summaryrefslogtreecommitdiff
path: root/drivers/infiniband
AgeCommit message (Collapse)AuthorFilesLines
7 daysRDMA/core: Prefer NLA_NUL_STRINGFlorian Westphal1-3/+3
[ Upstream commit 6ed3d14fc45d3da6025e7fe4a6a09066856698e2 ] These attributes are evaluated as c-string (passed to strcmp), but NLA_STRING doesn't check for the presence of a \0 terminator. Either this needs to switch to nla_strcmp() and needs to adjust printf fmt specifier to not use plain %s, or this needs to use NLA_NUL_STRING. As the code has been this way for long time, it seems to me that userspace does include the terminating nul, even tough its not enforced so far, and thus NLA_NUL_STRING use is the simpler solution. Fixes: 30dc5e63d6a5 ("RDMA/core: Add support for iWARP Port Mapper user space service") Link: https://patch.msgid.link/r/20260330122742.13315-1-fw@strlen.de Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
7 daysRDMA/umem: Use consistent DMA attributes when unmapping entriesLeon Romanovsky1-7/+6
[ Upstream commit 179b32095854d44749dd535502f05d95bbf43775 ] The DMA API expects that mapping and unmapping use the same DMA attributes. The RDMA umem code did not meet this requirement, so fix the mismatch. Fixes: f03d9fadfe13 ("RDMA/core: Add weak ordering dma attr to dma mapping") Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
2026-05-14RDMA/vmw_pvrdma: Fix double free on pvrdma_alloc_ucontext() error pathJason Gunthorpe1-1/+1
commit e38e86995df27f1f854063dab1f0c6a513db3faf upstream. Sashiko points out that pvrdma_uar_free() is already called within pvrdma_dealloc_ucontext(), so calling it before triggers a double free. Cc: stable@vger.kernel.org Fixes: 29c8d9eba550 ("IB: Add vmw_pvrdma driver") Link: https://sashiko.dev/#/patchset/0-v1-e911b76a94d1%2B65d95-rdma_udata_rep_jgg%40nvidia.com?part=4 Link: https://patch.msgid.link/r/10-v1-41f3135e5565+9d2-rdma_ai_fixes1_jgg@nvidia.com Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/rxe: Reject unknown opcodes before ICRC processingMichael Bommarito1-0/+11
commit 4c6f86d85d03cdb33addce86aa69aa795ca6c47a upstream. Even after applying commit 7244491dab34 ("RDMA/rxe: Validate pad and ICRC before payload_size() in rxe_rcv"), a single unauthenticated UDP packet can still trigger panic. That patch handled payload_size() underflow only for valid opcodes with short packets, not for packets carrying an unknown opcode. The unknown-opcode OOB read described below predates that commit and reaches back to the initial Soft RoCE driver. The check added there reads pkt->paylen < header_size(pkt) + bth_pad(pkt) + RXE_ICRC_SIZE where header_size(pkt) expands to rxe_opcode[pkt->opcode].length. The rxe_opcode[] array has 256 entries but is only populated for defined IB opcodes; any other entry (for example opcode 0xff) is zero-initialized, so length == 0 and the check degenerates to pkt->paylen < 0 + bth_pad(pkt) + RXE_ICRC_SIZE which does not constrain pkt->paylen enough. rxe_icrc_hdr() then computes rxe_opcode[pkt->opcode].length - RXE_BTH_BYTES which underflows when length == 0 and passes a huge value to rxe_crc32(), causing an out-of-bounds read of the skb payload. Reproduced on v7.0-rc7 with that fix applied, QEMU/KVM with CONFIG_RDMA_RXE=y and CONFIG_KASAN=y, after rdma link add rxe0 type rxe netdev eth0 A single 48-byte UDP packet to port 4791 with BTH opcode=0xff and QPN=IB_MULTICAST_QPN triggers: BUG: KASAN: slab-out-of-bounds in crc32_le+0x115/0x170 Read of size 1 at addr ... The buggy address is located 0 bytes to the right of allocated 704-byte region Call Trace: crc32_le+0x115/0x170 rxe_icrc_hdr.isra.0+0x226/0x300 rxe_icrc_check+0x13f/0x3a0 rxe_rcv+0x6e1/0x16e0 rxe_udp_encap_recv+0x20a/0x320 udp_queue_rcv_one_skb+0x7ed/0x12c0 Subsequent packets with the same shape fault on unmapped memory and panic the kernel. The trigger requires only module load and "rdma link add"; no QP, no connection, and no authentication. Fix this by rejecting packets whose opcode has no rxe_opcode[] entry, detected via the zero mask or zero length, before any length arithmetic runs. Cc: stable@vger.kernel.org Fixes: 8700e3e7c485 ("Soft RoCE driver") Link: https://patch.msgid.link/r/20260414111555.3386793-1-michael.bommarito@gmail.com Assisted-by: Claude:claude-opus-4-6 Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/rxe: Reject non-8-byte ATOMIC_WRITE payloadsMichael Bommarito1-1/+13
commit 1114c87aa6f195cf07da55a27b2122ae26557b26 upstream. atomic_write_reply() at drivers/infiniband/sw/rxe/rxe_resp.c unconditionally dereferences 8 bytes at payload_addr(pkt): value = *(u64 *)payload_addr(pkt); check_rkey() previously accepted an ATOMIC_WRITE request with pktlen == resid == 0 because the length validation only compared pktlen against resid. A remote initiator that sets the RETH length to 0 therefore reaches atomic_write_reply() with a zero-byte logical payload, and the responder reads sizeof(u64) bytes from past the logical end of the packet into skb->head tailroom, then writes those 8 bytes into the attacker's MR via rxe_mr_do_atomic_write(). That is a remote disclosure of 4 bytes of kernel tailroom per probe (the other 4 bytes are the packet's own trailing ICRC). IBA oA19-28 defines ATOMIC_WRITE as exactly 8 bytes. Anything else is protocol-invalid. Hoist a strict length check into check_rkey() so the responder never reaches the unchecked dereference, and keep the existing WRITE-family length logic for the normal RDMA WRITE path. Reproduced on mainline with an unmodified rxe driver: a sustained zero-length ATOMIC_WRITE probe repeatedly leaks adjacent skb head-buffer bytes into the attacker's MR, including recognisable kernel strings and partial kernel-direct-map pointer words. With this patch applied the responder rejects the PDU and the MR stays all-zero. Cc: stable@vger.kernel.org Fixes: 034e285f8b99 ("RDMA/rxe: Make responder support atomic write on RC service") Link: https://patch.msgid.link/r/20260418162141.3610201-1-michael.bommarito@gmail.com Assisted-by: Claude:claude-opus-4-7 Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/ocrdma: Don't NULL deref uctx on errors in ocrdma_copy_pd_uresp()Jason Gunthorpe1-2/+2
commit 34fbf48cf3b410d2a6e8c586fa952a36331ca5ba upstream. Sashiko points out that pd->uctx isn't initialized until late in the function so all these error flow references are NULL and will crash. Use the uctx that isn't NULL. Cc: stable@vger.kernel.org Fixes: fe2caefcdf58 ("RDMA/ocrdma: Add driver for Emulex OneConnect IBoE RDMA adapter") Link: https://sashiko.dev/#/patchset/0-v1-e911b76a94d1%2B65d95-rdma_udata_rep_jgg%40nvidia.com?part=4 Link: https://patch.msgid.link/r/9-v1-41f3135e5565+9d2-rdma_ai_fixes1_jgg@nvidia.com Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/mlx5: Fix error path fall-through in mlx5_ib_dev_res_srq_init()Junrui Luo1-0/+1
commit c488df06bd552bb8b6e14fa0cfd5ad986c6e9525 upstream. mlx5_ib_dev_res_srq_init() allocates two SRQs, s0 and s1. When ib_create_srq() fails for s1, the error branch destroys s0 but falls through and unconditionally assigns the freed s0 and the ERR_PTR s1 to devr->s0 and devr->s1. This leads to several problems: the lock-free fast path checks "if (devr->s1) return 0;" and treats the ERR_PTR as already initialised; users in mlx5_ib_create_qp() dereference the freed SRQ or ERR_PTR via to_msrq(devr->s0)->msrq.srqn; and mlx5_ib_dev_res_cleanup() dereferences the ERR_PTR and double-frees s0 on teardown. Fix by adding the same `goto unlock` in the s1 failure path. Cc: stable@vger.kernel.org Fixes: 5895e70f2e6e ("IB/mlx5: Allocate resources just before first QP/SRQ is created") Link: https://patch.msgid.link/r/SYBPR01MB7881E1E0970268BD69C0BA75AF2B2@SYBPR01MB7881.ausprd01.prod.outlook.com Reported-by: Yuhao Jiang <danisjiang@gmail.com> Signed-off-by: Junrui Luo <moonafterrain@outlook.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/mlx4: Fix resource leak on error in mlx4_ib_create_srq()Jason Gunthorpe1-1/+3
commit c54c7e4cb679c0aaa1cb489b9c3f2cd98e63a44c upstream. Sashiko points out that mlx4_srq_alloc() was not undone during error unwind, add the missing call to mlx4_srq_free(). Cc: stable@vger.kernel.org Fixes: 225c7b1feef1 ("IB/mlx4: Add a driver Mellanox ConnectX InfiniBand adapters") Link: https://sashiko.dev/#/patchset/0-v1-e911b76a94d1%2B65d95-rdma_udata_rep_jgg%40nvidia.com?part=8 Link: https://patch.msgid.link/r/11-v1-41f3135e5565+9d2-rdma_ai_fixes1_jgg@nvidia.com Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/mana: Validate rx_hash_key_lenJason Gunthorpe1-0/+3
commit 6dd2d4ad9c8429523b1c220c5132bd551c006425 upstream. Sashiko points out that rx_hash_key_len comes from a uAPI structure and is blindly passed to memcpy, allowing the userspace to trash kernel memory. Bounds check it so the memcpy cannot overflow. Cc: stable@vger.kernel.org Fixes: 0266a177631d ("RDMA/mana_ib: Add a driver for Microsoft Azure Network Adapter") Link: https://sashiko.dev/#/patchset/0-v2-1c49eeb88c48%2B91-rdma_udata_rep_jgg%40nvidia.com?part=1 Link: https://patch.msgid.link/r/4-v1-41f3135e5565+9d2-rdma_ai_fixes1_jgg@nvidia.com Reviewed-by: Long Li <longli@microsoft.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/mana: Remove user triggerable WARN_ON() in mana_ib_create_qp_rss()Jason Gunthorpe1-2/+3
commit 159f2efabc89d3f931d38f2d35876535d4abf0a3 upstream. Sashiko points out that the user can specify WQs sharing the same CQ as a part of the uAPI and this will trigger the WARN_ON() then go on to corrupt the kernel. Just reject it outright and fail the QP creation. Cc: stable@vger.kernel.org Fixes: c15d7802a424 ("RDMA/mana_ib: Add CQ interrupt support for RAW QP") Link: https://sashiko.dev/#/patchset/0-v2-1c49eeb88c48%2B91-rdma_udata_rep_jgg%40nvidia.com?part=1 Link: https://patch.msgid.link/r/5-v1-41f3135e5565+9d2-rdma_ai_fixes1_jgg@nvidia.com Reviewed-by: Long Li <longli@microsoft.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/mana: Fix mana_destroy_wq_obj() cleanup in mana_ib_create_qp_rss()Jason Gunthorpe1-5/+4
commit 34ecf795692ee57c393109f4a24ccc313091e137 upstream. Sashiko points out there are two bugs here in the error unwind flow, both related to how the WQ table is unwound. First there is a double i-- on the first failure path due to the while loop having a i--, remove it. Second if mana_ib_install_cq_cb() fails then mana_create_wq_obj() is not undone due to the above i--. Cc: stable@vger.kernel.org Fixes: c15d7802a424 ("RDMA/mana_ib: Add CQ interrupt support for RAW QP") Link: https://sashiko.dev/#/patchset/0-v2-1c49eeb88c48%2B91-rdma_udata_rep_jgg%40nvidia.com?part=1 Link: https://patch.msgid.link/r/6-v1-41f3135e5565+9d2-rdma_ai_fixes1_jgg@nvidia.com Reviewed-by: Long Li <longli@microsoft.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/mana: Fix error unwind in mana_ib_create_qp_rss()Jason Gunthorpe1-1/+3
commit 6aaa978c6b6218cfac15fe1dab17c76fe229ce3f upstream. Sashiko points out that mana_ib_cfg_vport_steering() is leaked, the normal destroy path cleans it up. Cc: stable@vger.kernel.org Fixes: 0266a177631d ("RDMA/mana_ib: Add a driver for Microsoft Azure Network Adapter") Link: https://sashiko.dev/#/patchset/0-v1-e911b76a94d1%2B65d95-rdma_udata_rep_jgg%40nvidia.com?part=4 Link: https://patch.msgid.link/r/7-v1-41f3135e5565+9d2-rdma_ai_fixes1_jgg@nvidia.com Reviewed-by: Long Li <longli@microsoft.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/ionic: Fix typo in format stringJason Gunthorpe1-1/+1
commit 70f780edcd1e86350202d8a409de026b2d2e2067 upstream. Applying the corrupted patch by hand mangled the format string, put the s in the right place. Cc: stable@vger.kernel.org Fixes: 654a27f25530 ("RDMA/ionic: bound node_desc sysfs read with %.64s") Link: https://patch.msgid.link/r/1-v1-41f3135e5565+9d2-rdma_ai_fixes1_jgg@nvidia.com Reported-by: Brad Spengler <brad.spengler@opensrcsec.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/ionic: bound node_desc sysfs read with %.64sKai Zen1-1/+1
commit 654a27f25530d052eeedf086e6c3e2d585c203bd upstream. node_desc[64] in struct ib_device is not guaranteed to be NUL- terminated. The core IB sysfs handler uses "%.64s" for exactly this reason (drivers/infiniband/core/sysfs.c:1307), since node_desc_store() performs a raw memcpy of up to IB_DEVICE_NODE_DESC_MAX bytes with no NUL termination: memcpy(desc.node_desc, buf, min_t(int, count, IB_DEVICE_NODE_DESC_MAX)); If exactly 64 bytes are written via the node_desc sysfs file, the array contains no NUL byte. The ionic hca_type_show() handler uses unbounded "%s" and will read past the end of node_desc into adjacent fields of struct ib_device until it encounters a NUL. ionic supports IB_DEVICE_MODIFY_NODE_DESC, so this is triggerable by userspace. Match the core handler and bound the format specifier. Cc: stable@vger.kernel.org Fixes: 2075bbe8ef03 ("RDMA/ionic: Register device ops for miscellaneous functionality") Link: https://patch.msgid.link/r/CALynFi7NAbhDCt1tdaDbf6TnLvAqbaHa6-Wqf6OkzREbA_PAfg@mail.gmail.com Signed-off-by: Kai Aizen <kai.aizen.dev@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-14RDMA/hns: Fix unlocked call to hns_roce_qp_remove()Jason Gunthorpe1-0/+7
commit 0c99acbc8b6c6dd526ae475a48ee1897b61072fb upstream. Sashiko points out that hns_roce_qp_remove() requires the caller to hold locks. The error flow in hns_roce_create_qp_common() doesn't hold those locks for the error unwind so it risks corrupting memory. Grab the same locks the other two callers use. Cc: stable@vger.kernel.org Fixes: e088a685eae9 ("RDMA/hns: Support rq record doorbell for the user space") Link: https://sashiko.dev/#/patchset/0-v2-1c49eeb88c48%2B91-rdma_udata_rep_jgg%40nvidia.com?part=9 Link: https://patch.msgid.link/r/15-v1-41f3135e5565+9d2-rdma_ai_fixes1_jgg@nvidia.com Reviewed-by: Junxian Huang <huangjunxian6@hisilicon.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-07IB/core: Fix zero dmac race in neighbor resolutionChen Zhao1-0/+3
commit 5e6de34d82b49cab9d8a42063e9cd0f22a4f31e5 upstream. dst_fetch_ha() checks nud_state without holding the neighbor lock, then copies ha under the seqlock. A race in __neigh_update() where nud_state is set to NUD_REACHABLE before ha is written allows dst_fetch_ha() to read a zero MAC address while the seqlock reports no concurrent writer. netevent_callback amplifies this by waking ALL pending addr_req workers when ANY neighbor becomes NUD_VALID. At scale (N peers resolving ARP concurrently), the hit probability scales as N^2, making it near-certain for large RDMA workloads. N(A): neigh_update(A) W(A): addr_resolve(A) | [sleep] | write_lock_bh(&A->lock) | | A->nud_state = NUD_REACHABLE | | // A->ha is still 0 | | [woken by netevent_cb() of | another neighbour] | | dst_fetch_ha(A) | | A->nud_state & NUD_VALID | | read_seqbegin(&A->ha_lock) | | snapshot = A->ha /* 0 */ | | read_seqretry(&A->ha_lock) | | return snapshot | seqlock(&A->ha_lock) | A->ha = mac_A /* too late */ | sequnlock(&A->ha_lock) | write_unlock_bh(&A->lock) The incorrect/zero mac is read and programmed in the device QP while it was not yet updated. This causes silent packet loss and eventual RETRY_EXC_ERR. Fix by holding the neighbor read lock across the nud_state check and ha copy in dst_fetch_ha(), ensuring it synchronizes with __neigh_update() which is updating while holding the write lock. Cc: stable@vger.kernel.org Fixes: 92ebb6a0a13a ("IB/cm: Remove now useless rcu_lock in dst_fetch_ha") Link: https://patch.msgid.link/r/20260405-fix-dmac-race-v1-1-cfa1ec2ce54a@nvidia.com Signed-off-by: Chen Zhao <chezhao@nvidia.com> Reviewed-by: Parav Pandit <parav@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-07RDMA/rxe: Validate pad and ICRC before payload_size() in rxe_rcvhkbinbin1-1/+2
commit 7244491dab347f648e661da96dc0febadd9daec3 upstream. rxe_rcv() currently checks only that the incoming packet is at least header_size(pkt) bytes long before payload_size() is used. However, payload_size() subtracts both the attacker-controlled BTH pad field and RXE_ICRC_SIZE from pkt->paylen: payload_size = pkt->paylen - offset[RXE_PAYLOAD] - bth_pad(pkt) - RXE_ICRC_SIZE This means a short packet can still make payload_size() underflow even if it includes enough bytes for the fixed headers. Simply requiring header_size(pkt) + RXE_ICRC_SIZE is not sufficient either, because a packet with a forged non-zero BTH pad can still leave payload_size() negative and pass an underflowed value to later receive-path users. Fix this by validating pkt->paylen against the full minimum length required by payload_size(): header_size(pkt) + bth_pad(pkt) + RXE_ICRC_SIZE. Cc: stable@vger.kernel.org Fixes: 8700e3e7c485 ("Soft RoCE driver") Link: https://patch.msgid.link/r/20260401121907.1468366-1-hkbinbinbin@gmail.com Signed-off-by: hkbinbin <hkbinbinbin@gmail.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-05-07RDMA/mana_ib: Disable RX steering on RSS QP destroyLong Li1-0/+15
commit dbeb256e8dd87233d891b170c0b32a6466467036 upstream. When an RSS QP is destroyed (e.g. DPDK exit), mana_ib_destroy_qp_rss() destroys the RX WQ objects but does not disable vPort RX steering in firmware. This leaves stale steering configuration that still points to the destroyed RX objects. If traffic continues to arrive (e.g. peer VM is still transmitting) and the VF interface is subsequently brought up (mana_open), the firmware may deliver completions using stale CQ IDs from the old RX objects. These CQ IDs can be reused by the ethernet driver for new TX CQs, causing RX completions to land on TX CQs: WARNING: mana_poll_tx_cq+0x1b8/0x220 [mana] (is_sq == false) WARNING: mana_gd_process_eq_events+0x209/0x290 (cq_table lookup fails) Fix this by disabling vPort RX steering before destroying RX WQ objects. Note that mana_fence_rqs() cannot be used here because the fence completion is delivered on the CQ, which is polled by user-mode (e.g. DPDK) and not visible to the kernel driver. Refactor the disable logic into a shared mana_disable_vport_rx() in mana_en, exported for use by mana_ib, replacing the duplicate code. The ethernet driver's mana_dealloc_queues() is also updated to call this common function. Fixes: 0266a177631d ("RDMA/mana_ib: Add a driver for Microsoft Azure Network Adapter") Cc: stable@vger.kernel.org Signed-off-by: Long Li <longli@microsoft.com> Link: https://patch.msgid.link/20260325194100.1929056-1-longli@microsoft.com Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-04-22x86-64: rename misleadingly named '__copy_user_nocache()' functionLinus Torvalds1-5/+3
