What can I do to solve the same problem "obj_persist_count/TEST1 failed" ?

49 views
Skip to first unread message

renzhix...@gmail.com

unread,
Mar 13, 2016, 11:18:40 AM3/13/16
to pmem
Hi everyone!
     I have tryed many times which I cannot count when I used the command “make check” in the nvml directory.
The result is always as follows:

FAIL: match: ./out1.log.match:8 did not match pattern

RUNTESTS: stopping: obj_persist_count/TEST1 failed, TEST=check FS=pmem BUILD=debug

make: *** [check] Error 1

---------------------------------------

I know the result supply that match pattern is wrong,but I don't know what the real reason is.

I have confusion for that,maybe you also have encountered this problem. If you know the truth,please tell me. Thank you.


best wishes

Renzhi

Marcin Ślusarz

unread,
Mar 13, 2016, 11:38:00 AM3/13/16
to renzhix...@gmail.com, pmem
2016-03-13 16:18 GMT+01:00 <renzhix...@gmail.com>:
> Hi everyone!
> I have tryed many times which I cannot count when I used the command
> “make check” in the nvml directory.
> The result is always as follows:
>
> FAIL: match: ./out1.log.match:8 did not match pattern
>
> RUNTESTS: stopping: obj_persist_count/TEST1 failed, TEST=check FS=pmem
> BUILD=debug

Please provide contents of src/test/obj_persist_count/out1.log and
commit id you tested this on.
Are you using "memmap" parameter to emulate pmem?

Cheers,
Marcin

renzhix...@gmail.com

unread,
Mar 14, 2016, 2:16:02 AM3/14/16
to pmem, renzhix...@gmail.com
Hi,
  1.The contents of src/test/obj_persist_count/out1.log is as follows:
obj_persist_count/TEST1: START: obj_persist_count
 ./obj_persist_count /mnt/pmem/test_obj_persist_count1/testfile
persist ;msync ;flush ;drain ;task
3 ;8 ;0 ;0 ;pool_create
6 ;0 ;1 ;0 ;root_alloc
2 ;0 ;1 ;1 ;atomic_alloc
1 ;0 ;0 ;0 ;atomic_free
16 ;0 ;6 ;1 ;tx_alloc
15 ;0 ;5 ;1 ;tx_free
18 ;0 ;5 ;1 ;tx_add
5 ;0 ;1 ;0 ;pmalloc
4 ;0 ;1 ;0 ;pfree
2 ;0 ;0 ;0 ;pmalloc_stack
1 ;0 ;0 ;0 ;pfree_stack
obj_persist_count/TEST1: Done

2.Commit id my tested this on
Sorry, I don't know what the id exactly means?!

3.Are you using "memmap" parameter to emulate pmem?
   I have the real equipment “NVDIMM”.Yes,I have used the "memmap"parameter in my system,but I don't if it is emulate.
The configuration of "memmap" is followed by website as below:  

Thanks
Renzhi

在 2016年3月13日星期日 UTC+8下午11:38:00,Marcin Ślusarz写道:
out1.log

Marcin Ślusarz

unread,
Mar 14, 2016, 5:56:05 AM3/14/16
to pmem, Renzhi XIAO
14 mar 2016 09:04 "Marcin Ślusarz" <marcin....@gmail.com> napisał(a):
>
>
> 14 mar 2016 04:18 "Renzhi XIAO" <renzhix...@gmail.com> napisał(a):
> >
> > Hi,
> > 1.The contents of src/test/obj_persist_count/out1.log is as follows:
> > obj_persist_count/TEST1: START: obj_persist_count
> > ./obj_persist_count /mnt/pmem/test_obj_persist_count1/testfile
> > persist ;msync ;flush ;drain ;task
> > 3 ;8 ;0 ;0 ;pool_create
> > 6 ;0 ;1 ;0 ;root_alloc
> > 2 ;0 ;1 ;1 ;atomic_alloc
> > 1 ;0 ;0 ;0 ;atomic_free
> > 16 ;0 ;6 ;1 ;tx_alloc
> > 15 ;0 ;5 ;1 ;tx_free
> > 18 ;0 ;5 ;1 ;tx_add
> > 5 ;0 ;1 ;0 ;pmalloc
> > 4 ;0 ;1 ;0 ;pfree
> > 2 ;0 ;0 ;0 ;pmalloc_stack
> > 1 ;0 ;0 ;0 ;pfree_stack
> > obj_persist_count/TEST1: Done

Ok, I know what is going on - it's a very recent regression (actually
forgotten "match" file update) and I'll fix it in a few minutes.


Thanks for reporting,

Marcin

renzhix...@gmail.com

unread,
Mar 14, 2016, 12:06:17 PM3/14/16
to pmem, renzhix...@gmail.com
Hi Marcin 

You are welcome. I have also encountered  many errors about vmem_stats 's match  in the command "make check" in the NVML directory. All the "grepX.log.match" file did't not match the file "grepX.log" and it was hint that uninitiated values.  
 Have you encountered this problem?

Best wishes
Renzhi

Marcin Ślusarz

unread,
Mar 14, 2016, 4:11:53 PM3/14/16
to Renzhi XIAO, pmem
Nope, please attach files that don't match.

Marcin

renzhix...@gmail.com

unread,
Mar 15, 2016, 8:07:33 AM3/15/16
to pmem, renzhix...@gmail.com
>>Nope, please attach files that don't match.

