Backports release 1.5.3 #38122

KristofferC · 2020-10-21T06:09:57Z

Backported PRs:

Need manual backport:

Fix === and objectid of object with potentially undef inline immutable field #37557 - Fix === and objectid of object with potentially undef inline immutable field
fix fields description of Module type #37645 - fix fields description of Module type
fix #37880, overflow in shift amount range check in codegen #37891 - fix Codegen bug with Base.lshr_int intrinsic #37880, overflow in shift amount range check in codegen

Contains multiple commits, manual intervention needed:

code coverage fixes #37243 - code coverage fixes

(cherry picked from commit 9b2c6b2)

(cherry picked from commit 05a0d54)

add an error check that the max size is not exceeded (cherry picked from commit 5f55b97)

This reverts commit cf0c3e0. (cherry picked from commit b18647e)

(cherry picked from commit 8c3f6e5)

yuyichao · 2020-10-21T14:27:44Z

#37557 introduced a bug that is fixed by #37896

Sacha0 · 2020-10-21T21:17:33Z

@KristofferC, I hope you don't mind: I took the liberty of pushing the patches we carry for

#37594 - dump: ensure Array eltype layout is initialized early
#37101 - make Profile more thread/signal-safe
#36821 - Fix function call to rec_backtrace_ctx when using USE_SYSTEM_LIBUNWIND=1

on Julia 1.5.2 and checked off / added those to your list of manual backports at top :). (Those patches have been exercised by our team for a few weeks each now.)

Sacha0 · 2020-10-22T02:24:13Z

Hm, would #38049 be a backport candidate once it merges, what with fixing precompilation failures?

KristofferC · 2020-10-22T06:06:25Z

Seems like a nice bug fix, so makes sense to me. Usually it is good to have PRs be on master for a while to shake out anything egregiously wrong but that should be pretty harmless and easy to test.

KristofferC · 2020-10-22T06:32:33Z

@Sacha0, seems like Windows is not happy with one of the backports?

martinholters · 2020-10-22T07:14:37Z

Include #37769?

dkarrasch · 2020-10-22T11:45:22Z

Also include #38002?

Sacha0 · 2020-10-22T20:54:34Z

@Sacha0, seems like Windows is not happy with one of the backports?

Hm, I don't think we have anyone running these patches on Windows, so perhaps that's not entirely surprising 😄. I suspect the patch for #37101; having a look at the logs now.

Sacha0 · 2020-10-22T21:28:00Z

Yes, it looks like #37101 introduces the failure in package_win{32,64}. Amusingly the failure is the Windows-equivalent of #36821:

In file included from /cygdrive/c/buildbot/worker/package_win64/build/src/signal-handling.c:117:
/cygdrive/c/buildbot/worker/package_win64/build/src/signals-win.c:354:36: error: too few arguments to function 'rec_backtrace_ctx'
  354 |                     bt_size_cur += rec_backtrace_ctx((jl_bt_element_t*)bt_data_prof + bt_size_cur,
      |                                    ^~~~~~~~~~~~~~~~~

i.e. the rec_backtrace_ctx calls introduced by #37101 need the trailing argument that #36821 adds to a call in signals-mach.c.

Sacha0 · 2020-10-22T21:41:13Z

I've pushed a commit that should sort out the error identified above. I'm not confident that fix will sort Windows out altogether, but let's ask the buildbot 🤞 :).

Sacha0 · 2020-10-22T21:49:24Z

It looks like package_win32 failed in a new and exciting way also related to #37101 (particularly its removal of jl_trygetUnwindInfo). Having a look now :).

Sacha0 · 2020-10-23T01:09:39Z

That last push seems to have fixed the package_ failures, but surfaced another new and exciting failure in tester_win32. I suspect that fixing the latter requires adding #37002 to the patch. I will give that a whirl tomorrow.

Add an `unwrapva` to ensure the returned type is not a `Vararg`. (cherry picked from commit 22b5d93)

This should make this previously non-atomic mutex threadsafe. Fixes #35117

Sacha0 · 2020-10-23T22:00:55Z

I think I've managed to backport #37002 with #37101 on top, by which I mean the result appears to function locally. Let's see what the buildbot thinks? :)

Sacha0 · 2020-10-23T22:02:17Z

(I should add that #37002 should obviate the patch from #36821, so I dropped the latter.)

