Re: [PATCH v3 06/10] migration/dirtyrate: Compare page hash results for recorded sampled page

["Dr. David Alan Gilbert" <dgilbert@redhat.com>]

* Zheng Chuan (zhengchuan@huawei.com) wrote:
> 
> 
> On 2020/8/21 1:36, Dr. David Alan Gilbert wrote:
> > * Chuan Zheng (zhengchuan@huawei.com) wrote:
> >> Compare page hash results for recorded sampled page.
> >>
> >> Signed-off-by: Chuan Zheng <zhengchuan@huawei.com>
> >> Signed-off-by: YanYing Zhuang <ann.zhuangyanying@huawei.com>
> >> ---
> >>  migration/dirtyrate.c | 76 +++++++++++++++++++++++++++++++++++++++++++++++++++
> >>  1 file changed, 76 insertions(+)
> >>
> >> diff --git a/migration/dirtyrate.c b/migration/dirtyrate.c
> >> index 62b6f69..3ce25f5 100644
> >> --- a/migration/dirtyrate.c
> >> +++ b/migration/dirtyrate.c
> >> @@ -215,6 +215,82 @@ static int record_ramblock_hash_info(struct RamblockDirtyInfo **block_dinfo,
> >>      return 0;
> >>  }
> >>  
> >> +static int calc_page_dirty_rate(struct RamblockDirtyInfo *info)
> >> +{
> >> +    uint8_t *md = NULL;
> >> +    int i;
> >> +    int ret = 0;
> >> +
> >> +    md = g_try_new0(uint8_t, qcrypto_hash_len);
> >> +    if (!md) {
> >> +        return -1;
> >> +    }
> > 
> > As previously asked; isn't this a nice small simple fixed length - no
> > need to allocate it?
> > 
> Yes, it could use QCRYPTO_HASH_LEN to define an array.
> >> +
> >> +    for (i = 0; i < info->sample_pages_count; i++) {
> >> +        ret = get_ramblock_vfn_hash(info, info->sample_page_vfn[i], &md);
> >> +        if (ret < 0) {
> >> +            goto out;
> >> +        }
> >> +
> >> +        if (memcmp(md, info->hash_result + i * qcrypto_hash_len,
> >> +                   qcrypto_hash_len) != 0) {
> >> +            info->sample_dirty_count++;
> >> +        }
> >> +    }
> >> +
> >> +out:
> >> +    g_free(md);
> >> +    return ret;
> >> +}
> >> +
> >> +static bool find_page_matched(RAMBlock *block, struct RamblockDirtyInfo *infos,
> >> +                              int count, struct RamblockDirtyInfo **matched)
> >> +{
> >> +    int i;
> >> +
> >> +    for (i = 0; i < count; i++) {
> >> +        if (!strcmp(infos[i].idstr, qemu_ram_get_idstr(block))) {
> >> +            break;
> >> +        }
> >> +    }
> >> +
> >> +    if (i == count) {
> >> +        return false;
> >> +    }
> >> +
> >> +    if (infos[i].ramblock_addr != qemu_ram_get_host_addr(block) ||
> >> +        infos[i].ramblock_pages !=
> >> +            (qemu_ram_get_used_length(block) >> 12)) {
> >> +        return false;
> > 
> > I previously asked how this happens.
> > Also this was DIRTYRATE_PAGE_SIZE_SHIFT
> > 
> Here, we want to find same ramblock we sampled before.
> We just ignore the ramblock if its hva address or size changed due to memory hot-plug during the measurement.

OK, just try and make the code consistent between '12',
'DIRTYRATE_PAGE_SIZE_SHIFT'
or TARGET_PAGE_SIZE/SHIFT.

Dave

> >> +    }
> >> +
> >> +    *matched = &infos[i];
> >> +    return true;
> >> +}
> >> +
> >> +static int compare_page_hash_info(struct RamblockDirtyInfo *info,
> >> +                                  int block_index)
> >> +{
> >> +    struct RamblockDirtyInfo *block_dinfo = NULL;
> >> +    RAMBlock *block = NULL;
> >> +
> >> +    RAMBLOCK_FOREACH_MIGRATABLE(block) {
> >> +        block_dinfo = NULL;
> > 
> > So you've removed the selction of only some RAMBlocks now?
> > 
> In next patch:), i add functions to skip sampling ramblock.
> 
> >> +        if (!find_page_matched(block, info, block_index + 1, &block_dinfo)) {
> >> +            continue;
> >> +        }
> >> +        if (calc_page_dirty_rate(block_dinfo) < 0) {
> >> +            return -1;
> >> +        }
> >> +        update_dirtyrate_stat(block_dinfo);
> >> +    }
> >> +    if (!dirty_stat.total_sample_count) {
> >> +        return -1;
> >> +    }
> >> +
> >> +    return 0;
> >> +}
> >> +
> >>  static void calculate_dirtyrate(struct DirtyRateConfig config)
> >>  {
> >>      /* todo */
> >> -- 
> >> 1.8.3.1
> >>
> 
-- 
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK