Please describe the problem:
In the Linux kernel 2.6.25 an additional field has been added to the smaps proc file: Pss. This leads to add_smaps() in sysdeps/linux/procmap.c to fail parsing the entry: it generates error messages like "Could not read smaps value Shared_Clean: ".
Here is an example of entry in /proc/self/smaps:

2ac8df322000-2ac8df406000 r--p 00000000 03:05 334328                     /usr/share/locale/UTF-8/LC_COLLATE
Size:                912 kB
Rss:                  12 kB
Pss:                   0 kB
Shared_Clean:         12 kB
Shared_Dirty:          0 kB
Private_Clean:         0 kB
Private_Dirty:         0 kB
Referenced:           12 kB


Steps to reproduce:
1. gnome-system-monitor
(do it in the console for more fun) 


Actual results:
Lots and lots of messages like "Could not read smaps value Shared_Clean:: Success" are displayed on the console. The Memory field in g-s-m has huge values (something like 4GB on x86, even quite a lot more on x86_64).  

Expected results:
No error message. The correct amount of memory is displayed in g-s-m (in the order of 100MB)

Does this happen every time?
Always, with a Linux kernel 2.6.25-rc2

Other information:
Probably the best way to solve it would be to have a more clever parser in add_smaps() which just skip field (that is, lines which starts with no whitespace and then a colon) which are unknown.

For more info on PSS: see commit ec4dd3eb35759f9fbeb5c1abb01403b2fde64cc9 . BTW, it is specifically made to represent the memory usage of a task, so it might be worth reading it :-)