[Linux] memory_full_info() incorrectly raise ZombieProcess

[giampaolo]

Summary

• OS: Linux
• Type: core

Description

...this happens, for example, with PID 2 (kthreadd):

>>> import psutil
>>> psutil.Process(2).memory_info()
pmem(rss=0, vms=0, shared=0, text=0, lib=0, data=0, dirty=0)
>>> psutil.Process(2).memory_full_info()
Traceback (most recent call last):
  File "/home/giampaolo/svn/psutil/psutil/_pslinux.py", line 1901, in _parse_smaps_rollup
    for line in f:
ProcessLookupError: [Errno 3] No such process

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/giampaolo/svn/psutil/psutil/__init__.py", line 1091, in memory_full_info
    return self._proc.memory_full_info()
  File "/home/giampaolo/svn/psutil/psutil/_pslinux.py", line 1948, in memory_full_info
    uss, pss, swap = self._parse_smaps_rollup()
  File "/home/giampaolo/svn/psutil/psutil/_pslinux.py", line 1653, in wrapper
    return fun(self, *args, **kwargs)
  File "/home/giampaolo/svn/psutil/psutil/_pslinux.py", line 1913, in _parse_smaps_rollup
    raise ZombieProcess(self.pid, self._name, self._ppid)
psutil.ZombieProcess: PID still exists but it's a zombie (pid=2)

The error originates from /proc/pid/smaps_rollup, which acts weirdly: it raises ESRCH / ENOENT for many PIDs, even if they're alive (also as root). In that case we'll have to use /proc/pid/smaps as fallback, which is slower but has a +50% success rate compared to /proc/pid/smaps_rollup.