[PATCH 3.10 040/319] xen/pciback: Do not install an IRQ handler for MSI interrupts.

From	Willy Tarreau <w@1wt.eu>
Newsgroups	linux.kernel
Subject	[PATCH 3.10 040/319] xen/pciback: Do not install an IRQ handler for MSI interrupts.
Date	2017-02-05 20:30 +0100
Message-ID	<t7AFE-32I-131@gated-at.bofh.it> (permalink)
References	<t7AFA-32I-9@gated-at.bofh.it>
Organization	linux.* mail to news gateway

Show all headers | View raw

From: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>

commit a396f3a210c3a61e94d6b87ec05a75d0be2a60d0 upstream.

Otherwise an guest can subvert the generic MSI code to trigger
an BUG_ON condition during MSI interrupt freeing:

 for (i = 0; i < entry->nvec_used; i++)
        BUG_ON(irq_has_action(entry->irq + i));

Xen PCI backed installs an IRQ handler (request_irq) for
the dev->irq whenever the guest writes PCI_COMMAND_MEMORY
(or PCI_COMMAND_IO) to the PCI_COMMAND register. This is
done in case the device has legacy interrupts the GSI line
is shared by the backend devices.

To subvert the backend the guest needs to make the backend
to change the dev->irq from the GSI to the MSI interrupt line,
make the backend allocate an interrupt handler, and then command
the backend to free the MSI interrupt and hit the BUG_ON.

Since the backend only calls 'request_irq' when the guest
writes to the PCI_COMMAND register the guest needs to call
XEN_PCI_OP_enable_msi before any other operation. This will
cause the generic MSI code to setup an MSI entry and
populate dev->irq with the new PIRQ value.

Then the guest can write to PCI_COMMAND PCI_COMMAND_MEMORY
and cause the backend to setup an IRQ handler for dev->irq
(which instead of the GSI value has the MSI pirq). See
'xen_pcibk_control_isr'.

Then the guest disables the MSI: XEN_PCI_OP_disable_msi
which ends up triggering the BUG_ON condition in 'free_msi_irqs'
as there is an IRQ handler for the entry->irq (dev->irq).

Note that this cannot be done using MSI-X as the generic
code does not over-write dev->irq with the MSI-X PIRQ values.

The patch inhibits setting up the IRQ handler if MSI or
MSI-X (for symmetry reasons) code had been called successfully.

P.S.
Xen PCIBack when it sets up the device for the guest consumption
ends up writting 0 to the PCI_COMMAND (see xen_pcibk_reset_device).
XSA-120 addendum patch removed that - however when upstreaming said
addendum we found that it caused issues with qemu upstream. That
has now been fixed in qemu upstream.

This is part of XSA-157

Reviewed-by: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>
---
 drivers/xen/xen-pciback/pciback_ops.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/xen/xen-pciback/pciback_ops.c b/drivers/xen/xen-pciback/pciback_ops.c
index 1765509..544e352 100644
--- a/drivers/xen/xen-pciback/pciback_ops.c
+++ b/drivers/xen/xen-pciback/pciback_ops.c
@@ -67,6 +67,13 @@ static void xen_pcibk_control_isr(struct pci_dev *dev, int reset)
 		enable ? "enable" : "disable");
 
 	if (enable) {
+		/*
+		 * The MSI or MSI-X should not have an IRQ handler. Otherwise
+		 * if the guest terminates we BUG_ON in free_msi_irqs.
+		 */
+		if (dev->msi_enabled || dev->msix_enabled)
+			goto out;
+
 		rc = request_irq(dev_data->irq,
 				xen_pcibk_guest_interrupt, IRQF_SHARED,
 				dev_data->irq_name, dev);
-- 
2.8.0.rc2.1.gbe9624a

