apis: update breaking driver existence check changes

[Samir-Rashid]

Bring libtock-rs into consistency with the changes in this PR.

Only apis/gpio/src/lib.rs needed to be changed. The rest is a grepped driver_check -> exists.

[ppannuto]

This is part of a trio of PRs across the kernel, libtock-c, and here. While it's a bit of churn, I think there's value in the consistency of the term "exists" across the tock ecosystem with these changes.

---

Blocked on pending kernel PR.

[ppannuto]

It might be a good idea to keep the latter half of this comment.

For another day—it's an interesting point more generally. What is the point of an "exists" check? I might think it's to make sure the driver will be available when you want to call it. Following that, there's a case that "exists" syscalls should do grant allocations, but that's a more fundamental design change.

[Samir-Rashid]

Added ppannuto's suggested change and resquashed.

diff

diff --git a/apis/gpio/src/lib.rs b/apis/gpio/src/lib.rs
index 8328358..bc2d500 100644
--- a/apis/gpio/src/lib.rs
+++ b/apis/gpio/src/lib.rs
@@ -57,7 +57,8 @@ pub struct Gpio<S: Syscalls>(S);
 
 impl<S: Syscalls> Gpio<S> {
     /// Returns Ok() if the driver was present.This does not necessarily mean
-    /// that the driver is working.
+    /// that the driver is working, as it may still fail to allocate grant
+    /// memory.
     pub fn exists() -> Result<(), ErrorCode> {
         S::command(DRIVER_NUM, EXISTS, 0, 0).to_result()
     }

[ppannuto]

It looks like some of those CI errors are real — unit tests that need to be updated to match the new expectations.

[Samir-Rashid]

I rebased to master. The tests are passing, so I think this PR should be merged.