From mboxrd@z Thu Jan 1 00:00:00 1970 Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.remote-ip=100.103.45.18 ARC-Seal: i=1; d=subspace.kernel.org; s=arc20261015; a=rsa-sha256; cv=none; t=1795010655; b=YxLbRV/lHhbSy7dIFqU6i8R0RB5M7JRsNeGOY5Opgz7h8yRTc12yYBECo5hUSqDKv7o+ lm6AQ8RorrxfgxPFMNPVRLXPAgJzOEtgT02wuKwzukpOn0sWM0yXs865ZvoLSIl6H7vpE C6CI86vO7btOXljwEkP9c8Alk/nbteczj295u935vSWtIJj/P3sTwDNdXlq+rRqmuS/zW 5hVras92lEpJRyTkJ9IQkVLu6ZsLWoNZbZRDFzgN286/Tiq5hljDWJhyjSMCtZOoRXw4L 5WLOzAcZzHjm2md/2P2WW9pivutcGhgImHQPGkzjNoWlsGmHt62x51+6A7FQxOcU+fw== ARC-Message-Signature: i=1; d=subspace.kernel.org; s=arc20261015; a=rsa-sha256; c=relaxed/relaxed; t=1795010655; h=Received:DKIM-Signature:Message-ID:From:Subject:Reply-To:To:Cc: In-Reply-To:References:Content-Type:Content-Transfer-Encoding:Date; bh=EQOQZFrkvwo1zsHF/5mHhzSx5Q7iG+U/S/puwp3Egu8=; b=nTujE9xtlYm4hAzirbffPcVXJLERcPGPh2r6L9ipUNI2lhjlCDcXVex9K/In9o2Te9aD Y5wgkHAswY5Ta72AZdObbks9PuwcA2wAnK1AAnxC02cuUVRbV1QzZye77tNQdiU/qYbD5 p9c9LtjPB/P/72ZkcFOWrKOAZgLAyvLiUIsH3PBKe8cj/pt3NfBPVlQncUlau1lqau92v nycuklNJfpymmff+2fs6oH+roVpOVn2ALf6UEnMsc0uo0QEZjLMwWdUUAH3tb3/iRKd3j IXgXG6u+OgeKccBz2KDIS7dGs1KVwsE6oVd2179lNJM8uRh1Lz/51JTliIRWpafxSKg== ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dkim=pass header.d=kernel.org header.i=@kernel.org header.a=rsa-sha256 header.s=k20260515 header.b=YrV2fjNT; arc=none smtp.remote-ip=100.103.45.18 Received: from smtp.kernel.org (quasi.space.kernel.org [100.103.45.18]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature ECDSA (prime256v1)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 87E01606A320 for ; Wed, 18 Nov 2026 14:04:14 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.a=rsa-sha256 header.s=k20260515 header.b=YrV2fjNT Received: by smtp.kernel.org (Postfix) with ESMTPSA id A65F61F000C4; Wed, 18 Nov 2026 14:04:14 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=kernel.org; s=k20260515; t=1795010654; bh=EQOQZFrkvwo1zsHF/5mHhzSx5Q7iG+U/S/puwp3Egu8=; h=From:Subject:Reply-To:To:Cc:In-Reply-To:References:Date; b=YrV2fjNTAvpiNz6XONrxESFJQdHVIkRzd6cjyv/n0olc7tdYqbQdnDdkcBL3gHW9o cEr6N/sGmVTwLIyESuwdl+1pXEyhoGddzyPYTz+HV00HEM2XpODfQ+WYwPXnPienUQ bpdPLESBxd+ZGK7FlnI5P6CNZ0VbfDYBTV2LZnbtwsnhLAw6JOrwBgpbb2LgijtYdY Y+KbimiNyOWXxoRVOVSDgP+LKiZJ7VizXfwornP8i2HL0evg1tGLHF/nxslP76PctL DHn9aUCwNKpwZwLyDk6XqW6WjVJM3cuE7HTfFM7zTvFspgsxkRkljQj5EqWKHJ7aJp 3k3jV5ST2jBng== Message-ID: From: sashiko-bot@kernel.org Subject: Re: [PATCH rdma-next 01/11] RDMA/mlx5: Enforce dev/mdev naming for device pointers Reply-To: sashiko-reviews@lists.linux.dev To: "Edward Srouji" Cc: linux-rdma@vger.kernel.org In-Reply-To: <20261117-mlx5-sd-prep-v1-1-8efd124c7895@nvidia.com> References: <20261117-mlx5-sd-prep-v1-0-8efd124c7895@nvidia.com> <20261117-mlx5-sd-prep-v1-1-8efd124c7895@nvidia.com> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Date: Wed, 18 Nov 2026 14:04:14 +0000 X-Mailing-List: linux-rdma@vger.kernel.org List-Id: List-Post: List-Help: List-Subscribe: List-Unsubscribe: List-Owner: Precedence: list Thank you for your contribution! Sashiko AI review found no new issues in this patch, but noticed 2 related pre-existing issues it does not introduce: - [Medium] RDMA/mlx5: out-of-bounds read in devx_get_obj_id() on short commands [1] - [High] RDMA/mlx5: deadlock in mlx5_poll_one() [2] -- commit 