Message ID | 20190204131228.25949-1-richard.henderson@linaro.org |
---|---|
Headers | show |
Series | target/arm: Implement ARMv8.5-BTI | expand |
On Mon, 4 Feb 2019 at 13:12, Richard Henderson <richard.henderson@linaro.org> wrote: > > One fixup to the set of patches applied to target-arm.next. > One adjustment to what is now patch 3 wrt EXCP_SEMIHOST. > > > r~ > > > Richard Henderson (4): > fixup! target/arm: Cache the GP bit for a page in MemTxAttrs > target/arm: Add x-guarded-pages cpu property for user-only > linux-user/aarch64: Reset btype for syscalls and signals > tests/tcg/aarch64: Add bti smoke test I'm pretty sure we talked about this on IRC at some point, but for the record: I believe the relevent ELF ABI spec docs are now public that have the changes that describe how ELF executables are marked up to indicate which parts should be loaded into guarded pages. The rest of the userspace ABI is likely to just be a new mmap() flag. So I think we're better off implementing that (possibly still guarded by an x- property if it's not finalized yet.) thanks -- PMM
On 3/26/19 6:57 AM, Peter Maydell wrote: > I'm pretty sure we talked about this on IRC at some point, > but for the record: I believe the relevent ELF ABI spec docs > are now public that have the changes that describe how ELF > executables are marked up to indicate which parts should be > loaded into guarded pages. The rest of the userspace ABI > is likely to just be a new mmap() flag. So I think we're > better off implementing that (possibly still guarded by > an x- property if it's not finalized yet.) We still don't know what the mmap flag will be. You're right that the elf abi is sufficient to build statically linked test cases, though, which is all we wanted here. r~