regrtest.py: spurious leaks with -R option

[skrah]

BPO	15632
Nosy	@skrah, @meadori
Files	warm_latin1_cache.diff test_pseudo_leak.py issue15632.diff

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2012-08-17.21:14:37.912>
created_at = <Date 2012-08-12.19:31:45.980>
labels = ['type-bug', 'tests']
title = 'regrtest.py: spurious leaks with -R option'
updated_at = <Date 2012-08-17.21:14:37.910>
user = 'https://github.com/skrah'

bugs.python.org fields:

activity = <Date 2012-08-17.21:14:37.910>
actor = 'python-dev'
assignee = 'none'
closed = True
closed_date = <Date 2012-08-17.21:14:37.912>
closer = 'python-dev'
components = ['Tests']
creation = <Date 2012-08-12.19:31:45.980>
creator = 'skrah'
dependencies = []
files = ['26777', '26778', '26798']
hgrepos = []
issue_num = 15632
keywords = ['patch']
message_count = 5.0
messages = ['168063', '168065', '168066', '168182', '168479']
nosy_count = 3.0
nosy_names = ['skrah', 'meador.inge', 'python-dev']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue15632'
versions = ['Python 3.2', 'Python 3.3']

[skrah]

With the -R option, test_buffer.py needs an additional warm-up for
the unicode_latin1 cache. Patch attached.

[meadori]

I know the title says spurious, but is there a way to reproduce this at all? Can I run a certain suite N times or something? How did you find the problem?

[skrah]

I'm attaching a test case. You're right, in test_buffer it's
hard to reproduce but I've encountered the "leak" several times
in the past months.

Today I realized that the "leak" always occurred with format code
'c'. There's this passage in test_buffer:

    x = randrange(*fmtdict[mode][char])
    if char == 'c':
        x = bytes(chr(x), 'latin1')

After some head scratching I looked into regrtest.py and found
the warm_char_cache() function. The whole thing makes sense: In
each repetition of the refleak mode new characters can be added
to the unicode_latin1 cache.

So, you should be able to reproduce the issue:

$ ./python -m test -uall -R 2:2 test_pseudo_leak
[1/1] test_pseudo_leak
beginning 4 repetitions
1234
....
test_pseudo_leak leaked [8, 8] references, sum=16
1 test failed:
    test_pseudo_leak
[151225 refs]

[skrah]

Should we go all the way and also explicitly fill the small int cache?

Integers 0-255 should be initialized already, but I'm not sure
about [-5..-1] and 256. I haven't been able to create a test case
for integers though.

[python-dev]

New changeset dc18d73e67a5 by Stefan Krah in branch 'default':
Closes bpo-15632: regrtest.py: fix spurious refleaks due to various caches
http://hg.python.org/cpython/rev/dc18d73e67a5