From: Zhouping Liu <zliu@redhat.com>
To: LTP List <ltp-list@lists.sourceforge.net>
Subject: Re: [LTP] [PATCH] numa/numa01.sh: fixed errors under noncontinuous numa nodes list
Date: Wed, 09 Jan 2013 20:03:06 +0800 [thread overview]
Message-ID: <50ED5C7A.2050301@redhat.com> (raw)
In-Reply-To: <d586a4f2cfdf91b3e480c8e275f53fae34a00705.1357732655.git.zliu@redhat.com>
On 01/09/2013 07:58 PM, Zhouping Liu wrote:
> The previous code numa01.sh thought all NUMA system had continuous
> numa nodes list, and node id started from 0, but in fact it does not.
>
> There are lots of system contains noncontinuous numa nodes list
> e.g:
> # numactl --hardware
> available: 4 nodes (0,2,4,7)
> node 0 cpus: 0 4 8 12 16 20 24 28
> node 0 size: 4061 MB
> node 0 free: 3756 MB
> node 2 cpus: 1 5 9 13 17 21 25 29
> node 2 size: 4096 MB
> node 2 free: 3196 MB
> node 4 cpus: 2 6 10 14 18 22 26 30
> node 4 size: 4096 MB
> node 4 free: 3885 MB
> node 7 cpus: 3 7 11 15 19 23 27 31
> node 7 size: 4095 MB
> node 7 free: 3947 MB
> node distances:
> node 0 2 4 7
> 0: 10 20 20 20
> 2: 20 10 20 20
> 4: 20 20 10 20
> 7: 20 20 20 10
>
> in such discontinuous nodes system, numa01.sh would fail with
> such error log:
> libnuma: Warning: node argument 1 is out of range
>
> The patch fixed it, and removed some useless/redundancy val/func,
> such as 'max_node' val and 'getmax' func, also make some cleanup.
I tested the patch on continuous/non-continuous NUMA system, also did on
non-NUMA system.
Thanks,
Zhouping
>
> Signed-off-by: Zhouping Liu <zliu@redhat.com>
> ---
> testcases/kernel/numa/numa01.sh | 179 ++++++++++++++--------------------------
> 1 file changed, 61 insertions(+), 118 deletions(-)
> mode change 100644 => 100755 testcases/kernel/numa/numa01.sh
>
> diff --git a/testcases/kernel/numa/numa01.sh b/testcases/kernel/numa/numa01.sh
> old mode 100644
> new mode 100755
> index 6ad8b11..19a8f71
> --- a/testcases/kernel/numa/numa01.sh
> +++ b/testcases/kernel/numa/numa01.sh
> @@ -68,49 +68,6 @@ chk_ifexists()
> return $RC
> }
>
> -
> -
> -# Function: getmax
> -#
> -# Description: - Returns the maximum available nodes if success.
> -#
> -# Input: - o/p of numactl --hardware command which is expected in the format
> -# shown below
> -# available: 2 nodes (0-1)
> -# node 0 size: 7808 MB
> -# node 0 free: 7457 MB
> -# node 1 size: 5807 MB
> -# node 1 free: 5731 MB
> -# node distances:
> -# node 0 1
> -# 0: 10 20
> -# 1: 20 10
> -#
> -# Return: - zero on success.
> -# - non-zero on failure.
> -getmax()
> -{
> - numactl --hardware > $LTPTMP/avail_nodes
> -
> - RC=$(awk '{ if ( NR == 1 ) {print $1;} }' $LTPTMP/avail_nodes)
> - if [ $RC = "available:" ]
> - then
> - RC=$(awk '{ if ( NR == 1 ) {print $3;} }' $LTPTMP/avail_nodes)
> - if [ $RC = "nodes" ]
> - then
> - RC=$(awk '{ if ( NR == 1 ) {print $2;} }' $LTPTMP/avail_nodes)
> - return 0;
> - else
> - tst_brkm TBROK NULL "the field nodes in the o/p of numactl --hardware seems to be different"
> - fi
> - else
> - tst_brkm TBROK NULL "the field available in the o/p of numactl --hardware seems to be different"
> - fi
> - return 1;
> -}
> -
> -
> -
> # Function: extract_numastat
> #
> # Description: - extract the value of given row,column from the numastat output .
