[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [RFC PATCH V2 2/2] msi: Handle remappable format interr
From: |
Anthony PERARD |
Subject: |
Re: [Qemu-devel] [RFC PATCH V2 2/2] msi: Handle remappable format interrupt request |
Date: |
Fri, 19 May 2017 12:57:33 +0100 |
User-agent: |
Mutt/1.8.2 (2017-04-18) |
On Thu, May 18, 2017 at 01:33:00AM -0400, Lan Tianyu wrote:
> From: Chao Gao <address@hidden>
>
> According to VT-d spec Interrupt Remapping and Interrupt Posting ->
> Interrupt Remapping -> Interrupt Request Formats On Intel 64
> Platforms, fields of MSI data register have changed. This patch
> avoids wrongly regarding a remappable format interrupt request as
> an interrupt binded with an event channel.
>
> Signed-off-by: Chao Gao <address@hidden>
> Signed-off-by: Lan Tianyu <address@hidden>
> ---
> hw/pci/msi.c | 5 +++--
> hw/pci/msix.c | 4 +++-
> hw/xen/xen_pt_msi.c | 2 +-
> include/hw/xen/xen.h | 2 +-
> xen-hvm-stub.c | 2 +-
> xen-hvm.c | 7 ++++++-
> 6 files changed, 15 insertions(+), 7 deletions(-)
>
> diff --git a/hw/pci/msi.c b/hw/pci/msi.c
> index a87b227..199cb47 100644
> --- a/hw/pci/msi.c
> +++ b/hw/pci/msi.c
> @@ -289,7 +289,7 @@ void msi_reset(PCIDevice *dev)
> static bool msi_is_masked(const PCIDevice *dev, unsigned int vector)
> {
> uint16_t flags = pci_get_word(dev->config + msi_flags_off(dev));
> - uint32_t mask, data;
> + uint32_t mask, data, addr_lo;
> bool msi64bit = flags & PCI_MSI_FLAGS_64BIT;
> assert(vector < PCI_MSI_VECTORS_MAX);
>
> @@ -298,7 +298,8 @@ static bool msi_is_masked(const PCIDevice *dev, unsigned
> int vector)
> }
>
> data = pci_get_word(dev->config + msi_data_off(dev, msi64bit));
> - if (xen_is_pirq_msi(data)) {
> + addr_lo = pci_get_long(dev->config + msi_address_lo_off(dev));
> + if (xen_is_pirq_msi(data, addr_lo)) {
> return false;
> }
>
> diff --git a/hw/pci/msix.c b/hw/pci/msix.c
> index bb54e8b..efe2982 100644
> --- a/hw/pci/msix.c
> +++ b/hw/pci/msix.c
> @@ -82,9 +82,11 @@ static bool msix_vector_masked(PCIDevice *dev, unsigned
> int vector, bool fmask)
> {
> unsigned offset = vector * PCI_MSIX_ENTRY_SIZE;
> uint8_t *data = &dev->msix_table[offset + PCI_MSIX_ENTRY_DATA];
> + uint8_t *addr_lo = &dev->msix_table[offset + PCI_MSIX_ENTRY_LOWER_ADDR];
> /* MSIs on Xen can be remapped into pirqs. In those cases, masking
> * and unmasking go through the PV evtchn path. */
> - if (xen_enabled() && xen_is_pirq_msi(pci_get_long(data))) {
> + if (xen_enabled() && xen_is_pirq_msi(pci_get_long(data),
> + pci_get_long(addr_lo))) {
> return false;
> }
> return fmask || dev->msix_table[offset + PCI_MSIX_ENTRY_VECTOR_CTRL] &
> diff --git a/hw/xen/xen_pt_msi.c b/hw/xen/xen_pt_msi.c
> index 5fab95e..45a9e9f 100644
> --- a/hw/xen/xen_pt_msi.c
> +++ b/hw/xen/xen_pt_msi.c
> @@ -114,7 +114,7 @@ static int msi_msix_setup(XenPCIPassthroughState *s,
>
> assert((!is_msix && msix_entry == 0) || is_msix);
>
> - if (xen_is_pirq_msi(data)) {
> + if (xen_is_pirq_msi(data, addr)) {
> *ppirq = msi_ext_dest_id(addr >> 32) | msi_dest_id(addr);
> if (!*ppirq) {
> /* this probably identifies an misconfiguration of the guest,
> diff --git a/include/hw/xen/xen.h b/include/hw/xen/xen.h
> index 09c2ce5..af759bc 100644
> --- a/include/hw/xen/xen.h
> +++ b/include/hw/xen/xen.h
> @@ -33,7 +33,7 @@ int xen_pci_slot_get_pirq(PCIDevice *pci_dev, int irq_num);
> void xen_piix3_set_irq(void *opaque, int irq_num, int level);
> void xen_piix_pci_write_config_client(uint32_t address, uint32_t val, int
> len);
> void xen_hvm_inject_msi(uint64_t addr, uint32_t data);
> -int xen_is_pirq_msi(uint32_t msi_data);
> +int xen_is_pirq_msi(uint32_t msi_data, uint32_t msi_addr_lo);
Maybe inverting the arguments would be better, so the arguments would be
the address first, then the data, like I think it is often the case.
What do you think?
>
> qemu_irq *xen_interrupt_controller_init(void);
>
> diff --git a/xen-hvm-stub.c b/xen-hvm-stub.c
> index c500325..dae421c 100644
> --- a/xen-hvm-stub.c
> +++ b/xen-hvm-stub.c
> @@ -31,7 +31,7 @@ void xen_hvm_inject_msi(uint64_t addr, uint32_t data)
> {
> }
>
> -int xen_is_pirq_msi(uint32_t msi_data)
> +int xen_is_pirq_msi(uint32_t msi_data, uint32_t msi_addr_lo)
> {
> return 0;
> }
> diff --git a/xen-hvm.c b/xen-hvm.c
> index 5043beb..db29121 100644
> --- a/xen-hvm.c
> +++ b/xen-hvm.c
> @@ -146,8 +146,13 @@ void xen_piix_pci_write_config_client(uint32_t address,
> uint32_t val, int len)
> }
> }
>
> -int xen_is_pirq_msi(uint32_t msi_data)
> +int xen_is_pirq_msi(uint32_t msi_data, uint32_t msi_addr_lo)
> {
> + /* If msi address is configurate to remapping format, the msi will not
> + * remapped into a pirq.
What do you think of: "If the MSI address is configured in remappable
format, the MSI will not be remapped into a pirq." ?
> + */
> + if (msi_addr_lo & MSI_ADDR_IF_MASK)
> + return 0;
> /* If vector is 0, the msi is remapped into a pirq, passed as
> * dest_id.
> */
Thanks,
--
Anthony PERARD
Re: [Qemu-devel] [RFC PATCH V2 0/2] Qemu: Add Xen vIOMMU interrupt remapping function., no-reply, 2017/05/18