commit d187a86de793f84766ea40b9ade7ac60aabbb4fe upstream. This function was a masterclass in bad naming, for various historical reasons. It claimed to be a non-cached user copy. It is literally _neither_ of those things. It's a specialty memory copy routine that uses non-temporal stores for the destination (but not the source), and that does exception handling for both source and destination accesses. Also note that while it works for unaligned targets, any unaligned parts (whether at beginning or end) will not use non-temporal stores, since only words and quadwords can be non-temporal on x86. The exception handling means that it _can_ be used for user space accesses, but not on its own - it needs all the normal "start user space access" logic around it. But typically the user space access would be the source, not the non-temporal destination. That was the original intention of this, where the destination was some fragile persistent memory target that needed non-temporal stores in order to catch machine check exceptions synchronously and deal with them gracefully. Thus that non-descriptive name: one use case was to copy from user space into a non-cached kernel buffer. However, the existing users are a mix of that intended use-case, and a couple of random drivers that just did this as a performance tweak. Some of those random drivers then actively misused the user copying version (with STAC/CLAC and all) to do kernel copies without ever even caring about the exception handling, _just_ for the non-temporal destination. Rename it as a first small step to actually make it halfway sane, and change the prototype to be more normal: it doesn't take a user pointer unless the caller has done the proper conversion, and the argument size is the full size_t (it still won't actually copy more than 4GB in one go, but there's also no reason to silently truncate the size argument in the caller). Finally, use this now sanely named function in the NTB code, which mis-used a user copy version (with STAC/CLAC and all) of this interface despite it not actually being a user copy at all. Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-04-09kernfs: pass struct ns_common instead of const void * for namespace tagsChristian Brauner2-5/+7
kernfs has historically used const void * to pass around namespace tags used for directory-level namespace filtering. The only current user of this is sysfs network namespace tagging where struct net pointers are cast to void *. Replace all const void * namespace parameters with const struct ns_common * throughout the kernfs, sysfs, and kobject namespace layers. This includes the kobj_ns_type_operations callbacks, kobject_namespace(), and all sysfs/kernfs APIs that accept or return namespace tags. Passing struct ns_common is needed because various codepaths require access to the underlying namespace. A struct ns_common can always be converted back to the concrete namespace type (e.g., struct net) via container_of() or to_ns_common() in the reverse direction. This is a preparatory change for switching to ns_id-based directory iteration to prevent a KASLR pointer leak through the current use of raw namespace pointers as hash seeds and comparison keys. Signed-off-by: Christian Brauner <brauner@kernel.org>
2026-03-27Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdmaLinus Torvalds8-100/+123
Pull rdma fixes from Jason Gunthorpe: - Quite a few irdma bug fixes, several user triggerable - Fix a 0 SMAC header in ionic - Tolerate FW errors for RAAS in bng_re - Don't UAF in efa when printing error events - Better handle pool exhaustion in the new bvec paths * tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: RDMA/irdma: Harden depth calculation functions RDMA/irdma: Return EINVAL for invalid arp index error RDMA/irdma: Fix deadlock during netdev reset with active connections RDMA/irdma: Remove reset check from irdma_modify_qp_to_err() RDMA/irdma: Clean up unnecessary dereference of event->cm_node RDMA/irdma: Remove a NOP wait_event() in irdma_modify_qp_roce() RDMA/irdma: Update ibqp state to error if QP is already in error state RDMA/irdma: Initialize free_qp completion before using it RDMA/efa: Fix possible deadlock RDMA/rw: Fix MR pool exhaustion in bvec RDMA READ path RDMA/rw: Fall back to direct SGE on MR pool exhaustion RDMA/efa: Fix use of completion ctx after free RDMA/bng_re: Fix silent failure in HWRM version query RDMA/ionic: Preserve and set Ethernet source MAC after ib_ud_header_init() RDMA/irdma: Fix double free related to rereg_user_mr
2026-03-26Merge tag 'dma-mapping-7.0-2026-03-25' of ↵Linus Torvalds1-2/+3
git://git.kernel.org/pub/scm/linux/kernel/git/mszyprowski/linux Pull dma-mapping fixes from Marek Szyprowski: "A set of fixes for DMA-mapping subsystem, which resolve false- positive warnings from KMSAN and DMA-API debug (Shigeru Yoshida and Leon Romanovsky) as well as a simple build fix (Miguel Ojeda)" * tag 'dma-mapping-7.0-2026-03-25' of git://git.kernel.org/pub/scm/linux/kernel/git/mszyprowski/linux: dma-mapping: add missing `inline` for `dma_free_attrs` mm/hmm: Indicate that HMM requires DMA coherency RDMA/umem: Tell DMA mapping that UMEM requires coherency iommu/dma: add support for DMA_ATTR_REQUIRE_COHERENT attribute dma-direct: prevent SWIOTLB path when DMA_ATTR_REQUIRE_COHERENT is set dma-mapping: Introduce DMA require coherency attribute dma-mapping: Clarify valid conditions for CPU cache line overlap dma-mapping: handle DMA_ATTR_CPU_CACHE_CLEAN in trace output dma-debug: Allow multiple invocations of overlapping entries dma: swiotlb: add KMSAN annotations to swiotlb_bounce()
2026-03-20RDMA/umem: Tell DMA mapping that UMEM requires coherencyLeon Romanovsky1-2/+3
The RDMA subsystem exposes DMA regions through the verbs interface, which assumes a coherent system. Use the DMA_ATTR_REQUIRE_COHERENCE attribute to ensure coherency and avoid taking the SWIOTLB path. The RDMA verbs programming model resembles HMM and assumes concurrent DMA and CPU access to userspace memory. The hardware and programming model support "one-sided" operations initiated remotely without any local CPU involvement or notification. These include ATOMIC compare/swap, READ, and WRITE. A remote CPU can use these operations to traverse data structures, manipulate locks, and perform similar tasks without the host CPU’s awareness. If SWIOTLB substitutes memory or DMA is not cache coherent, these use cases break entirely. In-kernel RDMA is fine with incoherent mappings because kernel users do not rely on one-sided operations in ways that would expose these issues. A given region may also be exported multiple times, which can trigger warnings about cacheline overlaps. These warnings are suppressed when the new attribute is used. infiniband rocep8s0f0: mlx5_ib_reg_user_mr:1592:(pid 5812): start 0x2b28c000, iova 0x2b28c000, length 0x1000, access_flags 0x1 infiniband rocep8s0f0: mlx5_ib_reg_user_mr:1592:(pid 5812): start 