memory_full_info() vs memory_maps()

[rohithvsm]

Hi,
Thanks for the great tool. I have an older version of psutil (2.1.0), which means I can't use memory_full_info(). And I can't upgrade to the latest version in my current situation. I was curious about the other function, memory_maps(). In order to get pss of a processs, can I add up the pss field of all the tuples in the list returned by memory_maps()? And will that be equal to memory_full_info().pss? I am sorry, I don't mean to open an issue here. But I am looking for an answer to if this is a reliable method. Thank you for your time.

[giampaolo]

Yes, that should do it:

import psutil

p = psutil.Process()
print p.memory_full_info().pss
print sum([x.pss for x in p.memory_maps()])

That should be logically equal.
On my machine that differs by 64K but I suppose it's because memory_maps() build and return an additional dict (which stays in memory).
Full implementation is here:

psutil/psutil/_pslinux.py

Line 1244 in a5beb29

@wrap_exceptions

....but I'd say you're better off using memory_maps().

[rohithvsm]

@giampaolo Thanks for your response. I was also trying to derive uss. On similar lines, is the below code snippet also true?

import psutil

proc = psutil.Process()
proc.memory_full_info().uss == sum(pmmap.private_clean for pmmap in proc.memory_maps()) + sum(pmmap.private_dirty for pmmap in proc.memory_maps())