This is the mail archive of the glibc-bugs@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug malloc/21273] insufficient memory with more then 10 GB free memory


https://sourceware.org/bugzilla/show_bug.cgi?id=21273

--- Comment #10 from Adhemerval Zanella <adhemerval.zanella at linaro dot org> ---
(In reply to dieter ferdinand from comment #9)
> hello,
> the problem is, that chrome/chromium starts a new process for every task, so
> it is impossible to trace all.
> 
> i can log the memory-usage of the system and the error-message or can try to
> check how many memory i can allocate with memtester.
> 
> i don't know, how i can get the information which you want.

So far you have indicate that *chrome* is misbehaving regarding memory
allocation (which is not an usual thing given its internet reputation). 
Nevertheless, you must indicate that this is indeed caused by a glibc issue
instead of something from other project.  The only indication you gave us so
far is you have updated your distro and it come with an update glibc, but also
potentially with multiple updates (other miscellaneous libraries and kernel). 
So this potentially issue can 1. chrome itself, 2. any potentially library
chrome uses (on my chrome installation it is linked against 81 libraries), 3.
libc (although as indicated by previous comments chrome itself does not use
libc allocator), 4. kernel, 5. mist (configuration issue, etc).

So I would recommend you to open an issue on chrome bug report with this
information.  If it is not an know issue, they will prob track it down and
report back to us if this is the case.  You can try to run chrome with a memory
profiler (valgrind) or build with memory instrumentation (asan), however this
is out of the scope of this bugzilla and it would to seek out this information
outside it.

Unfortunately without further information on how this is linked on glibc I am
afraid I can't help you out. 

> 
> this is not one process which makes problems. the problem is tat i get out
> of memory with more then 10 gb of availabe memory or if some process like
> chromium allocate much memory in a very short time!
> 
> at the moment, i have a virtual size for every chrome-process like this:
> 20876 dieter    20   0  807840 226216  84600 S   0,7  1,4  18:13.81 chrome  
> 
> 18853 dieter    20   0 1458940 186224  70656 S   0,0  1,1 173:52.01 chrome  
> 
> 21987 dieter    20   0  710472 179192  22340 S   0,3  1,1  41:39.29 chrome  
> 
> 20496 dieter    20   0  757240 128556  28204 S   0,0  0,8  14:13.67 chrome  
> 
> 21515 dieter    20   0  561164 120440  76568 S   0,0  0,7   1:07.86 chrome  
> 
> 12950 root      20   0 2673928  89100   7792 R  98,7  0,5   1858:49
> TMPGEnc.exe
>  5357 dieter    20   0  674432  76756  16264 S   0,0  0,5   7:19.20 chrome  
> 
> 22053 dieter    20   0  875196  56288  19956 S   1,3  0,3 145:04.11 chrome  
> 
> 20968 dieter    20   0  632740  55836  27860 S   0,0  0,3   5:42.53 chrome  
> 
> 18223 dieter    20   0  691952  55660  30648 S   0,0  0,3 127:21.16 chrome  
> 
>  5468 dieter    20   0  735424  55592  13964 S   0,3  0,3  78:08.28 chrome  
> 
> i think, the virtual size of chrome exceed 30GB
> 
> and this is the memory-status:
> KiB Mem : 16393256 total,   576408 free,  2220832 used, 13596016 buff/cache
> KiB Swap:  8380412 total,  4749532 free,  3630880 used. 13911028 avail Mem 
> 
> a tool, which i get from sombody report this for chrome:
>   1.0 GiB +  99.1 MiB =   1.1 GiB       chrome (67)
> 
> and pmap -X calculate for chrome 37 090 672 bytes.
> 
> and pmap -X for all:
> Size  Rss  Pss Referenced Anonymous ShmemPmdMapped Shared_Hugetlb
> Private_Hugetlb Swap SwapPss Locked Mapping
> 76480008 3511412 1884084    3220560   1706164              0              0 
> 0 4837468 2937521   2723 KB 
> 
> but what is the REAL memory-usage? the best value for total memory-usage
> (1.8 GiB) comes from the ps_mem-python-script.
> 
> goodby

-- 
You are receiving this mail because:
You are on the CC list for the bug.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]