Luca Greco [:rpl] [:luca] [:lgreco] Assignee
	Description • 6 years ago

When an extension has the "unlimitedStorage" permission, we should internally turn the storage persist mode on for the storageLocal principal that is used internally by ExtensionStorageIDB to open the extension IndexedDB storage on behalf of the extension.

	David Durst [:ddurst]
	Comment 2 • 6 years ago

I don't mean to pressure, but our window for getting this in Nightly is closing. Will you have a chance to look at this patch this week?

	Jan Varga [:janv]
	Comment 3 • 6 years ago

Sorry, I had to work on critical sec bugs last week.
Yeah, I'll look at the patch ASAP.

	Pulsebot
	Comment 4 • 6 years ago

Pushed by luca.greco@alcacoop.it:
https://hg.mozilla.org/integration/autoland/rev/2317749c5abf
Set persist mode on ExtensionStorageIDB for extensions with the unlimitedStorage permission. r=mixedpuppy,janv

	Bogdan Tara[:bogdan_tara | bogdant]
	Comment 5 • 6 years ago
bugherder

https://hg.mozilla.org/mozilla-central/rev/2317749c5abf

	Vlad Jiman
	Comment 6 • 6 years ago

Will manual testing be required on this bug? please provide some steps on how to correctly validate if so, or set the 'qe-validate-' flag, thank you

	Luca Greco [:rpl] [:luca] [:lgreco] Assignee
	Comment 7 • 6 years ago

(In reply to Vlad Jiman from comment #6)
> Will manual testing be required on this bug? please provide some steps on
> how to correctly validate if so, or set the 'qe-validate-' flag, thank you

Hi Vlad,
sorry for the lag time on this, I kept trying to check if it was doable to force IndexedDB to raise the errors related to storing too much data by using some of the QuotaManager's test preferences (in particular using dom.quotaManager.temporaryStorage.fixedLimit, with and without the dom.quotaManager.testing pref set to true at the same time). 

My test extension has been able to store 1.4Gb of storage without raising any error while storing more data into it (without the "persist mode" turned on by this patch), despite the limit I set in the `fixedLimit` perf, and so it doesn't seem that this quotaManager testing pref is making any actual difference.

The other part (testing that the persist mode is set on the storage principal as expected) is tested in automation,
and so I'm going to mark this as qe-verify-.