Spark SQL场景IO bpf readahead 优化同步OLK-6.6

【特性描述】
通过ebpf实现可编程内核，针对应用的读取方式调整内核的预读行为，达到spark sql场景的读性能优化，
【特性竞争力】
通过ebpf实现可编程内核，针对应用的读取方式调整内核的预读行为，避免反复下发小IO磁头跳动带来开销。
【硬件架构】
NoArch
【特性约束】
NA
【涉及仓库】
OLK-6.6
【验证记录】

打上验证补丁：

~~~diff
diff --git a/mm/readahead.c b/mm/readahead.c
index c3f45b710043..8599bb5147eb 100644
--- a/mm/readahead.c
+++ b/mm/readahead.c
@@ -717,7 +717,10 @@ void page_cache_sync_ra(struct readahead_control *ractl,
 		req_count = 1;
 		do_forced_ra = true;
 	}
-
+	if (ractl->file->f_ctl_mode & FMODE_CTL_RANDOM)
+		printk("FMODE_CTL_RANDOM was set\n");
+	if (ractl->file->f_ctl_mode & FMODE_CTL_WILLNEED)
+		printk("FMODE_CTL_WILLNEED was set\n");
 	/* try to read first READAHEAD_FIRST_SIZE into pagecache */
 	if (ractl->file && (ractl->file->f_ctl_mode & FMODE_CTL_WILLNEED) &&
 		page_cache_readahead_from_head(ractl->mapping, ractl->file,

~~~

执行测试用例：

~~~
root@debian:~/readahead# sh testcase00.sh
mke2fs 1.47.0 (5-Feb-2023)
Discarding device blocks: done                            
Creating filesystem with 2621440 4k blocks and 655360 inodes
Filesystem UUID: bf06d5b8-7e4b-4913-89e0-1dc1a80a8c1a
Superblock backups stored on blocks: 
        32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632

Allocating group tables: done                            
Writing inode tables: done                            
Creating journal (16384 blocks): done
Writing superblocks and filesystem accounting information: done

sleep 3s for file_read_pattern launch
2+0 records in
2+0 records out
2097152 bytes (2.1 MB, 2.0 MiB) copied, 0.0220223 s, 95.2 MB/s
test FMODE_WILLNEED
1+0 records in
1+0 records out
1048576 bytes (1.0 MB, 1.0 MiB) copied, 0.00451005 s, 232 MB/s
[ 1455.728379] FMODE_CTL_WILLNEED was set
test set FMODE_RANDOM
10+0 records in
10+0 records out
10485760 bytes (10 MB, 10 MiB) copied, 0.0714635 s, 147 MB/s
[ 1455.967379] FMODE_CTL_RANDOM was set
[ 1455.969469] FMODE_CTL_RANDOM was set
[ 1455.971596] FMODE_CTL_RANDOM was set
......
[ 1457.781893] FMODE_CTL_RANDOM was set
[ 1457.783646] FMODE_CTL_RANDOM was set
test successed!

root@debian:~/readahead# sh testcase01.sh
meta-data=/dev/sda               isize=512    agcount=4, agsize=655360 blks
         =                       sectsz=512   attr=2, projid32bit=1
         =                       crc=1        finobt=1, sparse=1, rmapbt=1
         =                       reflink=1    bigtime=1 inobtcount=1 nrext64=1
data     =                       bsize=4096   blocks=2621440, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0, ftype=1
log      =internal log           bsize=4096   blocks=16384, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0
Discarding blocks...Done.
sleep 3s for file_read_pattern launch
2+0 records in
2+0 records out
2097152 bytes (2.1 MB, 2.0 MiB) copied, 0.0200794 s, 104 MB/s
test FMODE_WILLNEED
1+0 records in
1+0 records out
1048576 bytes (1.0 MB, 1.0 MiB) copied, 0.0050548 s, 207 MB/s
[ 1581.737010] FMODE_CTL_WILLNEED was set
test set FMODE_RANDOM
testtest: (g=0): rw=randread, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=psync, iodepth=1
fio-3.36
Starting 1 thread
testtest: Laying out IO file (1 file / 50MiB)
100+0 records in
100+0 records out
104857600 bytes (105 MB, 100 MiB) copied, 0.502477 s, 209 MB/s

testtest: (groupid=0, jobs=1): err= 0: pid=3508: Mon Jan 15 14:08:20 2024
  read: IOPS=685, BW=2743KiB/s (2809kB/s)(50.0MiB/18665msec)
    clat (usec): min=389, max=18254, avg=1454.09, stdev=339.92
     lat (usec): min=389, max=18255, avg=1454.49, stdev=339.96
    clat percentiles (usec):
     |  1.00th=[ 1074],  5.00th=[ 1188], 10.00th=[ 1254], 20.00th=[ 1336],
     | 30.00th=[ 1369], 40.00th=[ 1401], 50.00th=[ 1434], 60.00th=[ 1467],
     | 70.00th=[ 1516], 80.00th=[ 1549], 90.00th=[ 1614], 95.00th=[ 1696],
     | 99.00th=[ 2024], 99.50th=[ 2114], 99.90th=[ 2671], 99.95th=[ 9503],
     | 99.99th=[16909]
   bw (  KiB/s): min= 2272, max= 3016, per=100.00%, avg=2747.03, stdev=144.66, samples=37
   iops        : min=  568, max=  754, avg=686.76, stdev=36.16, samples=37
  lat (usec)   : 500=0.01%, 750=0.05%, 1000=0.23%
  lat (msec)   : 2=98.50%, 4=1.12%, 10=0.05%, 20=0.03%
  cpu          : usr=0.14%, sys=57.88%, ctx=12800, majf=0, minf=0
  IO depths    : 1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=12800,0,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=1

Run status group 0 (all jobs):
   READ: bw=2743KiB/s (2809kB/s), 2743KiB/s-2743KiB/s (2809kB/s-2809kB/s), io=50.0MiB (52.4MB), run=18665-18665msec

Disk stats (read/write):
  sda: ios=12680/80, sectors=101440/204800, merge=0/0, ticks=7575/984, in_queue=8560, util=98.67%
[ 1582.126897] FMODE_CTL_RANDOM was set
[ 1582.129017] FMODE_CTL_RANDOM was set
[ 1582.131326] FMODE_CTL_RANDOM was set
......
[ 1600.774171] FMODE_CTL_RANDOM was set
[ 1600.775609] FMODE_CTL_RANDOM was set
[ 1600.776854] FMODE_CTL_RANDOM was set
test clear FMODE_RANDOM
test successed!
~~~

GVP openEuler/kernel
Closed

Content Risk Flag

Comments (1)

GVPopenEuler/kernelClosed

Content Risk Flag