Re: [tlug] procfs pseudo-file without a module load?

[Jens Oliver John <lists@example.com>]

On Mon, Jul 20, 2015 at 12:43:11PM +1000, Akira Kurogane wrote:
> As can so often happen writing an email helped me think of a new way to
> investigate the issue. And it turns out the answer must be no, you can't fiddle
> with the entries under /proc/<pid> directories.

I have used POSIX shared memory /dev/shm for things like this in the past. The
main inconvenience is that the memory objects cannot be created in a hierarchy
"/foo/bar" so I had to use prefices like in "/Foo_bar". Exposing and handling
the data is quite convenient since for example you can simply use ftruncate()
for managing the file/object size. A drawback could be that POSIX shared memory
can be disabled as a kernel option and thus is not guaranteed to always be
available.

Best regards,
Jens.