Re: [PATCH v3 1/4] btrfs: add read_policy latency

[Anand Jain <anand.jain@oracle.com>]

On 20/1/21 3:36 am, Josef Bacik wrote:
> On 1/11/21 4:41 AM, Anand Jain wrote:
>> The read policy type latency routes the read IO based on the historical
>> average wait-time experienced by the read IOs through the individual
>> device. This patch obtains the historical read IO stats from the kernel
>> block layer and calculates its average.
>>
>> Example usage:
>>   echo "latency" > /sys/fs/btrfs/$uuid/read_policy
>>
>> Signed-off-by: Anand Jain <anand.jain@oracle.com>
>> ---
>> v3: The block layer commit 0d02129e76ed (block: merge struct 
>> block_device and
>>      struct hd_struct) has changed the first argument in the function
>>      part_stat_read_all() in 5.11-rc1. So the compilation will fail. 
>> This patch
>>      fixes it.
>>      Commit log updated.
>>
>> v2: Use btrfs_debug_rl() instead of btrfs_info_rl()
>>      It is better we have this debug until we test this on at least few
>>      hardwares.
>>      Drop the unrelated changes.
>>      Update change log.
>>
>> v1: Drop part_stat_read_all instead use part_stat_read
>>      Drop inflight
>>
>>
>>   fs/btrfs/sysfs.c   |  3 ++-
>>   fs/btrfs/volumes.c | 38 ++++++++++++++++++++++++++++++++++++++
>>   fs/btrfs/volumes.h |  2 ++
>>   3 files changed, 42 insertions(+), 1 deletion(-)
>>
>> diff --git a/fs/btrfs/sysfs.c b/fs/btrfs/sysfs.c
>> index 19b9fffa2c9c..96ca7bef6357 100644
>> --- a/fs/btrfs/sysfs.c
>> +++ b/fs/btrfs/sysfs.c
>> @@ -915,7 +915,8 @@ static bool strmatch(const char *buffer, const 
>> char *string)
>>       return false;
>>   }
>> -static const char * const btrfs_read_policy_name[] = { "pid" };
>> +/* Must follow the order as in enum btrfs_read_policy */
>> +static const char * const btrfs_read_policy_name[] = { "pid", 
>> "latency" };
>>   static ssize_t btrfs_read_policy_show(struct kobject *kobj,
>>                         struct kobj_attribute *a, char *buf)
>> diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
>> index f4037a6bd926..f7a0a83d2cd4 100644
>> --- a/fs/btrfs/volumes.c
>> +++ b/fs/btrfs/volumes.c
>> @@ -14,6 +14,7 @@
>>   #include <linux/semaphore.h>
>>   #include <linux/uuid.h>
>>   #include <linux/list_sort.h>
>> +#include <linux/part_stat.h>
>>   #include "misc.h"
>>   #include "ctree.h"
>>   #include "extent_map.h"
>> @@ -5490,6 +5491,39 @@ int btrfs_is_parity_mirror(struct btrfs_fs_info 
>> *fs_info, u64 logical, u64 len)
>>       return ret;
>>   }
>> +static int btrfs_find_best_stripe(struct btrfs_fs_info *fs_info,
>> +                  struct map_lookup *map, int first,
>> +                  int num_stripe)
>> +{
>> +    u64 est_wait = 0;
>> +    int best_stripe = 0;
>> +    int index;
>> +
>> +    for (index = first; index < first + num_stripe; index++) {
>> +        u64 read_wait;
>> +        u64 avg_wait = 0;
>> +        unsigned long read_ios;
>> +        struct btrfs_device *device = map->stripes[index].dev;
>> +
>> +        read_wait = part_stat_read(device->bdev, nsecs[READ]);
>> +        read_ios = part_stat_read(device->bdev, ios[READ]);
>> +
>> +        if (read_wait && read_ios && read_wait >= read_ios)
>> +            avg_wait = div_u64(read_wait, read_ios);
>> +        else
>> +            btrfs_debug_rl(device->fs_devices->fs_info,
> 
> You can just use fs_info here, you already have it.  I'm not in love 
> with doing this check every time, I'd rather cache the results 
> somewhere.  However if we're read-only I can't think of a mechanism we 
> could piggy back on, RW we could just do it every transaction commit.  
> Fix the fs_info thing and you can add
> 

  Oh. I will fix it. Thanks for the review here and in other patches.
-Anand

> Reviewed-by: Josef Bacik <josef@toxicpanda.com>
> 
> Thanks,
> 
> Josef