Bug 250322 - lang/python3{6,7,8,9}: Backport close_range patches
Summary: lang/python3{6,7,8,9}: Backport close_range patches
Status: Closed FIXED
Alias: None
Product: Ports & Packages
Classification: Unclassified
Component: Individual Port(s) (show other bugs)
Version: Latest
Hardware: Any Any
: --- Affects Only Me
Assignee: freebsd-python (Nobody)
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2020-10-13 18:45 UTC by Kyle Evans
Modified: 2020-10-14 18:11 UTC (History)
2 users (show)

See Also:
lwhsu: maintainer-feedback+
kevans: merge-quarterly-


Attachments
svn(1) diff against the ports tree (39.61 KB, patch)
2020-10-13 18:45 UTC, Kyle Evans
no flags Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Kyle Evans freebsd_committer freebsd_triage 2020-10-13 18:45:05 UTC
Created attachment 218721 [details]
svn(1) diff against the ports tree

Worked out over BPO-40422 and BPO-40423, this is the culmination of months of work to coordinate with Linux and get close_range(2) added to FreeBSD, then the usage accepted into CPython. It has landed for Python 3.10 and here I've backported it locally to all the supported Python 3 versions we have.

Note that this does include and supercede our previous closefrom(2) patches. There was a lot of intersection between the work done, so this patch against the ports tree does remove those patches from each of the ports in favor of this patch. All the patches involved have been accepted and merged upstream.

This patch will bring a performance boost in some more situations on 12.2 and 13.0, as close_range exists there.

There is one additional patch sitting in an upstream PR that shuffles the _Py_closerange implementation into a different file -- this is not important for the backport, and the absence of that patch here will not realistically cause any issues.

Q/A:
 * portlint (pre-existing issues)
 * testport (11.4/amd64, 12.1/i386, -CURRENT/amd64) (note that only -CURRENT/amd64 currently has close_range, the other two did not)
Comment 1 Li-Wen Hsu freebsd_committer freebsd_triage 2020-10-14 10:14:52 UTC
LGTM, did poudriere tests with {11.3,12.1}x{amd64,i386} and simple run tests.

Just being curious about why from the BPO there seems no plan (yet?) to merge this to 3.6-3.7 branches?
Comment 2 Kyle Evans freebsd_committer freebsd_triage 2020-10-14 13:31:25 UTC
(In reply to Li-Wen Hsu from comment #1)

I think the upstream backport policy is that this won't get backported at all (to any version, so 3.10 will be the first release with it) since it's an enhancement.
Comment 3 commit-hook freebsd_committer freebsd_triage 2020-10-14 18:10:19 UTC
A commit references this bug:

Author: kevans
Date: Wed Oct 14 18:09:43 UTC 2020
New revision: 552351
URL: https://svnweb.freebsd.org/changeset/ports/552351

Log:
  lang/python3{6,7,8,9}: Backport close_range patches

  Worked out over BPO-40422 and BPO-40423, this is the culmination of months
  of work to coordinate with Linux and get close_range(2) added to FreeBSD,
  then the usage accepted into CPython. It has landed for Python 3.10 and here
  I've backported it locally to all the supported Python 3 versions we have.

  Note that this does include and supercede our previous closefrom(2) patches.
  There was a lot of intersection between the work done, so this patch against
  the ports tree does remove those patches from each of the ports in favor of
  this patch. All the patches involved have been accepted and merged upstream.

  This patch will bring a performance boost in some more situations on 12.2
  and 13.0, as close_range exists there.

  There is one additional patch sitting in an upstream PR that shuffles the
  _Py_closerange implementation into a different file -- this is not important
  for the backport, and the absence of that patch here will not realistically
  cause any issues.

  PR:		250322
  Approved by:	lwhsu (python)

Changes:
  head/lang/python36/Makefile
  head/lang/python36/files/patch-Modules___posixsubprocess.c
  head/lang/python36/files/patch-Modules_posixmodule.c
  head/lang/python36/files/patch-issue40422_issue40423
  head/lang/python37/Makefile
  head/lang/python37/files/patch-Modules___posixsubprocess.c
  head/lang/python37/files/patch-Modules_posixmodule.c
  head/lang/python37/files/patch-issue40422_issue40423
  head/lang/python38/Makefile
  head/lang/python38/files/patch-Modules___posixsubprocess.c
  head/lang/python38/files/patch-Modules_posixmodule.c
  head/lang/python38/files/patch-issue40422_issue40423
  head/lang/python39/Makefile
  head/lang/python39/files/patch-issue40422_issue40423
Comment 4 Kyle Evans freebsd_committer freebsd_triage 2020-10-14 18:11:27 UTC
Thanks!