> @@ -187,9 +144,6 @@ init()
> export TCID="Initnuma"
> export TST_COUNT=0
>
> - # Max. no. of Nodes
> - max_node=0
> -
> # Page Size
> page_size=0
>
> @@ -233,18 +187,6 @@ init()
> chk_ifexists INIT kill || return $RC
>
> RC=0
> - # set max_node
> - getmax || return 1
> - max_node=$RC
> -
> - if [ $max_node -eq 1 ]
> - then
> - tst_resm TCONF "non-NUMA aware kernel is running or your machine does not support numa policy or
> - your machine is not a NUMA machine"
> - exit 0
> - fi
> -
> - RC=0
> #Set pagesize
> support_numa $PRT_PG_SIZE > $LTPTMP/numaarg || RC=$?
> if [ $RC -ne 0 ]
> @@ -292,7 +234,6 @@ test01()
> Prev_value=0 # extracted from the numastat o/p
> Curr_value=0 # Current value extracted from numastat o/p
> Exp_incr=0 # 1 MB/ PAGESIZE
> - Node_num=0
> col=0
> MB=$[1024*1024]
>
> @@ -300,13 +241,12 @@ test01()
> Exp_incr=$[$MB/$page_size]
>
> COUNTER=1
> - while [ $COUNTER -le $max_node ]; do
> - Node_num=$[$COUNTER-1] #Node numbers start from 0
> + for node in `echo $nodes_list`; do
> col=$[$COUNTER+1] #Node number in numastat o/p
> numastat > $LTPTMP/numalog
> extract_numastat local_node $local_node $col || return 1
> Prev_value=$RC
> - numactl --cpunodebind=$Node_num --membind=$Node_num support_numa $ALLOC_1MB
> + numactl --cpunodebind=$node --membind=$node support_numa $ALLOC_1MB
> numastat > $LTPTMP/numalog
> extract_numastat local_node $local_node $col || return 1
> Curr_value=$RC
> @@ -314,7 +254,7 @@ test01()
> if [ $RC -lt $Exp_incr ]
> then
> tst_resm TFAIL \
> - "Test #1: NUMA hit and localnode increase in node$Node_num is less than expected"
> + "Test #1: NUMA hit and localnode increase in node$node is less than expected"
> return 1
> fi
> COUNTER=$[$COUNTER+1]
> @@ -340,7 +280,6 @@ test02()
> Prev_value=0 # extracted from the numastat o/p
> Curr_value=0 # Current value extracted from numastat o/p
> Exp_incr=0 # 1 MB/ PAGESIZE
> - Node_num=0
> col=0
> MB=$[1024*1024]
>
> @@ -348,29 +287,28 @@ test02()
> Exp_incr=$[$MB/$page_size]
>
> COUNTER=1
> - while [ $COUNTER -le $max_node ]; do
> + for node in `echo $nodes_list`; do
>
> - if [ $max_node -eq 1 ]
> + if [ $total_nodes -eq 1 ]
> then
> tst_brkm TBROK NULL "Preferred policy cant be applied for a single node machine"
> return 1
> fi
>
> - Node_num=$[$COUNTER-1] #Node numbers start from 0
> -
> - if [ $COUNTER -eq $max_node ] #wrap up for last node
> + if [ $COUNTER -eq $total_nodes ] #wrap up for last node
> then
> - Preferred_node=0
> + Preferred_node=`echo $nodes_list | cut -d ' ' -f 1`
> col=2 #column represents node0 in numastat o/p
> else
> - Preferred_node=$[$Node_num+1] #always next node is preferred node
> + # always next node is preferred node
> + Preferred_node=`echo $nodes_list | cut -d ' ' -f $[COUNTER+1]`
> col=$[$COUNTER+2] #Preferred Node number in numastat o/p
> fi
>
> numastat > $LTPTMP/numalog
> extract_numastat other_node $other_node $col || return 1
> Prev_value=$RC
> - numactl --cpunodebind=$Node_num --preferred=$Preferred_node support_numa $ALLOC_1MB
> + numactl --cpunodebind=$node --preferred=$Preferred_node support_numa $ALLOC_1MB
> sleep 2s #In RHEL collection of statistics takes more time.
