max thread per process in linux - linux

I wrote a simple program to calculate the maximum number of threads that a process can have in linux (Centos 5). here is the code:
int main()
{
pthread_t thrd[400];
for(int i=0;i<400;i++)
{
int err=pthread_create(&thrd[i],NULL,thread,(void*)i);
if(err!=0)
cout << "thread creation failed: " << i <<" error code: " << err << endl;
}
return 0;
}
void * thread(void* i)
{
sleep(100);//make the thread still alive
return 0;
}
I figured out that max number for threads is only 300!? What if i need more than that?
I have to mention that pthread_create returns 12 as error code.
Thanks before

There is a thread limit for linux and it can be modified runtime by writing desired limit to /proc/sys/kernel/threads-max. The default value is computed from the available system memory. In addition to that limit, there's also another limit: /proc/sys/vm/max_map_count which limits the maximum mmapped segments and at least recent kernels will mmap memory per thread. It should be safe to increase that limit a lot if you hit it.
However, the limit you're hitting is lack of virtual memory in 32bit operating system. Install a 64 bit linux if your hardware supports it and you'll be fine. I can easily start 30000 threads with a stack size of 8MB. The system has a single Core 2 Duo + 8 GB of system memory (I'm using 5 GB for other stuff in the same time) and it's running 64 bit Ubuntu with kernel 2.6.32. Note that memory overcommit (/proc/sys/vm/overcommit_memory) must be allowed because otherwise system would need at least 240 GB of committable memory (sum of real memory and swap space).
If you need lots of threads and cannot use 64 bit system your only choice is to minimize the memory usage per thread to conserve virtual memory. Start with requesting as little stack as you can live with.

Your system limits may not be allowing you to map the stacks of all the threads you require. Look at /proc/sys/vm/max_map_count, and see this answer. I'm not 100% sure this is your problem, because most people run into problems at much larger thread counts.

I had also encountered the same problem when my number of threads crosses some threshold.
It was because of the user level limit (number of process a user can run at a time) set to 1024 in /etc/security/limits.conf .
so check your /etc/security/limits.conf and look for entry:-
username -/soft/hard -nproc 1024
change it to some larger values to something 100k(requires sudo privileges/root) and it should work for you.
To learn more about security policy ,see http://linux.die.net/man/5/limits.conf.

check the stack size per thread with ulimit, in my case Redhat Linux 2.6:
ulimit -a
...
stack size (kbytes, -s) 10240
Each of your threads will get this amount of memory (10MB) assigned for it's stack. With a 32bit program and a maximum address space of 4GB, that is a maximum of only 4096MB / 10MB = 409 threads !!! Minus program code, minus heap-space will probably lead to your observed max. of 300 threads.
You should be able to raise this by compiling a 64bit application or setting ulimit -s 8192 or even ulimit -s 4096. But if this is advisable is another discussion...

