1Documentation for /proc/sys/fs/* kernel version 2.2.10 2 (c) 1998, 1999, Rik van Riel <riel@nl.linux.org> 3 4For general info and legal blurb, please look in README. 5 6============================================================== 7 8This file contains documentation for the sysctl files in 9/proc/sys/fs/ and is valid for Linux kernel version 2.2. 10 11The files in this directory can be used to tune and monitor 12miscellaneous and general things in the operation of the Linux 13kernel. Since some of the files _can_ be used to screw up your 14system, it is advisable to read both documentation and source 15before actually making adjustments. 16 17Currently, these files are in /proc/sys/fs: 18- dentry-state 19- dquot-max 20- dquot-nr 21- file-max 22- file-nr 23- inode-max 24- inode-nr 25- inode-state 26- nr_open 27- overflowuid 28- overflowgid 29- suid_dumpable 30- super-max 31- super-nr 32 33Documentation for the files in /proc/sys/fs/binfmt_misc is 34in Documentation/binfmt_misc.txt. 35 36============================================================== 37 38dentry-state: 39 40From linux/fs/dentry.c: 41-------------------------------------------------------------- 42struct { 43 int nr_dentry; 44 int nr_unused; 45 int age_limit; /* age in seconds */ 46 int want_pages; /* pages requested by system */ 47 int dummy[2]; 48} dentry_stat = {0, 0, 45, 0,}; 49-------------------------------------------------------------- 50 51Dentries are dynamically allocated and deallocated, and 52nr_dentry seems to be 0 all the time. Hence it's safe to 53assume that only nr_unused, age_limit and want_pages are 54used. Nr_unused seems to be exactly what its name says. 55Age_limit is the age in seconds after which dcache entries 56can be reclaimed when memory is short and want_pages is 57nonzero when shrink_dcache_pages() has been called and the 58dcache isn't pruned yet. 59 60============================================================== 61 62dquot-max & dquot-nr: 63 64The file dquot-max shows the maximum number of cached disk 65quota entries. 66 67The file dquot-nr shows the number of allocated disk quota 68entries and the number of free disk quota entries. 69 70If the number of free cached disk quotas is very low and 71you have some awesome number of simultaneous system users, 72you might want to raise the limit. 73 74============================================================== 75 76file-max & file-nr: 77 78The kernel allocates file handles dynamically, but as yet it 79doesn't free them again. 80 81The value in file-max denotes the maximum number of file- 82handles that the Linux kernel will allocate. When you get lots 83of error messages about running out of file handles, you might 84want to increase this limit. 85 86The three values in file-nr denote the number of allocated 87file handles, the number of unused file handles and the maximum 88number of file handles. When the allocated file handles come 89close to the maximum, but the number of unused file handles is 90significantly greater than 0, you've encountered a peak in your 91usage of file handles and you don't need to increase the maximum. 92 93============================================================== 94 95nr_open: 96 97This denotes the maximum number of file-handles a process can 98allocate. Default value is 1024*1024 (1048576) which should be 99enough for most machines. Actual limit depends on RLIMIT_NOFILE 100resource limit. 101 102============================================================== 103 104inode-max, inode-nr & inode-state: 105 106As with file handles, the kernel allocates the inode structures 107dynamically, but can't free them yet. 108 109The value in inode-max denotes the maximum number of inode 110handlers. This value should be 3-4 times larger than the value 111in file-max, since stdin, stdout and network sockets also 112need an inode struct to handle them. When you regularly run 113out of inodes, you need to increase this value. 114 115The file inode-nr contains the first two items from 116inode-state, so we'll skip to that file... 117 118Inode-state contains three actual numbers and four dummies. 119The actual numbers are, in order of appearance, nr_inodes, 120nr_free_inodes and preshrink. 121 122Nr_inodes stands for the number of inodes the system has 123allocated, this can be slightly more than inode-max because 124Linux allocates them one pageful at a time. 125 126Nr_free_inodes represents the number of free inodes (?) and 127preshrink is nonzero when the nr_inodes > inode-max and the 128system needs to prune the inode list instead of allocating 129more. 130 131============================================================== 132 133overflowgid & overflowuid: 134 135Some filesystems only support 16-bit UIDs and GIDs, although in Linux 136UIDs and GIDs are 32 bits. When one of these filesystems is mounted 137with writes enabled, any UID or GID that would exceed 65535 is translated 138to a fixed value before being written to disk. 139 140These sysctls allow you to change the value of the fixed UID and GID. 141The default is 65534. 142 143============================================================== 144 145suid_dumpable: 146 147This value can be used to query and set the core dump mode for setuid 148or otherwise protected/tainted binaries. The modes are 149 1500 - (default) - traditional behaviour. Any process which has changed 151 privilege levels or is execute only will not be dumped 1521 - (debug) - all processes dump core when possible. The core dump is 153 owned by the current user and no security is applied. This is 154 intended for system debugging situations only. Ptrace is unchecked. 1552 - (suidsafe) - any binary which normally would not be dumped is dumped 156 readable by root only. This allows the end user to remove 157 such a dump but not access it directly. For security reasons 158 core dumps in this mode will not overwrite one another or 159 other files. This mode is appropriate when administrators are 160 attempting to debug problems in a normal environment. 161 162============================================================== 163 164super-max & super-nr: 165 166These numbers control the maximum number of superblocks, and 167thus the maximum number of mounted filesystems the kernel 168can have. You only need to increase super-max if you need to 169mount more filesystems than the current value in super-max 170allows you to. 171 172============================================================== 173 174aio-nr & aio-max-nr: 175 176aio-nr shows the current system-wide number of asynchronous io 177requests. aio-max-nr allows you to change the maximum value 178aio-nr can grow to. 179 180============================================================== 181