Use queue-based RwLock on more platforms

[joboet]

This switches over Windows 7, SGX and Xous to the queue-based RwLock implementation added in #110211, thereby fixing #121949 for Windows 7 and partially resolving #114581 on SGX. TEEOS can't currently be switched because it doesn't have a good thread parking implementation.

CC @roblabla @raoulstrackx @xobs Could you help me test this, please?
r? @ChrisDenton the Windows stuff should be familiar to you

[ChrisDenton]

The Windows stuff is fine. I can't easily test on Windows 7 atm (so help is welcome) but testing the Win7 binary on Windows 10 worked..

I'll wait for the SGX and Xous experts to weigh in on their target. And if someone can test on Windows 7 proper that would be great. I don't foresee any issues though so I wouldn't block on it.

[xobs]

Looks good to me:

...
test sync::reentrant_lock::tests::smoke ... ok
test sync::reentrant_lock::tests::trylock_works ... ok
test sync::rwlock::tests::frob ... ok
test sync::rwlock::tests::panic_while_mapping_read_unlocked_no_poison ... ok
test sync::rwlock::tests::panic_while_mapping_write_unlocked_poison ... ok
test sync::rwlock::tests::smoke ... ok
test sync::rwlock::tests::test_get_mut ... ok
test sync::rwlock::tests::test_get_mut_poison ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:307:9:
test panic in inner thread to poison RwLock
ok
test sync::rwlock::tests::test_into_inner ... ok
test sync::rwlock::tests::test_into_inner_drop ... ok
test sync::rwlock::tests::test_into_inner_poison ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:283:9:
test panic in inner thread to poison RwLock
ok
test sync::rwlock::tests::test_mapped_read_guard_covariance ... ok
test sync::rwlock::tests::test_mapping_mapped_guard ... ok
test sync::rwlock::tests::test_read_guard_covariance ... ok
test sync::rwlock::tests::test_rw_arc ... ok
test sync::rwlock::tests::test_rw_arc_access_in_unwind ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:209:9:
explicit panic
ok
test sync::rwlock::tests::test_rw_arc_no_poison_mapped_r_r ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:122:9:
explicit panic
ok
test sync::rwlock::tests::test_rw_arc_no_poison_mapped_r_w ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:149:9:
explicit panic
ok
test sync::rwlock::tests::test_rw_arc_no_poison_rr ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:108:9:
explicit panic
ok
test sync::rwlock::tests::test_rw_arc_no_poison_rw ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:135:9:
explicit panic
ok
test sync::rwlock::tests::test_rw_arc_poison_mapped_w_r ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:68:9:
explicit panic
ok
test sync::rwlock::tests::test_rw_arc_poison_mapped_w_w ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:95:9:
explicit panic
ok
test sync::rwlock::tests::test_rw_arc_poison_wr ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:55:9:
explicit panic
ok
test sync::rwlock::tests::test_rw_arc_poison_ww ... PANIC in PID 6:
thread '<unnamed>' panicked at library/std/src/sync/rwlock/tests.rs:81:9:
explicit panic
ok
test sync::rwlock::tests::test_rwlock_try_write ... ok
test sync::rwlock::tests::test_rwlock_unsized ... ok
test sys::os_str::bytes::tests::display ... ok
test sys::os_str::bytes::tests::slice_debug_output ... ok
test sys::pal::common::tests::bench_heap_path_alloc ... ok
...

[roblabla]

Oh, awesome! I had resigned myself to thinking the win7 target would be stuck with a broken implementation of RwLocks, so this is great news!

I ran the full core/alloc/std test suite on Windows 7 32-bit and 64-bit, both are passing. LGTM!

[raoulstrackx]

The SGX part also looks good, but I had to apply this patch to keep the compiler happy:

diff --git a/library/std/src/sys/pal/sgx/waitqueue/mod.rs b/library/std/src/sys/pal/sgx/waitqueue/mod.rs
index ea49bb28034..9c106e1d960 100644
--- a/library/std/src/sys/pal/sgx/waitqueue/mod.rs
+++ b/library/std/src/sys/pal/sgx/waitqueue/mod.rs
@@ -64,7 +64,7 @@ pub fn lock_var_mut(&mut self) -> &mut T {
 #[derive(Copy, Clone)]
 pub enum NotifiedTcs {
     Single(Tcs),
-    All { count: NonZero<usize> },
+    All { _count: NonZero<usize> },
 }
 
 /// An RAII guard that will notify a set of target threads as well as unlock
@@ -232,7 +232,7 @@ pub fn notify_all<T>(
             }
 
             if let Some(count) = NonZero::new(count) {
-                Ok(WaitGuard { mutex_guard: Some(guard), notified_tcs: NotifiedTcs::All { count } })
+                Ok(WaitGuard { mutex_guard: Some(guard), notified_tcs: NotifiedTcs::All { _count: count } })
             } else {
                 Err(guard)
             }

[ChrisDenton]

Thank you everyone for testing this!

@bors r+

[bors]

📌 Commit 10b6ca1 has been approved by ChrisDenton

It is now in the queue for this repository.