Skip to contentSkip to conversation

SD scheduler testing hitch

DISCUSSION Al Boldi 3 participants 4 messages Apr 8, 2007 – Apr 9, 2007

lore
Share & tools
lore

Al Boldi

author
#0 lore raw
Quoted 8 lines · Mike Galbraith
Quoted 3 lines · Mike Galbraith

I lowered the time to 500us, and ran at nice -10.. it starves tenpercent here every time. (ran as taskset -c 1 nice -n -10 ./fairtest) The starving 10% duty cycle task has trouble getting 1% CPU.

Hmm. Playing with it some more today, it still happens, but it's not very repeatable. Something is odd. I wonder if any SD using readers will try it.

Tried it on mainline 2.6.20.3.
It's not easily repeatable, but it's got the same problem.

top - 21:21:45 up 27 min, 0 users, load average: 0.80, 0.43, 0.20

Tasks:  45 total,   3 running,  42 sleeping,   0 stopped,   0 zombie
Cpu(s):  24.3% user,   0.5% system,   0.0% nice,  75.0% idle,   0.2% IO-wait
Mem:    499488k total,    27352k used,   472136k free,     1996k buffers
Swap:  1020088k total,        0k used,  1020088k free,     9160k cached
  PID  PR  NI  VIRT  RES  SHR SWAP nFLT nDRT WCHAN     S %CPU    TIME+  Command 
  688  25   0  1804  412  352 1392    0    0 rest_init R 94.7   2:37.01 fairtest
  689  15   0  1804  264  204 1540    0    0 rest_init R  0.0   0:00.79 fairtest
    1  15   0  1440  500  444  940   15    0 rest_init S  0.0   0:00.73 init    

Thanks!

--
Al

#1 lore raw
Quoted 14 lines · Al Boldi
Quoted 8 lines · Mike Galbraith
Quoted 3 lines · Mike Galbraith

I lowered the time to 500us, and ran at nice -10.. it starves tenpercent here every time. (ran as taskset -c 1 nice -n -10 ./fairtest) The starving 10% duty cycle task has trouble getting 1% CPU.

Hmm. Playing with it some more today, it still happens, but it's not very repeatable. Something is odd. I wonder if any SD using readers will try it.

Tried it on mainline 2.6.20.3.
It's not easily repeatable, but it's got the same problem.

top - 21:21:45 up 27 min, 0 users, load average: 0.80, 0.43, 0.20

Quoted 8 lines
Tasks:  45 total,   3 running,  42 sleeping,   0 stopped,   0 zombie
Cpu(s):  24.3% user,   0.5% system,   0.0% nice,  75.0% idle,   0.2% IO-wait
Mem:    499488k total,    27352k used,   472136k free,     1996k buffers
Swap:  1020088k total,        0k used,  1020088k free,     9160k cached
  PID  PR  NI  VIRT  RES  SHR SWAP nFLT nDRT WCHAN     S %CPU    TIME+  Command 
  688  25   0  1804  412  352 1392    0    0 rest_init R 94.7   2:37.01 fairtest
  689  15   0  1804  264  204 1540    0    0 rest_init R  0.0   0:00.79 fairtest

Aha! Thanks a bunch for testing it. (thing was irritating me greatly)

-Mike

#2 lore raw
Quoted 7 lines · Mike Galbraith
Quoted 6 lines · Al Boldi
Quoted 5 lines · Mike Galbraith
Quoted 3 lines · Mike Galbraith

I lowered the time to 500us, and ran at nice -10.. it starves tenpercent here every time. (ran as taskset -c 1 nice -n -10 ./fairtest) The starving 10% duty cycle task has trouble getting 1% CPU.

Quoted 6 lines
Quoted 6 lines
Quoted 3 lines

Hmm. Playing with it some more today, it still happens, but it's not very repeatable. Something is odd. I wonder if any SD using readers will try it.

Tried it on mainline 2.6.20.3.
It's not easily repeatable, but it's got the same problem.

In my case, it's easily reproducible. Below is a mainline 2.6.19.

Look at both %CPU and TIME+ (sched_time).

top - 10:55:00 up 33 min, 8 users, load average: 1.07, 0.61, 0.51

Tasks:  76 total,   2 running,  74 sleeping,   0 stopped,   0 zombie
Cpu(s): 76.0% us,  0.3% sy,  0.0% ni, 23.7% id,  0.0% wa,  0.0% hi,  0.0% si
Mem:    256592k total,   251888k used,     4704k free,    46516k buffers
Swap:   272120k total,     4924k used,   267196k free,    90048k cached
 PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
8113 dimm      15   0  1460  436  360 R 74.4  0.2   1:43.63 fairtest_xx
8114 dimm      15   0  1460  108   32 S  0.3  0.0   0:00.15 fairtest_xx

If you take a look at my previous report, you'll also notice "strangeness" with both cpu% and sched_time.

sched_clock() is based on "tsc" and it's available for my CPU. So maybe something fishy is going on with accounting.

I just wonder, why it's so easily reproducible in my case for both mainline and SD (besides a 2.6.13 mainline, but I have to re-run all the tests letting tenp run longer for the "sched_time" to be accumulated)...

Signature · 2 lines
Best regards,
Dmitry Adamushko
#3 lore raw
Quoted 12 lines · Dmitry Adamushko
Quoted 11 lines · Mike Galbraith
Quoted 10 lines · Al Boldi
Quoted 8 lines · Mike Galbraith
Quoted 3 lines · Mike Galbraith

I lowered the time to 500us, and ran at nice -10.. it starves tenpercent here every time. (ran as taskset -c 1 nice -n -10 ./fairtest) The starving 10% duty cycle task has trouble getting 1% CPU.

Hmm. Playing with it some more today, it still happens, but it's not very repeatable. Something is odd. I wonder if any SD using readers will try it.

Quoted 6 lines
Quoted 2 lines
Quoted 2 lines

Tried it on mainline 2.6.20.3.
It's not easily repeatable, but it's got the same problem.

In my case, it's easily reproducible. Below is a mainline 2.6.19.

Look at both %CPU and TIME+ (sched_time).

Yeah, 23.7% idle. Hrm.

Quoted 11 lines

top - 10:55:00 up 33 min, 8 users, load average: 1.07, 0.61, 0.51

Tasks:  76 total,   2 running,  74 sleeping,   0 stopped,   0 zombie
Cpu(s): 76.0% us,  0.3% sy,  0.0% ni, 23.7% id,  0.0% wa,  0.0% hi,  0.0% si
Mem:    256592k total,   251888k used,     4704k free,    46516k buffers
Swap:   272120k total,     4924k used,   267196k free,    90048k cached
  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 8113 dimm      15   0  1460  436  360 R 74.4  0.2   1:43.63 fairtest_xx
 8114 dimm      15   0  1460  108   32 S  0.3  0.0   0:00.15 fairtest_xx
Quoted 5 lines

If you take a look at my previous report, you'll also notice "strangeness" with both cpu% and sched_time.

sched_clock() is based on "tsc" and it's available for my CPU. So maybe something fishy is going on with accounting.

Yeah. There's a patch floating around that is supposed to fix some of the error. I'm going to test with it, and I'm rummaging around a bit as well. I know for sure that frequency scaling will screw up the test proggy's calibration here, but I test with it locked to max, and in any case the hog half should eat any cpu not being used by the tenp half.

-Mike