1525136 - moz-phab exits with hg error about rebasing onto a hidden revision

Status: RESOLVED FIXED

(Conduit :: moz-phab, defect, P2)

Description

[Beth Rennie [:beth] (she/her)]

Attachment: submit.log

My latest run of moz-phab submit resulted in an error during the rebase phase:

.. [snip] ..

Waiting for JSON parameters on stdin...
rebasing 456834:f838931847d6 "Bug 1444449 - Implement CompositorD3D11::GetWindowRenderTarget API r?mstange,bas"
rebasing 456835:4704664b3959 "Bug 1444449 - Implement AsyncReadbackBuffer APIs for CompositorD3D11 r?mstange,bas"
rebasing 456836:2072bd4baea0 "Bug 1444449 - Implement CompositorD3D11::BlitRenderTarget r?bas" (dev/screenshots/d3d11)
abort: hidden revision '4704664b3959ff390c50172f57d4204e4f4718cd' was rewritten as: c6e1f719549c!
(use --hidden to access hidden revisions)
CommandError: command 'hg.exe' failed to complete successfully

(The entire log is attached as submit.log)

However, it appears that the rebase was actually successful. hg heads isn't reporting any orphan commits and my hg log -G looks correct:

@  changeset:   456840:15f7ab140d09
|  bookmark:    dev/screenshots/d3d11
|  tag:         tip
|  user:        Barret Rennie <[e-mail]>
|  date:        Wed Jan 16 13:24:42 2019 -0500
|  summary:     Bug 1444449 - Implement CompositorD3D11::BlitRenderTarget r?bas
|
o  changeset:   456839:c6e1f719549c
|  user:        Barret Rennie <[e-mail]>
|  date:        Tue Jan 15 17:30:59 2019 -0500
|  summary:     Bug 1444449 - Implement AsyncReadbackBuffer APIs for CompositorD3D11 r?mstange,bas
|
o  changeset:   456838:e5d0a24aa895
|  user:        Barret Rennie <[e-mail]>
|  date:        Tue Jan 15 17:19:48 2019 -0500
|  summary:     Bug 1444449 - Implement CompositorD3D11::GetWindowRenderTarget API r?mstange,bas
|
o  changeset:   456837:9db2aed0d962
|  parent:      456737:d58901c5036f
|  user:        Barret Rennie <[e-mail]>
|  date:        Fri Feb 01 15:38:31 2019 -0500
|  summary:     Bug 1444449 - Do not return raw pointers to window targets from compositors r?bas

Comment 1

[Steven MacLeod [:smacleod]]

Is there any chance that you had rewritten history (such as with hg amend), but had not fully evolved your branch before trying to submit? (meaning it was attempting to submit the middle obsolete commit, not the new version)?

Comment 2

[Beth Rennie [:beth] (she/her)]

It looks like it might be the case, but I can't tell from hg olog and my submitted log if that is for sure.

Comment 3

[Steven MacLeod [:smacleod]]

Going to mark as incomplete for now. If you hit this case again can you please re-open and check if it's the case I mentioned?

Comment 4

[Beth Rennie [:beth] (she/her)]

I just ran into this again.

I was at a commit that was not a tip and ran moz-phab submit:

