1198669 - Unable to get the original http header inside through nsIMultipartChannel

Status: RESOLVED FIXED

(Core :: Networking, defect)

Description

[Henry Chang [:hchang]]

In new security model, we need to check the integrity of each resource in the package. While we can get the original body for each subresource, there's no way to get the original http header. The closest way is to serialize the object nsHttpResponseHead which we obtain from nsIMultipartChannel. It could be different from the original one for a couple of reasons.

We need to either add API to nsIMultipartChannel to expose original http header.

Comment 1

[Dragana Damjanovic [:dragana]]

see bug 669259

Comment 2

[Henry Chang [:hchang]]

(In reply to Dragana Damjanovic [:dragana] from comment #1)
> see bug 669259

If you are referring to get them from cache, it would be a little difficult for packaged app since we will copy some headers from the base channel to the part channel and save to the cache. Sigh...

Comment 3

[Dragana Damjanovic [:dragana]]

(In reply to Henry Chang [:henry] from comment #2)
> (In reply to Dragana Damjanovic [:dragana] from comment #1)
> > see bug 669259
> 
> If you are referring to get them from cache, it would be a little difficult
> for packaged app since we will copy some headers from the base channel to
> the part channel and save to the cache. Sigh...

Sorry, I think, I understand now what you mean with "It could be different from the original one for a couple of reasons."  Because you have not really define the part "different to what" I have understood it wrong.

Comment 4

[Henry Chang [:hchang]]

Attachment: Bug1198669.diff

Comment 5

[Henry Chang [:hchang]]

Hi Valentin,

Sorry for bothering again :p

I just realised we need this patch to get rid of the header we copy from the base channel for verification. (since the integrity hash covers the header). 

I will provide the test case tomorrow. If you are busy or not able to review this, I may ask Jason to do the review. Just let you know first :) Thanks!

Comment 6

[Henry Chang [:hchang]]

By the way, I have tested the patch and it works pretty well.

Comment 7

[Valentin Gosu [:valentin]]

Comment on attachment 8657829 [details] [diff] [review]
Bug1198669.diff

This looks pretty nice. Add a test and it's good to go.

Comment 8

[Henry Chang [:hchang]]

Attachment: Bug1198669.patch (with test case)

Comment 9

[Henry Chang [:hchang]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=e3c344be9a05

Comment 10

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/ee49a5d7058c

Comment 11

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/ee49a5d7058c