Fix potentially massive memory leak in CPDF_DIBSource::LoadJpxBitmap().

Fix potentially massive memory leak in CPDF_DIBSource::LoadJpxBitmap().

Leaks can happen in several places. For this particular bug, it happens
when there is a colorspace component count mismatch.

BUG=497191
R=tsepez@chromium.org

Committed: https://pdfium.googlesource.com/pdfium/+/2a824f1c0ed786aed0dd15a0ea60dc90999e2b2c

[Lei Zhang, 2015-06-06 05:00:30]

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

[Tom Sepez, 2015-06-06 19:05:04]

lgtm

https://codereview.chromium.org/1153633009/diff/20001/core/src/fpdfapi/fpdf_r...
File core/src/fpdfapi/fpdf_render/fpdf_render_loadimage.cpp (right):

https://codereview.chromium.org/1153633009/diff/20001/core/src/fpdfapi/fpdf_r...
core/src/fpdfapi/fpdf_render/fpdf_render_loadimage.cpp:84: void
set_output_offsets(FX_LPBYTE output_offsets) {
nit: prefer void* to LPBYTE in new code.

https://codereview.chromium.org/1153633009/diff/20001/core/src/fpdfapi/fpdf_r...
core/src/fpdfapi/fpdf_render/fpdf_render_loadimage.cpp:753: delete
m_pCachedBitmap;
nit: can m_pCachedBitmap be a unique_ptr?

https://codereview.chromium.org/1153633009/diff/20001/core/src/fpdfapi/fpdf_r...
core/src/fpdfapi/fpdf_render/fpdf_render_loadimage.cpp:755: return;
Nit: presumably this is the return that leaks, right? maybe mention this in the
CL description ... "leak when JpxModule::Decode() fails" or some such.

[Lei Zhang, 2015-06-08 19:35:21]

Please take another look.

https://codereview.chromium.org/1153633009/diff/20001/core/src/fpdfapi/fpdf_r...
File core/src/fpdfapi/fpdf_render/fpdf_render_loadimage.cpp (right):

https://codereview.chromium.org/1153633009/diff/20001/core/src/fpdfapi/fpdf_r...
core/src/fpdfapi/fpdf_render/fpdf_render_loadimage.cpp:84: void
set_output_offsets(FX_LPBYTE output_offsets) {
On 2015/06/06 19:05:04, Tom Sepez wrote:
> nit: prefer void* to LPBYTE in new code.

LPBYTE is unsigned char*.

https://codereview.chromium.org/1153633009/diff/20001/core/src/fpdfapi/fpdf_r...
core/src/fpdfapi/fpdf_render/fpdf_render_loadimage.cpp:753: delete
m_pCachedBitmap;
On 2015/06/06 19:05:04, Tom Sepez wrote:
> nit: can m_pCachedBitmap be a unique_ptr?

Done.

https://codereview.chromium.org/1153633009/diff/20001/core/src/fpdfapi/fpdf_r...
core/src/fpdfapi/fpdf_render/fpdf_render_loadimage.cpp:755: return;
On 2015/06/06 19:05:04, Tom Sepez wrote:
> Nit: presumably this is the return that leaks, right? maybe mention this in
the
> CL description ... "leak when JpxModule::Decode() fails" or some such.

The failures are on line 700 above. "if (codestream_nComps !=
(FX_DWORD)m_pColorSpace->CountComponents())"

[Tom Sepez, 2015-06-08 19:42:12]

lgtm

[Lei Zhang, 2015-06-08 20:24:54]

Committed patchset #3 (id:40001) manually as
2a824f1c0ed786aed0dd15a0ea60dc90999e2b2c (presubmit successful).