From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?UTF-8?B?U8OpYmFzdGllbiBCb3VyZGVhdWR1Y3E=?= Subject: "Timeout waiting for hardware interrupt" on Lenovo Thinkpad Tablet 10 Date: Mon, 02 Feb 2015 20:29:51 +0800 Message-ID: <54CF6DBF.4080404@m-labs.hk> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from ssl.serverraum.org ([88.198.40.39]:38026 "EHLO ssl.serverraum.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754230AbbBBMmK (ORCPT ); Mon, 2 Feb 2015 07:42:10 -0500 Received: from localhost (localhost [127.0.0.1]) by ssl.serverraum.org (Postfix) with ESMTP id 382A63F12F for ; Mon, 2 Feb 2015 13:34:53 +0100 (CET) Received: from ssl.serverraum.org ([127.0.0.1]) by localhost (ssl.serverraum.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Cim7wZaLmQ2r for ; Mon, 2 Feb 2015 13:34:53 +0100 (CET) Received: from [192.168.0.14] (123202036015.ctinets.com [123.202.36.15]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ssl.serverraum.org (Postfix) with ESMTPSA id 4E1973F12E for ; Mon, 2 Feb 2015 13:34:51 +0100 (CET) Sender: linux-mmc-owner@vger.kernel.org List-Id: linux-mmc@vger.kernel.org To: linux-mmc@vger.kernel.org Hi, Under Linux, the most annoying problem that the Lenovo Thinkpad Tablet 10 has is the unreliable eMMC. I intermittently get an error "mmc0: Timeout waiting for hardware interrupt" followed by a crash of the machine (when the eMMC is used as rootfs). I have been able to work around this issue with kernel 3.13.0 by passin= g the debug_quirks2=3D0xc0 option to the sdhci module, which disables the HS200 and DDR50 transfer modes. The read data rate (as measured by hdparm -t) drops from 119MB/s to 42MB/s, but at least the computer no longer crashes every 20 minutes and my files are not corrupted anymore. Unfortunately, with more recent kernels such as 3.18.4, the sdhci modul= e (still with debug_quirks2=3D0xc0) appears to select another problematic transfer mode. The read data rate jumps to 65MB/s and the bug manifests itself again. What changes in sdhci between 3.13.0 and 3.18.4 could cause that? I'd like to be able to use the eMMC reliably at least at 42MB/s using recen= t kernels, which hopefully will introduce fixes for the gazillion other problems that this crippled machine has under Linux (absolutely nothing works correctly except the CPU, display and USB). Thanks, S=C3=A9bastien