819927 - mozilla::Base64{Encode,Decode}() should properly handle the empty string

Status: RESOLVED FIXED

(Core :: XPCOM, defect)

Description

[Trevor Saunders (:tbsaunde)]

PR_Base64{Encode,Decode}() call strlen() on the input if the length is 0, so if you call them with (null, 0, whatever) they crash.

Comment 1

[Trevor Saunders (:tbsaunde)]

Attachment: patch

Comment 2

[(no longer active)]

Comment on attachment 690365 [details] [diff] [review]
patch

Review of attachment 690365 [details] [diff] [review]:
-----------------------------------------------------------------

The nsJSEnvironment.cpp changes don't seem related to this patch.  Also, you should test this in xpcom/tests/TestBase64.cpp.

Comment 3

[Trevor Saunders (:tbsaunde)]

> The nsJSEnvironment.cpp changes don't seem related to this patch.  Also, you

oops, sorry about that

> should test this in xpcom/tests/TestBase64.cpp.

that is not exactly easy :( mozilla/Base64.h pulls in headers that are only supposed to be for libxul internal code.

Note that test also needed to be patched to work on !windows too.

Comment 4

[(no longer active)]

Ouch.  Well, nm then I guess.

Comment 5

[Trevor Saunders (:tbsaunde)]

Attachment: bug 819927 - don't ask PR_Base64{Encode,Decode}() to handle empty strings

Comment 6

[Trevor Saunders (:tbsaunde)]

remote:   https://hg.mozilla.org/integration/mozilla-inbound/rev/668141ff2577

Comment 7

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/668141ff2577

Comment 8

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

I'm confused. Neil tells me BeginReading should never return null, so why was this change necessary?