From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932311AbcELIwG (ORCPT ); Thu, 12 May 2016 04:52:06 -0400 Received: from hqemgate14.nvidia.com ([216.228.121.143]:17425 "EHLO hqemgate14.nvidia.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932088AbcELIwE convert rfc822-to-8bit (ORCPT ); Thu, 12 May 2016 04:52:04 -0400 X-PGP-Universal: processed; by hqnvupgp08.nvidia.com on Thu, 12 May 2016 01:51:08 -0700 From: Aniruddha Banerjee To: "levinsasha928@gmail.com" , "akpm@linux-foundation.org" , "msb@google.com" CC: "linux-kernel@vger.kernel.org" Subject: [RFC] hung_task: task white-listing for hung tasks Thread-Topic: [RFC] hung_task: task white-listing for hung tasks Thread-Index: AdGsKwBltX1rneXDRO2Nc7MnEAwz1Q== Date: Thu, 12 May 2016 08:51:58 +0000 Message-ID: <9a25d3183d2e4c99aa942d6beeb151f4@bgmail102.nvidia.com> Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-ms-exchange-transport-fromentityheader: Hosted x-originating-ip: [10.24.40.251] MIME-Version: 1.0 Content-Language: en-US Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hello, In the current hung_task framework, there is no way to white-list tasks for which a higher hang time out might be acceptable and expected even: Usecase: I want to keep a low hung_task timeout to catch more tasks which are hung-up. However, certain tasks (like sync, which flushes to a slow eMMC) may have a higher timeout. Problem: There is no way to implement the aforementioned usecase without increasing the hang task timeout, which will reduce the effictiveness of the hung_task framework. Solution: 1. Introduce a white-list of tasks that have a much higher timeout. 2. Create a simple sysfs interface with which it might be possible to dynamically add/remove any task in the white-list. Please let me know what you think about the problem and the proposed solution? Regards, Aniruddha