[OTS] Allow empty Private DICT inside CFF data

[OTS] Allow empty Private DICT inside CFF data

BUG=chromium:112414
TEST=ran test_{un,}malicious_fonts.sh

[bashi, 2012-02-23 10:12:40]

Hi,

I think it is ok to allow private_offset == table_length. (following checks
reject the case of private_length > 0 and private_offset == table_length)

Could you please take a look?

[Yusuke Sato, 2012-02-23 13:32:54]

lgtm

On 2012/02/23 10:12:40, bashik wrote:
> Hi,
> 
> I think it is ok to allow private_offset == table_length. (following checks
> reject the case of private_length > 0 and private_offset == table_length)
> 
> Could you please take a look?

[agl, 2012-02-23 15:30:55]

lgtm

http://codereview.chromium.org/9447013/diff/1/src/cff.cc
File src/cff.cc (right):

http://codereview.chromium.org/9447013/diff/1/src/cff.cc#newcode770
src/cff.cc:770: // does not overflow since table_length < 1GB
Note: this comment is nonsensical. It doesn't matter what value table_length
has, in order to avoid overflow, |private_length| and |private_offset| have to
be correctly bounded.

[bashi, 2012-02-23 23:36:34]

Thank you for review!

http://codereview.chromium.org/9447013/diff/1/src/cff.cc
File src/cff.cc (right):

http://codereview.chromium.org/9447013/diff/1/src/cff.cc#newcode770
src/cff.cc:770: // does not overflow since table_length < 1GB
On 2012/02/23 15:30:55, agl wrote:
> Note: this comment is nonsensical. It doesn't matter what value table_length
> has, in order to avoid overflow, |private_length| and |private_offset| have to
> be correctly bounded.

Done.