Sacha0 · 2020-10-24T00:31:56Z

The package_linuxaarch64 failure was a timeout, and the logs look otherwise clean. So we seem to be in good shape. I'll stop messing with your branch now @KristofferC 😄.

KristofferC · 2020-10-25T17:43:01Z

@nanosoldier runtests(ALL, vs = ":release-1.5")

nanosoldier · 2020-10-26T00:33:31Z

Your package evaluation job has completed - possible new issues were detected. A full report can be found here. cc @maleadt

NHDaly · 2020-10-26T18:47:20Z

Hm, would #38049 be a backport candidate once it merges, what with fixing precompilation failures?

Seems like a nice bug fix, so makes sense to me. Usually it is good to have PRs be on master for a while to shake out anything egregiously wrong but that should be pretty harmless and easy to test.

It sounds like we're planning to backport this? I've opened a PR against this branch, here:
#38183

KristofferC · 2020-10-26T18:50:11Z

@nanosoldier runtests(["Autologistic", "BlobTracking", "Cambrian", "Catalyst", "ChainRules", "ClimateERA", "CodeTransformation", "DiffEqOperators", "FlashWeave", "ImageDistances", "Oscar", "RiskAdjustedLinearizations", "Sched", "StochasticDelayDiffEq", "Tar", "Thorn", "TimeToLive", "Turmeric"], vs = ":release-1.5")

KristofferC · 2020-10-26T18:50:22Z

@nanosoldier runbenchmarks(ALL, vs = ":release-1.5")

nanosoldier · 2020-10-26T19:21:11Z

Your package evaluation job has completed - possible new issues were detected. A full report can be found here. cc @maleadt

nanosoldier · 2020-10-27T02:52:32Z

Your benchmark job has completed - possible performance regressions were detected. A full report can be found here. cc @christopher-dG

KristofferC · 2020-10-28T17:55:59Z

https://github.com/perrutquist/CodeTransformation.jl segfaults but that one heavily uses nternals so not sure we can do much about that.

Sacha0 · 2020-11-07T00:40:40Z

From a brief review of this thread, it seems like this pull request, and ergo 1.5.3, is ready to go? Is merge etc waiting on anything at this point? :)

KristofferC · 2020-11-07T15:24:39Z

I think this should be good to go from my pov as well.

dkarrasch · 2020-11-08T10:04:58Z

What about #38122 (comment)? It fixes something that got broken in v1.5.0.

racinmat · 2020-11-10T10:36:47Z

Hi, is there a chance #37243 will make it to some 1.5.* release? It seems it didn't make it to 1.5.2 nor 1.5.3.

KristofferC and others added 4 commits October 21, 2020 08:02

fix showing methods with unicode gensymed variable names (#36396)

1bb2631

(cherry picked from commit 9b2c6b2)

Add INNO_ARGS makefile var for signtool (#37236)

b32702d

(cherry picked from commit 05a0d54)

increase RELOC_TAG_OFFSET to handle larger system images (#37613)

7c7e8b6

add an error check that the max size is not exceeded (cherry picked from commit 5f55b97)

Revert "Improve typesubtract for tuples (#35600)" (#37562)

489ae0a

This reverts commit cf0c3e0. (cherry picked from commit b18647e)

KristofferC added the release Release management and versioning. label Oct 21, 2020

KristofferC and others added 2 commits October 21, 2020 08:39

backport part of a82bc01

a84f7cf

Bump libuv to latest commit (#37407)

628e8cb

(cherry picked from commit 8c3f6e5)

KristofferC force-pushed the backports-release-1.5 branch 2 times, most recently from 1ba7261 to 628e8cb Compare October 21, 2020 06:46

This was referenced Oct 21, 2020

Fix bug when inlining pending nodes #37634

Merged

dump: ensure Array eltype layout is initialized early #37594

Merged

fix fields description of Module type #37645

Merged

fix #37880, overflow in shift amount range check in codegen #37891

Merged

Backport of patch for #37594 to Julia 1.5.2 / for Julia 1.5.3.

cf8b9ec

Sacha0 force-pushed the backports-release-1.5 branch from fd03cb5 to a408c1d Compare October 22, 2020 23:59

Fix getfield_tfunc for PartialStruct with Vararg (#37769)

9bc7391

Add an `unwrapva` to ensure the returned type is not a `Vararg`. (cherry picked from commit 22b5d93)

benlorenz mentioned this pull request Oct 23, 2020

Improve libjulia, rebuild libjulia 1.3.1 with the improvements JuliaPackaging/Yggdrasil#1951

Merged

backport of: win32: fix in_stackwalk mutex (#37002)

45e9c22

This should make this previously non-atomic mutex threadsafe. Fixes #35117

Sacha0 force-pushed the backports-release-1.5 branch from a408c1d to bb83a29 Compare October 23, 2020 21:58

This comment has been minimized.

