Experiments show that on Android the current values result in too much
of the memory consumption for all quarantined chunks.
Details
Diff Detail
- Build Status
Buildable 2333 Build 2333: arc lint + arc unit
Event Timeline
Btw ASAN_LOW_MEMORY was set for all 32-bit targets in r171052 by kcc.
I don't understand why.
IMHO, it should be a property of environment (i.e. OS), not architecture. If necessary, we can always test for 32-bit platform directly. And we do, see all the FIRST_32_SECOND_64 macros!
I'd support removing "|| (SANITIZER_WORDSIZE == 32)" from this definition.
lib/asan/asan_allocator.cc | ||
---|---|---|
235 | Sure, but doesn't it make it a bit less cryptic using 64 and Kb (1 << 10) than 1 << 16? |
lib/asan/asan_allocator.cc | ||
---|---|---|
235 | it's a matter of taste, I guess |
The Darwin bots are failing the ThreadedQuarantineTest test: http://lab.llvm.org:8080/green/job/clang-stage1-configure-RA_check/25044/console
Can you please take a look?
1 << 16