[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH] libdiskfs: bugs
From: |
Thomas Bushnell, BSG |
Subject: |
Re: [PATCH] libdiskfs: bugs |
Date: |
25 Apr 2001 14:20:52 -0700 |
User-agent: |
Gnus/5.0803 (Gnus v5.8.3) Emacs/20.7 |
Neal H Walfield <neal@cs.uml.edu> writes:
> As the whole point of _diskfs_ncontrol_ports is debugging and
> _diskfs_ncontrol_ports is not surrounded by #ifdef DEBUG or something
> similar, I would like to see this added.
You're incorrect. The whole point of _diskfs_ncontrol_ports is to
facilitate the filesystem's exiting once the filesystem is idle and
has no users; control ports are deemed not to count against "no
users".
Take a look at libdiskfs/ports-idle.c.
That code is currently idle, because of the XXX comment; the actual
work isn't done. That's because of two things: first, the check isn't
quite right; only filesystems started as passive translators should
exit, and perhaps the test isn't quite right even then. And second,
there were persistent threads bugs in the implementation of
ports_manage_port_operations_multithread idle timeout code, and so the
calls to ports_notice_idle aren't made.
Indeed, this is not well evidenced by the code, which is slightly
inconsistent on the whole subject. But in any case, that's the point
of it all.