From af42b8d12f8adec6711cb824549a0edac6a4ae8f Mon Sep 17 00:00:00 2001 From: Stefano Stabellini Date: Wed, 1 Dec 2010 14:51:44 +0000 Subject: xen: fix MSI setup and teardown for PV on HVM guests When remapping MSIs into pirqs for PV on HVM guests, qemu is responsible for doing the actual mapping and unmapping. We only give qemu the desired pirq number when we ask to do the mapping the first time, after that we should be reading back the pirq number from qemu every time we want to re-enable the MSI. This fixes a bug in xen_hvm_setup_msi_irqs that manifests itself when trying to enable the same MSI for the second time: the old MSI to pirq mapping is still valid at this point but xen_hvm_setup_msi_irqs would try to assign a new pirq anyway. A simple way to reproduce this bug is to assign an MSI capable network card to a PV on HVM guest, if the user brings down the corresponding ethernet interface and up again, Linux would fail to enable MSIs on the device. Signed-off-by: Stefano Stabellini --- arch/x86/pci/xen.c | 27 ++++++++++++++++++++------- 1 file changed, 20 insertions(+), 7 deletions(-) (limited to 'arch/x86/pci') diff --git a/arch/x86/pci/xen.c b/arch/x86/pci/xen.c index d7b5109f7a9c..25cd4a07d09f 100644 --- a/arch/x86/pci/xen.c +++ b/arch/x86/pci/xen.c @@ -70,6 +70,9 @@ static int acpi_register_gsi_xen_hvm(struct device *dev, u32 gsi, struct xen_pci_frontend_ops *xen_pci_frontend; EXPORT_SYMBOL_GPL(xen_pci_frontend); +#define XEN_PIRQ_MSI_DATA (MSI_DATA_TRIGGER_EDGE | \ + MSI_DATA_LEVEL_ASSERT | (3 << 8) | MSI_DATA_VECTOR(0)) + static void xen_msi_compose_msg(struct pci_dev *pdev, unsigned int pirq, struct msi_msg *msg) { @@ -83,12 +86,7 @@ static void xen_msi_compose_msg(struct pci_dev *pdev, unsigned int pirq, MSI_ADDR_REDIRECTION_CPU | MSI_ADDR_DEST_ID(pirq); - msg->data = - MSI_DATA_TRIGGER_EDGE | - MSI_DATA_LEVEL_ASSERT | - /* delivery mode reserved */ - (3 << 8) | - MSI_DATA_VECTOR(0); + msg->data = XEN_PIRQ_MSI_DATA; } static int xen_hvm_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) @@ -98,8 +96,23 @@ static int xen_hvm_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) struct msi_msg msg; list_for_each_entry(msidesc, &dev->msi_list, list) { + __read_msi_msg(msidesc, &msg); + pirq = MSI_ADDR_EXT_DEST_ID(msg.address_hi) | + ((msg.address_lo >> MSI_ADDR_DEST_ID_SHIFT) & 0xff); + if (xen_irq_from_pirq(pirq) >= 0 && msg.data == XEN_PIRQ_MSI_DATA) { + xen_allocate_pirq_msi((type == PCI_CAP_ID_MSIX) ? + "msi-x" : "msi", &irq, &pirq, XEN_ALLOC_IRQ); + if (irq < 0) + goto error; + ret = set_irq_msi(irq, msidesc); + if (ret < 0) + goto error_while; + printk(KERN_DEBUG "xen: msi already setup: msi --> irq=%d" + " pirq=%d\n", irq, pirq); + return 0; + } xen_allocate_pirq_msi((type == PCI_CAP_ID_MSIX) ? - "msi-x" : "msi", &irq, &pirq); + "msi-x" : "msi", &irq, &pirq, (XEN_ALLOC_IRQ | XEN_ALLOC_PIRQ)); if (irq < 0 || pirq < 0) goto error; printk(KERN_DEBUG "xen: msi --> irq=%d, pirq=%d\n", irq, pirq); -- cgit v1.2.3 From d14125ecfee05473de46f06d992db109308c57a3 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Thu, 16 Dec 2010 10:38:31 -0700 Subject: Revert "x86/PCI: allocate space from the end of a region, not the beginning" This reverts commit dc9887dc02e37bcf83f4e792aa14b07782ef54cf. Acked-by: H. Peter Anvin Signed-off-by: Bjorn Helgaas Signed-off-by: Jesse Barnes --- arch/x86/pci/i386.c | 17 ++++++----------- 1 file changed, 6 insertions(+), 11 deletions(-) (limited to 'arch/x86/pci') diff --git a/arch/x86/pci/i386.c b/arch/x86/pci/i386.c index c4bb261c106e..8379c2c3d076 100644 --- a/arch/x86/pci/i386.c +++ b/arch/x86/pci/i386.c @@ -65,21 +65,16 @@ pcibios_align_resource(void *data, const struct resource *res, resource_size_t size, resource_size_t align) { struct pci_dev *dev = data; - resource_size_t start = round_down(res->end - size + 1, align); + resource_size_t start = res->start; if (res->flags & IORESOURCE_IO) { - - /* - * If we're avoiding ISA aliases, the largest contiguous I/O - * port space is 256 bytes. Clearing bits 9 and 10 preserves - * all 256-byte and smaller alignments, so the result will - * still be correctly aligned. - */ - if (!skip_isa_ioresource_align(dev)) - start &= ~0x300; + if (skip_isa_ioresource_align(dev)) + return start; + if (start & 0x300) + start = (start + 0x3ff) & ~0x3ff; } else if (res->flags & IORESOURCE_MEM) { if (start < BIOS_END) - start = res->end; /* fail; no space */ + start = BIOS_END; } return start; } -- cgit v1.2.3 From 30919b0bf356a8ee0ef4f7d38ca8ad99b96820b2 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Thu, 16 Dec 2010 10:38:51 -0700 Subject: x86: avoid low BIOS area when allocating address space This implements arch_remove_reservations() so allocate_resource() can avoid any arch-specific reserved areas. This currently just avoids the BIOS area (the first 1MB), but could be used for E820 reserved areas if that turns out to be necessary. We previously avoided this area in pcibios_align_resource(). This patch moves the test from that PCI-specific path to a generic path, so *all* resource allocations will avoid this area. Acked-by: H. Peter Anvin Signed-off-by: Bjorn Helgaas Signed-off-by: Jesse Barnes --- arch/x86/pci/i386.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'arch/x86/pci') diff --git a/arch/x86/pci/i386.c b/arch/x86/pci/i386.c index 8379c2c3d076..b1805b78842f 100644 --- a/arch/x86/pci/i386.c +++ b/arch/x86/pci/i386.c @@ -72,9 +72,6 @@ pcibios_align_resource(void *data, const struct resource *res, return start; if (start & 0x300) start = (start + 0x3ff) & ~0x3ff; - } else if (res->flags & IORESOURCE_MEM) { - if (start < BIOS_END) - start = BIOS_END; } return start; } -- cgit v1.2.3