shmfs/tmpfs/vm-fs

shmfs/tmpfs/vm-fs

[der erste Schuettler]

hi all,

How is SysV sharedmem implemented and how/why is
shmfs/tmpfs integrated in this topic?

A file in the shmfs/tmpfs is created with the required size and then with mmap
mapped into the prozessspace?

thanks lothar:x
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: shmfs/tmpfs/vm-fs

[Christoph Rohland]

[-- Attachment #1: Type: text/plain, Size: 498 bytes --]

Hi der,

On Thu, 6 Dec 2001, der erste Schuettler wrote:
> How is SysV sharedmem implemented and how/why is
> shmfs/tmpfs integrated in this topic?

shmem.c (^=tmpfs) is the implementation used by the SYSV SHM user
space API.

> A file in the shmfs/tmpfs is created with the required size and then
> with mmap mapped into the prozessspace?

Yes. shmget does create and size a tmpfs file and shmat maps it.

See the appended tmpfs.txt which was just included into 2.4.17-pre

Greetings
		Christoph


[-- Attachment #2: tmpfs.txt --]
[-- Type: text/plain, Size: 4181 bytes --]

Tmpfs is a file system which keeps all files in virtual memory.


Everything in tmpfs is temporary in the sense that no files will be
created on your hard drive. If you unmount a tmpfs instance,
everything stored therein is lost.

tmpfs puts everything into the kernel internal caches and grows and
shrinks to accommodate the files it contains and is able to swap
unneeded pages out to swap space. It has maximum size limits which can
be adjusted on the fly via 'mount -o remount ...'

If you compare it to ramfs (which was the template to create tmpfs)
you gain swapping and limit checking. Another similar thing is the RAM
disk (/dev/ram*), which simulates a fixed size hard disk in physical
RAM, where you have to create an ordinary filesystem on top. Ramdisks
cannot swap and you do not have the possibility to resize them. 

Since tmpfs lives completely in the page cache and on swap, all tmpfs
pages currently in memory will show up as cached. It will not show up
as shared or something like that. Further on you can check the actual
RAM+swap use of a tmpfs instance with df(1) and du(1).


tmpfs has the following uses:

1) There is always a kernel internal mount which you will not see at
   all. This is used for shared anonymous mappings and SYSV shared
   memory. 

   This mount does not depend on CONFIG_TMPFS. If CONFIG_TMPFS is not
   set, the user visible part of tmpfs is not build. But the internal
   mechanisms are always present.

2) glibc 2.2 and above expects tmpfs to be mounted at /dev/shm for
   POSIX shared memory (shm_open, shm_unlink). Adding the following
   line to /etc/fstab should take care of this:

	tmpfs	/dev/shm	tmpfs	defaults	0 0

   Remember to create the directory that you intend to mount tmpfs on
   if necessary (/dev/shm is automagically created if you use devfs).

   This mount is _not_ needed for SYSV shared memory. The internal
   mount is used for that. (In the 2.3 kernel versions it was
   necessary to mount the predecessor of tmpfs (shm fs) to use SYSV
   shared memory)

3) Some people (including me) find it very convenient to mount it
   e.g. on /tmp and /var/tmp and have a big swap partition. But be
   aware: loop mounts of tmpfs files do not work due to the internal
   design. So mkinitrd shipped by most distributions will fail with a
   tmpfs /tmp.

4) And probably a lot more I do not know about :-)


tmpfs has a couple of mount options:

size:	   The limit of allocated bytes for this tmpfs instance. The 
           default is half of your physical RAM without swap. If you
	   oversize your tmpfs instances the machine will deadlock
	   since the OOM handler will not be able to free that memory.
nr_blocks: The same as size, but in blocks of PAGECACHE_SIZE.
nr_inodes: The maximum number of inodes for this instance. The default
           is half of the number of your physical RAM pages.

These parameters accept a suffix k, m or g for kilo, mega and giga and
can be changed on remount.

