User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:56.0) Gecko/20100101 Firefox/56.0 Build ID: 20170926190823 Steps to reproduce: Tried to load backup created by Firefox 55.* from built-in Bookmarks manager. Actual results: "Unable to process the backup file" displayed, nothing else happens. Expected results: Bookmark file should be normally imported. When I create backup from Firefox 56, it's imported without issues.

Hi, would you allow me to have access to the broken backup for debugging purposes? A private mail to my bugzilla mail address would be fine.

you can also send it to my mozilla mail if you trust that more, mak AT mozilla.com

(In reply to Marco Bonardo [::mak] from comment #2) > you can also send it to my mozilla mail if you trust that more, mak AT > mozilla.com I have sent email with a sample bookmarks backup file attached.

Investigating.

Interesting, in the browser console I see this error: Error: Bookmarks.jsm: insertTree: Invalid value for property 'lastModified': 0 We should not be writing 0 here. It's possible there is a corruption in the database where either the lastModified or dateAdded field are zero or NULL, something that is unexpected.

(In reply to Marco Bonardo [::mak] from comment #5) > Interesting, in the browser console I see this error: Error: Bookmarks.jsm: > insertTree: Invalid value for property 'lastModified': 0 > > We should not be writing 0 here. It's possible there is a corruption in the > database where either the lastModified or dateAdded field are zero or NULL, > something that is unexpected. The original problem reported in this bug was happening on two separate Firefox installations (both working on Ubuntu, 14.04 and 16.04). If I understand correctly, even if database contained unexpected value, the result shouldn't be an exception, causing the process to abort.

Yes, we'll add a fixup procedure so that import/restore will try to do its best to insert "invalid" entries or skip that, so that most entries can still be added.

Comment on attachment 8922250 [details] Bug 1404631 - Bookmarks.insertTree should have an option to fixup input and skip broken entries. https://reviewboard.mozilla.org/r/193282/#review198724 Looking good. I found one small potential intermittent issue with the test, but I think that's probably an easy fix. ::: toolkit/components/places/tests/bookmarks/test_insertTree_fixupOrSkipInvalidEntries.js:12 (Diff revision 1) > + await Assert.throws(() => insertTree({guid, children: []}), > + /Should have a non-zero number of children to insert./); > + await Assert.throws(() => insertTree({guid: "invalid", children: [{}]}), > + /The parent guid is not valid/); > + > + let now = new Date(); I just tried this on my Mac and got: FAIL Found a valid dateAdded - false == true /Users/mark/dev/gecko/objdir-ff/_tests/xpcshell/toolkit/components/places/tests/bookmarks/test_insertTree_fixupOrSkipInvalidEntries.js:onItemAdded:25 resource://gre/modules/Bookmarks.jsm:notify:1206 resource://gre/modules/Bookmarks.jsm:insertTree/<:510 It looks like it takes a couple of seconds to run through, so `now` is 1509046068872000 but `dateAdded` recorded by the observer is 1509046068874000. I think you probably want to move `now` into the `onItemAdded` but also calculate it separately for the dates in the tree.

(In reply to Mark Banner (:standard8) from comment #9) > I think you probably want to move `now` into the `onItemAdded` but also > calculate it separately for the dates in the tree. I'll just replace the is_time_ordered with a >=, because I can't predict the time used by Bookmarks.jsm to fixup. And experience teaches me it's better to not trust timestamps guessing in tests :)

Pushed by mak77@bonardo.net: https://hg.mozilla.org/integration/autoland/rev/f4e6ab850229 Bookmarks.insertTree should have an option to fixup input and skip broken entries. r=standard8