Thread

[PATCH 3.10 030/319] PM / devfreq: Fix incorrect type issue. Willy Tarreau <w@1wt.eu> - 2017-02-05 20:30 +0100
  [PATCH 3.10 067/319] powerpc/64: Fix incorrect return value from __copy_tofrom_user Willy Tarreau <w@1wt.eu> - 2017-02-05 20:30 +0100
  [PATCH 3.10 040/319] xen/pciback: Do not install an IRQ handler for MSI interrupts. Willy Tarreau <w@1wt.eu> - 2017-02-05 20:30 +0100
  [PATCH 3.10 079/319] score: fix __get_user/get_user Willy Tarreau <w@1wt.eu> - 2017-02-05 20:30 +0100
  [PATCH 3.10 095/319] openrisc: fix the fix of copy_from_user() Willy Tarreau <w@1wt.eu> - 2017-02-05 20:30 +0100
  [PATCH 3.10 086/319] score: fix copy_from_user() and friends Willy Tarreau <w@1wt.eu> - 2017-02-05 20:30 +0100
  [PATCH 3.10 072/319] s390/dasd: fix hanging device after clear subchannel Willy Tarreau <w@1wt.eu> - 2017-02-05 20:30 +0100
  [PATCH 3.10 060/319] arm64: spinlocks: implement smp_mb__before_spinlock() as smp_mb() Willy Tarreau <w@1wt.eu> - 2017-02-05 20:40 +0100
  [PATCH 3.10 093/319] parisc: fix copy_from_user() Willy Tarreau <w@1wt.eu> - 2017-02-05 20:40 +0100
  [PATCH 3.10 087/319] sh: fix copy_from_user() Willy Tarreau <w@1wt.eu> - 2017-02-05 20:40 +0100
  [PATCH 3.10 094/319] openrisc: fix copy_from_user() Willy Tarreau <w@1wt.eu> - 2017-02-05 20:40 +0100
  [PATCH 3.10 063/319] MIPS: ptrace: Fix regs_return_value for kernel context Willy Tarreau <w@1wt.eu> - 2017-02-05 20:50 +0100
  [PATCH 3.10 059/319] arm64: Define AT_VECTOR_SIZE_ARCH for ARCH_DLINFO Willy Tarreau <w@1wt.eu> - 2017-02-05 20:50 +0100
  [PATCH 3.10 055/319] ARM: 8618/1: decompressor: reset ttbcr fields to use TTBR0 on ARMv7 Willy Tarreau <w@1wt.eu> - 2017-02-05 20:50 +0100
  [PATCH 3.10 034/319] xen-netback: don't use last request to determine minimum Tx credit Willy Tarreau <w@1wt.eu> - 2017-02-05 21:00 +0100
  [PATCH 3.10 091/319] alpha: fix copy_from_user() Willy Tarreau <w@1wt.eu> - 2017-02-05 21:00 +0100
  [PATCH 3.10 076/319] mn10300: failing __get_user() and get_user() should zero Willy Tarreau <w@1wt.eu> - 2017-02-05 21:10 +0100
  [PATCH 3.10 070/319] avr32: fix 'undefined reference to `___copy_from_user' Willy Tarreau <w@1wt.eu> - 2017-02-05 21:10 +0100
  [PATCH 3.10 049/319] x86/paravirt: Do not trace _paravirt_ident_*() functions Willy Tarreau <w@1wt.eu> - 2017-02-05 21:10 +0100
  [PATCH 3.10 068/319] powerpc/nvram: Fix an incorrect partition merge Willy Tarreau <w@1wt.eu> - 2017-02-05 21:10 +0100
  [PATCH 3.10 075/319] microblaze: fix copy_from_user() Willy Tarreau <w@1wt.eu> - 2017-02-05 21:10 +0100
  [PATCH 3.10 041/319] xen/pciback: For XEN_PCI_OP_disable_msi[|x] only disable if device has MSI(X) enabled. Willy Tarreau <w@1wt.eu> - 2017-02-05 21:20 +0100
  [PATCH 3.10 085/319] blackfin: fix copy_from_user() Willy Tarreau <w@1wt.eu> - 2017-02-05 21:20 +0100
  [PATCH 3.10 046/319] x86/mm: Disable preemption during CR3 read+write Willy Tarreau <w@1wt.eu> - 2017-02-05 21:20 +0100
  [PATCH 3.10 048/319] x86/mm/pat, /dev/mem: Remove superfluous error message Willy Tarreau <w@1wt.eu> - 2017-02-05 21:20 +0100
  [PATCH 3.10 081/319] ARC: uaccess: get_user to zero out dest in cause of fault Willy Tarreau <w@1wt.eu> - 2017-02-05 21:20 +0100
  [PATCH 3.10 037/319] xen/pciback: Save xen_pci_op commands before processing it Willy Tarreau <w@1wt.eu> - 2017-02-05 21:20 +0100
    Re: [PATCH 3.10 037/319] xen/pciback: Save xen_pci_op commands  before processing it Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> - 2017-02-06 15:40 +0100
      Re: [PATCH 3.10 037/319] xen/pciback: Save xen_pci_op commands  before processing it Willy Tarreau <w@1wt.eu> - 2017-02-06 23:40 +0100
  [PATCH 3.10 050/319] x86/build: Build compressed x86 kernels as PIE Willy Tarreau <w@1wt.eu> - 2017-02-05 21:20 +0100
  [PATCH 3.10 057/319] ARM: sa1111: fix pcmcia suspend/resume Willy Tarreau <w@1wt.eu> - 2017-02-05 21:20 +0100
  [PATCH 3.10 032/319] x86/mm/xen: Suppress hugetlbfs in PV guests Willy Tarreau <w@1wt.eu> - 2017-02-05 21:20 +0100
