Message ID | 1268736839-27371-2-git-send-email-tamura.yoshiaki@lab.ntt.co.jp |
---|---|
State | New |
Headers | show |
On 03/16/2010 12:53 PM, Yoshiaki Tamura wrote: > Replaces byte-based phys_ram_dirty bitmap with > three bit-based phys_ram_dirty bitmap. > On allocation, it sets all bits in the bitmap. > > Signed-off-by: Yoshiaki Tamura<tamura.yoshiaki@lab.ntt.co.jp> > Signed-off-by: OHMURA Kei<ohmura.kei@lab.ntt.co.jp> > --- > exec.c | 22 +++++++++++++++++----- > 1 files changed, 17 insertions(+), 5 deletions(-) > > diff --git a/exec.c b/exec.c > index 9bcb4de..ba334e7 100644 > --- a/exec.c > +++ b/exec.c > @@ -119,7 +119,9 @@ uint8_t *code_gen_ptr; > > #if !defined(CONFIG_USER_ONLY) > int phys_ram_fd; > -uint8_t *phys_ram_dirty; > +unsigned long *phys_ram_vga_dirty; > +unsigned long *phys_ram_code_dirty; > +unsigned long *phys_ram_migration_dirty; > Would be nice to make this an array.
Avi Kivity wrote: > On 03/16/2010 12:53 PM, Yoshiaki Tamura wrote: >> Replaces byte-based phys_ram_dirty bitmap with >> three bit-based phys_ram_dirty bitmap. >> On allocation, it sets all bits in the bitmap. >> >> Signed-off-by: Yoshiaki Tamura<tamura.yoshiaki@lab.ntt.co.jp> >> Signed-off-by: OHMURA Kei<ohmura.kei@lab.ntt.co.jp> >> --- >> exec.c | 22 +++++++++++++++++----- >> 1 files changed, 17 insertions(+), 5 deletions(-) >> >> diff --git a/exec.c b/exec.c >> index 9bcb4de..ba334e7 100644 >> --- a/exec.c >> +++ b/exec.c >> @@ -119,7 +119,9 @@ uint8_t *code_gen_ptr; >> >> #if !defined(CONFIG_USER_ONLY) >> int phys_ram_fd; >> -uint8_t *phys_ram_dirty; >> +unsigned long *phys_ram_vga_dirty; >> +unsigned long *phys_ram_code_dirty; >> +unsigned long *phys_ram_migration_dirty; > > Would be nice to make this an array. Thanks for pointing out. I have a question regarding the index of the array. From the compatibility perspective, I would prefer using the existing macros. #define VGA_DIRTY_FLAG 0x01 #define CODE_DIRTY_FLAG 0x02 #define MIGRATION_DIRTY_FLAG 0x08 However, if I use them as is, I'll get a sparse array... Is it acceptable to change these values like 0, 1, 2?
On 03/16/2010 03:01 PM, Yoshiaki Tamura wrote: >>> -uint8_t *phys_ram_dirty; >>> +unsigned long *phys_ram_vga_dirty; >>> +unsigned long *phys_ram_code_dirty; >>> +unsigned long *phys_ram_migration_dirty; >> >> Would be nice to make this an array. > > > Thanks for pointing out. > I have a question regarding the index of the array. > From the compatibility perspective, I would prefer using the existing > macros. > > #define VGA_DIRTY_FLAG 0x01 > #define CODE_DIRTY_FLAG 0x02 > #define MIGRATION_DIRTY_FLAG 0x08 > > However, if I use them as is, I'll get a sparse array... > Is it acceptable to change these values like 0, 1, 2? Sure.
diff --git a/exec.c b/exec.c index 9bcb4de..ba334e7 100644 --- a/exec.c +++ b/exec.c @@ -119,7 +119,9 @@ uint8_t *code_gen_ptr; #if !defined(CONFIG_USER_ONLY) int phys_ram_fd; -uint8_t *phys_ram_dirty; +unsigned long *phys_ram_vga_dirty; +unsigned long *phys_ram_code_dirty; +unsigned long *phys_ram_migration_dirty; uint8_t *bios_mem; static int in_migration; @@ -2659,10 +2661,20 @@ ram_addr_t qemu_ram_alloc(ram_addr_t size) new_block->next = ram_blocks; ram_blocks = new_block; - phys_ram_dirty = qemu_realloc(phys_ram_dirty, - (last_ram_offset + size) >> TARGET_PAGE_BITS); - memset(phys_ram_dirty + (last_ram_offset >> TARGET_PAGE_BITS), - 0xff, size >> TARGET_PAGE_BITS); + if (BITMAP_SIZE(last_ram_offset + size) != BITMAP_SIZE(last_ram_offset)) { + phys_ram_vga_dirty = qemu_realloc(phys_ram_vga_dirty, + BITMAP_SIZE(last_ram_offset + size)); + phys_ram_code_dirty = qemu_realloc(phys_ram_code_dirty, + BITMAP_SIZE(last_ram_offset + size)); + phys_ram_migration_dirty = qemu_realloc(phys_ram_migration_dirty, + BITMAP_SIZE(last_ram_offset + size)); + memset((uint8_t *)phys_ram_vga_dirty + + BITMAP_SIZE(last_ram_offset), 0xff, BITMAP_SIZE(size)); + memset((uint8_t *)phys_ram_code_dirty + + BITMAP_SIZE(last_ram_offset), 0xff, BITMAP_SIZE(size)); + memset((uint8_t *)phys_ram_migration_dirty + + BITMAP_SIZE(last_ram_offset), 0xff, BITMAP_SIZE(size)); + } last_ram_offset += size;