Following a discussion in gcc maillist for the libsanitizer merge I am now
trying to remove the SANITIZER_AARCH64_VMA flags for aarch64 port.
This patch makes ASAN for aarch64 use the same shadow offset for all
currently supported VMAs (39 and 42 bits). The shadow offset is the
same of 39-bit (36). To make it work on 42-bit llvm instrumentation will
require a change [1] on how to generate the memory address.
No regression tests found in 39 and 42-bit VMA. I have not tested on
48-bit VMA due lack of available platforms (but it is on my radar).
I'm pretty sure this won't work on 48 bit VMA.
Could you evaluate the cost of lifting the upper limit to 48? Last time I looked, this should only affect .bss size (so, no real RAM cost), and not really that much.