From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from e7.ny.us.ibm.com (e7.ny.us.ibm.com [32.97.182.137]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client CN "e7.ny.us.ibm.com", Issuer "GeoTrust SSL CA" (not verified)) by ozlabs.org (Postfix) with ESMTPS id 498D82C00C6 for ; Fri, 11 Oct 2013 18:56:53 +1100 (EST) Received: from /spool/local by e7.ny.us.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Fri, 11 Oct 2013 03:56:50 -0400 Received: from b01cxnp22035.gho.pok.ibm.com (b01cxnp22035.gho.pok.ibm.com [9.57.198.25]) by d01dlp02.pok.ibm.com (Postfix) with ESMTP id 05BD36E8040 for ; Fri, 11 Oct 2013 03:56:48 -0400 (EDT) Received: from d01av02.pok.ibm.com (d01av02.pok.ibm.com [9.56.224.216]) by b01cxnp22035.gho.pok.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id r9B7unUw44630222 for ; Fri, 11 Oct 2013 07:56:49 GMT Received: from d01av02.pok.ibm.com (loopback [127.0.0.1]) by d01av02.pok.ibm.com (8.14.4/8.13.1/NCO v10.0 AVout) with ESMTP id r9B7uldE020933 for ; Fri, 11 Oct 2013 04:56:48 -0300 Date: Fri, 11 Oct 2013 15:56:42 +0800 From: Gavin Shan To: Yijing Wang Subject: Re: [PATCH v2 3/6] powerpc/pci: use pci_is_pcie() to simplify code Message-ID: <20131011075642.GA20443@shangw.(null)> References: <1378367730-25996-1-git-send-email-wangyijing@huawei.com> <1378367730-25996-3-git-send-email-wangyijing@huawei.com> <20130906203035.GA27940@google.com> <1381470596.5630.61.camel@pasglop> <20131011061654.GA561@shangw.(null)> <52579BD6.40802@huawei.com> <20131011065329.GA5013@shangw.(null)> <5257A882.1070809@huawei.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <5257A882.1070809@huawei.com> Cc: Gavin Shan , Hanjun Guo , linux-kernel@vger.kernel.org, "James E.J. Bottomley" , Paul Mackerras , linux-pci@vger.kernel.org, Bjorn Helgaas , linuxppc-dev@lists.ozlabs.org Reply-To: Gavin Shan List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Fri, Oct 11, 2013 at 03:28:02PM +0800, Yijing Wang wrote: >On 2013/10/11 14:53, Gavin Shan wrote: >> On Fri, Oct 11, 2013 at 02:33:58PM +0800, Yijing Wang wrote: >>> On 2013/10/11 14:16, Gavin Shan wrote: >>>> On Fri, Oct 11, 2013 at 04:49:56PM +1100, Benjamin Herrenschmidt wrote: >>>>> On Fri, 2013-09-06 at 14:30 -0600, Bjorn Helgaas wrote: >>>>>> On Thu, Sep 05, 2013 at 03:55:27PM +0800, Yijing Wang wrote: >> >> .../... >> >>>>>>> Use pci_is_pcie() to simplify code. >>>>>>> diff --git a/arch/powerpc/kernel/eeh.c b/arch/powerpc/kernel/eeh.c >>>>>>> index 55593ee..6ebbe54 100644 >>>>>>> --- a/arch/powerpc/kernel/eeh.c >>>>>>> +++ b/arch/powerpc/kernel/eeh.c >>>>>>> @@ -189,8 +189,7 @@ static size_t eeh_gather_pci_data(struct eeh_dev *edev, char * buf, size_t len) >>>>>>> } >>>>>>> >>>>>>> /* If PCI-E capable, dump PCI-E cap 10, and the AER */ >>>>>>> - cap = pci_find_capability(dev, PCI_CAP_ID_EXP); >>>>>>> - if (cap) { >>>>>>> + if (pci_is_pcie(dev)) { >>>>>>> n += scnprintf(buf+n, len-n, "pci-e cap10:\n"); >>>>>>> printk(KERN_WARNING >>>>>>> "EEH: PCI-E capabilities and status follow:\n"); >>>>> >>>>> So we remove reading of "cap", but slightly further down the code does: >>>>> >>>>> for (i=0; i<=8; i++) { >>>>> eeh_ops->read_config(dn, cap+4*i, 4, &cfg); >>>>> n += scnprintf(buf+n, len-n, "%02x:%x\n", 4*i, cfg); >>>>> printk(KERN_WARNING "EEH: PCI-E %02x: %08x\n", i, cfg); >>>>> } >>>>> >>>>> Which actually *uses* the value of "cap" ... oops :-) >>>>> >>>> >>>> It's my fault and I should have looked into the changes more closely. >>>> How about changing it like this: >>>> >>>> cap = pci_is_pcie(dev) ? pci_pcie_cap(dev) : >>>> pci_find_capability(dev, PCI_CAP_ID_EXP); >>>> if (cap) { >>>> ... >>>> } >>>> >>>> It would save some PCI-CFG access cycles for most cases :-) >>> >>> Hi Gavin, it's not your fault, it's my fault. :) >>> >>> Because pci_pcie_cap(dev) == dev->pcie_cap == pci_find_capability(dev, PCI_CAP_ID_EXP); >>> >>> so I think it's ok to use dev->pcie_cap instead of stale "cap". >>> >> >> Yijing, There has one exception: dev->pcie_cap isn't updated yet. > >In my idea, dev->pcie_cap(here is pci_dev->pcie_cap) will update in set_pcie_port_type() function, >and this function always be called after allocate pci device. We get pci_dev by eeh_dev_to_pci_dev(), >I think pci_dev has been initialized completely. > >> This function has possibility to be invoked before that. However, >> we don't have the binding (eeh device <-> PCI device) for the case. >> So the piece of code shouldn't be running > >In PCI core, I knew > >pci_scan_device() > pci_setup_device() > set_pcie_port_type() > pci_dev->pcie_cap = pci_find_capability(pdev, PCI_CAP_ID_EXP); > >In powerpc, I also found > >of_scan_pci_dev() > of_create_pci_dev() > set_pcie_port_type() > pci_dev->pcie_cap = pci_find_capability(pdev, PCI_CAP_ID_EXP); >> >> However, it's a bit safer to have pci_find_capability(dev, PCI_CAP_ID_EXP) >> as well even though we needn't it for 99.9% cases if you agree :-) > >I agree, this function is not the performance bottleneck, >safety is more important. :) >So if Bjorn and Benjamin think it's not safe, it's ok to drop it. :) > No, it's not what I mean. Anyway, "v3" looks good to me. At least, it can save PCI-CFG access cycles find locate the PCIe capability position :-) Thanks, Gavin