>Marcin

Hi Marcin,
     The error detail is as below:
Use of uninitialized value in printf at ../match line 174, <F> line 2.
./grep0.log.match:1   ___ Begin jemalloc statistics ___
./grep0.log:1         
FAIL: match: ./grep0.log.match:1 did not match pattern
RUNTESTS: stopping: vmem_stats/TEST0 failed, TEST=check FS=none BUILD=debug
make[4]: *** [TEST0] Error 1
make[3]: *** [vmem_stats] Error 2
make[2]: *** [pcheck] Error 2
make[1]: *** [pcheck] Error 2
error: Bad exit status from /var/tmp/rpm-tmp.hq4evh (%check)

The log file content,please see the attached file.
Best wishes 
Renzhi

grep0.log
grep0.log.match

Marcin Ślusarz

unread,
Mar 15, 2016, 10:16:12 AM3/15/16
to Renzhi XIAO, pmem
2016-03-15 13:07 GMT+01:00 <renzhix...@gmail.com>:
>> >>Nope, please attach files that don't match.
>>
>> >Marcin
>
>
> Hi Marcin,
> The error detail is as below:
> Use of uninitialized value in printf at ../match line 174, <F> line 2.
> ./grep0.log.match:1 ___ Begin jemalloc statistics ___
> ./grep0.log:1
> FAIL: match: ./grep0.log.match:1 did not match pattern
> RUNTESTS: stopping: vmem_stats/TEST0 failed, TEST=check FS=none BUILD=debug
> make[4]: *** [TEST0] Error 1
> make[3]: *** [vmem_stats] Error 2
> make[2]: *** [pcheck] Error 2
> make[1]: *** [pcheck] Error 2
> error: Bad exit status from /var/tmp/rpm-tmp.hq4evh (%check)

There are 2 unrelated bugs here.
1) "match" tool does not handle empty lines well
2) vmem_stats test produces file that contains only empty lines

First bug is easily fixable, but I don't understand the second one.
Can I see vmem0.log? What distribution do you use? Do you use any
special CFLAGS?
Full build log wouldn't hurt.

Marcin

renzhix...@gmail.com

unread,
Mar 15, 2016, 10:43:31 PM3/15/16
to pmem, renzhix...@gmail.com
》There are 2 unrelated bugs here.
>1) "match" tool does not handle empty lines well
>2) vmem_stats test produces file that contains only empty lines

>First bug is easily fixable, but I don't understand the second one.
>Can I see vmem0.log? What distribution do you use? Do you use any
>special CFLAGS?
>Full build log wouldn't hurt.

>Marcin
Hi marcin,
   1. The distribution version is 0.4+b3-74-gcb367d9;
   2. No,I don't use any special CFLAGS.
   3. I don't know what exact the build log mean, but I guess you maybe want the log of 
command "make check" in nvml directory ,which I put the log in the makecheck.txt.
please see the attached files.

Thanks
Renzhi

 
makecheck.txt
vmem0.log

Marcin Ślusarz

unread,
Mar 16, 2016, 6:54:15 AM3/16/16
to Renzhi XIAO, pmem
2016-03-16 3:43 GMT+01:00 <renzhix...@gmail.com>:
>> 》There are 2 unrelated bugs here.
>> >1) "match" tool does not handle empty lines well
>> >2) vmem_stats test produces file that contains only empty lines
>>
>> >First bug is easily fixable, but I don't understand the second one.
>> >Can I see vmem0.log? What distribution do you use? Do you use any
>> >special CFLAGS?
>> >Full build log wouldn't hurt.
>>
>
> Hi marcin,
> 1. The distribution version is 0.4+b3-74-gcb367d9;

I wanted to know what Linux distribution you use (RHEL, Fedora,
Debian, Ubuntu, etc) and which version.

> 2. No,I don't use any special CFLAGS.
> 3. I don't know what exact the build log mean, but I guess you maybe want
> the log of
> command "make check" in nvml directory ,which I put the log in the
> makecheck.txt.
> please see the attached files.

I wanted to see the output of "make" on clean tree (eg after make
clobber), but I think it's not going to give us any more information.

Just by looking at vmem0.log, I can infer that the stacktrace looks like this:
main
vmem_stats_print
je_vmem_pool_malloc_stats_print
stats_print
je_mallctl
ctl_byname
ctl_init
ctl_init_pool
base_alloc
...

or

main
vmem_stats_print
je_vmem_pool_malloc_stats_print
stats_print
je_mallctl
ctl_byname
ctl_init
ctl_init_pool
ctl_arena_init
base_alloc
...

Without being able to reproduce it locally, I'm not sure how to proceed.
Please describe the environment (Linux distribution, kernel version,
is it a distribution kernel or is it self-compiled (any patches?
.config?), do you enable any security features that might affect
mmap/memory allocation, etc)

Marcin

Renzhi XIAO

unread,
Mar 16, 2016, 10:49:51 AM3/16/16
to pmem, renzhix...@gmail.com
>>I wanted to know what Linux distribution you use (RHEL, Fedora,
>>Debian, Ubuntu, etc) and which version. 
>Without being able to reproduce it locally, I'm not sure how to proceed.
>Please describe the environment (Linux distribution, kernel version,
>is it a distribution kernel or is it self-compiled (any patches?
>>.config?), do you enable any security features that might affect
>mmap/memory allocation, etc)
 
Hi Marcin ,
      I am sorry to misunderstand the meaning as you want.Now I will tell you that my original linux version
is RHEL6.4 which kernel is 2.6.32. After then, I have self-compiled it to the kernel 4.4.4.(You can download the
source package linux-4.4.4.tar.xz) I just add some argument necessary in the .config file to support new NVDIMM.
 I have created the linux's .config file as default except the parameter related with NVDIMMs,please see the website:
 https://nvdimm.wiki.kernel.org/ .   I have not enabled any security features.

Best wishes,
Renzhi


Reply all
Reply to author
Forward
0 new messages