85981 - nsStdURL::SchemeIs() method doesn't check for null mScheme

Status: VERIFIED FIXED

(Core :: Networking, defect, P3)

Description

[Charles Manske]

If you make an nsIURI with a URL that doesn't exist, it seems mScheme is null.
It seems that in this line of nsStdURL::SchemeIs():
    if ( *i_Scheme == *mScheme || *i_Scheme == (*mScheme - ('a' - 'A')) ) {

we should always check for null mScheme.
This is hard to demonstrate, but I discovered it testing the fix to bug 78351.
It's kind of messy, because you have to apply fixes to bug 78351, bug 76177,
and bug 82435 before you can get the crash! If you do that, here's how to show
it:
Bring up a composer window, click on Image button to bring up dialog.
Type "xxxxxxxxx.gif"  and BAM!
I don't know anything about this code, but looking at nsStdURL.cpp, it
certainly looks like there's plenty of opportunity to have a null mScheme,
so checking it certainly seems the right thing to do!

Comment 1

[Charles Manske]

Since this is a crasher and blocker, I'm going to take the liberty of marking
it "0.9.2"

Comment 2

[Charles Manske]

Attachment: Simple fix: check for null mScheme

Comment 3

[Charles Manske]

Valeski is not around. Reassigning to Radha

Comment 4

[kinmoz]

sr=[e-mail] if you get an r= from a module owner/peer.

Comment 5

[Stuart Parmenter]

this is necko.  it should be assigned to gagan or dougt or darin.

Comment 6

[Radha on family leave (not reading bugmail)]

I'm filling in for Jud, when he is not here, but I'm not the module owner for
this code.  Though the changes look OK to me, It think it is not PC for me to
give a review. I also suggest you try dougt/darin/gagan for a review.

Comment 7

[Gagan]

r=gagan, get an sr/a and land this in soon...

Comment 8

[kinmoz]

Already sr'd just get a= from [e-mail].

Comment 9

[Asa Dotzler [:asa]]

a= [e-mail] for checkin to the trunk.
(on behalf of drivers)

Comment 10

[Charles Manske]

Might as well assign to me since I fixed it.

Comment 11

[Charles Manske]

Fix checked in.

Comment 12

[Doug Turner (:dougt)]

WAIT!!!! This is NOT something that we want to do!!!  The scheme should never be 
null.  


I am workin on changing the callers so that this is impossible.  Now, I am going 
to have to back this out....  reopening and reassigning...

Comment 13

[Doug Turner (:dougt)]

dougt, back this out.

Comment 14

[Charles Manske]

Ok, as long a mScheme is assured of being non-null. Maybe adding an assertion
would be a good idea?

Comment 15

[Doug Turner (:dougt)]

fixed checked in.

Comment 16

[Jan Carpenter]

Verified on:
Win2K:  2001-07-27-00-0.9.2
        2001-07-27-12-trunk
MacOS:  2001-07-26-21-0.9.2
Linux:  2001-07-26-23-0.9.2