Re: Floating Point Exception

[Samuel Thibault <samuel.thibault@eu.citrix.com>]

[-- Attachment #1: Type: text/plain, Size: 877 bytes --]

Cody Jarrett, le Fri 11 Jul 2008 11:27:35 -0500, a écrit :
> It seems to happen randomly when executing the 'rpm' or 'yum'  
> commands. It happens on centos 5, 5.1, and 5.2 machines while using  
> the xen kernel and on 3 of my completely different servers. When it  
> does happen, all the strace outputs seem to be the same, encountering  
> the SIGFPE after the second gettimeofday. What should I provide to  
> help track this issue down?

Well, I had a look at the rpm code, and it's quite bogus: it uses
gettimeofday() to measure the time taken by nanosleep(20ms). There is
no way that can reliably work in all environments, virtualized or not,
since the resolution of gettimeofday may be >20ms (posix even says it is
unspecified). The attached patch should fix that, but I guess it'll be
hard to get it applied to all rpm-based distributions :/

Samuel

[-- Attachment #2: patch --]
[-- Type: text/plain, Size: 298 bytes --]

--- rpmio/rpmsw.c.orig	2008-07-11 18:19:43.000000000 +0100
+++ rpmio/rpmsw.c	2008-07-11 18:19:57.000000000 +0100
@@ -220,7 +220,7 @@
 	rpmsw_type = 1;
 
 	/* Compute cycles/usec */
-	rpmsw_cycles = sum_cycles/sum_usecs;
+	rpmsw_cycles = sum_cycles/(sum_usecs + 1);
 #else
 	rpmsw_type = 0;
 #endif

[-- Attachment #3: Type: text/plain, Size: 138 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel