Re: [Qemu-devel] [PATCH arm-devs v1 05/13] net/cadence_gem: Prefetch rx descriptors ASAP

[Peter Crosthwaite]

On Mon, Dec 2, 2013 at 10:14 PM, Peter Maydell <address@hidden> wrote:
> On 2 December 2013 07:11, Peter Crosthwaite
> <address@hidden> wrote:
>> The real hardware prefetches rx buffer descriptors ASAP and
>> potentially throws relevant interrupts following the fetch
>> even in the absence of a recieved packet.
>>
>> Reported-by: Deepika Dhamija <address@hidden>
>> Signed-off-by: Peter Crosthwaite <address@hidden>
>> ---
>>
>>  hw/net/cadence_gem.c | 64 
>> +++++++++++++++++++++++++++++-----------------------
>>  1 file changed, 36 insertions(+), 28 deletions(-)
>>
>> diff --git a/hw/net/cadence_gem.c b/hw/net/cadence_gem.c
>> index 73ac0d8..e5a6d87 100644
>> --- a/hw/net/cadence_gem.c
>> +++ b/hw/net/cadence_gem.c
>> @@ -346,6 +346,8 @@ typedef struct GemState {
>>      uint32_t rx_desc_addr;
>>      uint32_t tx_desc_addr;
>>
>> +    unsigned rx_desc[2];
>> +
>>  } GemState;
>>
>>  /* The broadcast MAC address: 0xFFFFFFFFFFFF */
>> @@ -579,13 +581,30 @@ static int gem_mac_address_filter(GemState *s, const 
>> uint8_t *packet)
>>      return GEM_RX_REJECT;
>>  }
>>
>> +static void gem_get_rx_desc(GemState *s)
>> +{
>> +        DB_PRINT("read descriptor 0x%x\n", (unsigned)s->rx_desc_addr);
>> +        /* read current descriptor */
>> +        cpu_physical_memory_read(s->rx_desc_addr,
>> +                                 (uint8_t *)s->rx_desc, sizeof(s->rx_desc));
>> +
>> +        /* Descriptor owned by software ? */
>> +        if (rx_desc_get_ownership(s->rx_desc) == 1) {
>> +            DB_PRINT("descriptor 0x%x owned by sw.\n",
>> +                     (unsigned)s->rx_desc_addr);
>> +            s->regs[GEM_RXSTATUS] |= GEM_RXSTATUS_NOBUF;
>> +            s->regs[GEM_ISR] |= GEM_INT_RXUSED & ~(s->regs[GEM_IMR]);
>> +            /* Handle interrupt consequences */
>> +            gem_update_int_status(s);
>> +        }
>> +}
>> +
>
> Looks OK codewise but your indent here is wrong...
>

Fixed,

Regards,
Peter

> thanks
> -- PMM
>