Fix up rowbytes vs. width desynchronization, and fix failure to initialize en...

Fix up rowbytes vs. width desynchronization, and fix failure to initialize entire bitmap memory. To fix up the rowbytes value properly, we simply don't send it via IPC any more, and recalulate it from width and depth in the trusted code. It's a cheap calculation.
Also one bonus fix: don't use an unintialized IconInfo if deserialization fails.

BUG=31307
TEST=Manual; ran with breakpoints on the failure paths.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=35371

[jschuh, 2009-12-30 03:44:08]

LGTM, but with two nits that don't change functionality.

http://codereview.chromium.org/517023/diff/1/2
File chrome/common/common_param_traits.cc (right):

http://codereview.chromium.org/517023/diff/1/2#newcode204
chrome/common/common_param_traits.cc:204: for (size_t i = 0; i < icon_count &&
result; ++i) {
Do we need to check result as a loop condition, since we use it as an immediate
return condition inside the loop?

http://codereview.chromium.org/517023/diff/1/2#newcode214
chrome/common/common_param_traits.cc:214: return result;
Can't we just return true, since it looks like result must be true if we got
here?

[Chris Evans, 2009-12-30 04:34:20]

Thanks.

http://codereview.chromium.org/517023/diff/1/2
File chrome/common/common_param_traits.cc (right):

http://codereview.chromium.org/517023/diff/1/2#newcode204
chrome/common/common_param_traits.cc:204: for (size_t i = 0; i < icon_count &&
result; ++i) {
On 2009/12/30 03:44:09, jschuh wrote:
> Do we need to check result as a loop condition, since we use it as an
immediate
> return condition inside the loop?

Done.

http://codereview.chromium.org/517023/diff/1/2#newcode214
chrome/common/common_param_traits.cc:214: return result;
On 2009/12/30 03:44:09, jschuh wrote:
> Can't we just return true, since it looks like result must be true if we got
> here?

Done.