[lkml]   [1997]   [Jan]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: Swap Files

On Sun, 19 Jan 1997 13:19:00 -0500 (EST), "Andrew E. Mileski"
<> said:

>> Fix the distributions, fix swapoff but don't keep dumping on the kernel to
>> fix user mode problems.
>> Alan

> You obviously don't understand the problem. That is my fault for not
> explaining it better.

I think we _do_ understand the problem, but it is still a problem with
the local system setup, not with the kernel...

> Swap files are created on-the-fly by a daemon. At shutdown, the daemon
> is killed off before it has a chance to get rid of all the swap files
> it created.

That's a bug. If you want that sort of behaviour, better to make the
daemon do a swapoff of all of its own swap files before it dies.
That's why shutdown delivers two separate signals to outstanding
daemons: a SIGTERM which the daemons can trap and die cleanly on, and
then a SIGKILL to clean up any remaining daemons still lurking.

> The task of getting rid of the swap files falls to the shutdown scripts,
> which have no way to track which swap files are in use.

Nope, it falls to whoever set up the swap files. An alternative
solution would be to let the swapfile daemon write its current state
to /etc/swaptab (similar to the way mount writes to /etc/mtab), and
use that to swapoff at shutdown. Either way, this is a user-space
problem, not a kernel space one.

> Adding a /proc/swap entry that shows all swap areas allows a modified
> swapoff command to properly turn off all swapping.

Yup. However, I tend to dislike bloating the kernel with things like
this which can be done perfectly well from user space. At the very
least, you should never rely on kernel extensions to do a job which
can be done fine without any kernel help.


 \ /
  Last update: 2005-03-22 13:38    [W:0.030 / U:0.920 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site