0x2b28c001, iova 0x2b28c001, length 0xfff, access_flags 0x1 ------------[ cut here ]------------ DMA-API: mlx5_core 0000:08:00.0: cacheline tracking EEXIST, overlapping mappings aren't supported WARNING: kernel/dma/debug.c:620 at add_dma_entry+0x1bb/0x280, CPU#6: ibv_rc_pingpong/5812 Modules linked in: veth xt_conntrack xt_MASQUERADE nf_conntrack_netlink nfnetlink iptable_nat nf_nat xt_addrtype br_netfilter rpcsec_gss_krb5 auth_rpcgss oid_registry overlay mlx5_fwctl zram zsmalloc mlx5_ib fuse rpcrdma rdma_ucm ib_uverbs ib_iser libiscsi scsi_transport_iscsi ib_umad rdma_cm ib_ipoib iw_cm ib_cm mlx5_core ib_core CPU: 6 UID: 2733 PID: 5812 Comm: ibv_rc_pingpong Tainted: G W 6.19.0+ #129 PREEMPT Tainted: [W]=WARN Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014 RIP: 0010:add_dma_entry+0x1be/0x280 Code: 8b 7b 10 48 85 ff 0f 84 c3 00 00 00 48 8b 6f 50 48 85 ed 75 03 48 8b 2f e8 ff 8e 6a 00 48 89 c6 48 8d 3d 55 ef 2d 01 48 89 ea <67> 48 0f b9 3a 48 85 db 74 1a 48 c7 c7 b0 00 2b 82 e8 9c 25 fd ff RSP: 0018:ff11000138717978 EFLAGS: 00010286 RAX: ffffffffa02d7831 RBX: ff1100010246de00 RCX: 0000000000000000 RDX: ff110001036fac30 RSI: ffffffffa02d7831 RDI: ffffffff82678650 RBP: ff110001036fac30 R08: ff11000110dcb4a0 R09: ff11000110dcb478 R10: 0000000000000000 R11: ffffffff824b30a8 R12: 0000000000000000 R13: 00000000ffffffef R14: 0000000000000202 R15: ff1100010246de00 FS: 00007f59b411c740(0000) GS:ff110008dcc99000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007ffe538f7000 CR3: 000000010e066005 CR4: 0000000000373eb0 Call Trace: <TASK> debug_dma_map_sg+0x1b4/0x390 __dma_map_sg_attrs+0x6d/0x1a0 dma_map_sgtable+0x19/0x30 ib_umem_get+0x254/0x380 [ib_uverbs] mlx5_ib_reg_user_mr+0x68/0x2a0 [mlx5_ib] ib_uverbs_reg_mr+0x17f/0x2a0 [ib_uverbs] ib_uverbs_handler_UVERBS_METHOD_INVOKE_WRITE+0xc2/0x130 [ib_uverbs] ib_uverbs_cmd_verbs+0xa0b/0xae0 [ib_uverbs] ? ib_uverbs_handler_UVERBS_METHOD_QUERY_PORT_SPEED+0xe0/0xe0 [ib_uverbs] ? mmap_region+0x7a/0xb0 ? do_mmap+0x3b8/0x5c0 ib_uverbs_ioctl+0xa7/0x110 [ib_uverbs] __x64_sys_ioctl+0x14f/0x8b0 ? ksys_mmap_pgoff+0xc5/0x190 do_syscall_64+0x8c/0xbf0 entry_SYSCALL_64_after_hwframe+0x4b/0x53 RIP: 0033:0x7f59b430aeed Code: 04 25 28 00 00 00 48 89 45 c8 31 c0 48 8d 45 10 c7 45 b0 10 00 00 00 48 89 45 b8 48 8d 45 d0 48 89 45 c0 b8 10 00 00 00 0f 05 <89> c2 3d 00 f0 ff ff 77 1a 48 8b 45 c8 64 48 2b 04 25 28 00 00 00 RSP: 002b:00007ffe538f9430 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 RAX: ffffffffffffffda RBX: 00007ffe538f94c0 RCX: 00007f59b430aeed RDX: 00007ffe538f94e0 RSI: 00000000c0181b01 RDI: 0000000000000003 RBP: 00007ffe538f9480 R08: 0000000000000028 R09: 00007ffe538f9684 R10: 0000000000000001 R11: 0000000000000246 R12: 00007ffe538f9684 R13: 000000000000000c R14: 000000002b28d170 R15: 000000000000000c </TASK> ---[ end trace 0000000000000000 ]--- Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> Link: https://lore.kernel.org/r/20260316-dma-debug-overlap-v3-7-1dde90a7f08b@nvidia.com
2026-03-18RDMA/irdma: Harden depth calculation functionsShiraz Saleem1-17/+22
An issue was exposed where OS can pass in U32_MAX for SQ/RQ/SRQ size. This can cause integer overflow and truncation of SQ/RQ/SRQ depth returning a success when it should have failed. Harden the functions to do all depth calculations and boundary checking in u64 sizes. Fixes: 563e1feb5f6e ("RDMA/irdma: Add SRQ support") Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com> Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-18RDMA/irdma: Return EINVAL for invalid arp index errorTatyana Nikolova1-7/+10
When rdma_connect() fails due to an invalid arp index, user space rdma core reports ENOMEM which is confusing. Modify irdma_make_cm_node() to return the correct error code. Fixes: 146b9756f14c ("RDMA/irdma: Add connection manager") Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-18RDMA/irdma: Fix deadlock during netdev reset with active connectionsAnil Samal1-1/+2
Resolve deadlock that occurs when user executes netdev reset while RDMA applications (e.g., rping) are active. The netdev reset causes ice driver to remove irdma auxiliary driver, triggering device_delete and subsequent client removal. During client removal, uverbs_client waits for QP reference count to reach zero while cma_client holds the final reference, creating circular dependency and indefinite wait in iWARP mode. Skip QP reference count wait during device reset to prevent deadlock. Fixes: c8f304d75f6c ("RDMA/irdma: Prevent QP use after free") Signed-off-by: Anil Samal <anil.samal@intel.com> Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-18RDMA/irdma: Remove reset check from irdma_modify_qp_to_err()Tatyana Nikolova1-2/+0
During reset, irdma_modify_qp() to error should be called to disconnect the QP. Without this fix, if not preceded by irdma_modify_qp() to error, the API call irdma_destroy_qp() gets stuck waiting for the QP refcount to go to zero, because the cm_node associated with this QP isn't disconnected. Fixes: 915cc7ac0f8e ("RDMA/irdma: Add miscellaneous utility definitions") Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-18RDMA/irdma: Clean up unnecessary dereference of event->cm_nodeIvan Barrera1-6/+6
The cm_node is available and the usage of cm_node and event->cm_node seems arbitrary. Clean up unnecessary dereference of event->cm_node. Fixes: 146b9756f14c ("RDMA/irdma: Add connection manager") Signed-off-by: Ivan Barrera <ivan.d.barrera@intel.com> Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-18RDMA/irdma: Remove a NOP wait_event() in irdma_modify_qp_roce()Tatyana Nikolova1-2/+0
Remove a NOP wait_event() in irdma_modify_qp_roce() which is relevant for iWARP and likely a copy and paste artifact for RoCEv2. The wait event is for sending a reset on a TCP connection, after the reset has been requested in irdma_modify_qp(), which occurs only in iWarp mode. Fixes: b48c24c2d710 ("RDMA/irdma: Implement device supported verb APIs") Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-18RDMA/irdma: Update ibqp state to error if QP is already in error stateTatyana Nikolova1-0/+2
In irdma_modify_qp() update ibqp state to error if the irdma QP is already in error state, otherwise the ibqp state which is visible to the consumer app remains stale. Fixes: b48c24c2d710 ("RDMA/irdma: Implement device supported verb APIs") Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-18RDMA/irdma: Initialize free_qp completion before using itJacob Moroni1-1/+1