> numastat > $LTPTMP/numalog
> extract_numastat other_node $other_node $col || return 1
> @@ -379,7 +317,7 @@ test02()
> if [ $RC -lt $Exp_incr ]
> then
> tst_resm TFAIL \
> - "Test #2: NUMA hit and othernode increase in node$Node_num is less than expected"
> + "Test #2: NUMA hit and othernode increase in node$node is less than expected"
> return 1
> fi
> COUNTER=$[$COUNTER+1]
> @@ -405,19 +343,18 @@ test03()
> declare -a parray # array contains previous value of all nodes
> Curr_value=0 # Current value extracted from numastat o/p
> Exp_incr=0 # 1 MB/ (PAGESIZE*num_of_nodes)
> - Node_num=0
> col=0
> MB=$[1024*1024]
>
> # Increase in numastat o/p is interms of pages
> Exp_incr=$[$MB/$page_size]
> # Pages will be allocated using round robin on nodes.
> - Exp_incr=$[$Exp_incr/$max_node]
> + Exp_incr=$[$Exp_incr/$total_nodes]
>
> # Check whether the pages are equally distributed among available nodes
> numastat > $LTPTMP/numalog
> COUNTER=1
> - while [ $COUNTER -le $max_node ]; do
> + for node in `echo $nodes_list`; do
> col=$[$COUNTER+1] #Node number in numastat o/p
> extract_numastat interleave_hit $interleave_hit $col || return 1
> Prev_value=$RC
> @@ -430,16 +367,15 @@ test03()
>
> numastat > $LTPTMP/numalog
> COUNTER=1
> - while [ $COUNTER -le $max_node ]; do
> + for node in `echo $nodes_list`; do
> col=$[$COUNTER+1] #Node number in numastat o/p
> - Node_num=$[$COUNTER-1] #Node numbers start from 0
> extract_numastat interleave_hit $interleave_hit $col || return 1
> Curr_value=$RC
> comparelog ${parray[$COUNTER]} $Curr_value
> if [ $RC -lt $Exp_incr ]
> then
> tst_resm TFAIL \
> - "Test #3: NUMA interleave hit in node$Node_num is less than expected"
> + "Test #3: NUMA interleave hit in node$node is less than expected"
> return 1
> fi
> COUNTER=$[$COUNTER+1]
> @@ -511,7 +447,6 @@ test05()
> Prev_value=0 # extracted from the numastat o/p
> Curr_value=0 # Current value extracted from numastat o/p
> Exp_incr=0 # 1 MB/ PAGESIZE
> - Node_num=0
> col=0
> MB=$[1024*1024]
>
> @@ -519,13 +454,12 @@ test05()
> Exp_incr=$[$MB/$page_size]
>
> COUNTER=1
> - while [ $COUNTER -le $max_node ]; do
> - Node_num=$[$COUNTER-1] #Node numbers start from 0
> + for node in `echo $nodes_list`; do
> col=$[$COUNTER+1] #Node number in numastat o/p
> numastat > $LTPTMP/numalog
> extract_numastat local_node $local_node $col || return 1
> Prev_value=$RC
> - numactl --cpunodebind=$Node_num --localalloc support_numa $ALLOC_1MB
> + numactl --cpunodebind=$node --localalloc support_numa $ALLOC_1MB
> numastat > $LTPTMP/numalog
> extract_numastat local_node $local_node $col || return 1
> Curr_value=$RC
> @@ -533,7 +467,7 @@ test05()
> if [ $RC -lt $Exp_incr ]
> then
> tst_resm TFAIL \
> - "Test #5: NUMA hit and localnode increase in node$Node_num is less than expected"
> + "Test #5: NUMA hit and localnode increase in node$node is less than expected"
> return 1
> fi
> COUNTER=$[$COUNTER+1]
> @@ -560,19 +494,18 @@ test06()
> declare -a parray # array contains previous value of all nodes
> Curr_value=0 # Current value extracted from numastat o/p
> Exp_incr=0 # 1 MB/ (PAGESIZE*num_of_nodes)
> - Node_num=0
> col=0
> MB=$[1024*1024]
>
> # Increase in numastat o/p is interms of pages
> Exp_incr=$[$MB/$page_size]
> # Pages will be allocated using round robin on nodes.
