From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com ([209.132.183.28]:44792 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932252AbcASVEc (ORCPT ); Tue, 19 Jan 2016 16:04:32 -0500 Message-ID: <1453237471.32741.267.camel@redhat.com> Subject: Re: [PATCH] pci: Add support for multiple DMA aliases From: Alex Williamson To: Bjorn Helgaas , Jacek Lawrynowicz Cc: linux-pci@vger.kernel.org, bhelgaas@google.com, dwmw2@infradead.org, jroedel@suse.de Date: Tue, 19 Jan 2016 14:04:31 -0700 In-Reply-To: <20160119201254.GB6510@localhost> References: <1453118346-76897-1-git-send-email-jacek.lawrynowicz@intel.com> <1453133267-79756-1-git-send-email-jacek.lawrynowicz@intel.com> <20160119033315.GA6510@localhost> <20160119201254.GB6510@localhost> Content-Type: text/plain; charset="UTF-8" Mime-Version: 1.0 Sender: linux-pci-owner@vger.kernel.org List-ID: On Tue, 2016-01-19 at 14:12 -0600, Bjorn Helgaas wrote: > [+cc Alex] > > On Mon, Jan 18, 2016 at 09:33:15PM -0600, Bjorn Helgaas wrote: > > On Mon, Jan 18, 2016 at 05:07:47PM +0100, Jacek Lawrynowicz wrote: > > > This patch solves IOMMU support issues with PCIe non-transparent bridges > > > that use Requester ID look-up tables (LUT), e.g. PEX8733. Before exiting > > > the bridge, packet's RID is rewritten according to LUT programmed by > > > a driver. Modified packets are then passed to a destination bus and > > > processed upstream. The problem is that such packets seem to come from > > > non-existent nodes that are hidden behind NTB and are not discoverable > > > by a destination node, so IOMMU discards them. Adding DMA alias for a > > > given LUT entry allows IOMMU to create a proper mapping that enables > > > inter-node communication. > > > > > > The current DMA alias implementation supports only single alias, so it's > > > not possible to connect more than two nodes when IOMMU is enabled. This > > > implementation enables all possible aliases on a given bus (256) that > > > are stored in a bitset. Alias devfn is directly translated to a bit > > > number. The bitset is not allocated for devices that have no need for > > > DMA aliases. My only concern here is that pci_add_dma_alias() makes aliases seem more dynamic than they really are.  For instance, when we add a device to an IOMMU domain, we evaluate the aliases at that point, if an NTB later adds a new lookup entry and specifies a new alias, it's still not going to work.  Similarly, IOMMU groups are evaluated as the device is added, so if an alias is to a physical device and we need the cross reference to bind them together into a single group, calling pci_add_dma_alias() from a driver isn't going to work. The existing code had this problem too, it's just more obvious now that we have a helper function and that the helper is exported for use outside of the PCI core.  Thanks, Alex > > > > > > More details can be found in following article: > > > http://www.plxtech.com/files/pdf/technical/expresslane/RTC_Enabling%20MulitHostSystemDesigns.pdf > > > > > > Signed-off-by: Jacek Lawrynowicz > > > Acked-by: David Woodhouse > > > Acked-by: Joerg Roedel > > > > I applied this to pci/iommu and intend to merge it for v4.5. > > > > I made the following change because the kcalloc() failure warning in > > your patch looks wrong: > > > > diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c > > index b0d6a0a..29cfe1a 100644 > > --- a/drivers/pci/pci.c > > +++ b/drivers/pci/pci.c > > @@ -4577,13 +4577,15 @@ int pci_set_vga_state(struct pci_dev *dev, bool decode, > >   */ > >  void pci_add_dma_alias(struct pci_dev *dev, u8 devfn) > >  { > > - if (!dev->dma_alias_mask) { > > + if (!dev->dma_alias_mask) > >   dev->dma_alias_mask = kcalloc(BITS_TO_LONGS(U8_MAX), > >         sizeof(long), GFP_KERNEL); > > - dev_warn(&dev->dev, "Unable to allocate DMA alias mask.