From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 85B642FA0C6; Wed, 4 Mar 2026 01:26:14 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=10.30.226.201 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1772587574; cv=none; b=KR+e9CTTjUAtD3QxxfeDbBhUFY95J2qW6NeWyFJgMx6RIMzZM3e0FMiXEcNEmmSmI4U+W9uDgxQtSSD9xkO/T0CDC39J8+aIxtU6JthqmovSl0qzMHSMxjM6eQiK1hcx10YPrvDIfRrtHP6D17Ouih2qe9k3NZLo3pFONGlsxhQ= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1772587574; c=relaxed/simple; bh=/GzBSdollLP1oWlbirW3cObED2Mti3E2VhxF9n4062g=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=hk45eamosnBAJDm8Rhz/Cly6zOPqETdOIgNznJcLMWHedQr7z/UEzAxESc9C41+7Ovrr1TgoQ5UeqN4EXeX375n5cO5BuVsf3MK+wEyjatXRYWbgWTH2UQm0zYoj3NGixGL2UZga62OwbiGI8JHXM0+YXY0lJpF7nKGR8Vahcho= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=nPLs/puc; arc=none smtp.client-ip=10.30.226.201 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="nPLs/puc" Received: by smtp.kernel.org (Postfix) with ESMTPSA id A4989C116C6; Wed, 4 Mar 2026 01:26:13 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1772587574; bh=/GzBSdollLP1oWlbirW3cObED2Mti3E2VhxF9n4062g=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=nPLs/puc0eofNR6Zb9n8ISui3nCYsj5pbcSGS9n79ziK/iAeE5oBDtQ51emrWEkiz U3TtwyOCKIpnX1JdOzDbPMNpC76P0WF3NfHu8kcscI1vzi00datHkKf7SEV4n+Iv4x UCLQ5tUmSEixG1WQ2UphsDB4atqZuWZ9u/zFtJ1g4qNDaKw1Cx0d2itc2x3J+d8Msx zwcwMJR9YXK+zT8uAkd6IsR8rhWqC6pPvngIlTb5iUVKrtWLybq5NbMkPvIuMuZj/q FLApbiaNv45AsWG7ceHdGoepd2n5tSCO8U0QzP4g8A8oKLwgvj75Izfux8xKv83Qld CU65NPgkGU6fA== From: Jakub Kicinski To: shuangpeng.kernel@gmail.com Cc: Jakub Kicinski , davem@davemloft.net, edumazet@google.com, linux-kernel@vger.kernel.org, pabeni@redhat.com, netdev@vger.kernel.org, horms@kernel.org, jiayuan.chen@shopee.com, jirislaby@kernel.org, andrew+netdev@lunn.ch, gregkh@linuxfoundation.org, shaojijie@huawei.com Subject: Re: [net,v4] serial: caif: hold tty->link reference in ldisc_open and ser_release Date: Tue, 3 Mar 2026 17:26:12 -0800 Message-ID: <20260304012612.2592839-1-kuba@kernel.org> X-Mailer: git-send-email 2.53.0 In-Reply-To: <20260301220525.1546355-1-shuangpeng.kernel@gmail.com> References: <20260301220525.1546355-1-shuangpeng.kernel@gmail.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit This is an AI-generated review of your patch. The human sending this email has considered the AI review valid, or at least plausible. For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html --- serial: caif: hold tty->link reference in ldisc_open and ser_release This commit addresses a KASAN slab-use-after-free in pty_write_room() triggered when caif_serial's TX path calls tty_write_room(). The fix holds an extra reference on tty->link throughout the lifetime of the caif_serial line discipline. Link: https://lore.kernel.org/all/20260228094741.1e248271@kernel.org/ Link: https://gist.github.com/shuangpengbai/c898debad6bdf170a84be7e6b3d8707f The Fixes tag points to commit 308e7e4d0a84 ("serial: caif: fix use-after-free in caif_serial ldisc_close()"), but the regression was actually introduced by commit e31d5a05948e ("caif: tty's are kref objects so take a reference"). That commit added tty_kref_get(tty) to hold a reference on the tty, but failed to also hold a reference on tty->link, which is accessed via tty_write_room() in