You will run out of memory too unless u shrink the default thread stack size. Its 10MB on our version of linux.
EDIT:
Error code 12 = out of memory, so I think the 1mb stack is still too big for you. Compiled for 32 bit, I can get a 100k stack to give me 30k threads. Beyond 30k threads I get Error code 11 which means no more threads allowed. A 1MB stack gives me about 4k threads before error code 12. 10MB gives me 427 threads. 100MB gives me 42 threads. 1 GB gives me 4... We have 64 bit OS with 64 GB ram. Is your OS 32 bit? When I compile for 64bit, I can use any stack size I want and get the limit of threads.
Also I noticed if i turn the profiling stuff (Tools|Profiling) on for netbeans and run from the ide...I only can get 400 threads. Weird. Netbeans also dies if you use up all the threads.
Here is a test app you can run:
#include <stdio.h>
#include <stdlib.h>
#include <pthread.h>
#include <signal.h>
// this prevents the compiler from reordering code over this COMPILER_BARRIER
// this doesnt do anything
#define COMPILER_BARRIER() __asm__ __volatile__ ("" ::: "memory")
sigset_t _fSigSet;
volatile int _cActive = 0;
pthread_t thrd[1000000];
void * thread(void *i)
{
int nSig, cActive;
cActive = __sync_fetch_and_add(&_cActive, 1);
COMPILER_BARRIER(); // make sure the active count is incremented before sigwait
// sigwait is a handy way to sleep a thread and wake it on command
sigwait(&_fSigSet, &nSig); //make the thread still alive
COMPILER_BARRIER(); // make sure the active count is decrimented after sigwait
cActive = __sync_fetch_and_add(&_cActive, -1);
//printf("%d(%d) ", i, cActive);
return 0;
}
int main(int argc, char** argv)
{
pthread_attr_t attr;
int cThreadRequest, cThreads, i, err, cActive, cbStack;
cbStack = (argc > 1) ? atoi(argv[1]) : 0x100000;
cThreadRequest = (argc > 2) ? atoi(argv[2]) : 30000;
sigemptyset(&_fSigSet);
sigaddset(&_fSigSet, SIGUSR1);
sigaddset(&_fSigSet, SIGSEGV);
printf("Start\n");
pthread_attr_init(&attr);
if ((err = pthread_attr_setstacksize(&attr, cbStack)) != 0)
printf("pthread_attr_setstacksize failed: err: %d %s\n", err, strerror(err));
for (i = 0; i < cThreadRequest; i++)
{
if ((err = pthread_create(&thrd[i], &attr, thread, (void*)i)) != 0)
{
printf("pthread_create failed on thread %d, error code: %d %s\n",
i, err, strerror(err));
break;
}
}
cThreads = i;
printf("\n");
// wait for threads to all be created, although we might not wait for
// all threads to make it through sigwait
while (1)
{
cActive = _cActive;
if (cActive == cThreads)
break;
printf("Waiting A %d/%d,", cActive, cThreads);
sched_yield();
}
// wake em all up so they exit
for (i = 0; i < cThreads; i++)
pthread_kill(thrd[i], SIGUSR1);
// wait for them all to exit, although we might be able to exit before
// the last thread returns
while (1)
{
cActive = _cActive;
if (!cActive)
break;
printf("Waiting B %d/%d,", cActive, cThreads);
sched_yield();
}
printf("\nDone. Threads requested: %d. Threads created: %d. StackSize=%lfmb\n",
cThreadRequest, cThreads, (double)cbStack/0x100000);
return 0;
}