  [PATCH 3.10 098/319] ppc32: fix copy_from_user() Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 053/319] iommu/amd: Free domain id when free a domain of struct dma_ops_domain Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 099/319] ia64: copy_from_user() should zero the destination on access_ok() failure Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 092/319] metag: copy_from_user() should zero the destination on access_ok() failure Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 047/319] x86/apic: Do not init irq remapping if ioapic is disabled Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 066/319] powerpc/powernv: Use CPU-endian PEST in pnv_pci_dump_p7ioc_diag_data() Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 058/319] arm64: avoid returning from bad_mode Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 065/319] powerpc/vdso64: Use double word compare on pointers Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 031/319] ppp: defer netns reference release for ppp channel Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 064/319] powerpc/mm: Don't alias user region to other regions below PAGE_OFFSET Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 035/319] xen-netback: use RING_COPY_REQUEST() throughout Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 062/319] MIPS: Malta: Fix IOCU disable switch read for MIPS64 Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 033/319] xen: Add RING_COPY_REQUEST() Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 078/319] sh64: failing __get_user() should zero Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 089/319] mips: copy_from_user() must zero the destination on access_ok() failure Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 036/319] xen-blkback: only read request operation from shared ring once Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 083/319] frv: fix clear_user() Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 080/319] s390: get_user() should zero on failure Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 073/319] parisc: Ensure consistent state when switching to kernel stack at syscall entry Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 038/319] xen/pciback: Return error on XEN_PCI_OP_enable_msi when device has MSI or MSI-X enabled Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 054/319] ARM: 8616/1: dt: Respect property size when parsing CPUs Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 088/319] hexagon: fix strncpy_from_user() error return Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 052/319] iommu/amd: Update Alias-DTE in update_device_table() Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 039/319] xen/pciback: Return error on XEN_PCI_OP_enable_msix when device has MSI or MSI-X enabled Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 090/319] asm-generic: make copy_from_user() zero the destination properly Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 044/319] x86/xen: fix upper bound of pmd loop in xen_cleanhighmap() Willy Tarreau <w@1wt.eu> - 2017-02-05 21:30 +0100
  [PATCH 3.10 045/319] x86/traps: Ignore high word of regs->cs in early_idt_handler_common Willy Tarreau <w@1wt.eu> - 2017-02-05 21:40 +0100
  [PATCH 3.10 042/319] xen/pciback: Don't allow MSI-X ops if PCI_COMMAND_MEMORY is not set. Willy Tarreau <w@1wt.eu> - 2017-02-05 21:40 +0100
  [PATCH 3.10 061/319] arm64: debug: avoid resetting stepping state machine when TIF_SINGLESTEP Willy Tarreau <w@1wt.eu> - 2017-02-05 21:40 +0100
  [PATCH 3.10 097/319] sparc32: fix copy_from_user() Willy Tarreau <w@1wt.eu> - 2017-02-05 21:40 +0100
  [PATCH 3.10 051/319] x86/um: reuse asm-generic/barrier.h Willy Tarreau <w@1wt.eu> - 2017-02-05 21:40 +0100
  [PATCH 3.10 043/319] xen-pciback: Add name prefix to global 'permissive' variable Willy Tarreau <w@1wt.eu> - 2017-02-05 21:40 +0100

csiph-web