On Jul 17, 2014, at 9:37 AM, Jaromir Capik <jcapik@xxxxxxxxxx> wrote: >> I would plan on keeping the graphs the same. > > Well, that would probably be even more challenging as the particular > sets overlap. You would have to add one more mode to the 'm' key loop, > where the graph would be divided into 'used' + 'unavailable'. The empty > space on the right side would be 'available'. The 'unavailable' would > have to be calculated as total - used - available. But it's up to you > whether it's worthy to add such mode. If I keep the graphs the same, and if I’m visualizing your workstation properly, your graph *might* seem alarming with almost all memory in use. Yet the right-most portion would hint that much/most of that use may be reclaimable cache. And if there was a doubt, toggling back to view the detail figures could confirm that assumption via the new 'avail' figure. I kind of like the idea of two distinct interpretations being at your disposal. And speaking of available, what’s to trigger a recalculation of kb_main_available once the fallback method has been employed. Don’t we need something like this: diff --git a/proc/sysinfo.c b/proc/sysinfo.c index f6cc162..3ce5fc8 100644 --- a/proc/sysinfo.c +++ b/proc/sysinfo.c @@ -661,6 +661,7 @@ void meminfo(void){ FILE_TO_BUF(MEMINFO_FILE,meminfo_fd); kb_inactive = ~0UL; + kb_low_total = kb_main_available = 0; head = buf; for(;;){ Regards, Jim