Don't use PROT_EXEC with mmap on macos
#1110
Conversation
On macos I was running `./miniruby --mmtk -e 'puts "Hello world!"'` and seeing a panic: ``` thread '<unnamed>' panicked at /src/util/raw_memory_freelist.rs:202:9: Can't get more space with mmap() note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace fatal runtime error: failed to initiate panic, error 5 Abort trap: 6 ``` On macos, you can't combine `PROT_EXEC` with `PROT_WRITE` and `PROT_READ` in `mmap`, it will return a permission denied error. It's not clear from the `Can't get more space with mmap()` that's what was happening, but I could see it when inspecting the `ret` from `mmap_fixed` in `dzmmap_noreplace`. I found a thread on apple confirming that `PROT_EXEC` doesn't make sense for macos in this context. https://forums.developer.apple.com/forums/thread/740017 Note: there are still failing tests for `mmap` on macos. The main change with this PR is that it no longer panics when runnning a simple hello world with Ruby on macos. Previously the build wasn't finishing and would throw the following error: ``` process didn't exit successfully: `/mmtk-core/target/debug/deps/mmtk-ef0f14e36a03fbba` (signal: 11, SIGSEGV: invalid memory reference) ``` Now the tests do finish, with 25 failures.
eileencodes
force-pushed
the
remove-PROT_EXEC-on-macos
branch
from
0b2b442
to
c34c568
Compare
| const MMAP_PROT: libc::c_int = PROT_READ | PROT_WRITE | PROT_EXEC; | ||
| #[cfg(target_os = "macos")] | ||
| // PROT_EXEC cannot be used with PROT_READ on Apple Silicon | ||
| const MMAP_PROT: libc::c_int = PROT_READ | PROT_WRITE; |
There was a problem hiding this comment.
Thanks for the PR. I think for now getting something to work is better than nothing. But we do need to consider how MMTk handles code allocation for JIT.
MMTk might need to do this on macOS https://developer.apple.com/documentation/apple-silicon/porting-just-in-time-compilers-to-apple-silicon#Enable-the-JIT-entitlements-for-the-Hardened-Runtime
There was a problem hiding this comment.
Thank you for the PR! I wonder if we should do things differently for Intel Macs vs Apple Silicon? Though maybe it makes more sense to have the same mechanism regardless of the underlying architecture. @qinsoon Do you have any thoughts?
There was a problem hiding this comment.
I wonder if we should do things differently for Intel Macs vs Apple Silicon?
Maybe. But our support for intel macOS is minimal as well so this is not a big concern for now.
On macos I was running
./miniruby --mmtk -e 'puts "Hello world!"'and seeing a panic:On macos, you can't combine
PROT_EXECwithPROT_WRITEandPROT_READinmmap, it will return a permission denied error. It's not clear from theCan't get more space with mmap()that's what was happening, but I could see it when inspecting theretfrommmap_fixedindzmmap_noreplace.I found a thread on apple confirming that
PROT_EXECdoesn't make sense for macos in this context. https://forums.developer.apple.com/forums/thread/740017Note: there are still failing tests for
mmapon macos. The main change with this PR is that it no longer panics when runnning a simple hello world with Ruby on macos. Previously the build wasn't finishing and would throw the following error:Now the tests do finish, with 25 failures.