handle_tx(). Commit 308e7e4d0a84 merely exposed this existing regression by changing the timing. Should the Fixes tag be updated to: Fixes: e31d5a05948e ("caif: tty's are kref objects so take a reference") > diff --git a/drivers/net/caif/caif_serial.c b/drivers/net/caif/caif_serial.c > index b90890030751..1873d8287bb9 100644 > --- a/drivers/net/caif/caif_serial.c > +++ b/drivers/net/caif/caif_serial.c > @@ -297,6 +297,7 @@ static void ser_release(struct work_struct *work) > dev_close(ser->dev); > unregister_netdevice(ser->dev); > debugfs_deinit(ser); > + tty_kref_put(tty->link); ^^^^^^^^^^^^^^^^^^^^ Can this dereference NULL? The tty->link field is only set for PTY (pseudo-terminal) pairs. For regular serial ports and other non-PTY devices, tty->link is NULL. The CAIF line discipline can be attached to any TTY device via the TIOCSETD ioctl. When ser_release() is called to clean up a ser_device that was created for a non-PTY TTY device, tty->link would be NULL. Should this check for NULL before calling tty_kref_put()? > tty_kref_put(tty); > } > rtnl_unlock(); > @@ -331,6 +332,7 @@ static int ldisc_open(struct tty_struct *tty) > > ser = netdev_priv(dev); > ser->tty = tty_kref_get(tty); > + tty_kref_get(tty->link); ^^^^^^^^^^^^^^^^^^^^^^^^ Can this dereference NULL? When a user attaches the CAIF line discipline to a non-PTY TTY device via TIOCSETD, ldisc_open() is called with a tty where tty->link is NULL. While ldisc_open() performs capability checks (CAP_SYS_ADMIN), there is no code restricting this to PTY devices only. Should this check for NULL before calling tty_kref_get()? > ser->dev = dev; > debugfs_init(ser, tty); > tty->receive_room = 4096; > @@ -339,6 +341,7 @@ static int ldisc_open(struct tty_struct *tty) > rtnl_lock(); > result = register_netdevice(dev); > if (result) { > + tty_kref_put(tty->link); ^^^^^^^^^^^^^^^^^^^^^^^^ Can this dereference NULL in the error path? If register_netdevice() fails after the CAIF line discipline was attached to a non-PTY TTY device, the error path would attempt to release a reference on tty->link when it is NULL. Should this check for NULL before calling tty_kref_put()? > tty_kref_put(tty); > rtnl_unlock(); > free_netdev(dev); Simon Horman raised a concern in v2 about a NULL pointer dereference in update_tty_status(). The v4 patch does not use the spinlock approach from v2, and does not address this race condition. After ser_release() sets ser->tty = NULL, a concurrent handle_tx() that already passed the NULL check can reach update_tty_status(ser). Looking at update_tty_status(): drivers/net/caif/caif_serial.c:update_tty_status() { inline_decode(ser->common.flowctrl, ser->tty->flow.stopped << 5 | ser->tty->flow.tco_stopped << 3 | ser->tty->ctrl.packet << 2); } It dereferences ser->tty directly (ser->tty->flow.stopped, ser->tty->flow.tco_stopped, ser->tty->ctrl.packet). Can this cause a NULL pointer dereference? Simon also raised a concern about potential deadlock with spinlock usage in v2. The v4 patch removes the spinlock approach entirely. handle_tx() is called from caif_xmit(), which is the ndo_start_xmit handler. This can run in softirq context (NET_TX softirq). ser_release() runs in process context (workqueue). How does the current synchronization between handle_tx() and ser_release() prevent races? -- pw-bot: cr