rebasing 466734:dfe034cfcc7a "Bug 1510569 - Add missing includes and using namespace declarations r?ehsan"
rebasing 466735:ea63178ecd24 "Bug 1510569 - Reimplement RemoteWebProgressRequest as an XPCOM component in C++ r?ehsan"
rebasing 466736:9d4c0ae2205f "Bug 1510569 - Implement nsIWebProgressListener for RemoteWebProgressManager r?ehsan"
rebasing 466737:7019b5c68410 "Bug 1510569 - Reconstruct nsIWebProgress and nsIRequest for onContentBlockingEvent in TabParent r?ehsan"
rebasing 466738:77426c1ff55d "Bug 1510569 - Port onStatusChange notifications inside WebProgressChild.jsm to C++ r?baku"
rebasing 466739:fa05303ac799 "Bug 1510569 - Implement nsIWebProgressListener2 for TabChild r?baku"
rebasing 466740:7dedf6830615 "Bug 1510569 - Port onProgressChange notifications inside WebProgressChild.jsm to C++ r?baku"
rebasing 466741:046c88b158a4 "Bug 1510569 - Port onRefreshAttempted from WebProgressChild.jsm to C++ r?baku"
rebasing 466742:47174f8cc3bb "Bug 1510569 - Keep track of whether we are navigating to a new URI on nsIDocShell r?smaug,mconley"
rebasing 466743:a10531ea68c8 "Bug 1510569 - Port onStateChange notifications inside WebProgressChild.jsm to C++ r?baku,kmag" (dev/rewrite-web-progress-child)
rebasing 466744:d92d00b7b165 "Bug 1510569 - Port onLocationChange notifications inside WebProgressChild.jsm to C++ r?baku"
rebasing 466745:df63b9612ee2 "Bug 1510569 - Move Content:BrowserChildReady from WebProgressChild.jsm into the browser child frame script r?mconley"
rebasing 466746:e732bffa365e "Bug 1510569 - Move Content:LoadURIResult message from WebProgressChild.jsm to WebNavigationChild.jsm r?mconley"
rebasing 466747:8e4ad25cfc23 "Bug 1510569 - Prevent missing symbol errors during builds when adding new IPDL headers r?ehsan"
rebasing 466748:d15d3ff2f226 "Bug 1510569 - Implement serializers for nsITransportSecurityInfo, nsIX509Cert, and nsIX509CertList r?froydnj,mccr8,keeler"
rebasing 466749:70ff5ee22456 "Bug 1510569 - Port OnSecurityChange from WebProgressChild.jsm to C++; remove WebProgressChild r?baku"
rebasing 466750:9365e96fbdfa "Bug 1510569 - Remove nsIRemoteWebProgressRequest XPCOM component r?ehsan"
rebasing 466752:8939dd2f38ae "Bug 1510569 - Use Maybe<T> instead of IPDL unions for TabChild-TabParent WebProgress messages r?nika"
rebasing 466753:458773ff4afd "Bug 1510569 - Ensure TabChild's IPC actor is not dead before sending OnContentBlockingEvent message to TabParent r?froydnj"
rebasing 466754:db041a047017 "Bug 1510569 - Include innerDOMWindowID in the WebProgressData sent from TabChild to TabParent r?baku"
rebasing 466755:1d7f01b09d4a "Bug 1510569 - QueryInterface received nsIRequest in promiseBrowserLoaded to an nsIChannel r?kmag"
rebasing 466756:bb760892f710 "Bug 1510569 - Add missing includes to WindowGlobalParent.cpp r?ehsan"
rebasing 466757:96ac27776dc0 "Bug 1510569 - Reimplement RemoteWebProgress as an XPCOM component in C++ r?ehsan"
rebasing 466758:57c491eac6bf "Bug 1510569 - Add missing includes and using namespace declarations r?ehsan"
rebasing 466759:a9978c2f2db6 "Bug 1510569 - Reimplement RemoteWebProgressRequest as an XPCOM component in C++ r?ehsan"
rebasing 466760:b8ed5427a94f "Bug 1510569 - Implement nsIWebProgressListener for RemoteWebProgressManager r?ehsan"
rebasing 466761:2b27aeeefb8e "Bug 1510569 - Reconstruct nsIWebProgress and nsIRequest for onContentBlockingEvent in TabParent r?ehsan"
rebasing 466762:198547e6f7c5 "Bug 1510569 - Port onStatusChange notifications inside WebProgressChild.jsm to C++ r?baku"
rebasing 466763:d6e75497b00a "Bug 1510569 - Implement nsIWebProgressListener2 for TabChild r?baku"
rebasing 466764:191eb9585107 "Bug 1510569 - Port onProgressChange notifications inside WebProgressChild.jsm to C++ r?baku"
rebasing 466765:3a9740c635f5 "Bug 1510569 - Port onRefreshAttempted from WebProgressChild.jsm to C++ r?baku"
rebasing 466766:607f7f7d27ea "Bug 1510569 - Keep track of whether we are navigating to a new URI on nsIDocShell r?smaug,mconley"
rebasing 466767:617af983feeb "Bug 1510569 - Port onStateChange notifications inside WebProgressChild.jsm to C++ r?baku,kmag" (dev/rewrite-web-progress-child)
rebasing 466768:2aaa964e5931 "Bug 1510569 - Port onLocationChange notifications inside WebProgressChild.jsm to C++ r?baku"
rebasing 466769:c200a73a32e1 "Bug 1510569 - Move Content:BrowserChildReady from WebProgressChild.jsm into the browser child frame script r?mconley"
rebasing 466770:133f5b02519b "Bug 1510569 - Move Content:LoadURIResult message from WebProgressChild.jsm to WebNavigationChild.jsm r?mconley"
rebasing 466771:8a587ef4fdc9 "Bug 1510569 - Prevent missing symbol errors during builds when adding new IPDL headers r?ehsan"
rebasing 466772:30386ebdf154 "Bug 1510569 - Implement serializers for nsITransportSecurityInfo, nsIX509Cert, and nsIX509CertList r?froydnj,mccr8,keeler"
rebasing 466773:31a59bea9d01 "Bug 1510569 - Port OnSecurityChange from WebProgressChild.jsm to C++; remove WebProgressChild r?baku"
rebasing 466774:e76b83f1ad12 "Bug 1510569 - Remove nsIRemoteWebProgressRequest XPCOM component r?ehsan"
abort: hidden revision 'dfe034cfcc7ae9539df02dd93c23c4af065feb86' was rewritten as: 172d29881a3e!
(use --hidden to access hidden revisions)
CalledProcessError: Command '['hg.exe', '--config', 'extensions.rebase=', 'rebase', '--source', 'dfe034cfcc7ae9539df02dd93c23c4af065feb86', '--dest', '21c9d5b1aa043c4285ee39148ef12a50ea498b71']' returned non-zero exit status 255

It appears that is trying to rebase a commit twice. The commit it was trying to rebase and failed at (dfe034) is the first commit that was rebased.

Comment 5

[Beth Rennie [:beth] (she/her)]

Also I can confirm that before I ran moz-phab that I had no unstable changesets.

Comment 6

[Axel Hecht [:Pike]]

I just hit this, too.

Here's what I did: I had an existing patch with a D, https://phabricator.services.mozilla.com/D20465.

I rebased that on a new landing, https://phabricator.services.mozilla.com/D29000, and added another new on top, D29001.

The one it complained about was the top-most. It also failed to set D29000 as parent of D20465, I had to do that by hand.

Comment 9

[:glob ✱]

Attachment: Bug 1525136 - moz-phab exits with hg error about rebasing onto a hidden revision r=zalun!

Fix our handling of non-stack children that were orphaned during an evolve
amend. Instead of keeping an up-front list of children to rebase after
submission, rebase non-stack children immediately after amending a commit.

Comment 10

[:glob ✱]

https://github.com/mozilla-conduit/review/commit/b6f812bd0cf10a6125145940aa3569c1ea8d4fe9