[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH 8/8] sun4u: move in-built devices behind PCI bri
From: |
Artyom Tarasenko |
Subject: |
Re: [Qemu-devel] [PATCH 8/8] sun4u: move in-built devices behind PCI bridge A |
Date: |
Wed, 12 Jul 2017 11:59:28 +0200 |
On Tue, Jul 11, 2017 at 11:53 PM, Mark Cave-Ayland
<address@hidden> wrote:
> This switches the sun4u model to being much closer to a real Ultra 5.
>
> Since the existing code previously bypassed the PCI bridge interrupt
> swizzling, reorganise the interrupt mapping functions so that
> pci_pbm_map_irq() is used by the PCI bridges and pci_apb_map_irq() is
> used by the PCI host bridge.
>
> As part of this change we also combine the "onboard" NIC and the ebus into
> a single multi-function device as done on a real Ultra 5.
While they are combined on a real Ultra 5, it has a different NIC.
Are the guest OSes smart enough to use NE2000 as an onboard device?
> Finally we mark the physically unavailable slots (plus slot 0 in busA) as
> reserved to ensure that users can't plug devices into non-existent slots
> which will break interrupt routing.
>
> Signed-off-by: Mark Cave-Ayland <address@hidden>
> ---
> hw/pci-host/apb.c | 35 +++++++++++++++++++++++++++--------
> hw/sparc64/sun4u.c | 25 ++++++++++++++++++++-----
> 2 files changed, 47 insertions(+), 13 deletions(-)
>
> diff --git a/hw/pci-host/apb.c b/hw/pci-host/apb.c
> index f9badad..5000432 100644
> --- a/hw/pci-host/apb.c
> +++ b/hw/pci-host/apb.c
> @@ -601,16 +601,35 @@ static uint64_t apb_pci_config_read(void *opaque,
> hwaddr addr,
> /* The APB host has an IRQ line for each IRQ line of each slot. */
> static int pci_apb_map_irq(PCIDevice *pci_dev, int irq_num)
> {
> - return ((pci_dev->devfn & 0x18) >> 1) + irq_num;
> + /* Return the irq as swizzled by the PBM */
> + return irq_num;
> }
>
> static int pci_pbm_map_irq(PCIDevice *pci_dev, int irq_num)
> {
> + PBMPCIBridge *br = PBM_PCI_BRIDGE(pci_bridge_get_device(
> + PCI_BUS(qdev_get_parent_bus(DEVICE(pci_dev)))));
> +
> int bus_offset;
> - if (pci_dev->devfn & 1)
> - bus_offset = 16;
> - else
> - bus_offset = 0;
> + if (br->busA) {
> + bus_offset = 0x0;
> +
> + /* The on-board devices have fixed (legacy) OBIO intnos */
> + switch (PCI_SLOT(pci_dev->devfn)) {
> + case 1:
> + /* Onboard NIC */
> + return 0x21;
> + case 3:
> + /* Onboard IDE */
> + return 0x20;
> +
> + default:
> + /* Normal intno, fall through */
> + break;
> + }
> + } else {
> + bus_offset = 0x10;
> + }
> return (bus_offset + (PCI_SLOT(pci_dev->devfn) << 2) + irq_num) & 0x1f;
> }
>
> @@ -692,7 +711,7 @@ PCIBus *pci_apb_init(hwaddr special_base,
> d = APB_DEVICE(dev);
> phb = PCI_HOST_BRIDGE(dev);
> phb->bus = pci_register_bus(DEVICE(phb), "pci",
> - pci_apb_set_irq, pci_pbm_map_irq, d,
> + pci_apb_set_irq, pci_apb_map_irq, d,
> &d->pci_mmio,
> get_system_io(),
> 0, 32, TYPE_PCI_BUS);
> @@ -726,14 +745,14 @@ PCIBus *pci_apb_init(hwaddr special_base,
> pci_dev = pci_create_multifunction(phb->bus, PCI_DEVFN(1, 0), true,
> TYPE_PBM_PCI_BRIDGE);
> br = PCI_BRIDGE(pci_dev);
> - pci_bridge_map_irq(br, "pciB", pci_apb_map_irq);
> + pci_bridge_map_irq(br, "pciB", pci_pbm_map_irq);
> qdev_init_nofail(&pci_dev->qdev);
> *busB = pci_bridge_get_sec_bus(br);
>
> pci_dev = pci_create_multifunction(phb->bus, PCI_DEVFN(1, 1), true,
> TYPE_PBM_PCI_BRIDGE);
> br = PCI_BRIDGE(pci_dev);
> - pci_bridge_map_irq(br, "pciA", pci_apb_map_irq);
> + pci_bridge_map_irq(br, "pciA", pci_pbm_map_irq);
> qdev_prop_set_bit(DEVICE(pci_dev), "busA", true);
> qdev_init_nofail(&pci_dev->qdev);
> *busA = pci_bridge_get_sec_bus(br);
