Store pinned static fields in the Pinned Object Heap.

[teo-tsirpanis]

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes#66043

[ghost]

Tagging subscribers to this area: @vitek-karas, @agocke, @VSadov
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes #66043

Author:	teo-tsirpanis
Assignees:	-
Labels:	area-AssemblyLoader-coreclr
Milestone:	-

[janvorli]

LGTM, thank you!

[janvorli]

@teo-tsirpanis you'll also need to remove the library negative test for the FixedAddressValueType attribute and unloadability - it is causing the CI failures.

[MichalPetryka]

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields.

Worth noting is that this will unload the assembly when the user uses Unsafe.AsPointer on the field since nothing with keep it alive and then such access could corrupt the process.

[mangod9]

@teo-tsirpanis, could you please rebase the PR and then we could run the CI again to get it merged? Thanks

[teo-tsirpanis]

@mangod9 done. CI is green.

[jkotas]

Thank you!

[teo-tsirpanis]

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes#66043

[ghost]

Tagging subscribers to this area: @vitek-karas, @agocke, @VSadov
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes #66043

Author:	teo-tsirpanis
Assignees:	-
Labels:	area-AssemblyLoader-coreclr
Milestone:	-

[janvorli]

LGTM, thank you!

[janvorli]

@teo-tsirpanis you'll also need to remove the library negative test for the FixedAddressValueType attribute and unloadability - it is causing the CI failures.

[MichalPetryka]

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields.

Worth noting is that this will unload the assembly when the user uses Unsafe.AsPointer on the field since nothing with keep it alive and then such access could corrupt the process.

[mangod9]

@teo-tsirpanis, could you please rebase the PR and then we could run the CI again to get it merged? Thanks

[teo-tsirpanis]

@mangod9 done. CI is green.

[jkotas]

Thank you!

[teo-tsirpanis]

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes#66043

[ghost]

Tagging subscribers to this area: @vitek-karas, @agocke, @VSadov
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes #66043

Author:	teo-tsirpanis
Assignees:	-
Labels:	area-AssemblyLoader-coreclr
Milestone:	-

[janvorli]

LGTM, thank you!

[janvorli]

@teo-tsirpanis you'll also need to remove the library negative test for the FixedAddressValueType attribute and unloadability - it is causing the CI failures.

[MichalPetryka]

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields.

Worth noting is that this will unload the assembly when the user uses Unsafe.AsPointer on the field since nothing with keep it alive and then such access could corrupt the process.

[mangod9]

@teo-tsirpanis, could you please rebase the PR and then we could run the CI again to get it merged? Thanks

[teo-tsirpanis]

@mangod9 done. CI is green.

[jkotas]

Thank you!

[teo-tsirpanis]

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes#66043

[ghost]

Tagging subscribers to this area: @vitek-karas, @agocke, @VSadov
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes #66043

Author:	teo-tsirpanis
Assignees:	-
Labels:	area-AssemblyLoader-coreclr
Milestone:	-

[janvorli]

LGTM, thank you!

[janvorli]

@teo-tsirpanis you'll also need to remove the library negative test for the FixedAddressValueType attribute and unloadability - it is causing the CI failures.

[MichalPetryka]

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields.

Worth noting is that this will unload the assembly when the user uses Unsafe.AsPointer on the field since nothing with keep it alive and then such access could corrupt the process.

[mangod9]

@teo-tsirpanis, could you please rebase the PR and then we could run the CI again to get it merged? Thanks

[teo-tsirpanis]

@mangod9 done. CI is green.

[jkotas]

Thank you!

[teo-tsirpanis]

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes#66043

[ghost]

Tagging subscribers to this area: @vitek-karas, @agocke, @VSadov
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes #66043

Author:	teo-tsirpanis
Assignees:	-
Labels:	area-AssemblyLoader-coreclr
Milestone:	-

[janvorli]

LGTM, thank you!

[janvorli]

@teo-tsirpanis you'll also need to remove the library negative test for the FixedAddressValueType attribute and unloadability - it is causing the CI failures.

[MichalPetryka]

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields.

Worth noting is that this will unload the assembly when the user uses Unsafe.AsPointer on the field since nothing with keep it alive and then such access could corrupt the process.

[mangod9]

@teo-tsirpanis, could you please rebase the PR and then we could run the CI again to get it merged? Thanks

[teo-tsirpanis]

@mangod9 done. CI is green.

[jkotas]

Thank you!

[teo-tsirpanis]

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes#66043

[ghost]

Tagging subscribers to this area: @vitek-karas, @agocke, @VSadov
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes #66043

Author:	teo-tsirpanis
Assignees:	-
Labels:	area-AssemblyLoader-coreclr
Milestone:	-

[janvorli]

LGTM, thank you!

[janvorli]

@teo-tsirpanis you'll also need to remove the library negative test for the FixedAddressValueType attribute and unloadability - it is causing the CI failures.

[MichalPetryka]

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields.

Worth noting is that this will unload the assembly when the user uses Unsafe.AsPointer on the field since nothing with keep it alive and then such access could corrupt the process.

[mangod9]

@teo-tsirpanis, could you please rebase the PR and then we could run the CI again to get it merged? Thanks

[teo-tsirpanis]

@mangod9 done. CI is green.

[jkotas]

Thank you!

[teo-tsirpanis]

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes#66043

[ghost]

Tagging subscribers to this area: @vitek-karas, @agocke, @VSadov
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes #66043

Author:	teo-tsirpanis
Assignees:	-
Labels:	area-AssemblyLoader-coreclr
Milestone:	-

[janvorli]

LGTM, thank you!

[janvorli]

@teo-tsirpanis you'll also need to remove the library negative test for the FixedAddressValueType attribute and unloadability - it is causing the CI failures.

[MichalPetryka]

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields.

Worth noting is that this will unload the assembly when the user uses Unsafe.AsPointer on the field since nothing with keep it alive and then such access could corrupt the process.

[mangod9]

@teo-tsirpanis, could you please rebase the PR and then we could run the CI again to get it merged? Thanks

[teo-tsirpanis]

@mangod9 done. CI is green.

[jkotas]

Thank you!

[teo-tsirpanis]

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes#66043

[ghost]

Tagging subscribers to this area: @vitek-karas, @agocke, @VSadov
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR allocates fields marked with [FixedAddressValueType] to the Pinned Object Heap. We introduce a new overload of AllocateObject that accepts a GC_ALLOC_FLAGS, and when allocating the static fields we pass GC_ALLOC_PINNED_OBJECT_HEAP if the field has that attribute.

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields. All we have to do is remove the check that blocked these assemblies, and unmark the relevant unit test as incompatible with unloadability. I also removed a now-unused tracking mechanism for thread-static pinned fields.

Fixes #66043

Author:	teo-tsirpanis
Assignees:	-
Labels:	area-AssemblyLoader-coreclr
Milestone:	-

[janvorli]

LGTM, thank you!

[janvorli]

@teo-tsirpanis you'll also need to remove the library negative test for the FixedAddressValueType attribute and unloadability - it is causing the CI failures.

[MichalPetryka]

Because we don't use a pinned handle to pin the fields anymore, the object is not permanently held in memory, which enables unloadability for assemblies with pinned static fields.

Worth noting is that this will unload the assembly when the user uses Unsafe.AsPointer on the field since nothing with keep it alive and then such access could corrupt the process.

[mangod9]

@teo-tsirpanis, could you please rebase the PR and then we could run the CI again to get it merged? Thanks

[teo-tsirpanis]

@mangod9 done. CI is green.

[jkotas]

Thank you!