On 2018/4/13 11:12 AM, tang.junhui@xxxxxxxxxx wrote:
> Hi Coly,
>
> Hello Coly,
>
>> On 2018/4/12 2:38 PM, tang.junhui@xxxxxxxxxx wrote:
>>> From: Tang Junhui <tang.junhui@xxxxxxxxxx>
>>>
>>> In GC thread, we record the latest GC key in gc_done, which is expected
>>> to be used for incremental GC, but in currently code, we didn't realize
>>> it. When GC runs, front side IO would be blocked until the GC over, it
>>> would be a long time if there is a lot of btree nodes.
>>>
>>> This patch realizes incremental GC, the main ideal is that, when there
>>> are front side I/Os, after GC some nodes (100), we stop GC, release locker
>>> of the btree node, and go to process the front side I/Os for some times
>>> (100 ms), then go back to GC again.
>>>
>>> By this patch, when we doing GC, I/Os are not blocked all the time, and
>>> there is no obvious I/Os zero jump problem any more.
>>>
>>> Signed-off-by: Tang Junhui <tang.junhui@xxxxxxxxxx>
>>
>> Hi Junhui,
>>
>> I reply my comments in line,
> Thanks for your comments in advance.
>>
>>> ---
>>> drivers/md/bcache/bcache.h | 5 +++++
>>> drivers/md/bcache/btree.c | 15 ++++++++++++++-
>>> drivers/md/bcache/request.c | 3 +++
>>> 3 files changed, 22 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/md/bcache/bcache.h b/drivers/md/bcache/bcache.h
>>> index 843877e..ab4c9ca 100644
>>> --- a/drivers/md/bcache/bcache.h
>>> +++ b/drivers/md/bcache/bcache.h
>>> @@ -445,6 +445,7 @@ struct cache {
>>>
>>> struct gc_stat {
>>> size_t nodes;
>>> + size_t nodes_pre;
>>> size_t key_bytes;
>>>
>>> size_t nkeys;
>>> @@ -568,6 +569,10 @@ struct cache_set {
>>> */
>>> atomic_t rescale;
>>> /*
>>> + * used for GC, identify if any front side I/Os is inflight
>>> + */
>>> + atomic_t io_inflight;
>>
>> Could you please to rename the above variable to something like
>> search_inflight ? Just to be more explicit, considering we have many
>> types of io requests.
> OK, It looks better.
>>
>>> + /*
>>> * When we invalidate buckets, we use both the priority and the amount
>>> * of good data to determine which buckets to reuse first - to weight
>>> * those together consistently we keep track of the smallest nonzero
>>> diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c
>>> index 81e8dc3..b36d276 100644
>>> --- a/drivers/md/bcache/btree.c
>>> +++ b/drivers/md/bcache/btree.c
>>> @@ -90,6 +90,9 @@
>>>
>>> #define MAX_NEED_GC 64
>>> #define MAX_SAVE_PRIO 72
>>> +#define MIN_GC_NODES 100
>>> +#define GC_SLEEP_TIME 100
>>
>> How about naming the above macro as GC_SLEEP_MS ? So people may
>> understand the sleep time is 100ms without checking more context.
> OK, It looks better.
>>> +
>>>
>>> #define PTR_DIRTY_BIT (((uint64_t) 1 << 36))
>>>
>>> @@ -1581,6 +1584,13 @@ static int btree_gc_recurse(struct btree *b, struct btree_op *op,
>>> memmove(r + 1, r, sizeof(r[0]) * (GC_MERGE_NODES - 1));
>>> r->b = NULL;
>>>
>>> + if (atomic_read(&b->c->io_inflight) &&
>>> + gc->nodes >= gc->nodes_pre + MIN_GC_NODES) {
>>
>> On 32bit machines, gc->nodes is a 32bit count, if it is overflowed the
>> above check will be false for a very long time, and in some condition it
>> might be always false after gc->nodes overflowed.
>>
>> How about adding the following check before the if() statement ?
>
>> /* in case gc->nodes is overflowed */
>> if (gc->nodes_pre < gc->nodes)
>> gc->noeds_pre = gc->nodes;
>>
> I think 32bit is big enough, which can express a value of billions,
> but the number of nodes is usually around tens of thousands.
> Also gc->nodes and gc->nodes_pre were set to zero when GC begin each time.
>
> How do you think about?
Oh, I see. Then I don't worry here. The patch is OK to me.
Thanks :-)
Coly Li
[code snipped]
