tracking") Link: https://patch.msgid.link/r/4-v1-29ebd2c229b5+fd5-ib_mr_pd_jgg@nvidia.com Signed-off-by: Jason Gunthorpe <jgg(a)nvidia.com> Conflicts: drivers/infiniband/core/restrack.c drivers/infiniband/core/uverbs_cmd.c include/rdma/ib_verbs.h [In restrack.c, rdma_restrack_count() in target tree lacks show_details parameter and RESTRACK_DD mark logic; only added xa_is_zero() check required by rdma_restrack_sync(). Removed RESTRACK_DD xa_get_mark() check in rdma_restrack_sync() as target tree does not have driver-specific resource tracking (RESTRACK_DD mark). In uverbs_cmd.c, kept downstream #include "uverbs_profile.h" alongside new #include "restrack.h".] Signed-off-by: Chen Jinghuang <chenjinghuang2(a)huawei.com> --- drivers/infiniband/core/nldev.c | 15 ++++---- drivers/infiniband/core/restrack.c | 52 +++++++++++++++++++++++++++- drivers/infiniband/core/restrack.h | 1 + drivers/infiniband/core/uverbs_cmd.c | 10 ++++-- include/rdma/ib_verbs.h | 5 +++ 5 files changed, 74 insertions(+), 9 deletions(-) diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c index 9d6627bde61a..6c4008a1e18f 100644 --- a/drivers/infiniband/core/nldev.c +++ b/drivers/infiniband/core/nldev.c @@ -644,7 +644,7 @@ static int fill_res_mr_entry(struct sk_buff *msg, bool has_cap_net_admin, struct rdma_restrack_entry *res, uint32_t port) { struct ib_mr *mr = container_of(res, struct ib_mr, res); - struct ib_device *dev = mr->pd->device; + struct ib_device *dev = mr->device; if (has_cap_net_admin) { if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_RKEY, mr->rkey)) @@ -660,9 +660,12 @@ static int fill_res_mr_entry(struct sk_buff *msg, bool has_cap_net_admin, if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_MRN, res->id)) return -EMSGSIZE; - if (!rdma_is_kernel_res(res) && - nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_PDN, mr->pd->res.id)) - return -EMSGSIZE; + if (!rdma_is_kernel_res(res)) { + struct ib_pd *pd = READ_ONCE(mr->pd); + + if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_PDN, pd->res.id)) + return -EMSGSIZE; + } if (fill_res_name_pid(msg, res)) return -EMSGSIZE; @@ -676,7 +679,7 @@ static int fill_res_mr_raw_entry(struct sk_buff *msg, bool has_cap_net_admin, struct rdma_restrack_entry *res, uint32_t port) { struct ib_mr *mr = container_of(res, struct ib_mr, res); - struct ib_device *dev = mr->pd->device; + struct ib_device *dev = mr->device; if (!dev->ops.fill_res_mr_entry_raw) return -EINVAL; @@ -966,7 +969,7 @@ static int fill_stat_mr_entry(struct sk_buff *msg, bool has_cap_net_admin, struct rdma_restrack_entry *res, uint32_t port) { struct ib_mr *mr = container_of(res, struct ib_mr, res); - struct ib_device *dev = mr->pd->device; + struct ib_device *dev = mr->device; if (nla_put_u32(msg, RDMA_NLDEV_ATTR_RES_MRN, res->id)) goto err; diff --git a/drivers/infiniband/core/restrack.c b/drivers/infiniband/core/restrack.c index 438ed3588175..19a813e75486 100644 --- a/drivers/infiniband/core/restrack.c +++ b/drivers/infiniband/core/restrack.c @@ -68,8 +68,11 @@ int rdma_restrack_count(struct ib_device *dev, enum rdma_restrack_type type) u32 cnt = 0; xa_lock(&rt->xa); - xas_for_each(&xas, e, U32_MAX) + xas_for_each(&xas, e, U32_MAX) { + if (xa_is_zero(e)) + continue; cnt++; + } xa_unlock(&rt->xa); return cnt; } @@ -266,6 +269,53 @@ int rdma_restrack_put(struct rdma_restrack_entry *res) } EXPORT_SYMBOL(rdma_restrack_put); +/** + * rdma_restrack_sync() - Fence concurrent netlink dumps on an entry + * @res: resource entry + * + * After this returns any concurrent netlink dump threads will see the current + * value of the object. This is useful if the object has to be changed and there + * is not locking to protect the nl side. Eg for mr->pd. This effectively + * destroys the object from a kref/xarray perspective and then immediately + * restores it. The kref is acting like a lock to barrier concurrent nl threads. + * Callers must ensure rdma_restrack_del() is not concurrently called. + */ +void rdma_restrack_sync(struct rdma_restrack_entry *res) +{ + struct rdma_restrack_entry *old; + struct rdma_restrack_root *rt; + struct task_struct *task; + struct ib_device *dev; + + if (!res->valid || res->no_track) + return; + + dev = res_to_dev(res); + if (WARN_ON(!dev)) + return; + + rt = &dev->res[res->type]; + if (WARN_ON(xa_get_mark(&rt->xa, res->id, RESTRACK_DD))) + return; + + old = xa_cmpxchg(&rt->xa, res->id, res, XA_ZERO_ENTRY, GFP_KERNEL); + if (WARN_ON(old != res)) + return; + + task = res->task; + if (task) + get_task_struct(task); + rdma_restrack_put(res); + wait_for_completion(&res->comp); + reinit_completion(&res->comp); + if (task) + res->task = task; + kref_init(&res->kref); + + xa_cmpxchg(&rt->xa, res->id, XA_ZERO_ENTRY, res, GFP_KERNEL); +} +EXPORT_SYMBOL(rdma_restrack_sync); + /** * rdma_restrack_del() - delete object from the reource tracking database * @res: resource entry diff --git a/drivers/infiniband/core/restrack.h b/drivers/infiniband/core/restrack.h index 6a04fc41f738..75b8d1005a98 100644 --- a/drivers/infiniband/core/restrack.h +++ b/drivers/infiniband/core/restrack.h @@ -27,6 +27,7 @@ int rdma_restrack_init(struct ib_device *dev); void rdma_restrack_clean(struct ib_device *dev); void rdma_restrack_add(struct rdma_restrack_entry *res); void rdma_restrack_del(struct rdma_restrack_entry *res); +void rdma_restrack_sync(struct rdma_restrack_entry *res); void rdma_restrack_new(struct rdma_restrack_entry *res, enum rdma_restrack_type type); void rdma_restrack_set_name(struct rdma_restrack_entry *res, diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c index 2e4265ba35b7..f442cca3c05c 100644 --- a/drivers/infiniband/core/uverbs_cmd.c +++ b/drivers/infiniband/core/uverbs_cmd.c @@ -46,6 +46,7 @@ #include "uverbs.h" #include "core_priv.h" +#include "restrack.h" /* * Copy a response to userspace. If the provided 'resp' is larger than the @@ -813,6 +814,10 @@ static int ib_uverbs_rereg_mr(struct uverbs_attr_bundle *attrs) ret = PTR_ERR(new_pd); goto put_uobjs; } + if (new_pd == orig_pd) { + uobj_put_obj_read(new_pd); + cmd.flags &= ~IB_MR_REREG_PD; + } } else { new_pd = mr->pd; } @@ -858,9 +863,10 @@ static int ib_uverbs_rereg_mr(struct uverbs_attr_bundle *attrs) mr = new_mr; } else { if (cmd.flags & IB_MR_REREG_PD) { - atomic_dec(&orig_pd->usecnt); - mr->pd = new_pd; atomic_inc(&new_pd->usecnt); + WRITE_ONCE(mr->pd, new_pd); + rdma_restrack_sync(&mr->res); + atomic_dec(&orig_pd->usecnt); } if (cmd.flags & IB_MR_REREG_TRANS) { mr->iova = cmd.hca_va; diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index 3eea95e51928..05b101693cd5 100644 --- a/include/rdma/ib_verbs.h +++ b/include/rdma/ib_verbs.h @@ -1833,6 +1833,11 @@ struct ib_dm { struct ib_mr { struct ib_device *device; + /* + * Due to IB_MR_REREG_PD pd is not a fixed pointer and can change. For a + * user MR, this value should only be read from a system call that holds + * the uobject lock, or the driver should disable in-place REREG_PD. + */ struct ib_pd *pd; u32 lkey; u32 rkey; -- 2.34.1

2 1

[PTACH openEuler-1.0-LTS] RDMA/mlx5: Release the HW‑provided UAR index rather than the SW one
by Chen Jinghuang 27 Aug '26

27 Aug '26

From: Leon Romanovsky <leonro(a)nvidia.com> mainline inclusion from mainline-v7.2-rc1 commit 449ae7927152e46acbe5f19f97eafdae6d3a96b1 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/18162 CVE: CVE-2026-74296 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… -------------------------------- Free the UAR index returned by the hardware. Fixes: 4ed131d0bb15 ("IB/mlx5: Expose dynamic mmap allocation") Link: https://patch.msgid.link/r/20260611-fix-uar-release-v1-1-f5464d845dbf@nvidi… Signed-off-by: Leon Romanovsky <leonro(a)nvidia.com> Signed-off-by: Jason Gunthorpe <jgg(a)nvidia.com> Conflicts: drivers/infiniband/hw/mlx5/main.c [ Context conflict ] Signed-off-by: Chen jinghuang <chenjinghuang2(a)huawei.com> --- drivers/infiniband/hw/mlx5/main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index 4f340d6db582..051de9f3ed82 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -2148,7 +2148,7 @@ static int uar_mmap(struct mlx5_ib_dev *dev, enum mlx5_ib_mmap_cmd cmd, if (!dyn_uar) return err; - mlx5_cmd_free_uar(dev->mdev, idx); + mlx5_cmd_free_uar(dev->mdev, uar_index); free_bfreg: mlx5_ib_free_bfreg(dev, bfregi, bfreg_dyn_idx); -- 2.34.1

1 0