Related

munmap() failure with ENOMEM with private anonymous mapping

I have recently discovered that Linux does not guarantee that memory allocated with mmap can be freed with munmap if this leads to situation when number of VMA (Virtual Memory Area) structures exceed vm.max_map_count. Manpage states this (almost) clearly:
ENOMEM The process's maximum number of mappings would have been exceeded.
This error can also occur for munmap(), when unmapping a region
in the middle of an existing mapping, since this results in two
smaller mappings on either side of the region being unmapped.
The problem is that Linux kernel always tries to merge VMA structures if possible, making munmap fail even for separately created mappings. I was able to write a small program to confirm this behavior:
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <sys/mman.h>
// value of vm.max_map_count
#define VM_MAX_MAP_COUNT (65530)
// number of vma for the empty process linked against libc - /proc/<id>/maps
#define VMA_PREMAPPED (15)
#define VMA_SIZE (4096)
#define VMA_COUNT ((VM_MAX_MAP_COUNT - VMA_PREMAPPED) * 2)
int main(void)
{
static void *vma[VMA_COUNT];
for (int i = 0; i < VMA_COUNT; i++) {
vma[i] = mmap(0, VMA_SIZE, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
if (vma[i] == MAP_FAILED) {
printf("mmap() failed at %d\n", i);
return 1;
}
}
for (int i = 0; i < VMA_COUNT; i += 2) {
if (munmap(vma[i], VMA_SIZE) != 0) {
printf("munmap() failed at %d (%p): %m\n", i, vma[i]);
}
}
}
It allocates a large number of pages (twice the default allowed maximum) using mmap, then munmaps every second page to create separate VMA structure for each remaining page. On my machine the last munmap call always fails with ENOMEM.
Initially I thought that munmap never fails if used with the same values for address and size that were used to create mapping. Apparently this is not the case on Linux and I was not able to find information about similar behavior on other systems.
At the same time in my opinion partial unmapping applied to the middle of a mapped region is expected to fail on any OS for every sane implementation, but I haven't found any documentation that says such failure is possible.
I would usually consider this a bug in the kernel, but knowing how Linux deals with memory overcommit and OOM I am almost sure this is a "feature" that exists to improve performance and decrease memory consumption.
Other information I was able to find:
Similar APIs on Windows do not have this "feature" due to their design (see MapViewOfFile, UnmapViewOfFile, VirtualAlloc, VirtualFree) - they simply do not support partial unmapping.
glibc malloc implementation does not create more than 65535 mappings, backing off to sbrk when this limit is reached: https://code.woboq.org/userspace/glibc/malloc/malloc.c.html. This looks like a workaround for this issue, but it is still possible to make free silently leak memory.
jemalloc had trouble with this and tried to avoid using mmap/munmap because of this issue (I don't know how it ended for them).
Do other OS's really guarantee deallocation of memory mappings? I know Windows does this, but what about other Unix-like operating systems? FreeBSD? QNX?
EDIT: I am adding example that shows how glibc's free can leak memory when internal munmap call fails with ENOMEM. Use strace to see that munmap fails:
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <sys/mman.h>
// value of vm.max_map_count
#define VM_MAX_MAP_COUNT (65530)
#define VMA_MMAP_SIZE (4096)
#define VMA_MMAP_COUNT (VM_MAX_MAP_COUNT)
// glibc's malloc default mmap_threshold is 128 KiB
#define VMA_MALLOC_SIZE (128 * 1024)
#define VMA_MALLOC_COUNT (VM_MAX_MAP_COUNT)
int main(void)
{
static void *mmap_vma[VMA_MMAP_COUNT];
for (int i = 0; i < VMA_MMAP_COUNT; i++) {
mmap_vma[i] = mmap(0, VMA_MMAP_SIZE, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
if (mmap_vma[i] == MAP_FAILED) {
printf("mmap() failed at %d\n", i);
return 1;
}
}
for (int i = 0; i < VMA_MMAP_COUNT; i += 2) {
if (munmap(mmap_vma[i], VMA_MMAP_SIZE) != 0) {
printf("munmap() failed at %d (%p): %m\n", i, mmap_vma[i]);
return 1;
}
}
static void *malloc_vma[VMA_MALLOC_COUNT];
for (int i = 0; i < VMA_MALLOC_COUNT; i++) {
malloc_vma[i] = malloc(VMA_MALLOC_SIZE);
if (malloc_vma[i] == NULL) {
printf("malloc() failed at %d\n", i);
return 1;
}
}
for (int i = 0; i < VMA_MALLOC_COUNT; i += 2) {
free(malloc_vma[i]);
}
}
One way to work around this problem on Linux is to mmap more that 1 page at once (e.g. 1 MB at a time), and also map a separator page after it. So, you actually call mmap on 257 pages of memory, then remap the last page with PROT_NONE, so that it cannot be accessed. This should defeat the VMA merging optimization in the kernel. Since you are allocating many pages at once, you should not run into the max mapping limit. The downside is you have to manually manage how you want to slice the large mmap.
As to your questions:
System calls can fail on any system for a variety of reasons. Documentation is not always complete.
You are allowed to munmap a part of a mmapd region as long as the address passed in lies on a page boundary, and the length argument is rounded up to the next multiple of the page size.

Linux clock_gettime() elapse spikes?

