RESTRICT AUTOMERGE: Cherry-pick "begin cleanup of malloc porting layer"
Test: Not feasible
1. Merge some of the allocators into sk_malloc_flags by redefining a flag to mean zero-init
2. Add more private helpers to simplify our call-sites (and handle some overflow mul checks)
3. The 2-param helpers rely on the saturating SkSafeMath::Mul to pass max_size_t as the request,
which should always fail.
Commit-Queue: Mike Reed <email@example.com>
Reviewed-by: Robert Phillips <firstname.lastname@example.org>
Reviewed-by: Stephan Altmueller <email@example.com>
Simply removed the old definitions of SK_MALLOC_TEMP and SK_MALLOC_THROW.
Copied SK_SUPPORT_LEGACY_MALLOC_PORTING_LAYER into the old defines.
Drop a change where we were not using malloc yet.
These files weren't yet using malloc (and SkBmpBaseCodec hadn't been
These were still using New rather than Make (return raw pointer). Leave
them unchanged, as sk_malloc_flags is still valid.
Leave this unchanged; sk_malloc_flags is still valid
In addition, pull in SkSafeMath.h, which was originally introduced in
https://skia-review.googlesource.com/c/skia/+/33721. This is required
for the new sk_malloc calls.
Also pull in SkSafeMath::Add and SkSafeMath::Mul, introduced in
Also add SK_MaxSizeT, which the above depends on, introduced in
Also, modify NewFromStream to use sk_malloc_canfail, matching pi and
avoiding a build break
(cherry picked from commit a12cc3e14ea6734c7efe76aa6a19239909830b28)
15 files changed