Re: [Qemu-devel] [PATCH v3 4/6] spapr_pci: enumerate and add PCI device tree

[Thomas Huth]

On Tue,  5 May 2015 14:23:54 +0530
Nikunj A Dadhania <address@hidden> wrote:

> All the PCI enumeration and device node creation was off-loaded to
> SLOF. With PCI hotplug support, code needed to be added to add device
> node. This creates multiple copy of the code one in SLOF and other in
> hotplug code. To unify this, the patch adds the pci device node
> creation in Qemu. For backward compatibility, a flag
> "qemu,phb-enumerated" is added to the phb, suggesting to SLOF to not
> do device node creation.
> 
> Signed-off-by: Nikunj A Dadhania <address@hidden>
> ---
>  hw/ppc/spapr_pci.c | 108 
> ++++++++++++++++++++++++++++++++++++++++++++++++++---
>  1 file changed, 103 insertions(+), 5 deletions(-)
> 
> diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
> index 8b02a3e..103284a 100644
> --- a/hw/ppc/spapr_pci.c
> +++ b/hw/ppc/spapr_pci.c
> @@ -23,6 +23,7 @@
>   * THE SOFTWARE.
>   */
>  #include "hw/hw.h"
> +#include "hw/sysbus.h"
>  #include "hw/pci/pci.h"
>  #include "hw/pci/msi.h"
>  #include "hw/pci/msix.h"
> @@ -35,6 +36,7 @@
>  #include "qemu/error-report.h"
>  #include "qapi/qmp/qerror.h"
>  
> +#include "hw/pci/pci_bridge.h"
>  #include "hw/pci/pci_bus.h"
>  #include "hw/ppc/spapr_drc.h"
>  #include "sysemu/device_tree.h"
> @@ -945,7 +947,10 @@ static int spapr_populate_pci_child_dt(PCIDevice *dev, 
> void *fdt, int offset,
>       * processed by OF beforehand
>       */
>      _FDT(fdt_setprop_string(fdt, offset, "name", "pci"));
> -    _FDT(fdt_setprop(fdt, offset, "ibm,loc-code", drc_name, 
> strlen(drc_name)));
> +    if (drc_name) {
> +        _FDT(fdt_setprop(fdt, offset, "ibm,loc-code", drc_name,
> +                         strlen(drc_name)));
> +    }
>      _FDT(fdt_setprop_cell(fdt, offset, "ibm,my-drc-index", drc_index));
>  
>      _FDT(fdt_setprop_cell(fdt, offset, "#address-cells",
> @@ -1001,10 +1006,6 @@ static void spapr_phb_add_pci_device(sPAPRDRConnector 
> *drc,
>      void *fdt = NULL;
>      int fdt_start_offset = 0;
>  
> -    /* boot-time devices get their device tree node created by SLOF, but for
> -     * hotplugged devices we need QEMU to generate it so the guest can fetch
> -     * it via RTAS
> -     */
>      if (dev->hotplugged) {
>          fdt = spapr_create_pci_child_dt(phb, pdev, drc_index, drc_name,
>                                          &fdt_start_offset);
> @@ -1482,6 +1483,89 @@ PCIHostState *spapr_create_phb(sPAPREnvironment 
> *spapr, int index)
>      return PCI_HOST_BRIDGE(dev);
>  }
>  
> +typedef struct sPAPRFDT {
> +    void *fdt;
> +    int node_off;
> +    uint32_t index;
> +} sPAPRFDT;
> +
> +static void spapr_populate_pci_devices_dt(PCIBus *bus, PCIDevice *pdev,
> +                                          void *opaque)
> +{
> +    PCIBus *sec_bus;
> +    sPAPRFDT *p = opaque;
> +    int ret, offset;
> +    int slot = PCI_SLOT(pdev->devfn);
> +    int func = PCI_FUNC(pdev->devfn);
> +    char nodename[512];

That's quite a big array ....

> +    sPAPRFDT s_fdt;
> +
> +    if (func) {
> +        sprintf(nodename, "address@hidden,%d", slot, func);
> +    } else {
> +        sprintf(nodename, "address@hidden", slot);
> +    }

... just for holding such a small string. Could you maybe use
a smaller array size for nodename (especially since you call this
function recursively)?

> +    offset = fdt_add_subnode(p->fdt, p->node_off, nodename);
> +    ret = spapr_populate_pci_child_dt(pdev, p->fdt, offset, p->index, 0, 
> NULL);

The above code sequence looks pretty much similar to
spapr_create_pci_child_dt() ... maybe it's worth the effort to merge
the common code of both functions into a separate helper function
to avoid the duplication? ... not sure if this is worth the effort,
it's just a suggestion.

> +    g_assert(!ret);

You know that assert statements can simply be disabled by a
preprocessor switch - and in that case there is no more error handling
here at all and the code continues silently with a partial initialized
node! So it's maybe better to do a proper error handling here instead?

