Guard against null image data in CJBig2_GRRDProc.

Guard against null image data in CJBig2_GRRDProc.

Credit to karl at skomski.com for the initial version of the CL.

BUG=527174
R=tsepez@chromium.org

Committed: https://pdfium.googlesource.com/pdfium/+/89d8b4681ce98894a1ee1a6cf4bae77e00d28797

[Lei Zhang, 2015-09-04 23:49:38]

thestig@chromium.org changed reviewers:
+ tsepez@chromium.org

[Lei Zhang, 2015-09-04 23:49:39]

Slightly different way of achieving the same thing as Karl's suggestion. In most
other places, the callee checks for NULL m_pData, so I'm just following that
pattern.

[Tom Sepez, 2015-09-08 16:00:39]

Add test case as an embeddertest or such?

[Lei Zhang, 2015-09-09 03:37:26]

On 2015/09/08 16:00:39, Tom Sepez wrote:
> Add test case as an embeddertest or such?

Should I put a new embeddertest in core/src/fxcodec/jbig2 all by itself? How
about a pixel test?

In either case, I have a hand-written test case ready to go based on Karl's test
PDF. Just wanted to make sure he does not mind me using it.

[skomski, 2015-09-09 09:29:13]

On 2015/09/09 03:37:26, Lei Zhang wrote:
> On 2015/09/08 16:00:39, Tom Sepez wrote:
> > Add test case as an embeddertest or such?
> 
> Should I put a new embeddertest in core/src/fxcodec/jbig2 all by itself? How
> about a pixel test?
> 
> In either case, I have a hand-written test case ready to go based on Karl's
test
> PDF. Just wanted to make sure he does not mind me using it.

Sure you can use my test case.

Regarding the CR I don't really understand why not also to bail out early
like in my original patch because there will never be a valid image after that,
only new bugs.

[Tom Sepez, 2015-09-09 17:52:43]

On 2015/09/09 03:37:26, Lei Zhang wrote:
> On 2015/09/08 16:00:39, Tom Sepez wrote:
> > Add test case as an embeddertest or such?
> 
> Should I put a new embeddertest in core/src/fxcodec/jbig2 all by itself? How
> about a pixel test?
> 
> In either case, I have a hand-written test case ready to go based on Karl's
test
> PDF. Just wanted to make sure he does not mind me using it.
Pixel test seems reasonable.

[Lei Zhang, 2015-09-09 23:14:01]

On 2015/09/09 09:29:13, skomski wrote:
> On 2015/09/09 03:37:26, Lei Zhang wrote:
> > On 2015/09/08 16:00:39, Tom Sepez wrote:
> > > Add test case as an embeddertest or such?
> > 
> > Should I put a new embeddertest in core/src/fxcodec/jbig2 all by itself? How
> > about a pixel test?
> > 
> > In either case, I have a hand-written test case ready to go based on Karl's
> test
> > PDF. Just wanted to make sure he does not mind me using it.
> 
> Sure you can use my test case.
> 
> Regarding the CR I don't really understand why not also to bail out early
> like in my original patch because there will never be a valid image after
that,
> only new bugs.

I suppose we can take the belt and suspender approach. I do want to fix
CJBig2_GRRDProc::decode_Template[01]_opt() as well, since they may be reachable
via other paths as well.

[Lei Zhang, 2015-09-09 23:41:07]

Onwards to patch set 4.

https://codereview.chromium.org/1305033006/diff/60001/core/src/fxcodec/jbig2/...
File core/src/fxcodec/jbig2/JBig2_Context.cpp (right):

https://codereview.chromium.org/1305033006/diff/60001/core/src/fxcodec/jbig2/...
core/src/fxcodec/jbig2/JBig2_Context.cpp:472: m_ProcessiveStatus =
FXCODEC_STATUS_ERROR;
This will make CJBig2_Context::parseSegmentData() break out rather than
continuing.

I didn't set |m_bBufSpecified| as Karl did in his patch. That may fix the
|m_pPage| leak, but we can deal with that separately. Also
s/processive/progressive/ ...

[Tom Sepez, 2015-09-10 22:31:58]

LGTM, but reitveld wouldn't show the test files.  I'll assume they're OK.

[Lei Zhang, 2015-09-10 23:13:35]

Committed patchset #5 (id:80001) manually as
89d8b4681ce98894a1ee1a6cf4bae77e00d28797 (presubmit successful).