In irdma_create_qp, if ib_copy_to_udata fails, it will call irdma_destroy_qp to clean up which will attempt to wait on the free_qp completion, which is not initialized yet. Fix this by initializing the completion before the ib_copy_to_udata call. Fixes: b48c24c2d710 ("RDMA/irdma: Implement device supported verb APIs") Signed-off-by: Jacob Moroni <jmoroni@google.com> Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-17RDMA/efa: Fix possible deadlockEthan Tidmore1-0/+1
In the error path for efa_com_alloc_comp_ctx() the semaphore assigned to &aq->avail_cmds is not released. Detected by Smatch: drivers/infiniband/hw/efa/efa_com.c:662 efa_com_cmd_exec() warn: inconsistent returns '&aq->avail_cmds' Add release for &aq->avail_cmds in efa_com_alloc_comp_ctx() error path. Fixes: ef3b06742c8a2 ("RDMA/efa: Fix use of completion ctx after free") Signed-off-by: Ethan Tidmore <ethantidmore06@gmail.com> Link: https://patch.msgid.link/20260314045730.1143862-1-ethantidmore06@gmail.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-17RDMA/rw: Fix MR pool exhaustion in bvec RDMA READ pathChuck Lever1-7/+9
When IOVA-based DMA mapping is unavailable (e.g., IOMMU passthrough mode), rdma_rw_ctx_init_bvec() falls back to checking rdma_rw_io_needs_mr() with the raw bvec count. Unlike the scatterlist path in rdma_rw_ctx_init(), which passes a post-DMA-mapping entry count that reflects coalescing of physically contiguous pages, the bvec path passes the pre-mapping page count. This overstates the number of DMA entries, causing every multi-bvec RDMA READ to consume an MR from the QP's pool. Under NFS WRITE workloads the server performs RDMA READs to pull data from the client. With the inflated MR demand, the pool is rapidly exhausted, ib_mr_pool_get() returns NULL, and rdma_rw_init_one_mr() returns -EAGAIN. svcrdma treats this as a DMA mapping failure, closes the connection, and the client reconnects -- producing a cycle of 71% RPC retransmissions and ~100 reconnections per test run. RDMA WRITEs (NFS READ direction) are unaffected because DMA_TO_DEVICE never triggers the max_sgl_rd check. Remove the rdma_rw_io_needs_mr() gate from the bvec path entirely, so that bvec RDMA operations always use the map_wrs path (direct WR posting without MR allocation). The bvec caller has no post-DMA-coalescing segment count available -- xdr_buf and svc_rqst hold pages as individual pointers, and physical contiguity is discovered only during DMA mapping -- so the raw page count cannot serve as a reliable input to rdma_rw_io_needs_mr(). iWARP devices, which require MRs unconditionally, are handled by an earlier check in rdma_rw_ctx_init_bvec() and are unaffected. Fixes: bea28ac14cab ("RDMA/core: add MR support for bvec-based RDMA operations") Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Link: https://patch.msgid.link/20260313194201.5818-3-cel@kernel.org Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-17RDMA/rw: Fall back to direct SGE on MR pool exhaustionChuck Lever1-3/+18
When IOMMU passthrough mode is active, ib_dma_map_sgtable_attrs() produces no coalescing: each scatterlist page maps 1:1 to a DMA entry, so sgt.nents equals the raw page count. A 1 MB transfer yields 256 DMA entries. If that count exceeds the device's max_sgl_rd threshold (an optimization hint from mlx5 firmware), rdma_rw_io_needs_mr() steers the operation into the MR registration path. Each such operation consumes one or more MRs from a pool sized at max_rdma_ctxs -- roughly one MR per concurrent context. Under write-intensive workloads that issue many concurrent RDMA READs, the pool is rapidly exhausted, ib_mr_pool_get() returns NULL, and rdma_rw_init_one_mr() returns -EAGAIN. Upper layer protocols treat this as a fatal DMA mapping failure and tear down the connection. The max_sgl_rd check is a performance optimization, not a correctness requirement: the device can handle large SGE counts via direct posting, just less efficiently than with MR registration. When the MR pool cannot satisfy a request, falling back to the direct SGE (map_wrs) path avoids the connection reset while preserving the MR optimization for the common case where pool resources are available. Add a fallback in rdma_rw_ctx_init() so that -EAGAIN from rdma_rw_init_mr_wrs() triggers direct SGE posting instead of propagating the error. iWARP devices, which mandate MR registration for RDMA READs, and force_mr debug mode continue to treat -EAGAIN as terminal. Fixes: 00bd1439f464 ("RDMA/rw: Support threshold for registration vs scattering to local pages") Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Link: https://patch.msgid.link/20260313194201.5818-2-cel@kernel.org Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-08RDMA/efa: Fix use of completion ctx after freeYonatan Nachum1-48/+39
On admin queue completion handling, if the admin command completed with error we print data from the completion context. The issue is that we already freed the completion context in polling/interrupts handler which means we print data from context in an unknown state (it might be already used again). Change the admin submission flow so alloc/dealloc of the context will be symmetric and dealloc will be called after any potential use of the context. Fixes: 68fb9f3e312a ("RDMA/efa: Remove redundant NULL pointer check of CQE") Reviewed-by: Daniel Kranzdorf <dkkranzd@amazon.com> Reviewed-by: Michael Margolin <mrgolin@amazon.com> Signed-off-by: Yonatan Nachum <ynachum@amazon.com> Link: https://patch.msgid.link/20260308165350.18219-1-ynachum@amazon.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-05RDMA/bng_re: Fix silent failure in HWRM version queryKamal Heib1-5/+9
If the firmware version query fails, the driver currently ignores the error and continues initializing. This leaves the device in a bad state. Fix this by making bng_re_query_hwrm_version() return the error code and update the driver to check for this error and stop the setup process safely if it happens. Fixes: 745065770c2d ("RDMA/bng_re: Register and get the resources from bnge driver") Signed-off-by: Kamal Heib <kheib@redhat.com> Link: https://patch.msgid.link/20260303043645.425724-1-kheib@redhat.com Reviewed-by: Siva Reddy Kallam <siva.kallam@broadcom.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-04RDMA/ionic: Preserve and set Ethernet source MAC after ib_ud_header_init()Abhijit Gangurde1-1/+3
