diff options
author | 2018-05-10 17:22:13 -0400 | |
---|---|---|
committer | 2018-05-10 22:43:56 +0000 | |
commit | 5c560c90f7345cb0bc57571eb3215ad25c1a1b10 (patch) | |
tree | 924ff1a19f4e478c7cf1190e55b92ac9b2ff1e22 /third_party/skcms/src/PortableMath.h | |
parent | e5f39624b69e95b08837311c58553685131bafe1 (diff) |
Fix unchecked allocation in kImageFilter::Common::unflatten()
The bad news is we don't have a good way to validate the count against
the available data size.
The good news is we don't have to: most filters only use two inputs,
which will fit in the stack-reserved storage; for those who don't,
growing the input vector during deserialization should not be
problematic.
The CL drops the prea-allocation, and instead appends inputs iteratively.
Also switches the storage to SkSTArray for append/push_back() support.
Bug: oss-fuzz:8219
Change-Id: Ic6bf6e435e6e18b9e026aeb393612210c9e6e57a
Reviewed-on: https://skia-review.googlesource.com/127306
Commit-Queue: Florin Malita <fmalita@chromium.org>
Reviewed-by: Mike Reed <reed@google.com>
Diffstat (limited to 'third_party/skcms/src/PortableMath.h')
0 files changed, 0 insertions, 0 deletions