\n"); > > + if (!dev->dma_alias_mask) { > > + dev_warn(&dev->dev, "Unable to allocate DMA alias mask\n"); > > + return; > >   } > > - if (dev->dma_alias_mask) > > - set_bit(devfn, dev->dma_alias_mask); > > + > > + set_bit(devfn, dev->dma_alias_mask); > >  } > >  EXPORT_SYMBOL_GPL(pci_add_dma_alias); > >   > > > > > --- > > >  drivers/iommu/iommu.c |  8 ++++---- > > >  drivers/pci/pci.c     | 19 +++++++++++++++++++ > > >  drivers/pci/probe.c   |  1 + > > >  drivers/pci/quirks.c  | 15 ++++++--------- > > >  drivers/pci/search.c  | 14 +++++++++----- > > >  include/linux/pci.h   | 15 ++++++++------- > > >  6 files changed, 47 insertions(+), 25 deletions(-) > > > > > > diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c > > > index abae363..98ae7ff 100644 > > > --- a/drivers/iommu/iommu.c > > > +++ b/drivers/iommu/iommu.c > > > @@ -686,10 +686,10 @@ static struct iommu_group *get_pci_alias_group(struct pci_dev *pdev, > > >   continue; > > >   > > >   /* We alias them or they alias us */ > > > - if (((pdev->dev_flags & PCI_DEV_FLAGS_DMA_ALIAS_DEVFN) && > > > -      pdev->dma_alias_devfn == tmp->devfn) || > > > -     ((tmp->dev_flags & PCI_DEV_FLAGS_DMA_ALIAS_DEVFN) && > > > -      tmp->dma_alias_devfn == pdev->devfn)) { > > > + if ((pdev->dma_alias_mask && > > > +      test_bit(tmp->devfn, pdev->dma_alias_mask)) || > > > +     ((tmp->dma_alias_mask && > > > +      test_bit(pdev->devfn, tmp->dma_alias_mask)))) { > > >   > > >   group = get_pci_alias_group(tmp, devfns); > > >   if (group) { > > > diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c > > > index 49e3715..5b27d65 100644 > > > --- a/drivers/pci/pci.c > > > +++ b/drivers/pci/pci.c > > > @@ -4568,6 +4568,25 @@ int pci_set_vga_state(struct pci_dev *dev, bool decode, > > >   return 0; > > >  } > > >   > > > +/** > > > + * pci_add_dma_alias - Allows to add multiple devfn aliases for given device > > > + * @dev: the PCI device for which alias is added > > > + * @devfn: alias slot and function > > > + * > > > + * This helper encodes 8-bit devfn as bit number in dma_alias_mask > > > + */ > > > +void pci_add_dma_alias(struct pci_dev *dev, u8 devfn) > > > +{ > > > + if (!dev->dma_alias_mask) { > > > + dev->dma_alias_mask = kcalloc(BITS_TO_LONGS(U8_MAX), > > > +       sizeof(long), GFP_KERNEL); > > > + dev_warn(&dev->dev, "Unable to allocate DMA alias mask.\n"); > > > + } > > > + if (dev->dma_alias_mask) > > > + set_bit(devfn, dev->dma_alias_mask); > > > +} > > > +EXPORT_SYMBOL_GPL(pci_add_dma_alias); > > > + > > >  bool pci_device_is_present(struct pci_dev *pdev) > > >  { > > >   u32 v; > > > diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c > > > index 32b9f1b..5da4dd3 100644 > > > --- a/drivers/pci/probe.c > > > +++ b/drivers/pci/probe.c > > > @@ -1501,6 +1501,7 @@ static void pci_release_dev(struct device *dev) > > >   pcibios_release_device(pci_dev); > > >   pci_bus_put(pci_dev->bus); > > >   kfree(pci_dev->driver_override); > > > + kfree(pci_dev->dma_alias_mask); > > >   kfree(pci_dev); > > >  } > > >   > > > diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c > > > index 83e93d7..3ed1f9a 100644 > > > --- a/drivers/pci/quirks.c > > > +++ b/drivers/pci/quirks.c > > > @@ -3579,8 +3579,7 @@ int pci_dev_specific_reset(struct pci_dev *dev, int probe) > > >  static void quirk_dma_func0_alias(struct pci_dev *dev) > > >  { > > >   if (PCI_FUNC(dev->devfn) != 0) { > > > - dev->dma_alias_devfn = PCI_DEVFN(PCI_SLOT(dev->devfn), 0); > > > - dev->dev_flags |= PCI_DEV_FLAGS_DMA_ALIAS_DEVFN; > > > + pci_add_dma_alias(dev, PCI_DEVFN(PCI_SLOT(dev->devfn), 0)); > > >   } > > >  } > > >   > > > @@ -3595,8 +3594,7 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_RICOH, 0xe476, quirk_dma_func0_alias); > > >  static void quirk_dma_func1_alias(struct pci_dev *dev) > > >  { > > >   if (PCI_FUNC(dev->devfn) != 1) { > > > - dev->dma_alias_devfn = PCI_DEVFN(PCI_SLOT(dev->devfn), 1); > > > - dev->dev_flags |= PCI_DEV_FLAGS_DMA_ALIAS_DEVFN; > > > + pci_add_dma_alias(dev, PCI_DEVFN(PCI_SLOT(dev->devfn), 1)); > > >   } > > >  } > > >   > > > @@ -3660,11 +3658,10 @@ static void quirk_fixed_dma_alias(struct pci_dev *dev) > > >   > > >   id = pci_match_id(fixed_dma_alias_tbl, dev); > > >   if (id) { > > > - dev->dma_alias_devfn = id->driver_data; > > > - dev->dev_flags |= PCI_DEV_FLAGS_DMA_ALIAS_DEVFN; > > > - dev_info(&dev->dev, "Enabling fixed DMA alias to %02x.