mprotect() failed: Cannot allocate memory

遇到这个问题是在測试项目的性能时发现的，每一个对象分配一页大小的空间然后mprotect() 保护起来，当系统分配3W多个页的时候会出现这个问题。

google到在一份邮件列表中也曾提到该问题.https://sourceware.org/ml/libc-help/2010-04/msg00026.html



摘抄部分例如以下：

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

On Wed, 2010-04-21 at 01:05 +0200, PaweÅ Sikora wrote:

> hi,

> 

> i'm trying to debug an ugly application with ElectricFence.

> in fact, on x86-64 box with 8GB ram and 16GB swap i'm getting following error:

> 

> "ElectricFence Exiting: mprotect() failed: Cannot allocate memory"

> 

> the program has been compiled with gcc-4.5, glibc-2.11.1, kernel-2.6.32.

> did you ever come across such (kernel/glibc) limitations?

> 

> here's a simple testcase which triggs -ENOMEM in mprotect().



You probably depleted the max map count, see:

  /proc/sys/vm/max_map_count



We have a limit on the number of maps you can have, those mprotect()

calls split you maps like crazy, see also /proc/$pid/maps.

eg. change your second test program to include something like:

char buf[128];

  snprintf(buf, sizeof(buf), "cat /proc/%d/maps", (int)getpid());

  system(buf);



at the end after lowering your NN count to fit, and observe the result

of those mprotect() calls.



yes, that is the clue :)



the limit in /proc/sys/vm/max_map_count was set to 65530.

with `echo 128000 > /proc/sys/vm/max_map_count` the testcase passes.

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

所以 解决方式是设置/proc/sys/vm/max_map_count为更大的值。

查看 max_map_count值：

more /proc/sys/vm/max_map_count

vm.max_map_count=128000直接写到/etc/sysctl.conf中,

然后运行

sysctl -p

或者直接运行

echo 128000 > /proc/sys/vm/max_map_count

OK 到这里问题就攻克了。总结一下原因：

proc文件系统给用户提供了非常多内核信息帮助，使得用户能够通过改动内核參数达到提高系统性能的目的.

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

wangtao@wangtao-Z87X-UD5H:/proc/sys/vm$ ls

block_dump                  legacy_va_layout           oom_dump_tasks

compact_memory              lowmem_reserve_ratio       oom_kill_allocating_task

dirty_background_bytes      max_map_count              overcommit_memory

dirty_background_ratio      memory_failure_early_kill  overcommit_ratio

dirty_bytes                 memory_failure_recovery    page-cluster

dirty_expire_centisecs      min_free_kbytes            panic_on_oom

dirty_ratio                 min_slab_ratio             percpu_pagelist_fraction

dirty_writeback_centisecs   min_unmapped_ratio         scan_unevictable_pages

drop_caches                 mmap_min_addr              stat_interval

extfrag_threshold           nr_hugepages               swappiness

extra_free_kbytes           nr_hugepages_mempolicy     vfs_cache_pressure

hugepages_treat_as_movable  nr_overcommit_hugepages    would_have_oomkilled

hugetlb_shm_group           nr_pdflush_threads         zone_reclaim_mode

laptop_mode                 numa_zonelist_order

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

/proc/sys/vm 下每一个文件的信息详见http://www.linuxinsight.com/proc_sys_vm_hierarchy.html

max_map_count :  This file contains the maximum number of memory map areas a process may have. Memory map areas are used as a side-effect of calling
malloc, directly by mmap and mprotect, and also when loading shared libraries. While most applications need less than a thousand maps, certain programs, particularly malloc debuggers, may consume lots of them, e.g., up to one or two maps per allocation. The
default value is 65536.