> diff --git a/hw/sparc64/sun4u.c b/hw/sparc64/sun4u.c
> index 3bb3bf2..b8b96be 100644
> --- a/hw/sparc64/sun4u.c
> +++ b/hw/sparc64/sun4u.c
> @@ -27,6 +27,7 @@
> #include "cpu.h"
> #include "hw/hw.h"
> #include "hw/pci/pci.h"
> +#include "hw/pci/pci_bus.h"
> #include "hw/pci-host/apb.h"
> #include "hw/i386/pc.h"
> #include "hw/char/serial.h"
> @@ -42,6 +43,7 @@
> #include "hw/nvram/fw_cfg.h"
> #include "hw/sysbus.h"
> #include "hw/ide.h"
> +#include "hw/ide/pci.h"
> #include "hw/loader.h"
> #include "elf.h"
> #include "qemu/cutils.h"
> @@ -448,10 +450,17 @@ static void sun4uv_init(MemoryRegion *address_space_mem,
> ivec_irqs = qemu_allocate_irqs(sparc64_cpu_set_ivec_irq, cpu, IVEC_MAX);
> pci_bus = pci_apb_init(APB_SPECIAL_BASE, APB_MEM_BASE, ivec_irqs,
> &pci_busA,
> &pci_busB, &pbm_irqs);
> - pci_vga_init(pci_bus);
>
> - /* XXX Should be pci_busA */
> - ebus = pci_create_simple(pci_bus, -1, "ebus");
> + /* Only in-built Simba PBMs can exist on the root bus, slot 0 on busA is
> + reserved (leaving no slots free after on-board devices) leaving slots
> + 0-3 are free on busB 4*/
> + pci_bus->slot_reserved_mask = 0xfffffffc;
> + pci_busA->slot_reserved_mask = 0xfffffff1;
> + pci_busB->slot_reserved_mask = 0xfffffff0;
> +
> + ebus = pci_create_multifunction(pci_busA, PCI_DEVFN(1, 0), true, "ebus");
> + qdev_init_nofail(DEVICE(ebus));
> +
> isa_bus = pci_ebus_init(ebus, pbm_irqs);
>
> i = 0;
> @@ -464,14 +473,20 @@ static void sun4uv_init(MemoryRegion *address_space_mem,
> serial_hds_isa_init(isa_bus, i, MAX_SERIAL_PORTS);
> parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
>
> - pci_dev = pci_create(pci_bus, -1, "ne2k_pci");
> + pci_dev = pci_create_simple(pci_busA, PCI_DEVFN(2, 0), "VGA");
> +
> + pci_dev = pci_create_multifunction(pci_busA, PCI_DEVFN(1, 1), true,
> + "ne2k_pci");
> dev = &pci_dev->qdev;
> qdev_set_nic_properties(dev, &nd_table[0]);
> qdev_init_nofail(dev);
>
> ide_drive_get(hd, ARRAY_SIZE(hd));
>
> - pci_cmd646_ide_init(pci_bus, hd, 1);
> + pci_dev = pci_create(pci_busA, PCI_DEVFN(3, 0), "cmd646-ide");
> + qdev_prop_set_uint32(&pci_dev->qdev, "secondary", 1);
> + qdev_init_nofail(&pci_dev->qdev);
> + pci_ide_create_devs(pci_dev, hd);
>
> isa_create_simple(isa_bus, "i8042");
>
> --
> 1.7.10.4
>
--
Regards,
Artyom Tarasenko
SPARC and PPC PReP under qemu blog: http://tyom.blogspot.com/search/label/qemu
- [Qemu-devel] [PATCH 0/8] sun4u: change PCI topology to better match a real Ultra 5, Mark Cave-Ayland, 2017/07/11
- [Qemu-devel] [PATCH 2/8] sun4u: switch to using qdev to instantiate fw_cfg interface, Mark Cave-Ayland, 2017/07/11
- [Qemu-devel] [PATCH 4/8] apb: fix up PCI bus nomenclature, Mark Cave-Ayland, 2017/07/11
- [Qemu-devel] [PATCH 3/8] sun4u: expose fw_cfg and NVRAM on ebus PCI IO address space, Mark Cave-Ayland, 2017/07/11
- [Qemu-devel] [PATCH 7/8] sun4u: create single default onboard ne2k_pci NIC for machine, Mark Cave-Ayland, 2017/07/11
- [Qemu-devel] [PATCH 6/8] apb: add busA qdev property to PBM PCI bridge, Mark Cave-Ayland, 2017/07/11
- [Qemu-devel] [PATCH 8/8] sun4u: move in-built devices behind PCI bridge A, Mark Cave-Ayland, 2017/07/11
- Re: [Qemu-devel] [PATCH 8/8] sun4u: move in-built devices behind PCI bridge A,
Artyom Tarasenko <=
- [Qemu-devel] [PATCH 1/8] sun4u: pass PCIDevice into pci_ebus_init() instead of PCIBus, Mark Cave-Ayland, 2017/07/11
- [Qemu-devel] [PATCH 5/8] apb: fix endianness for APB and PCI config accesses, Mark Cave-Ayland, 2017/07/11
- Re: [Qemu-devel] [PATCH 0/8] sun4u: change PCI topology to better match a real Ultra 5, no-reply, 2017/07/11