661583 - xhr.response needs to hold on to the cache token when the responseType is blob.

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

From nsICachingChannel:

123     /**
124      * Get the "file" where the cached data can be found.  This is valid for
125      * as long as a reference to the cache token is held.  This may return
126      * an error if cacheAsFile is false.
127      */
128     readonly attribute nsIFile cacheFile;

If we're not holding the cache token alive, this file might be removed from the disk.  The easiest way forward here is probably to make nsDOMFile able to hold onto an opaque nsISupports pointer, and then give the constructor the cacheToken from the caching channel.

Comment 1

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Also, we have to ensure that slices hold onto the cache token as well.

Comment 2

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Masatoshi: Do you think you'd be able to get to this? I think we should fix this for FF6 or disable the feature.

Comment 3

[Masatoshi Kimura [:emk]]

Attachment: patch

Please apply this on top of the patch of bug 661582.

Comment 4

[Masatoshi Kimura [:emk]]

Attachment: patch v2

Comment 5

[Masatoshi Kimura [:emk]]

Attachment: patch v2.1

Comment 6

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Comment on attachment 537313 [details] [diff] [review]
patch v2.1

Looks good, but add the new aCacheToken argument to the first constructor instead, as per the review of bug 661582.

Comment 7

[Masatoshi Kimura [:emk]]

Attachment: patch v3

Rebased to the latest bug 661583 patch.
Carrying forward r+.

Comment 8

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Attachment: Reordered patch

This is Masatoshi's patch reordered to not depend on the patch in Bug 661582.

Comment 9

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Comment on attachment 543237 [details] [diff] [review]
Reordered patch

Drivers, we will want to take this patch for Firefox 6 to ensure that a File object's actual file on disk is not removed from the cache depending on the whims of GC behavior.  I believe the risk of this patch is very low.

Comment 10

[LegNeato]

Comment on attachment 543237 [details] [diff] [review]
Reordered patch

Approved for releases/mozilla-aurora. Please land asap before 2011-07-05 @ 9:00 am PDT.

Comment 11

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

http://hg.mozilla.org/mozilla-central/rev/c3d3292fbd1e
http://hg.mozilla.org/releases/mozilla-aurora/rev/c660b06ec3b3

Comment 12

[Matt Brubeck (:mbrubeck)]

Backed out from mozilla-central during investigation of Android browser-chrome test failures:
http://hg.mozilla.org/mozilla-central/rev/00bb08972e46

Comment 13

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

http://hg.mozilla.org/mozilla-central/rev/8c8d3b22ea79

Comment 14

[Virgil Dicu [:virgil] [QA]]

Mozilla/5.0 (X11; Linux x86_64; rv:6.0) Gecko/20100101 Firefox/6.0

Could you please provide a test case in order to have this issue verified?