%d\n", > > > -  PCI_SLOT(dev->dma_alias_devfn), > > > -  PCI_FUNC(dev->dma_alias_devfn)); > > > + pci_add_dma_alias(dev, id->driver_data); > > > + dev_info(&dev->dev, "Enabling fixed DMA alias to %02lx.%ld\n", > > > +  PCI_SLOT(id->driver_data), > > > +  PCI_FUNC(id->driver_data)); > > >   } > > >  } > > >   > > > diff --git a/drivers/pci/search.c b/drivers/pci/search.c > > > index a20ce7d..33e0f03 100644 > > > --- a/drivers/pci/search.c > > > +++ b/drivers/pci/search.c > > > @@ -40,11 +40,15 @@ int pci_for_each_dma_alias(struct pci_dev *pdev, > > >    * If the device is broken and uses an alias requester ID for > > >    * DMA, iterate over that too. > > >    */ > > > - if (unlikely(pdev->dev_flags & PCI_DEV_FLAGS_DMA_ALIAS_DEVFN)) { > > > - ret = fn(pdev, PCI_DEVID(pdev->bus->number, > > > -  pdev->dma_alias_devfn), data); > > > - if (ret) > > > - return ret; > > > + if (unlikely(pdev->dma_alias_mask)) { > > > + u8 devfn; > > > + > > > + for_each_set_bit(devfn, pdev->dma_alias_mask, U8_MAX) { > > > + ret = fn(pdev, PCI_DEVID(pdev->bus->number, devfn), > > > +  data); > > > + if (ret) > > > + return ret; > > > + } > > >   } > > >   > > >   for (bus = pdev->bus; !pci_is_root_bus(bus); bus = bus->parent) { > > > diff --git a/include/linux/pci.h b/include/linux/pci.h > > > index f9f79ad..6200175 100644 > > > --- a/include/linux/pci.h > > > +++ b/include/linux/pci.h > > > @@ -172,16 +172,14 @@ enum pci_dev_flags { > > >   PCI_DEV_FLAGS_ASSIGNED = (__force pci_dev_flags_t) (1 << 2), > > >   /* Flag for quirk use to store if quirk-specific ACS is enabled */ > > >   PCI_DEV_FLAGS_ACS_ENABLED_QUIRK = (__force pci_dev_flags_t) (1 << 3), > > > - /* Flag to indicate the device uses dma_alias_devfn */ > > > - PCI_DEV_FLAGS_DMA_ALIAS_DEVFN = (__force pci_dev_flags_t) (1 << 4), > > >   /* Use a PCIe-to-PCI bridge alias even if !pci_is_pcie */ > > > - PCI_DEV_FLAG_PCIE_BRIDGE_ALIAS = (__force pci_dev_flags_t) (1 << 5), > > > + PCI_DEV_FLAG_PCIE_BRIDGE_ALIAS = (__force pci_dev_flags_t) (1 << 4), > > >   /* Do not use bus resets for device */ > > > - PCI_DEV_FLAGS_NO_BUS_RESET = (__force pci_dev_flags_t) (1 << 6), > > > + PCI_DEV_FLAGS_NO_BUS_RESET = (__force pci_dev_flags_t) (1 << 5), > > >   /* Do not use PM reset even if device advertises NoSoftRst- */ > > > - PCI_DEV_FLAGS_NO_PM_RESET = (__force pci_dev_flags_t) (1 << 7), > > > + PCI_DEV_FLAGS_NO_PM_RESET = (__force pci_dev_flags_t) (1 << 6), > > >   /* Get VPD from function 0 VPD */ > > > - PCI_DEV_FLAGS_VPD_REF_F0 = (__force pci_dev_flags_t) (1 << 8), > > > + PCI_DEV_FLAGS_VPD_REF_F0 = (__force pci_dev_flags_t) (1 << 7), > > >  }; > > >   > > >  enum pci_irq_reroute_variant { > > > @@ -279,7 +277,7 @@ struct pci_dev { > > >   u8 rom_base_reg; /* which config register controls the ROM */ > > >   u8 pin; /* which interrupt pin this device uses */ > > >   u16 pcie_flags_reg; /* cached PCIe Capabilities Register */ > > > - u8 dma_alias_devfn;/* devfn of DMA alias, if any */ > > > + unsigned long *dma_alias_mask;/* mask of enabled devfn aliases */ > > >   > > >   struct pci_driver *driver; /* which driver has allocated this device */ > > >   u64 dma_mask; /* Mask of the bits of bus address this > > > @@ -1229,6 +1227,9 @@ resource_size_t pcibios_iov_resource_alignment(struct pci_dev *dev, int resno); > > >   > > >  int pci_set_vga_state(struct pci_dev *pdev, bool decode, > > >         unsigned int command_bits, u32 flags); > > > + > > > +void pci_add_dma_alias(struct pci_dev *dev, u8 devfn); > > > + > > >  /* kmem_cache style wrapper around pci_alloc_consistent() */ > > >   > > >  #include