I'm try to get high resolution timestamp on linux. Using clock_gettime(), as below, I got "spike" elapses that looks pretty horrible at almost 26 micro second elapse. Most of the "dt"'s are around 30 ns. I was on linux 2.6.32, Red Hat 4.4.6. 'lscpu' shows CPU MHz=2666.121. I thought that means each each clock tick needs about 2 ns. So, asking for ns resolution didn't see like too unreasonable here.
output of program (sorry wasn't able to post this without making it a list. It thinks it's code some how)
1397534268,40823395 1397534268,40827950,dt=4555
1397534268,41233555 1397534268,41236716,dt=3161
1397534268,41389902 1397534268,41392922,dt=3020
1397534268,46488430 1397534268,46491674,dt=3244
1397534268,46531297 1397534268,46534279,dt=2982
1397534268,46823368 1397534268,46849336,dt=25968
1397534268,46915657 1397534268,46918663,dt=3006
1397534268,51488643 1397534268,51491791,dt=3148
1397534268,51530490 1397534268,51533496,dt=3006
1397534268,51823307 1397534268,51826904,dt=3597
1397534268,55823359 1397534268,55827826,dt=4467
1397534268,60531184 1397534268,60534183,dt=2999
1397534268,60823381 1397534268,60844866,dt=21485
1397534268,60913003 1397534268,60915998,dt=2995
1397534268,65823269 1397534268,65827742,dt=4473
1397534268,70823376 1397534268,70835280,dt=11904
1397534268,75823489 1397534268,75828872,dt=5383
1397534268,80823503 1397534268,80859500,dt=35997
1397534268,86823381 1397534268,86831907,dt=8526
Any ideas? thanks
#include <vector>
#include <iostream>
#include <time.h>
long long elapse( const timespec& t1, const timespec& t2 )
{
return ( t2.tv_sec * 1000000000L + t2.tv_nsec ) -
t1.tv_sec * 1000000000L + t1.tv_nsec );
}
int main()
{
const unsigned n=30000;
timespec ts;
std::vector<timespec> t( n );
for( unsigned i=0; i < n; ++i )
{
clock_gettime( CLOCK_REALTIME, &ts );
t[i] = ts;
}
std::vector<long> dt( n );
for( unsigned i=1; i < n; ++i )
{
dt[i] = elapse( t[i-1], t[i] );
if( dt[i] > 1000 )
{
std::cerr <<
t[i-1].tv_sec << ","
<< t[i-1].tv_nsec << " "
<< t[i].tv_sec << ","
<< t[i].tv_nsec
<< ",dt=" << dt[i] << std::endl;
}
else
{
//normally I get dt[i] = approx 30-35 nano secs
}
}
return 0;
}
The numbers you quoted are in the 3 to 30 microsecond range (3,000 to 30,000 nanoseconds). That is too short a time to be a context switch to another thread/process, let the other thread run, and context switch back to your thread. Most likely the core where your process was running was used by the kernel to service an external interrupt (e.g. network card, disk, timer), then returned to running your process.
You can watch the linux interrupt counters (per CPU core and per source) with this command
watch -d -n 0.2 cat /proc/interrupts
The -n 0.2 will cause the command to be issued at 5Hz, the -d flag will highlight what has changed.
The source of the interrupt could also be a TLB shootdown, which results in an IPI (Inter-Processor Interrupt). You can read more about TLB shootdowns here.
If you want to reduce the number of interrupts serviced by the core running your thread/process, you need to set the interrupt affinity. You can learn more about Red Hat Interrupts and IRQ (Interrupt requests) tuning here, and here.
Worth noting is that you are using CLOCK_REALTIME which isn't guaranteed to be "smooth", it could jump around as the system clock is "disciplined" to keep accurate time by a service like NTP (Network Time Protocol) or PTP (Precision Time Protocol). For your purposes it is better to use CLOCK_MONOTONIC, you can read more about the difference here. When a clock is "disciplined" the clock can jump by a "step" - this is unusual and certainly not the cause of the many spikes you see.
Could you check the resolution with clock_getres()?
I suspect what you are measuring here is called "OS Noise". This is often caused by your program getting pre-empted by the operating system. The operating system then performs other work. There are numerous causes, but commonly it is: other runnable tasks, hardware interrupts, or timer events.
The FTQ/FWQ benchmarks were designed to measure this characteristic and the summary contains some further information:
https://asc.llnl.gov/sequoia/benchmarks/FTQ_summary_v1.1.pdf

malloc large memory never returns NULL

when I run this, it seems to have no problem with keep allocating memory with cnt going over thousands. I don't understand why -- aren't I supposed to get a NULL at some point? Thanks!
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <math.h>
int main(void)
{
long C = pow(10, 9);
int cnt = 0;
int conversion = 8 * 1024 * 1024;
int *p;
while (1)
{
p = (int *)malloc(C * sizeof(int));
if (p != NULL)
cnt++;
else break;
if (cnt % 10 == 0)
printf("number of successful malloc is %d with %ld Mb\n", cnt, cnt * C / conversion);
}
return 0;
}
Are you running this on Linux? Linux has a highly surprising feature known as overcommit. It doesn't actually allocate memory when you call malloc(), but rather when you actually use that memory. malloc() will happily let you allocate as much memory as your heart desires, never returning a NULL pointer.
It's only when you actually access the memory that Linux takes you seriously and goes out searching for free memory to give you. Of course there may not actually be enough memory to meet the promise it gave your program. You say, "Give me 8GB," and malloc() says, "Sure." Then you try to write to your pointer and Linux says, "Oops! I lied. How bout I just kill off processes (probably yours) until I I free up enough memory?"
You're allocating virtual memory. On a 64-bit OS, virtual memory is available in almost unlimited supply.

