[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Nmh-workers] Lock gone mad

From: Neil W Rickert
Subject: [Nmh-workers] Lock gone mad
Date: Fri, 06 Apr 2007 09:53:50 -0500

I recently installed nmh from a vendor package, on my home system.  This
is SuSE 10.1.

        % scan -version
        scan -- nmh-1.1-RC4 [compiled on e52 at Tue May  2 07:43:00 UTC 2006]

Certain commands have long pauses before they do anything.  In
particular, I notice this for "send" and for "whom".

This morning, I ran "whom" under "strace" to try to see what is

The delay appears to be due to multiple retries of the following (with

 open("/dev/,LCK.YcMuXZ", O_RDWR|O_CREAT|O_EXCL, 0600) = -1 EACCES (Permission 
 stat64("/dev/null.lock", 0xbff4ed5c)    = -1 ENOENT (No such file or directory)

WTF!  It seems that nmh is attempting to lock "/dev/null".

Personal opinion:  The locking code should be backed out and tossed
into /dev/null without first locking /dev/null :( .  MH and nmh managed
quite well for years without any locking.  What can be lost (mostly
defined sequences) is not all that serious anyway.  We are better off
without the locking code.  Moreover, if nmh ignores the lock failure
and goes ahead anyway (after sleeps and retries), then it doesn't
solve the problem that locks are supposed to solve anyway.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]