Re: [Qemu-devel] [PATCH 6/7] aspeed: add a max_ram_size property to the memory controller

[Cédric Le Goater]

On 08/16/2018 02:48 PM, Peter Maydell wrote:
> On 7 August 2018 at 08:57, Joel Stanley <address@hidden> wrote:
>> From: Cédric Le Goater <address@hidden>
>>
>> This will be used to construct a memory region beyond the RAM region
>> to let firmwares scan the address space with load/store to guess how
>> much RAM the SoC has.
>>
>> Signed-off-by: Cédric Le Goater <address@hidden>
>> Signed-off-by: Joel Stanley <address@hidden>
>> ---
>>  hw/arm/aspeed.c               | 31 +++++++++++++++++++++++++++++++
>>  hw/arm/aspeed_soc.c           |  2 ++
>>  hw/misc/aspeed_sdmc.c         |  3 +++
>>  include/hw/misc/aspeed_sdmc.h |  1 +
>>  4 files changed, 37 insertions(+)
>>
>> diff --git a/hw/arm/aspeed.c b/hw/arm/aspeed.c
>> index bb9d33848d3f..e078269266bc 100644
>> --- a/hw/arm/aspeed.c
>> +++ b/hw/arm/aspeed.c
>> @@ -31,6 +31,7 @@ static struct arm_boot_info aspeed_board_binfo = {
>>  typedef struct AspeedBoardState {
>>      AspeedSoCState soc;
>>      MemoryRegion ram;
>> +    MemoryRegion max_ram;
>>  } AspeedBoardState;
>>
>>  typedef struct AspeedBoardConfig {
>> @@ -127,6 +128,27 @@ static const AspeedBoardConfig aspeed_boards[] = {
>>      },
>>  };
>>
>> +/*
>> + * The max ram region is for firmwares that scan the address space
>> + * with load/store to guess how much RAM the SoC has.
>> + */
>> +static uint64_t max_ram_read(void *opaque, hwaddr offset, unsigned size)
>> +{
>> +    return 0;
>> +}
>> +
>> +static void max_ram_write(void *opaque, hwaddr offset, uint64_t value,
>> +                           unsigned size)
>> +{
>> +    /* Disacard writes */
>> +}
>> +
>> +static const MemoryRegionOps max_ram_ops = {
>> +    .read = max_ram_read,
>> +    .write = max_ram_write,
>> +    .endianness = DEVICE_NATIVE_ENDIAN,
>> +};
>> +
>>  #define FIRMWARE_ADDR 0x0
>>
>>  static void write_boot_rom(DriveInfo *dinfo, hwaddr addr, size_t rom_size,
>> @@ -187,6 +209,7 @@ static void aspeed_board_init(MachineState *machine,
>>      AspeedBoardState *bmc;
>>      AspeedSoCClass *sc;
>>      DriveInfo *drive0 = drive_get(IF_MTD, 0, 0);
>> +    ram_addr_t max_ram_size;
>>
>>      bmc = g_new0(AspeedBoardState, 1);
>>      object_initialize(&bmc->soc, (sizeof(bmc->soc)), cfg->soc_name);
>> @@ -226,6 +249,14 @@ static void aspeed_board_init(MachineState *machine,
>>      object_property_add_const_link(OBJECT(&bmc->soc), "ram", 
>> OBJECT(&bmc->ram),
>>                                     &error_abort);
>>
>> +    max_ram_size = object_property_get_uint(OBJECT(&bmc->soc), 
>> "max-ram-size",
>> +                                            &error_abort);
>> +    memory_region_init_io(&bmc->max_ram, NULL, &max_ram_ops, NULL,
>> +                          "max_ram", max_ram_size  - ram_size);
>> +    memory_region_add_subregion(get_system_memory(),
>> +                                sc->info->sdram_base + ram_size,
>> +                                &bmc->max_ram);
> 
> I'm surprised that you need the IO ops, ie that it doesn't work
> just to define an empty container region with memory_region_init().

Initially, there was some logging in the IO ops, which was a nice to have.
But that was dropped in this patch. No a big issue I think.

Thanks,

C.