From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-0.7 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id C355CC31E49 for ; Wed, 19 Jun 2019 09:48:52 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id A34FA2084A for ; Wed, 19 Jun 2019 09:48:52 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1731579AbfFSJsv (ORCPT ); Wed, 19 Jun 2019 05:48:51 -0400 Received: from szxga07-in.huawei.com ([45.249.212.35]:43392 "EHLO huawei.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1726958AbfFSJsu (ORCPT ); Wed, 19 Jun 2019 05:48:50 -0400 Received: from DGGEMS406-HUB.china.huawei.com (unknown [172.30.72.60]) by Forcepoint Email with ESMTP id 9CE4EEB9E44137C3E136; Wed, 19 Jun 2019 17:48:45 +0800 (CST) Received: from [127.0.0.1] (10.202.227.238) by DGGEMS406-HUB.china.huawei.com (10.3.19.206) with Microsoft SMTP Server id 14.3.439.0; Wed, 19 Jun 2019 17:48:37 +0800 Subject: PCI host bridge hotplug test question (was Re: [PATCH v2] bus: hisi_lpc: Don't use devm_kzalloc() to allocate logical PIO range) To: Bjorn Helgaas References: <1560507893-42553-1-git-send-email-john.garry@huawei.com> <82840955-6365-0b95-6d69-8a2f7c7880af@huawei.com> <9e8b6971-3189-9d4b-de9a-ff09f859f4f6@huawei.com> <539835d3-770c-285c-0c49-ae15ceaa3079@huawei.com> CC: , , Linux Kernel Mailing List , Joe Perches , , "Rafael J. Wysocki" , "Mika Westerberg" , Jiang Liu , "liudongdong (C)" From: John Garry Message-ID: <3ada51e7-7cb9-e1e8-7ab7-859799232aa5@huawei.com> Date: Wed, 19 Jun 2019 10:48:32 +0100 User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:45.0) Gecko/20100101 Thunderbird/45.3.0 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset="utf-8"; format=flowed Content-Transfer-Encoding: 7bit X-Originating-IP: [10.202.227.238] X-CFilter-Loop: Reflected Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 18/06/2019 18:50, Bjorn Helgaas wrote: > [+cc Rafael, Mika, Jiang, linux-pci for ACPI host bridge hotplug test question] > > On Tue, Jun 18, 2019 at 3:44 AM John Garry wrote: > >>>>> Could you just move the logic_pio_register_range() call farther down >>>>> in hisi_lpc_probe()? IIUC, once logic_pio_register_range() returns, >>>>> an inb() with the right port number will try to access that port, so >>>>> we should be prepared for that, i.e., maybe this in the wrong order to >>>>> begin with? >>>> >>>> No, unfortunately we can't. The reason is that we need the logical PIO >>>> base for that range before we enumerate the children of that host. We >>>> need that base address for "translating" the child bus addresses to >>>> logical PIO addresses. > >>> Ah, yeah, that makes sense. I think. We do assume that we know all >>> the MMIO and I/O port translations before enumerating devices. It's >>> *conceivable* that could be changed someday since we don't actually >>> need the translations until a driver claims the device, >> >> We actually need them before a driver claims the device. >> >> The reason is that when we create that child platform device we set the >> device's IORESOURCE_IO resources according to the translated logic PIO >> addresses, and not the host bus address. This is what makes the host >> transparent to the child device driver. > > I think you need it to set pdev->resource[], which is currently done > long before the driver claims the device (though one could imagine > delaying it even as far as pci_enable_device()-time). I don't think > the translation is actually *used* until the driver claims the device > because only the driver knows how to do any inb/outb to the device. > > But of course, that's all speculative and doesn't change what you need > to do now. The current code assumes we know the translations during > enumeration, so you need to do the logic_pio registration before > enumerating. > >>> and it would >>> gain some flexibility if we didn't have to program the host bridge >>> windows until we know how much space is required. But I don't see >>> that happening anytime soon. > >> My problem is that I need to ensure that the new logical PIO unregister >> function works ok for hot-pluggable host bridges. I need to get some way >> to test this. Advice? > Hi Bjorn, > Good question. The ACPI host bridge driver (drivers/acpi/pci_root.c) > should support hotplug, but I'm not sure if there's a manual way to > trigger it via sysfs or something similar. If there is, and you have > a machine with more than one host bridge, you might be able to remove > one that leads to non-essential devices. For one of our earlier boards I don't think that it had any essential devices on the host bridge. But I need to find out about possibility of removal. Hmmm. > > Bjorn > Further to the topic of supporting hotplug and unregistering IO port regions, we don't even release IO port regions in the error path of PCI host enumeration. We have pci_register_io_range(), but no unregister equivalent. Looking at the history here, pci_register_io_range() was originally in OF code. And in the OF code, calling pci_register_io_range() is a side-effect of parsing the device tree. As such, I can see why there was no unregister function. It would be worth noting this discussion, where the same was mentioned: https://lore.kernel.org/linux-pci/20180403140410.GE27789@ulmo/ The tegra PCI host probe can defer, but, since there is no tidy-up of pci_register_io_range() when deferring, we need to ensure that the port IO management code can handle re-attempts to register the same range. It looks like this can be cleaned up also. Thanks, John > . >