Re: [Qemu-devel] [PATCH v5 2/3] pc & q35: Add new machine opt max-ram-below-4g

[Michael S. Tsirkin]

On Fri, Jun 06, 2014 at 01:52:05PM -0400, Don Slutz wrote:
> This is a pc & q35 only machine opt.  One use is to allow for more
> ram in a 32bit guest for example:
> 
> -machine pc,max-ram-below-4g=3.75G
> 
> If you add enough PCI devices then all mmio for them will not fit
> below 4G which may not be the layout the user wanted. This allows
> you to increase the below 4G address space that PCI devices can use
> (aka decrease ram below 4G) and therefore in more cases not have any
> mmio that is above 4G.
> 
> For example using "-machine pc,max-ram-below-4g=2G" on the command
> line will limit the amount of ram that is below 4G to 2G.
> 
> Signed-off-by: Don Slutz <address@hidden>
> ---
> v5:
>   Re-work based on:
> 
>   https://github.com/imammedo/qemu/commits/memory-hotplug-v11
> 
> 
>  hw/i386/pc.c         | 38 ++++++++++++++++++++++++++++++++++++++
>  hw/i386/pc_piix.c    | 15 ++++++++++++---
>  hw/i386/pc_q35.c     | 15 ++++++++++++---
>  include/hw/i386/pc.h |  3 +++
>  vl.c                 |  4 ++++
>  5 files changed, 69 insertions(+), 6 deletions(-)
> 
> diff --git a/hw/i386/pc.c b/hw/i386/pc.c
> index 7cdba10..bccb746 100644
> --- a/hw/i386/pc.c
> +++ b/hw/i386/pc.c
> @@ -1644,11 +1644,49 @@ pc_machine_get_hotplug_memory_region_size(Object 
> *obj, Visitor *v, void *opaque,
>      visit_type_int(v, &value, name, errp);
>  }
>  
> +static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
> +                                         void *opaque, const char *name,
> +                                         Error **errp)
> +{
> +    PCMachineState *pcms = PC_MACHINE(obj);
> +    uint64_t value = pcms->max_ram_below_4g;
> +
> +    visit_type_size(v, &value, name, errp);
> +}
> +
> +static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
> +                                         void *opaque, const char *name,
> +                                         Error **errp)
> +{
> +    PCMachineState *pcms = PC_MACHINE(obj);
> +    Error *error = NULL;
> +    uint64_t value;
> +
> +    visit_type_size(v, &value, name, &error);
> +    if (error) {
> +        error_propagate(errp, error);
> +        return;
> +    }
> +    if (value > (1ULL << 32)) {
> +        error_set(&error, ERROR_CLASS_GENERIC_ERROR,
> +                  "Machine option 'max-ram-below-4g=%"PRIu64
> +                  "' expects size less then or equal to 4G", value);

less than

> +        error_propagate(errp, error);
> +        return;
> +    }
> +
> +    pcms->max_ram_below_4g = value;
> +}
> +
>  static void pc_machine_initfn(Object *obj)
>  {
>      object_property_add(obj, PC_MACHINE_MEMHP_REGION_SIZE, "int",
>                          pc_machine_get_hotplug_memory_region_size,
>                          NULL, NULL, NULL, NULL);
> +    object_property_add(obj, PC_MACHINE_MAX_RAM_BELOW_4G,  "size",
> +                        pc_machine_get_max_ram_below_4g,
> +                        pc_machine_set_max_ram_below_4g,
> +                        NULL, NULL, NULL);
>  }
>  
>  static void pc_machine_class_init(ObjectClass *oc, void *data)
> diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c
> index 40f6eaf..25f4727 100644
> --- a/hw/i386/pc_piix.c
> +++ b/hw/i386/pc_piix.c
> @@ -98,6 +98,13 @@ static void pc_init1(MachineState *machine,
>      DeviceState *icc_bridge;
>      FWCfgState *fw_cfg = NULL;
>      PcGuestInfo *guest_info;
> +    Object *mo = qdev_get_machine();
> +    PCMachineState *pcms = PC_MACHINE(mo);
> +    ram_addr_t lowmem = 0xe0000000;
> +
> +    if (pcms && pcms->max_ram_below_4g) {

Is pcms ever NULL? If yes why?

