From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from www.linutronix.de ([62.245.132.108]:44586 "EHLO Galois.linutronix.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752602Ab3LBNnu (ORCPT ); Mon, 2 Dec 2013 08:43:50 -0500 Received: from bigeasy by Galois.linutronix.de with local (Exim 4.72) (envelope-from ) id 1VnTmv-0005NV-EJ for linux-pci@vger.kernel.org; Mon, 02 Dec 2013 14:43:49 +0100 Date: Mon, 2 Dec 2013 14:43:49 +0100 From: Sebastian Andrzej Siewior To: linux-pci@vger.kernel.org Subject: Could enabling MSI interrupts lead to problems? Message-ID: <20131202134349.GA19436@linutronix.de> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Sender: linux-pci-owner@vger.kernel.org List-ID: I've been playing with PCIe parallel port cards. Both of them have the MSI flag set and it is possible to enable MSI interrupts. The advantage is that the interrupt is no longer shared with other devices and it becomes edge instead of level. Now I'm thinking about posting a patch which always calls pci_enable_msi() on init so we have MSI support in parport_pc driver if possible. I know that atleast XHCI has a flag not to enable MSI on some FRESCO_LOGIC devices because they announce it but can't generate interrupts with MSI enabled. Is this something "usual" and I should only enable MSI based on a whitelist or is this more something exceptional? Sebastian