ionic_build_hdr() populated the Ethernet source MAC (hdr->eth.smac_h) by passing the header’s storage directly to rdma_read_gid_l2_fields(). However, ib_ud_header_init() is called after that and re-initializes the UD header, which wipes the previously written smac_h. As a result, packets are emitted with an zero source MAC address on the wire. Correct the source MAC by reading the GID-derived smac into a temporary buffer and copy it after ib_ud_header_init() completes. Fixes: e8521822c733 ("RDMA/ionic: Register device ops for control path") Cc: stable@vger.kernel.org # 6.18 Signed-off-by: Abhijit Gangurde <abhijit.gangurde@amd.com> Link: https://patch.msgid.link/20260227061809.2979990-1-abhijit.gangurde@amd.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-03-04RDMA/irdma: Fix double free related to rereg_user_mrJacob Moroni1-0/+1
If IB_MR_REREG_TRANS is set during rereg_user_mr, the umem will be released and a new one will be allocated in irdma_rereg_mr_trans. If any step of irdma_rereg_mr_trans fails after the new umem is allocated, it releases the umem, but does not set iwmr->region to NULL. The problem is that this failure is propagated to the user, who will then call ibv_dereg_mr (as they should). Then, the dereg_mr path will see a non-NULL umem and attempt to call ib_umem_release again. Fix this by setting iwmr->region to NULL after ib_umem_release. Fixed: 5ac388db27c4 ("RDMA/irdma: Add support to re-register a memory region") Signed-off-by: Jacob Moroni <jmoroni@google.com> Link: https://patch.msgid.link/20260227152743.1183388-1-jmoroni@google.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-02-26RDMA/uverbs: Import DMA-BUF module in uverbs_std_types_dmabuf fileLeon Romanovsky1-0/+2
Fix the following compilation error: ERROR: modpost: module ib_uverbs uses symbol dma_buf_move_notify from namespace DMA_BUF, but does not import it. Fixes: 0ac6f4056c4a ("RDMA/uverbs: Add DMABUF object type and operations") Link: https://patch.msgid.link/20260225-fix-uverbs-compilation-v1-1-acf7b3d0f9fa@nvidia.com Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
2026-02-25RDMA/umem: Fix double dma_buf_unpin in failure pathJacob Moroni1-3/+1
In ib_umem_dmabuf_get_pinned_with_dma_device(), the call to ib_umem_dmabuf_map_pages() can fail. If this occurs, the dmabuf is immediately unpinned but the umem_dmabuf->pinned flag is still set. Then, when ib_umem_release() is called, it calls ib_umem_dmabuf_revoke() which will call dma_buf_unpin() again. Fix this by removing the immediate unpin upon failure and just let the ib_umem_release/revoke path handle it. This also ensures the proper unmap-unpin unwind ordering if the dmabuf_map_pages call happened to fail due to dma_resv_wait_timeout (and therefore has a non-NULL umem_dmabuf->sgt). Fixes: 1e4df4a21c5a ("RDMA/umem: Allow pinned dmabuf umem usage") Signed-off-by: Jacob Moroni <jmoroni@google.com> Link: https://patch.msgid.link/20260224234153.1207849-1-jmoroni@google.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-02-25RDMA/core: Check id_priv->restricted_node_type in cma_listen_on_dev()Stefan Metzmacher1-1/+5
When listening on wildcard addresses we have a global list for the application layer rdma_cm_id and for any existing device or any device added in future we try to listen on any wildcard listener. When the listener has a restricted_node_type we should prevent listening on devices with a different node type. While there fix the documentation comment of rdma_restrict_node_type() to include rdma_resolve_addr() instead of having rdma_bind_addr() twice. Fixes: a760e80e90f5 ("RDMA/core: introduce rdma_restrict_node_type()") Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: Leon Romanovsky <leon@kernel.org> Cc: Steve French <smfrench@gmail.com> Cc: Namjae Jeon <linkinjeon@kernel.org> Cc: Tom Talpey <tom@talpey.com> Cc: Long Li <longli@microsoft.com> Cc: linux-rdma@vger.kernel.org Cc: linux-cifs@vger.kernel.org Cc: samba-technical@lists.samba.org Signed-off-by: Stefan Metzmacher <metze@samba.org> Link: https://patch.msgid.link/20260224165951.3582093-2-metze@samba.org Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-02-24RDMA/ionic: Fix kernel stack leak in ionic_create_cq()Jason Gunthorpe1-1/+1
struct ionic_cq_resp resp { __u32 cqid[2]; // offset 0 - PARTIALLY SET (see below) __u8 udma_mask; // offset 8 - SET (resp.udma_mask = vcq->udma_mask) __u8 rsvd[7]; // offset 9 - NEVER SET <- LEAK }; rsvd[7]: 7 bytes of stack memory leaked unconditionally. cqid[2]: The loop at line 1256 iterates over udma_idx but skips indices where !(vcq->udma_mask & BIT(udma_idx)). The array has 2 entries but udma_count could be 1, meaning cqid[1] might never be written via ionic_create_cq_common(). If udma_mask only has bit 0 set, cqid[1] (4 bytes) is also leaked. So potentially 11 bytes leaked. Cc: stable@vger.kernel.org Fixes: e8521822c733 ("RDMA/ionic: Register device ops for control path") Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Link: https://patch.msgid.link/4-v1-83e918d69e73+a9-rdma_udata_rc_jgg@nvidia.com Acked-by: Abhijit Gangurde <abhijit.gangurde@amd.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-02-24RDMA/irdma: Fix kernel stack leak in irdma_create_user_ah()Jason Gunthorpe1-1/+1
struct irdma_create_ah_resp { // 8 bytes, no padding __u32 ah_id; // offset 0 - SET (uresp.ah_id = ah->sc_ah.ah_info.ah_idx) __u8 rsvd[4]; // offset 4 - NEVER SET <- LEAK }; rsvd[4]: 4 bytes of stack memory leaked unconditionally. Only ah_id is assigned before ib_respond_udata(). The reserved members of the structure were not zeroed. Cc: stable@vger.kernel.org Fixes: b48c24c2d710 ("RDMA/irdma: Implement device supported verb APIs") Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Link: https://patch.msgid.link/3-v1-83e918d69e73+a9-rdma_udata_rc_jgg@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-02-24IB/mthca: Add missed mthca_unmap_user_db() for mthca_create_srq()Jason Gunthorpe1-2/+3
Fix a user triggerable leak on the system call failure path. Cc: stable@vger.kernel.org Fixes: ec34a922d243 ("[PATCH] IB/mthca: Add SRQ implementation") Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Link: https://patch.msgid.link/2-v1-83e918d69e73+a9-rdma_udata_rc_jgg@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-02-24RDMA/efa: Fix typo in efa_alloc_mr()Jason Gunthorpe1-1/+1