> +    if ((pci_default_read_config(pdev, PCI_HEADER_TYPE, 1) !=
> +         PCI_HEADER_TYPE_BRIDGE)) {
> +        return;
> +    }
> +
> +    sec_bus = pci_bridge_get_sec_bus(PCI_BRIDGE(pdev));
> +    if (!sec_bus) {
> +        return;
> +    }
> +
> +    s_fdt.fdt = p->fdt;
> +    s_fdt.node_off = offset;
> +    s_fdt.index = p->index;
> +    pci_for_each_device(sec_bus, pci_bus_num(sec_bus),
> +                        spapr_populate_pci_devices_dt,
> +                        &s_fdt);
> +}
> +
> +static void spapr_phb_pci_enumerate_bridge(PCIBus *bus, PCIDevice *pdev,
> +                                           void *opaque)
> +{
> +    unsigned short *bus_no = (unsigned short *) opaque;

opaque is a void pointer, so I think you don't need the typecast here.

> +    unsigned short primary = *bus_no;
> +    unsigned short secondary;
> +    unsigned short subordinate = 0xff;

Is there a special reason for using unsigned short variables here?
"unsigned int" would sound much more natural to me.

> +    if ((pci_default_read_config(pdev, PCI_HEADER_TYPE, 1) ==
> +         PCI_HEADER_TYPE_BRIDGE)) {
> +        PCIBus *sec_bus = pci_bridge_get_sec_bus(PCI_BRIDGE(pdev));
> +        secondary = *bus_no + 1;
> +        pci_default_write_config(pdev, PCI_PRIMARY_BUS, primary, 1);
> +        pci_default_write_config(pdev, PCI_SECONDARY_BUS, secondary, 1);
> +        pci_default_write_config(pdev, PCI_SUBORDINATE_BUS, secondary, 1);
> +        *bus_no = *bus_no + 1;
> +        if (sec_bus) {
> +            pci_default_write_config(pdev, PCI_PRIMARY_BUS, primary, 1);
> +            pci_default_write_config(pdev, PCI_SECONDARY_BUS, secondary, 1);
> +            pci_default_write_config(pdev, PCI_SUBORDINATE_BUS, subordinate, 
> 1);

You write all value here again ... I think you could either drop the
write to the PRIMARY and SECONDARY BUS registers, or you could use a
proper if-else instead.

> +            pci_for_each_device(sec_bus, pci_bus_num(sec_bus),
> +                                spapr_phb_pci_enumerate_bridge,
> +                                bus_no);
> +            pci_default_write_config(pdev, PCI_SUBORDINATE_BUS, *bus_no, 1);
> +        }
> +    }
> +}
> +
> +static void spapr_phb_pci_enumerate(sPAPRPHBState *phb)
> +{
> +    PCIBus *bus = PCI_HOST_BRIDGE(phb)->bus;
> +    unsigned short bus_no = 0;

Again, why "short" ?

> +    pci_for_each_device(bus, pci_bus_num(bus),
> +                        spapr_phb_pci_enumerate_bridge,
> +                        &bus_no);
> +
> +}
> +
>  int spapr_populate_pci_dt(sPAPRPHBState *phb,
>                            uint32_t xics_phandle,
>                            void *fdt)
> @@ -1521,6 +1605,8 @@ int spapr_populate_pci_dt(sPAPRPHBState *phb,
>          cpu_to_be32(b_ddddd(-1)|b_fff(0)), 0x0, 0x0, cpu_to_be32(-1)};
>      uint32_t interrupt_map[PCI_SLOT_MAX * PCI_NUM_PINS][7];
>      sPAPRTCETable *tcet;
> +    PCIBus *bus = PCI_HOST_BRIDGE(phb)->bus;
> +    sPAPRFDT s_fdt;
>  
>      /* Start populating the FDT */
>      sprintf(nodename, "address@hidden" PRIx64, phb->buid);
> @@ -1570,6 +1656,18 @@ int spapr_populate_pci_dt(sPAPRPHBState *phb,
>                   tcet->liobn, tcet->bus_offset,
>                   tcet->nb_table << tcet->page_shift);
>  
> +    /* Walk the bridges and program the bus numbers*/
> +    spapr_phb_pci_enumerate(phb);
> +    _FDT(fdt_setprop_cell(fdt, bus_off, "qemu,phb-enumerated", 0x1));
> +
> +    /* Populate tree nodes with PCI devices attached */
> +    s_fdt.fdt = fdt;
> +    s_fdt.node_off = bus_off;
> +    s_fdt.index = phb->index;
> +    pci_for_each_device(bus, pci_bus_num(bus),
> +                        spapr_populate_pci_devices_dt,
> +                        &s_fdt);
> +
>      ret = spapr_drc_populate_dt(fdt, bus_off, OBJECT(phb),
>                                  SPAPR_DR_CONNECTOR_TYPE_PCI);
>      if (ret) {

 Thomas