From nobody Sun Jun 14 18:59:06 2026 Received: from mail-wm1-f44.google.com (mail-wm1-f44.google.com [209.85.128.44]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 378272DECBA for ; Sat, 4 Apr 2026 23:08:37 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.128.44 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1775344118; cv=none; b=Jm/x2btX6ll3tR4ygUk6Ro0P4pAuqRjIplZCr5FzGrdQEE+CivRXL0XoW5676idNe+8mcz6ofDN8O9CRu8x3QoH15P46n1dAIS97g36pJ0Vdx6tJeSldVzv46AfTYEzz9yy8TGX+U1FsP5FfBQjQ0xDuWzlkIxzhYyilX7h5YIM= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1775344118; c=relaxed/simple; bh=xpg9Q/ksUK8DcRTOQbzD6p0gQtdWyK+r1oMqC0fUAnQ=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=Rxwfs7VbWltqst1lu5U54vOhkmWPL/Sllr0Qw18ELdwRLGzGKuck161JDfDGb1jFGkolYkeoeQLrN36poT4E7VTcxdM45Y2Y6DEaXwqmoqMo86xapStE690FHfUAvIsaPDYFCRxbKBUTpkRonQfjHy67nEGuoOez+4rlhHSvaVg= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com; spf=pass smtp.mailfrom=gmail.com; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b=ZrCrIgZN; arc=none smtp.client-ip=209.85.128.44 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=gmail.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b="ZrCrIgZN" Received: by mail-wm1-f44.google.com with SMTP id 5b1f17b1804b1-482f454be5bso42765275e9.0 for ; Sat, 04 Apr 2026 16:08:36 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20251104; t=1775344115; x=1775948915; darn=vger.kernel.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:from:to:cc:subject:date :message-id:reply-to; bh=0hZbwZb7uVS7mrdKS76KPEmzeWc8xmna+sIeGDuLXBQ=; b=ZrCrIgZN39uCemqCafF+Zin1EqzmquEAFoz45v8A+enRp+LEvm+H28ZDfXdaQWqYRG +bHtMICUuys2ewQckunmJkptA78yp1tBmGWxcNga1wSIDP0sIi9E0TbjeUf8QRUgxpO2 0DYvKBvUtdFZMMgKZpsmNvRhQ4grA2/eHyiJ1ZTtoPqXPj5N7qkWe2PkrAY9YmsjZDvJ RkKLkmR77fvlyM8VcrYOQ1/y7KOSruFjhYag+bSJ35Fbs2LUGgBW1JD4AbimWByzE//k 1yyquh3WzvcR98o3tz+UyVhb2i5uXbIL1DNVYoMROY0ISmEodBC57Uv8KecIYLKgGh45 gGgw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20251104; t=1775344115; x=1775948915; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-gg:x-gm-message-state:from :to:cc:subject:date:message-id:reply-to; bh=0hZbwZb7uVS7mrdKS76KPEmzeWc8xmna+sIeGDuLXBQ=; b=YxhOU0lVQPTH8v3JjRDQUGS11RtndDlHMWrz4kLcLFbqiDRqY7psrFt3G5+WnECQjN 2uw9lQcm3C2tvRhInpCo/ryrn6YrFGVnh/3iVCG8XhVpQixUZfGblBBDZifPF1kqeE9V vQhOEbbiPdP8lKGRYs3u7T4AVaD7yoPwP4vDGTlS+vZTAk6/2BLbKLC0g7P1pp5XlUF9 wsS4J8Rxo8XUEpZhfULaXV9ik8j+fDsn2gBzl4EO9bHpWf6AjCsiPHU778ZlmfO53QEo Vs//oN7Hbb8X/A/9hWvEmP4fiz2pSbHX+WBX3CqAsbMZEM6CpfVcJ9X0sfc69sibYnab rOzw== X-Forwarded-Encrypted: i=1; AJvYcCV6IxI8+jN7oxf+MsvbQUO5MWVxkdZzDvJB4Cjekdw112adNfDjo3mEHHWVFVedR+AM6TeoBcCFdKCBsAw=@vger.kernel.org X-Gm-Message-State: AOJu0YzxlqAzPHml9nPr+hRNUNAj+B01zwt5dohrQJQUdKaM1oqKo4j9 r018yV72nZDoLldNkQVFDUrg6Gg/eDxtb/rnIA4HmQsp/yesMEXRJW5x X-Gm-Gg: AeBDievu9hjG3KcKvTYyhsnsRgkS1MLHO5ejUTci65mfYYJ9qZht4/QdsIinph4PYPk YdCNh+HzLw/HOoc3Zilt7xD9OepAqQXJFXyA8ilhdlPkub74k5dKTOXc/B9SRx7soLlX797J8Xi 0xEHaVxYX56uXVvy6kDx65V2C6KBYyGg+YifR0TmgP3IS9hLcm1Wh0EgaQ8VBbkYMNIur81yr2r