Re: [Qemu-devel] [PATCH 36/39] memory: move bitmap synchronization to its own function

[Juan Quintela]

Paolo Bonzini <address@hidden> wrote:
> Il 06/11/2013 14:04, Juan Quintela ha scritto:
>> We want to have all the functions that handle directly the dirty
>> bitmap near.  We will change it later.
>
> Please move it to exec.c instead.
>
>> Signed-off-by: Juan Quintela <address@hidden>
>> ---
>>  include/exec/memory-physical.h | 31 +++++++++++++++++++++++++++++++
>>  kvm-all.c                      | 27 ++-------------------------
>>  2 files changed, 33 insertions(+), 25 deletions(-)
>> 
>> diff --git a/include/exec/memory-physical.h b/include/exec/memory-physical.h
>> index 610c55f..72faf06 100644
>> --- a/include/exec/memory-physical.h
>> +++ b/include/exec/memory-physical.h
>> @@ -72,6 +72,37 @@ static inline void 
>> cpu_physical_memory_set_dirty_range(ram_addr_t start,
>>      xen_modified_memory(start, length);
>>  }
>> 
>> +static inline void cpu_physical_memory_set_dirty_lebitmap(unsigned long 
>> *bitmap,
>
> Why "le"?

little endian,  name is already too long (cpu_physical_memory_ is too
big as a preffix.)

>
> Paolo
>
>> +                                                          ram_addr_t start,
>> +                                                          ram_addr_t pages)
>> +{
>> +    unsigned int i, j;
>> +    unsigned long page_number, c;
>> +    hwaddr addr;
>> +    ram_addr_t ram_addr;
>> +    unsigned int len = (pages + HOST_LONG_BITS - 1) / HOST_LONG_BITS;
>> +    unsigned long hpratio = getpagesize() / TARGET_PAGE_SIZE;
>> +
>> +    /*
>> +     * bitmap-traveling is faster than memory-traveling (for addr...)
>> +     * especially when most of the memory is not dirty.
>> +     */
>> +    for (i = 0; i < len; i++) {
>> +        if (bitmap[i] != 0) {
>> +            c = leul_to_cpu(bitmap[i]);

this is the important part.  KVM maintains the bitmap is Little Endian
mode,  and PPC?  I assme needs this.


>> +            do {
>> +                j = ffsl(c) - 1;
>> +                c &= ~(1ul << j);
>> +                page_number = (i * HOST_LONG_BITS + j) * hpratio;
>> +                addr = page_number * TARGET_PAGE_SIZE;
>> +                ram_addr = start + addr;
>> +                cpu_physical_memory_set_dirty_range(ram_addr,
>> +                                                    TARGET_PAGE_SIZE
>> * hpratio);

I checked,  and it appears that ppc really use the hpratio thing.  I
hope that the bitmap optimization also works for ppc,  but its
architecture is weird^Winteresting O:-)

Alex?