The pattern is to check the entire driver request space, not just sizeof something unrelated. Fixes: 40909f664d27 ("RDMA/efa: Add EFA verbs implementation") Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Link: https://patch.msgid.link/1-v1-83e918d69e73+a9-rdma_udata_rc_jgg@nvidia.com Acked-by: Michael Margolin <mrgolin@amazon.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-02-24RDMA/ionic: Fix potential NULL pointer dereference in ionic_query_portKamal Heib1-0/+2
The function ionic_query_port() calls ib_device_get_netdev() without checking the return value which could lead to NULL pointer dereference, Fix it by checking the return value and return -ENODEV if the 'ndev' is NULL. Fixes: 2075bbe8ef03 ("RDMA/ionic: Register device ops for miscellaneous functionality") Signed-off-by: Kamal Heib <kheib@redhat.com> Link: https://patch.msgid.link/20260220222125.16973-2-kheib@redhat.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-02-24RDMA/bng_re: Unwind bng_re_dev_init properlySiva Reddy Kallam1-14/+15
Fix below smatch warning: drivers/infiniband/hw/bng_re/bng_dev.c:270 bng_re_dev_init() warn: missing unwind goto? Current bng_re_dev_init function is not having clear unwinding code. So, added proper unwinding with ladder. Fixes: 4f830cd8d7fe ("RDMA/bng_re: Add infrastructure for enabling Firmware channel") Reported-by: Simon Horman <horms@kernel.org> Reported-by: kernel test robot <lkp@intel.com> Reported-by: Dan Carpenter <error27@gmail.com> Closes: https://lore.kernel.org/r/202601010413.sWadrQel-lkp@intel.com/ Signed-off-by: Siva Reddy Kallam <siva.kallam@broadcom.com> Link: https://patch.msgid.link/20260218091246.1764808-3-siva.kallam@broadcom.com Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
2026-02-24RDMA/bng_re: Remove unnessary validity checksSiva Reddy Kallam1-23/+4
Fix below smatch warning: drivers/infiniband/hw/bng_re/bng_dev.c:113 bng_re_net_ring_free() warn: variable dereferenced before check 'rdev' (see line 107) current driver has unnessary validity checks. So, removing these unnessary validity checks. Fixes: 4f830cd8d7fe ("RDMA/bng_re: Add infrastructure for enabling Firmware channel") Fixes: 745065770c2d ("RDMA/bng_re: Register and get the resources from bnge driver") Fixes: 04e031ff6e60 ("RDMA/bng_re: Initialize the Firmware and Hardware") Fixes: d0da769c19d0 ("RDMA/bng_re: Add Auxiliary interface") Reported-by: Simon Horman <horms@kernel.org> Reported-by: kernel test robot <lkp@intel.com> Reported-by: Dan Carpenter <error27@gmail.com> Closes: https://lore.kernel.org/r/202601010413.sWadrQel-lkp@intel.com/ Signed-off-by: Siva Reddy Kallam <siva.kallam@broadcom.com> Link: https://patch.msgid.link/20260218091246.1764808-2-siva.kallam@broadcom.com Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
2026-02-24RDMA/core: Fix stale RoCE GIDs during netdev events at registrationJiri Pirko3-1/+49
RoCE GID entries become stale when netdev properties change during the IB device registration window. This is reproducible with a udev rule that sets a MAC address when a VF netdev appears: ACTION=="add", SUBSYSTEM=="net", KERNEL=="eth4", \ RUN+="/sbin/ip link set eth4 address 88:22:33:44:55:66" After VF creation, show_gids displays GIDs derived from the original random MAC rather than the configured one. The root cause is a race between netdev event processing and device registration: CPU 0 (driver) CPU 1 (udev/workqueue) ────────────── ────────────────────── ib_register_device() ib_cache_setup_one() gid_table_setup_one() _gid_table_setup_one() ← GID table allocated rdma_roce_rescan_device() ← GIDs populated with OLD MAC ip link set eth4 addr NEW_MAC NETDEV_CHANGEADDR queued netdevice_event_work_handler() ib_enum_all_roce_netdevs() ← Iterates DEVICE_REGISTERED ← Device NOT marked yet, SKIP! enable_device_and_get() xa_set_mark(DEVICE_REGISTERED) ← Too late, event was lost The netdev event handler uses ib_enum_all_roce_netdevs() which only iterates devices marked DEVICE_REGISTERED. However, this mark is set late in the registration process, after the GID cache is already populated. Events arriving in this window are silently dropped. Fix this by introducing a new xarray mark DEVICE_GID_UPDATES that is set immediately after the GID table is allocated and initialized. Use the new mark in ib_enum_all_roce_netdevs() function to iterate devices instead of DEVICE_REGISTERED. This is safe because: - After _gid_table_setup_one(), all required structures exist (port_data, immutable, cache.gid) - The GID table mutex serializes concurrent access between the initial rescan and event handlers - Event handlers correctly update stale GIDs even when racing with rescan - The mark is cleared in ib_cache_cleanup_one() before teardown This also fixes similar races for IP address events (inetaddr_event, inet6addr_event) which use the same enumeration path. Fixes: 0df91bb67334 ("RDMA/devices: Use xarray to store the client_data") Signed-off-by: Jiri Pirko <jiri@nvidia.com> Link: https://patch.msgid.link/20260127093839.126291-1-jiri@resnulli.us Reported-by: syzbot+881d65229ca4f9ae8c84@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=881d65229ca4f9ae8c84 Signed-off-by: Leon Romanovsky <leon@kernel.org>
2026-02-23RDMA/uverbs: select CONFIG_DMA_SHARED_BUFFERArnd Bergmann1-0/+1
The addition of dmabuf support in uverbs means that it is no longer possible to build infiniband support if that is disabled: arm-linux-gnueabi-ld: drivers/infiniband/core/ib_core_uverbs.o: in function `rdma_user_mmap_entry_remove.part.0': ib_core_uverbs.c:(.text+0x508): undefined reference to `dma_buf_move_notify' (dma_buf_move_notify): Unknown destination type (ARM/Thumb) in drivers/infiniband/core/ib_core_uverbs.o ib_core_uverbs.c:(.text+0x518): undefined reference to `dma_resv_wait_timeout' (dma_resv_wait_timeout): Unknown destination type (ARM/Thumb) in drivers/infiniband/core/ib_core_uverbs.o Select this from Kconfig, as we do for the other users. Fixes: 0ac6f4056c4a ("RDMA/uverbs: Add DMABUF object type and operations") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Link: https://patch.msgid.link/20260216121213.2088910-1-arnd@kernel.org Signed-off-by: Leon Romanovsky <leon@kernel.org>
generated by cgit v1.2.3 (git 2.39.0) at 2026-05-30 17:36:01 +0300