> - Exp_incr=$[$Exp_incr/$max_node]
> + Exp_incr=$[$Exp_incr/$total_nodes]
>
> # Check whether the pages are equally distributed among available nodes
> numastat > $LTPTMP/numalog
> COUNTER=1
> - while [ $COUNTER -le $max_node ]; do
> + for node in `echo $nodes_list`; do
> col=$[$COUNTER+1] #Node number in numastat o/p
> extract_numastat numa_hit $numa_hit $col || return 1
> Prev_value=$RC
> @@ -585,16 +518,15 @@ test06()
>
> numastat > $LTPTMP/numalog
> COUNTER=1
> - while [ $COUNTER -le $max_node ]; do
> + for node in `echo $nodes_list`; do
> col=$[$COUNTER+1] #Node number in numastat o/p
> - Node_num=$[$COUNTER-1] #Node numbers start from 0
> extract_numastat numa_hit $numa_hit $col || return 1
> Curr_value=$RC
> comparelog ${parray[$COUNTER]} $Curr_value
> if [ $RC -lt $Exp_incr ]
> then
> tst_resm TFAIL \
> - "Test #6: NUMA numa_hit for shm file numa_shm in node$Node_num is less than expected"
> + "Test #6: NUMA numa_hit for shm file numa_shm in node$node is less than expected"
> return 1
> fi
> COUNTER=$[$COUNTER+1]
> @@ -626,19 +558,18 @@ test07()
> declare -a parray # array contains previous value of all nodes
> Curr_value=0 # Current value extracted from numastat o/p
> Exp_incr=0 # 1 MB/ (PAGESIZE*num_of_nodes)
> - Node_num=0
> col=0
> msize=1000
> KB=1024
> # Increase in numastat o/p is interms of pages
> Exp_incr=$[($KB * $msize)/$page_size]
> # Pages will be allocated using round robin on nodes.
> - Exp_incr=$[$Exp_incr/$max_node]
> + Exp_incr=$[$Exp_incr/$total_nodes]
>
> # Check whether the pages are equally distributed among available nodes
> numastat > $LTPTMP/numalog
> COUNTER=1
> - while [ $COUNTER -lt $max_node ]; do
> + for node in `echo $nodes_list`; do
> col=$[$COUNTER+1] #Node number in numastat o/p
> extract_numastat interleave_hit $interleave_hit $col || return 1
> Prev_value=$RC
> @@ -646,16 +577,14 @@ test07()
> COUNTER=$[$COUNTER+1]
> done
>
> -
> numademo -c msize > $LTPTMP/demolog
> sleep 2s #In RHEL collection of statistics takes more time.
>
> numastat > $LTPTMP/numalog
> COUNTER=1
> x=0
> - while [ $COUNTER -le $max_node ]; do
> + for node in `echo $nodes_list`; do
> col=$[$COUNTER+1] #Node number in numastat o/p
> - Node_num=$[$COUNTER-1] #Node numbers start from 0
> extract_numastat interleave_hit $interleave_hit $col || return 1
> Curr_value=$RC
> comparelog ${parray[$COUNTER]} $Curr_value
> @@ -693,19 +622,18 @@ test08()
> declare -a parray # array contains previous value of all nodes
> Curr_value=0 # Current value extracted from numastat o/p
> Exp_incr=0 # 1 MB/ (PAGESIZE*num_of_nodes)
> - Node_num=0
> col=0
> MB=$[1024*1024]
>
> # Increase in numastat o/p is interms of pages
> Exp_incr=$[$MB/$page_size]
> # Pages will be allocated using round robin on nodes.
> - Exp_incr=$[$Exp_incr/$max_node]
> + Exp_incr=$[$Exp_incr/$total_nodes]
>
> # Check whether the pages are equally distributed among available nodes
> numastat > $LTPTMP/numalog
> COUNTER=1
> - while [ $COUNTER -le $max_node ]; do
> + for node in `echo $nodes_list`; do
> col=$[$COUNTER+1] #Node number in numastat o/p
> extract_numastat interleave_hit $interleave_hit $col || return 1
> Prev_value=$RC
> @@ -717,16 +645,15 @@ test08()
>
> numastat > $LTPTMP/numalog
> COUNTER=1
> - while [ $COUNTER -le $max_node ]; do
> + for node in `echo #nodes_list`; do
> col=$[$COUNTER+1] #Node number in numastat o/p
> - Node_num=$[$COUNTER-1] #Node numbers start from 0
> extract_numastat interleave_hit $interleave_hit $col || return 1
> Curr_value=$RC
> comparelog ${parray[$COUNTER]} $Curr_value
> if [ $RC -lt $Exp_incr ]
> then
> tst_resm TFAIL \
> - "Test #8: NUMA interleave hit in node$Node_num is less than expected"
> + "Test #8: NUMA interleave hit in node$node is less than expected"