high resource usage program stalls/crashes linux

I have a program that reads about 1000 images and creates a statistical summary of their contents. Each image is processed in its own thread using OpenMP, and I have the thread limit set to match my number of processors.
Until about two weeks ago, the program ran fine. Now, however, if I run the program more than once, my system slows down and eventually freezes up.
In order to troubleshoot, I wrote the simple code listed below that emulates what my program is doing. This code will freeze my system, just as my original program does, after trying to read only a few files at line 35.
I ran the program, successively reverting to an earlier kernel after each failure, and found that it fails with all 3.6 kernels up to version 3.6.8.
However, when I go back to kernel 3.5.6, it works.
test.cc:
1 #include <cstdio>
2 #include <iostream>
3 #include <vector>
4 #include <unistd.h>
5
6 using namespace std;
7
8 int main ()
9 {
10 // number of files
11 const size_t N = 1000;
12 // total system memory
13 const size_t MEM = sysconf (_SC_PHYS_PAGES) * sysconf (_SC_PAGE_SIZE);
14 // file size
15 const size_t SZ = MEM/N;
16
17 // create temp filenames
18 vector<string> fn (N);
19 for (size_t i = 0; i < fn.size (); ++i)
20 fn[i] = string (tmpnam (NULL));
21
22 // write a bunch of files to disk
23 for (size_t i = 0; i < fn.size (); ++i)
24 {
25 vector<char> a (SZ);
26 FILE *fp = fopen (fn[i].c_str (), "wb");
27 fwrite (&a[0], a.size (), 1, fp);
28 clog << fn[i] << " written" << endl;
29 }
30
31 // read a bunch of files from disk
32 #pragma omp parallel for
33 for (size_t i = 0; i < fn.size (); ++i)
34 {
35 vector<char> a (SZ);
36 FILE *fp = fopen (fn[i].c_str (), "rb");
37 fread (&a[0], a.size (), 1, fp);
38 clog << fn[i] << " read" << endl;
39 }
40
41 return 0;
42 }
Makefile:
1 a:$
2 g++ -fopenmp -Wall -o test -g test.cc$
3 ./test$
My question is: What is different about kernel 3.6 that would cause this program to fail, but does not cause it to fail in version 3.5?
Without going through the code, if you want to set some limits to your processes, have a look at cgroups for limiting resource usage.
As for the freezing - you are trying to read/write GBs of data to disk at once. Given the speeds of ~100MB/s of today's hard-drives, I would expect a freeze at the time the kernel decides to flush the caches to the disk - which will probably occur as soon as you try to read a reasonably sized chunk of data from the disk under memory pressure (since you allocated lots of memory, the space for caches is limited).
You can try to mmap() the files or change kernel I/O scheduler.
I haven't look in deep at your code, but I realised some bad practices (at least, I thing they're) :
First, the critical section inside the openmp loop. That is a synchronism point, and putting it in every iteration sounds kind of problematic to me. Since each thread must be sure no other one has entered there, probably the overhead that synchronism introduces increases with the number of threads.
Second: I am not very used to C++, but I guess that every time vector<char> a (SZ) is executed memory is allocated (and freed at the end of the block). Excuse me if I am wrong. Since you know beforehand the value of SZ, you'll better allocate a vector<vector<char> > with as many elements as threads before the parallel region. Then, in the parallel region, you'd make each thread access its vector<char>.

Zero bytes lost in Valgrind

What does it mean when Valgrind reports o bytes lost, like here:
==27752== 0 bytes in 1 blocks are definitely lost in loss record 2 of 1,532
I suspect it is just an artifact from creative use of malloc, but it is good to be sure (-;
EDIT: Of course the real question is whether it can be ignored or it is an effective leak that should be fixed by freeing those buffers.
Yes, this is a real leak, and it should be fixed.
When you malloc(0), malloc may either give you NULL, or an address that is guaranteed to be different from that of any other object.
Since you are likely on Linux, you get the second. There is no space wasted for the allocated buffer itself, but libc has to do some housekeeping, and that does waste space, so you can't go on doing malloc(0) indefinitely.
You can observe it with:
#include <stdio.h>
#include <stdlib.h>
int main() {
unsigned long i;
for (i = 0; i < (size_t)-1; ++i) {
void *p = malloc(0);
if (p == NULL) {
fprintf(stderr, "Ran out of memory on %ld iteration\n", i);
break;
}
}
return 0;
}
gcc t.c && bash -c 'ulimit -v 10240 && ./a.out'
Ran out of memory on 202751 iteration
It looks like you allocated a block with 0 size and then didn't subsequently free it.

Resources