Refactor - Simplify CoW in MemoryMapping #37
Conversation
6da3754 to
cceb6ba
Compare
cceb6ba to
5feb4c3
Compare
src/memory_region.rs
Outdated
| pub type AccessViolationHandler = Box<dyn Fn(u32) -> Result<u64, ()>>; | ||
| /// Fail always | ||
| #[allow(clippy::result_unit_err)] | ||
| pub fn default_access_violation_handler(_: u32) -> Result<u64, ()> { | ||
| Err(()) |
There was a problem hiding this comment.
The usage of the function is to try to write to the memory and copy if it is readonly, right? I believe the old name MemoryCowCallback (or something related) is clearer on the intentions of this function.
There was a problem hiding this comment.
I can split the renaming into a separate PR. But in the future it is not only for CoW (readonly to writable) but also to map in unmapped address space (for growing the realloc padding).
There was a problem hiding this comment.
Yes, another PR is good.
but also to map in unmapped address space
MemoryCowCallback isn't the best option considering that, but I still prefer it to describe what the function does, instead of the error it might return when it can't do what it is supposed to.
There was a problem hiding this comment.
Well the naming scheme is similar to the signal segmentation fault (SIGSEGV) signal handler. That one is also named after the signal / error message.
5feb4c3 to
1b8335b
Compare
1b8335b to
6920d29
Compare
MemoryCowCallbackis alwaysSome()in production, thus remove theOption<>.cow_cbparameter to be last innew_with_cowconstructors for consistency with thenewconstructors.cow_callback_payloadindependent of theMemoryState. This will later allow writable regions to be identifiable inside theMemoryCowCallbackas well.MemoryStateby boolean flagwritable.cow_callback_payloadfromu64tou32.