> return 1
> fi
> COUNTER=$[$COUNTER+1]
> @@ -763,13 +690,12 @@ test09()
> Prev_value=0 # extracted from the numastat o/p
> Curr_value=0 # Current value extracted from numastat o/p
> Exp_incr=0 # 1 MB/ PAGESIZE
> - Node_num=0
> col=0
> MB=$[1024*1024]
> # Increase in numastat o/p is interms of pages
> Exp_incr=$[$MB/$page_size]
>
> - numactl --hardware > $LTPTMP/avail_nodes
> + numactl --hardware > $LTPTMP/avail_nodes
> RC=$(awk '{ if ( NR == 1 ) {print $1;} }' $LTPTMP/avail_nodes)
> if [ $RC = "available:" ]
> then
> @@ -797,34 +723,38 @@ test010()
> {
> TCID=numa10
> TST_COUNT=10
> + RC=0
> + Prev_value=0
> + Curr_value=0
>
> - while [ $COUNTER -le $max_node ]; do
> + COUNTER=1
> + for node in `echo $nodes_list`; do
>
> - if [ $max_node -eq 1 ]
> - then
> + if [ $total_nodes -eq 1 ]
> + then
> tst_brkm TBROK NULL "Preferred policy cant be applied for a single node machine"
> return 1
> fi
>
> - col=2 #column represents node0 in numastat o/p
> - numnodes=$(ls /sys/devices/system/node | grep -Ec "node[0-9]*")
> - Preferred_node=$[$[$numnodes/2]-1]
> - col=$[$Preferred_node+2]
> + if [ $COUNTER -eq $total_nodes ]; then
> + Preferred_node=`echo $nodes_list | cut -d ' ' -f 1`
> + col=2
> + else
> + Preferred_node=`echo $nodes_list | cut -d ' ' -f $[$COUNTER+1]`
> + col=$[$COUNTER+2]
> + fi
> +
> numastat > $LTPTMP/numalog
> extract_numastat other_node $other_node $col || return 1
> Prev_value=$RC
> - #echo $Preferred_node
> - #numactl --cpunodebind=$Preferred_node --localalloc support_numa $ALLOC_1MB
> - #numactl --preferred=$Preferred_node support_numa $ALLOC_1MB
> - numactl --preferred=$Preferred_node support_numa $PAUSE &
> + numactl --preferred=$node support_numa $PAUSE &
> pid=$!
> - migratepages $pid $Preferred_node $[$Preferred_node + 1]
> + migratepages $pid $node $Preferred_node
> numastat > $LTPTMP/numalog
> extract_numastat other_node $other_node $col || return 1
> Curr_value=$RC
> kill -INT $pid
> - if [ $RC -lt $Prev_value ]
> - then
> + if [ $RC -lt $Prev_value ]; then
> tst_resm TFAIL \
> "Test #10: NUMA migratepages is not working fine"
> return 1
> @@ -855,6 +785,19 @@ test010()
> exit $RC
> fi
>
> + total_nodes=0 # total no. of numa nodes
> + # all availiable nodes id list
> + nodes_list=`numactl --show | grep nodebind | cut -d ':' -f 2`
> + for node in `echo $nodes_list`; do
> + total_nodes=$[$total_nodes+1]
> + done
> + tst_resm TINFO "The system contains $total_nodes nodes: $nodes_list"
> + if [ $total_nodes -le 1 ]; then
> + tst_resm TCONF "your machine does not support numa policy or
> + your machine is not a NUMA machine"
> + exit 0
> + fi
> +
> # call each testcases sequentially
> COUNT=1
> while [ $COUNT -le $no_of_test ]; do
------------------------------------------------------------------------------
Master Java SE, Java EE, Eclipse, Spring, Hibernate, JavaScript, jQuery
and much more. Keep your Java skills current with LearnJavaNow -
200+ hours of step-by-step video tutorials by Java experts.
SALE $49.99 this month only -- learn more at:
http://p.sf.net/sfu/learnmore_122612
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list
next prev parent reply other threads:[~2013-01-09 12:04 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-01-09 11:58 [LTP] [PATCH] numa/numa01.sh: fixed errors under noncontinuous numa nodes list Zhouping Liu
2013-01-09 12:03 ` Zhouping Liu [this message]
2013-01-16 7:16 ` Wanlong Gao
2013-01-14 1:25 ` Wanlong Gao
2013-01-14 1:55 ` Zhouping Liu
2013-01-14 2:08 ` Wanlong Gao
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=50ED5C7A.2050301@redhat.com \
--to=zliu@redhat.com \
--cc=ltp-list@lists.sourceforge.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.