VXcdDsxhzi5e4BTuUE9q+2J2jcqG51O0+oyoiYibX5/fiEIe6X9HqP3dxHABmsubyoyVXdwfzOb thhe8fEjRSIP5VB7xCqPoN8Kcd4/C9KZWsROFmNkHXdqY8dElRafbZ6SjhcUxREQjNpZy6m0f60 u5joUmeO2U8qiEl1rkCU6LciJZB8x080exvnyfeZFFj2P+0AeCSkHxaRQRLw1AjeXmodKFCChia BXuYhvb+PNqH9TDwYw9682JajLPCu1l/ide3WWCjqYrihgAWy6jen5Id5GdwPD9ld+B34el0jop trTZ2YDLJ/aC/UFkrXw7MMVqy48E7u3YSO7F3PU0uXz37On X-Received: by 2002:a05:600c:5289:b0:487:22ad:403e with SMTP id 5b1f17b1804b1-488994b34b4mr127115975e9.14.1775344115272; Sat, 04 Apr 2026 16:08:35 -0700 (PDT) Received: from DESKTOP-NQ2T5I7.localdomain (heme-13-b2-v4wan-167795-cust403.vm32.cable.virginm.net. [81.108.45.148]) by smtp.gmail.com with ESMTPSA id 5b1f17b1804b1-488a3d6944bsm82743405e9.11.2026.04.04.16.08.33 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Sat, 04 Apr 2026 16:08:34 -0700 (PDT) From: Prathamesh Deshpande To: prathameshdeshpande7@gmail.com Cc: dledford@redhat.com, haggaie@mellanox.com, jgg@ziepe.ca, leon@kernel.org, linux-kernel@vger.kernel.org, linux-rdma@vger.kernel.org Subject: [PATCH v6] IB/mlx5: Fix loopback enablement state and resource leaks Date: Sun, 5 Apr 2026 00:07:58 +0100 Message-ID: <20260404230759.15131-1-prathameshdeshpande7@gmail.com> X-Mailer: git-send-email 2.43.0 In-Reply-To: <20260404215155.13254-1-prathameshdeshpande7@gmail.com> References: <20260404215155.13254-1-prathameshdeshpande7@gmail.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" In mlx5_ib_alloc_transport_domain(), an early success path was incorrectly returning a variable instead of a literal 0. Additionally, the loopback (LB) infrastructure required updates to address initialization and concurrency issues: 1. Initialization: The LB mutex was initialized conditionally, but paths like create_raw_packet_qp_tir() in qp.c call LB functions regardless of those conditions. Move initialization to stage_init_init() to ensure the lock is always valid. 2. State Integrity: Update mlx5_ib_enable_lb() to roll back reference counters if the hardware command fails, ensuring software state matches hardware. 3. Concurrency: Move 'force_enable' checks inside the mutex in both enable/disable paths to prevent potential counter underflows. 4. Resource Leak: Deallocate the transport domain if LB enablement fails in mlx5_ib_alloc_transport_domain(). Signed-off-by: Prathamesh Deshpande --- v6: - Always update software counters regardless of force_enable to prevent=20 underflows during dynamic unbinding [Sashiko]. - Updated disable condition to user_td <=3D 1 to prevent HW state leaks=20 on systems without transport domains [Sashiko]. - Rebased on rdma/for-next to resolve baseline application failures. v5: - Moved mutex_init to stage_init_init to prevent crashes on non-ETH hardwar= e. - Implemented 'goto unlock' for concurrency safety in enable/disable paths. - Added atomic rollback and fixed tdn leak. v4: - Moved rollback logic into mlx5_ib_enable_lb() to ensure atomicity within the mutex and prevent race conditions [Sashiko]. v3: - Also call mlx5_ib_disable_lb() on failure to roll back software state/cou= nters [Sashiko]. v2: - Added deallocation of tdn if mlx5_ib_enable_lb() fails [Sashiko]. - Reworded commit message to reflect the functional fix and credit the tool. drivers/infiniband/hw/mlx5/main.c | 39 +++++++++++++++++++------------ 1 file changed, 24 insertions(+), 15 deletions(-) diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5= /main.c index b74bf2697655..c2fafa00b60c 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -2009,23 +2009,31 @@ int mlx5_ib_enable_lb(struct mlx5_ib_dev *dev, bool= td, bool qp) { int err =3D 0; =20 - if (dev->lb.force_enable) - return 0; - mutex_lock(&dev->lb.mutex); if (td) dev->lb.user_td++; if (qp) dev->lb.qps++; =20 - if (dev->lb.user_td =3D=3D 2 || + if (dev->lb.force_enable) + goto unlock; + + if (dev->lb.user_td =3D=3D 1 || dev->lb.qps =3D=3D 1) { if (!dev->lb.enabled) { err =3D mlx5_nic_vport_update_local_lb(dev->mdev, true); - dev->lb.enabled =3D true; + if (err) { + if (td) + dev->lb.user_td--; + if (qp) + dev->lb.qps--; + } else { + dev->lb.enabled =3D true; + } } } =20 +unlock: mutex_unlock(&dev->lb.mutex); =20 return err; @@ -2033,16 +2041,16 @@ int mlx5_ib_enable_lb(struct mlx5_ib_dev *dev, bool= td, bool qp) =20 void mlx5_ib_disable_lb(struct mlx5_ib_dev *dev, bool td, bool qp) { + mutex_lock(&dev->lb.mutex); if (dev->lb.force_enable) - return; + goto unlock; =20 - mutex_lock(&dev->lb.mutex); if (td) dev->lb.user_td--; if (qp) dev->lb.qps--; =20 - if (dev->lb.user_td =3D=3D 1 && + if (dev->lb.user_td <=3D 1 && dev->lb.qps =3D=3D 0) { if (dev->lb.enabled) { mlx5_nic_vport_update_local_lb(dev->mdev, false); @@ -2050,6 +2058,7 @@ void mlx5_ib_disable_lb(struct mlx5_ib_dev *dev, bool= td, bool qp) } } =20 +unlock: mutex_unlock(&dev->lb.mutex); } =20 @@ -2068,9 +2077,13 @@ static int mlx5_ib_alloc_transport_domain(struct mlx= 5_ib_dev *dev, u32 *tdn, if ((MLX5_CAP_GEN(dev->mdev, port_type) !=3D MLX5_CAP_PORT_TYPE_ETH) || (!MLX5_CAP_GEN(dev->mdev, disable_local_lb_uc) && !MLX5_CAP_GEN(dev->mdev, disable_local_lb_mc))) - return err; + return 0; + + err =3D mlx5_ib_enable_lb(dev, true, false); + if (err) + mlx5_cmd_dealloc_transport_domain(dev->mdev, *tdn, uid); =20 - return mlx5_ib_enable_lb(dev, true, false); + return err; } =20 static void mlx5_ib_dealloc_transport_domain(struct mlx5_ib_dev *dev, u32 = tdn, @@ -4515,6 +4528,7 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_dev= *dev) mutex_init(&dev->data_direct_lock); INIT_LIST_HEAD(&dev->qp_list); spin_lock_init(&dev->reset_flow_resource_lock); + mutex_init(&dev->lb.mutex); xa_init(&dev->odp_mkeys); xa_init(&dev->sig_mrs); atomic_set(&dev->mkey_var, 0); @@ -4786,11 +4800,6 @@ static int mlx5_ib_stage_caps_init(struct mlx5_ib_de= v *dev) if (err) return err; =20 - if ((MLX5_CAP_GEN(dev->mdev, port_type) =3D=3D MLX5_CAP_PORT_TYPE_ETH) && - (MLX5_CAP_GEN(dev->mdev, disable_local_lb_uc) || - MLX5_CAP_GEN(dev->mdev, disable_local_lb_mc))) - mutex_init(&dev->lb.mutex); - if (MLX5_CAP_GEN_64(dev->mdev, general_obj_types) & MLX5_GENERAL_OBJ_TYPES_CAP_VIRTIO_NET_Q) { err =3D mlx5_ib_init_var_region(dev); --=20 2.43.0