Sign in to view

NHDaly mentioned this pull request Oct 26, 2020

Make static_show() print valid identifiers (#38049) #38183

Merged

Make static_show() print valid identifiers (#38049) (#38183)

58cd4b8

fredrikekre and others added 2 commits October 27, 2020 19:22

Update Pkg for 1.5.3.

37c1096

Backport prepare_for_deletion() (#38159)

0305b7c

KristofferC merged commit 006853a into release-1.5 Nov 7, 2020

KristofferC deleted the backports-release-1.5 branch November 7, 2020 15:28

musm mentioned this pull request Dec 22, 2020

win: Set correct folder permissions after folder creation #38942

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Backports release 1.5.3 #38122

Backports release 1.5.3 #38122

KristofferC commented Oct 21, 2020 •

edited

Loading

yuyichao commented Oct 21, 2020

Sacha0 commented Oct 21, 2020

Sacha0 commented Oct 22, 2020

KristofferC commented Oct 22, 2020

KristofferC commented Oct 22, 2020

martinholters commented Oct 22, 2020

dkarrasch commented Oct 22, 2020

Sacha0 commented Oct 22, 2020 •

edited

Loading

Sacha0 commented Oct 22, 2020 •

edited

Loading

Sacha0 commented Oct 22, 2020 •

edited

Loading

Sacha0 commented Oct 22, 2020 •

edited

Loading

Sacha0 commented Oct 23, 2020

Sacha0 commented Oct 23, 2020

Sacha0 commented Oct 23, 2020

Sacha0 commented Oct 24, 2020

KristofferC commented Oct 25, 2020

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

nanosoldier commented Oct 26, 2020

This comment has been minimized.

This comment has been minimized.

NHDaly commented Oct 26, 2020

KristofferC commented Oct 26, 2020

KristofferC commented Oct 26, 2020

nanosoldier commented Oct 26, 2020

nanosoldier commented Oct 27, 2020

KristofferC commented Oct 28, 2020

Sacha0 commented Nov 7, 2020

KristofferC commented Nov 7, 2020

dkarrasch commented Nov 8, 2020

racinmat commented Nov 10, 2020

Backports release 1.5.3 #38122

Backports release 1.5.3 #38122

Conversation

KristofferC commented Oct 21, 2020 • edited Loading

yuyichao commented Oct 21, 2020

Sacha0 commented Oct 21, 2020

Sacha0 commented Oct 22, 2020

KristofferC commented Oct 22, 2020

KristofferC commented Oct 22, 2020

martinholters commented Oct 22, 2020

dkarrasch commented Oct 22, 2020

Sacha0 commented Oct 22, 2020 • edited Loading

Sacha0 commented Oct 22, 2020 • edited Loading

Sacha0 commented Oct 22, 2020 • edited Loading

Sacha0 commented Oct 22, 2020 • edited Loading

Sacha0 commented Oct 23, 2020

Sacha0 commented Oct 23, 2020

Sacha0 commented Oct 23, 2020

Sacha0 commented Oct 24, 2020

KristofferC commented Oct 25, 2020

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

nanosoldier commented Oct 26, 2020

This comment has been minimized.

This comment has been minimized.

NHDaly commented Oct 26, 2020

KristofferC commented Oct 26, 2020

KristofferC commented Oct 26, 2020

nanosoldier commented Oct 26, 2020

nanosoldier commented Oct 27, 2020

KristofferC commented Oct 28, 2020

Sacha0 commented Nov 7, 2020

KristofferC commented Nov 7, 2020

dkarrasch commented Nov 8, 2020

racinmat commented Nov 10, 2020

KristofferC commented Oct 21, 2020 •

edited

Loading

Sacha0 commented Oct 22, 2020 •

edited

Loading

Sacha0 commented Oct 22, 2020 •

edited

Loading

Sacha0 commented Oct 22, 2020 •

edited

Loading

Sacha0 commented Oct 22, 2020 •

edited

Loading