To specify the initial root directory you can use the following mount
options:

mode:	The permissions as an octal number
uid:	The user id 
gid:	The group id

These options do not have any effect on remount. You can change these
parameters with chmod(1), chown(1) and chgrp(1) on a mounted filesystem.


So 'mount -t tmpfs -o size=10G,nr_inodes=10k,mode=700 tmpfs /mytmpfs'
will give you tmpfs instance on /mytmpfs which can allocate 10GB
RAM/SWAP in 10240 inodes and it is only accessible by root.


TODOs:

1) give the size option a percent semantic: If you give a mount option
   size=50% the tmpfs instance should be able to grow to 50 percent of
   RAM + swap. So the instance should adapt automatically if you add
   or remove swap space.
2) loop mounts: This is difficult since loop.c relies on the readpage
   operation. This operation gets a page from the caller to be filled
   with the content of the file at that position. But tmpfs always has
   the page and thus cannot copy the content to the given page. So it
   cannot provide this operation. The VM had to be changed seriously
   to achieve this.
3) Show the number of tmpfs RAM pages. (As shared?)

Author:
   Christoph Rohland <cr@sap.com>, 1.12.01

Re: shmfs/tmpfs/vm-fs

[der erste Schuettler]

hi

thanks for the paper
but, with sysvipc,  are msyncs still needed, to keep the shared pages in sync 
with the file on the tmpfs? You could use the same pages for all...
tmpfs ist cool, is it possible to change the permissions on an shared object
or istead of shmctl IPC_RMID just use rm /de/shm/SYSVblablub?


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: shmfs/tmpfs/vm-fs

[Christoph Rohland]

Hi Lothar,

On Fri, 7 Dec 2001, der erste Schuettler wrote:
> thanks for the paper but, with sysvipc, are msyncs still needed, to
> keep the shared pages in sync with the file on the tmpfs? 

Of course not. sync on tmpfs does nothing.

> You could use the same pages for all...  tmpfs ist cool, is it
> possible to change the permissions on an shared object or istead of
> shmctl IPC_RMID just use rm /de/shm/SYSVblablub?

It was possible in some 2.3 kernels, but this had to be removed with
the cleanup :-(

Greetings
		Christoph


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: shmfs/tmpfs/vm-fs

[H. Peter Anvin]

Christoph Rohland wrote:

> 
> It was possible in some 2.3 kernels, but this had to be removed with
> the cleanup :-(
> 


I guess I really still don't understand why.  I certainly can understand 
that it would be highly undesirable if it had to be supported before 
/dev/shm is mounted, but I don't see any reason to allow SysV shared 
memory before mounting /dev/shm.

	-hpa


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: shmfs/tmpfs/vm-fs

[Christoph Rohland]

Hi hpa,

On Fri, 07 Dec 2001, H. Peter Anvin wrote:
> Christoph Rohland wrote:
> 
>> It was possible in some 2.3 kernels, but this had to be removed
>> with the cleanup :-(
> 
> I guess I really still don't understand why.  

It is mainly a matter of simplicity: Filesystem semantics differ
considerably from the SYSV semantics and I can assure you that the old
implementation was a nightmare to maintain and keep compatible with
all the little oddities of SYSV shm.

E.g. there is a creator id which you cannot change and
which is always allowed to control the opject. Also the Linux feature,
that you can SHM_RMID a segment but still access is with its id as
long as there are references is pretty unusual for filesystems.

So I had to realise that the SYSV shm API does _not_ correspond to
files in a directory, but it more an open struct file without a linked
directory entry. This struct file is managed by the SYSV ipc module
and the user can access it via shmat. This model works out quite
naturally.

The whole interface between ipc/shm.c and mm/shmem.c is
shmem_file_setup, shmem_nopage and shmem_lock!  shmem does not know
anything about SYSV and its weird semantics.

> I certainly can understand that it would be highly undesirable if it
> had to be supported before /dev/shm is mounted, but I don't see any
> reason to allow SysV shared memory before mounting /dev/shm.

I would like this behaviour as well, but it would mean to complicate
the whole thing more than it's worth.

Greetings
		Christoph


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/