Restore sndrcv behaviour from before 53afe84 #4538
Conversation
|
Aims to fix #4531 |
|
Thanks a lot for figuring this out, this is great work ! |
|
I was trying, if this would immediately fix the unit tests. But my next guess is, that we may need to remove the daemon flag of the send thread.
--
Mit freundlichen Grüßen
Dr. Nils Weiß
dissecto GmbH
Franz-Mayer-Str. 1
93053 Regensburg
________________________________
From: gpotter2 ***@***.***>
Sent: Sunday, September 22, 2024 2:48:59 PM
To: secdev/scapy ***@***.***>
Cc: Dr. Nils Weiß ***@***.***>; Author ***@***.***>
Subject: Re: [secdev/scapy] Restore sndrcv behaviour from before 53afe84 (PR #4538)
@gpotter2 commented on this pull request.
________________________________
In scapy/sendrecv.py<#4538 (comment)>:
@@ -357,6 +357,7 @@ def sndrcv(*args, **kwargs):
WARNING: This is an internal function. Using sr/srp/sr1/srp is
more appropriate in many cases.
"""
+ kwargs['threaded'] = kwargs.pop('threaded', True) or True
Why this commit?
—
Reply to this email directly, view it on GitHub<#4538 (review)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AAMZGFZ4CDPX6JVXG3VMTWDZX24DXAVCNFSM6AAAAABOUIRDESVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZDGMRQHA2DQMRSG4>.
You are receiving this because you authored the thread.Message ID: ***@***.***>
|
gpotter2
force-pushed
the
issue_4531
branch
from
450d941
to
56e82b2
Compare
| @@ -323,9 +330,7 @@ def _process_packet(self, r): | |||
| self.noans += 1 | |||
| sentpkt._answered = 1 | |||
| break | |||
| if self._send_done and self.noans >= self.notans and not self.multi: | |||
There was a problem hiding this comment.
To explain this code can be reached before the sending thread exists (because of the new time.sleep) in which case the sniffer is never closed
gpotter2
force-pushed
the
issue_4531
branch
3 times, most recently
from
f4e06ce
to
9e5d6e6
Compare
|
Seems to really be failing on the |
| # In threaded mode, timeout. | ||
| if self.threaded and self.timeout is not None and not self.breakout: | ||
| t = time.monotonic() + self.timeout | ||
| while time.monotonic() < t: |
There was a problem hiding this comment.
And while we're at it, this code is disgusting and has terrible perfs. Should have used a lock or something, so I just did that.
Try to get it debug as soon as possible |
|
Could I get #4533 merged before I start debugging? |
|
Sure thing ! |
polybassa
force-pushed
the
issue_4531
branch
from
6a5faf5
to
65100e7
Compare
polybassa
force-pushed
the
issue_4531
branch
from
a05c650
to
1b12525
Compare
This PR rolls back the internal behaviour of sndrcv from before 53afe84
Two major changes happend in 53afe84
interis only forwarded totime.sleepif it is not0. However, callingtime.sleep(0)gives the python threading system the possibility to do a context switch.srwas always executed inthreadedmode.