From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sinan Kaya Subject: Re: [PATCH V2] PCI: Do not enable extended tags on pre-dated (v1.x) systems Date: Mon, 10 Jul 2017 20:20:49 -0400 Message-ID: <5f5119fd-64d5-5e7c-fa50-54ac6e6a36b6@codeaurora.org> References: <1499439193-16628-1-git-send-email-okaya@codeaurora.org> <20170710230911.GH5610@bhelgaas-glaptop.roam.corp.google.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Return-path: Received: from smtp.codeaurora.org ([198.145.29.96]:34512 "EHLO smtp.codeaurora.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750827AbdGKAUw (ORCPT ); Mon, 10 Jul 2017 20:20:52 -0400 In-Reply-To: <20170710230911.GH5610@bhelgaas-glaptop.roam.corp.google.com> Content-Language: en-US Sender: linux-arm-msm-owner@vger.kernel.org List-Id: linux-arm-msm@vger.kernel.org To: Bjorn Helgaas Cc: linux-pci@vger.kernel.org, timur@codeaurora.org, wim.ten.have@oracle.com, linux-arm-msm@vger.kernel.org, Bjorn Helgaas , linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org Hi Bjorn, On 7/10/2017 7:09 PM, Bjorn Helgaas wrote: > The PCI_EXP_DEVCTL_EXT_TAG bit controls the behavior of the function as a > Requester. As far as I can see, it has nothing to do with whether the > function supports 8-bit tags as a *Completer*, so the implicit assumption > of the spec is that all Completers always support 8-bit tags. My guess is > that's why the ECN thought it would be safe to enable extended tags by > default. > > If that's the case, this is just a defect in the device (the Completer), > and we should blacklist it. Looking at the PCIe Capability version might > happen to correlate with Completer support for 8-bit tags, but that looks > like just a coincidence to me. Sure, I can change to blacklist. I'll not enable it unless the device is blacklisted via quirks. Sinan -- Sinan Kaya Qualcomm Datacenter Technologies, Inc. as an affiliate of Qualcomm Technologies, Inc. Qualcomm Technologies, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project. From mboxrd@z Thu Jan 1 00:00:00 1970 From: okaya@codeaurora.org (Sinan Kaya) Date: Mon, 10 Jul 2017 20:20:49 -0400 Subject: [PATCH V2] PCI: Do not enable extended tags on pre-dated (v1.x) systems In-Reply-To: <20170710230911.GH5610@bhelgaas-glaptop.roam.corp.google.com> References: <1499439193-16628-1-git-send-email-okaya@codeaurora.org> <20170710230911.GH5610@bhelgaas-glaptop.roam.corp.google.com> Message-ID: <5f5119fd-64d5-5e7c-fa50-54ac6e6a36b6@codeaurora.org> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Hi Bjorn, On 7/10/2017 7:09 PM, Bjorn Helgaas wrote: > The PCI_EXP_DEVCTL_EXT_TAG bit controls the behavior of the function as a > Requester. As far as I can see, it has nothing to do with whether the > function supports 8-bit tags as a *Completer*, so the implicit assumption > of the spec is that all Completers always support 8-bit tags. My guess is > that's why the ECN thought it would be safe to enable extended tags by > default. > > If that's the case, this is just a defect in the device (the Completer), > and we should blacklist it. Looking at the PCIe Capability version might > happen to correlate with Completer support for 8-bit tags, but that looks > like just a coincidence to me. Sure, I can change to blacklist. I'll not enable it unless the device is blacklisted via quirks. Sinan -- Sinan Kaya Qualcomm Datacenter Technologies, Inc. as an affiliate of Qualcomm Technologies, Inc. Qualcomm Technologies, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project.