Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932771AbdDRS20 (ORCPT ); Tue, 18 Apr 2017 14:28:26 -0400 Received: from mail.kernel.org ([198.145.29.136]:58944 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932512AbdDRS2U (ORCPT ); Tue, 18 Apr 2017 14:28:20 -0400 Date: Tue, 18 Apr 2017 13:28:16 -0500 From: Bjorn Helgaas To: David Woodhouse Cc: linux-pci@vger.kernel.org, linux-arch@vger.kernel.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH v2 00/27] PCI resource mmap cleanup Message-ID: <20170418182816.GB25295@bhelgaas-glaptop.roam.corp.google.com> References: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 5625 Lines: 112 On Wed, Apr 12, 2017 at 01:25:49PM +0100, David Woodhouse wrote: > This pursues my previous patch set all the way to its logical conclusion. > > It kills off the legacy arch-provided pci_mmap_page_range() completely, > along with its vile 'address converted by pci_resource_to_user()' API > and the various bugs and other strange behaviour that various > architectures had. > > In some cases like IA64 I've killed off the odd behaviour *first* in > separately reviewable patches, before the final switch over to the > generic code which should then basically have no functional effect. > > To accommodate the ARM64 maintainers' desire *not* to support mmap > through /proc/bus/pci I have separated HAVE_PCI_MMAP from the sysfs > implementation, which (as of the last patch in the series) ends up > present on *all* platforms with an MMU. I still don't think that's a > *good* idea though; it's a generic part of /proc, and nobody ever did > answer my question about what *else* we can delete from /proc on ARM64 > "because it's a new platform"... > > I would very much like the interesting parts of this to go through the > arch maintainers' trees. I'd suggest that we can pull everything up to > and including patch 19 ("pci: Add I/O BAR support to generic > pci_mmap_resource_range()") through the PCI tree, which includes the > relatively trivial architectures. Then the others can go through the > appropriate arch tree with more careful review and testing. > > Once everything's done, then we can apply something like the final > patch ("pci: Kill ARCH_GENERIC_PCI_MMAP_RESOURCE"). But that's mostly > just there for now as a demonstration of the intended end point. > > This is in > git://git.infradead.org/users/dwmw2/random-2.6.git pcimmap > http://git.infradead.org/users/dwmw2/random-2.6.git/shortlog/refs/heads/pcimmap Applied to pci/resource-mmap for v4.12, thanks! This consists of the email patches + the acks/etc so far. It doesn't include any updates from your git branch (but it doesn't look like it's been updated recently). > David Woodhouse (27): > pci: Fix pci_mmap_fits() for HAVE_PCI_RESOURCE_TO_USER platforms > pci: Fix another sanity check bug in /proc/pci mmap > pci: Only allow WC mmap on prefetchable resources > xtensa: Do not mmap PCI BARs to userspace as write-through > pci: Add arch_can_pci_mmap_wc() macro > pci: Move multiple declarations of pci_mmap_page_range() to > > pci: Add arch_can_pci_mmap_io() on architectures which can mmap() I/O > space > pci: Use BAR index in sysfs attr->private instead of resource pointer > pci: Add BAR index argument to pci_mmap_page_range() > pci: Add pci_mmap_resource_range() and use it for ARM64 > arm: Use generic pci_mmap_resource_range() > cris: Use generic pci_mmap_resource_range() > mips: Use generic pci_mmap_resource_range() > mn10300: Use generic pci_mmap_resource_range() > parisc: Use generic pci_mmap_resource_range() > sh: Use generic pci_mmap_resource_range() > unicore: Use generic pci_mmap_resource_range() > x86: Use generic pci_mmap_resource_range() > pci: Add I/O BAR support to generic pci_mmap_resource_range() > powerpc: Use generic pci_mmap_resource_range() > microblaze: Use generic pci_mmap_resource_range() > xtensa: Use generic pci_mmap_resource_range() > ia64: Remove redundant valid_mmap_phys_addr_range() from > pci_mmap_page_range() > ia64: Remove redundant checks for WC in pci_mmap_page_range() > ia64: Use generic pci_mmap_resource_range() > sparc: Use generic pci_mmap_resource_range() > pci: Kill ARCH_GENERIC_PCI_MMAP_RESOURCE > > Documentation/filesystems/sysfs-pci.txt | 12 ++- > arch/arm/include/asm/pci.h | 2 - > arch/arm/kernel/bios32.c | 19 ---- > arch/cris/arch-v32/drivers/pci/bios.c | 22 ----- > arch/cris/include/asm/pci.h | 3 - > arch/ia64/include/asm/pci.h | 4 +- > arch/ia64/pci/pci.c | 46 ---------- > arch/microblaze/include/asm/pci.h | 8 +- > arch/microblaze/pci/pci-common.c | 99 ++------------------ > arch/mips/include/asm/pci.h | 4 - > arch/mips/pci/pci.c | 24 ----- > arch/mn10300/include/asm/pci.h | 3 - > arch/mn10300/unit-asb2305/pci-asb2305.c | 23 ----- > arch/parisc/include/asm/pci.h | 3 - > arch/parisc/kernel/pci.c | 28 ------ > arch/powerpc/include/asm/pci.h | 9 +- > arch/powerpc/kernel/pci-common.c | 105 +++------------------- > arch/sh/drivers/pci/pci.c | 21 ----- > arch/sh/include/asm/pci.h | 3 +- > arch/sparc/include/asm/pci_64.h | 5 +- > arch/sparc/kernel/pci.c | 155 -------------------------------- > arch/unicore32/include/asm/pci.h | 2 - > arch/unicore32/kernel/pci.c | 23 ----- > arch/x86/include/asm/pci.h | 6 +- > arch/x86/pci/i386.c | 47 ---------- > arch/xtensa/include/asm/pci.h | 9 +- > arch/xtensa/kernel/pci.c | 112 ++--------------------- > drivers/pci/Makefile | 1 + > drivers/pci/mmap.c | 67 ++++++++++++++ > drivers/pci/pci-sysfs.c | 76 +++++++--------- > drivers/pci/pci.h | 4 +- > drivers/pci/proc.c | 41 ++++++--- > include/linux/pci.h | 30 +++++++ > 33 files changed, 211 insertions(+), 805 deletions(-) > create mode 100644 drivers/pci/mmap.c > > -- > 2.9.3 >