3bdcdff9fe049e25d51894257ec6c6716c3a9770 Author: Akiva Goldberger RDMA/mlx5: Enforce dev/mdev naming for device pointers This commit renames variables, parameters, and struct fields across mlx5_ib= so that dev exclusively refers to struct mlx5_ib_dev pointers and mdev exclusively refers to struct mlx5_core_dev pointers. > diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/ml= x5/devx.c > index fc6f793dcc658..7e84dd15ac1f9 100644 > --- a/drivers/infiniband/hw/mlx5/devx.c > +++ b/drivers/infiniband/hw/mlx5/devx.c [ ... ] > @@ -1652,7 +1652,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_M= ODIFY)( > MLX5_IB_ATTR_DEVX_OBJ_MODIFY_HANDLE); > struct mlx5_ib_ucontext *c =3D rdma_udata_to_drv_context( > &attrs->driver_udata, struct mlx5_ib_ucontext, ibucontext); > - struct mlx5_ib_dev *mdev =3D to_mdev(c->ibucontext.device); > + struct mlx5_ib_dev *dev =3D to_mdev(c->ibucontext.device); > void *cmd_out; > int err, err2; > int uid; > @@ -1677,7 +1677,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_M= ODIFY)( > MLX5_SET(general_obj_in_cmd_hdr, cmd_in, uid, uid); > devx_set_umem_valid(cmd_in); > =20 > - err =3D mlx5_cmd_do(mdev->mdev, cmd_in, > + err =3D mlx5_cmd_do(dev->mdev, cmd_in, > uverbs_attr_get_len(attrs, MLX5_IB_ATTR_DEVX_OBJ_MODIFY_CMD_IN), > cmd_out, cmd_out_len); [Severity: Medium] This problem wasn't introduced by this patch, but can devx_get_obj_id() read past the end of cmd_in [1] when called from MLX5_IB_METHOD_DEVX_OBJ_MODIFY, MLX5_IB_METHOD_DEVX_OBJ_QUERY, or MLX5_IB_METHOD_DEVX_OBJ_ASYNC_QUERY? The uverbs attribute spec only enforces a 16-byte minimum (general_obj_in_cmd_hdr) on cmd_in, and these handlers call devx_is_valid_obj_id() -> devx_get_obj_id() without checking that the buffer is large enough for the opcode-specific object ID field. For an opcode such as MLX5_CMD_OP_MODIFY_FLOW_TABLE, reading table_id in modify_flow_table_in = at byte offset 20 performs a 4-byte out-of-bounds read when userspace supplies= a 16-byte cmd_in. [ ... ] > diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/ml= x5/main.c > index 65455aedd559b..1ecd79929f8cf 100644 > --- a/drivers/infiniband/hw/mlx5/main.c > +++ b/drivers/infiniband/hw/mlx5/main.c [ ... ] > @@ -4483,7 +4483,7 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_d= ev *dev) > atomic_set(&dev->mkey_var, 0); > =20 > spin_lock_init(&dev->dm.lock); > - dev->dm.dev =3D mdev; > + dev->dm.mdev =3D mdev; [Severity: High] This problem wasn't introduced by this patch, but can dev->sig_mrs deadlock in mlx5_poll_one() [2] because mlx5_ib_stage_init_init() initializes it with xa_init() instead of xa_init_flags(..., XA_FLAGS_LOCK_IRQ)? Process-context callers such as mlx5_alloc_integrity_descs() and __mlx5_ib_dereg_mr() take xa_lock on dev->sig_mrs via xa_store() or xa_erase() with local interrupts enabled, while mlx5_poll_one() takes the same xa_lock from CQ interrupt context on a MLX5_CQE_SIG_ERR completion: mlx5_alloc_integrity_descs() xa_store(&dev->sig_mrs, ...) spin_lock(&xa->xa_lock) ... mlx5_poll_one() xa_lock(&dev->sig_mrs) <-- spins forever [1] https://sashiko.dev/#/bug/linux-2b4a0b34-fea8-4d2a-87be-350c461b8f4f [2] https://sashiko.dev/#/bug/linux-4a13b647-d8f7-45a3-8a1e-c17db57fefbb --=20 Sashiko AI review =C2=B7 https://sashiko.dev/#/patchset/20261117-mlx5-sd-pr= ep-v1-0-8efd124c7895@nvidia.com?part=3D1