> +        lowmem = pcms->max_ram_below_4g;
> +    }
>  
>      /* Check whether RAM fits below 4G (leaving 1/2 GByte for IO memory).
>       * If it doesn't, we need to split it in chunks below and above 4G.
> @@ -106,8 +113,10 @@ static void pc_init1(MachineState *machine,
>       * For old machine types, use whatever split we used historically to 
> avoid
>       * breaking migration.
>       */
> -    if (machine->ram_size >= 0xe0000000) {
> -        ram_addr_t lowmem = gigabyte_align ? 0xc0000000 : 0xe0000000;
> +    if (machine->ram_size >= lowmem) {
> +        if (!(pcms && pcms->max_ram_below_4g) && gigabyte_align) {
> +            lowmem = 0xc0000000;
> +        }
>          above_4g_mem_size = machine->ram_size - lowmem;
>          below_4g_mem_size = lowmem;
>      } else {


So why do we need gigabyte_align anymore?
Can't we set property to 0xc0000000 by default, and
override for old machine types?

Also, a value that isn't a multiple of 1G will lead to bad
performance for large machines which do have above_4g_mem_size.
Let's detect and print a warning.



> @@ -122,7 +131,7 @@ static void pc_init1(MachineState *machine,
>      }
>  
>      icc_bridge = qdev_create(NULL, TYPE_ICC_BRIDGE);
> -    object_property_add_child(qdev_get_machine(), "icc-bridge",
> +    object_property_add_child(mo, "icc-bridge",
>                                OBJECT(icc_bridge), NULL);
>  
>      pc_cpus_init(machine->cpu_model, icc_bridge);
> diff --git a/hw/i386/pc_q35.c b/hw/i386/pc_q35.c
> index e28ce40..155cdf1 100644
> --- a/hw/i386/pc_q35.c
> +++ b/hw/i386/pc_q35.c
> @@ -85,6 +85,13 @@ static void pc_q35_init(MachineState *machine)
>      PCIDevice *ahci;
>      DeviceState *icc_bridge;
>      PcGuestInfo *guest_info;
> +    Object *mo = qdev_get_machine();
> +    PCMachineState *pcms = PC_MACHINE(mo);
> +    ram_addr_t lowmem = 0xb0000000;
> +
> +    if (pcms && pcms->max_ram_below_4g) {
> +        lowmem = pcms->max_ram_below_4g;
> +    }
>  
>      /* Check whether RAM fits below 4G (leaving 1/2 GByte for IO memory
>       * and 256 Mbytes for PCI Express Enhanced Configuration Access Mapping
> @@ -95,8 +102,10 @@ static void pc_q35_init(MachineState *machine)
>       * For old machine types, use whatever split we used historically to 
> avoid
>       * breaking migration.
>       */
> -    if (machine->ram_size >= 0xb0000000) {
> -        ram_addr_t lowmem = gigabyte_align ? 0x80000000 : 0xb0000000;
> +    if (machine->ram_size >= lowmem) {
> +        if (!(pcms && pcms->max_ram_below_4g) && gigabyte_align) {
> +            lowmem = 0x800000000;
> +        }
>          above_4g_mem_size = machine->ram_size - lowmem;
>          below_4g_mem_size = lowmem;
>      } else {
> @@ -111,7 +120,7 @@ static void pc_q35_init(MachineState *machine)
>      }
>  
>      icc_bridge = qdev_create(NULL, TYPE_ICC_BRIDGE);
> -    object_property_add_child(qdev_get_machine(), "icc-bridge",
> +    object_property_add_child(mo, "icc-bridge",
>                                OBJECT(icc_bridge), NULL);
>  
>      pc_cpus_init(machine->cpu_model, icc_bridge);
> diff --git a/include/hw/i386/pc.h b/include/hw/i386/pc.h
> index 19530bd..2d8b562 100644
> --- a/include/hw/i386/pc.h
> +++ b/include/hw/i386/pc.h
> @@ -32,10 +32,13 @@ struct PCMachineState {
>      MemoryRegion hotplug_memory;
>  
>      HotplugHandler *acpi_dev;
> +
> +    uint64_t max_ram_below_4g;
>  };
>  
>  #define PC_MACHINE_ACPI_DEVICE_PROP "acpi-device"
>  #define PC_MACHINE_MEMHP_REGION_SIZE "hotplug-memory-region-size"
> +#define PC_MACHINE_MAX_RAM_BELOW_4G "max-ram-below-4g"
>  
>  /**
>   * PCMachineClass:
> diff --git a/vl.c b/vl.c
> index 5e77a27..cffb9c5 100644
> --- a/vl.c
> +++ b/vl.c
> @@ -382,6 +382,10 @@ static QemuOptsList qemu_machine_opts = {
>              .name = "kvm-type",
>              .type = QEMU_OPT_STRING,
>              .help = "Specifies the KVM virtualization mode (HV, PR)",
> +        },{
> +            .name = PC_MACHINE_MAX_RAM_BELOW_4G,
> +            .type = QEMU_OPT_SIZE,
> +            .help = "maximum ram below the 4G boundary (32bit boundary)",
>          },
>          { /* End of list */ }
>      },
> -- 
> 1.8.4