refactor: rename uffd_valid_handler to uffd_on_demand_handler

This better describes what the handler actually does (technically,
fault_all_handler is also "valid"), and now that some tests emit the
name of the uffd handler as a metric, the names should be descriptive
from the get-go to avoid confusion and future renamings.

Signed-off-by: Patrick Roy <[email protected]>

Author: roypat

docs/snapshotting/handling-page-faults-on-snapshot-resume.md

@@ -162,7 +162,7 @@ connect/send data.
 ### Example
 
 An example of a handler process can be found
-[here](../../src/firecracker/examples/uffd/valid_handler.rs). The process is
+[here](../../src/firecracker/examples/uffd/on_demand_handler.rs). The process is
 designed to tackle faults on a certain address by loading into memory the entire
 region that the address belongs to, but users can choose any other behavior that
 suits their use case best.

src/firecracker/Cargo.toml

@@ -57,8 +57,8 @@ name = "uffd_malicious_handler"
 path = "examples/uffd/malicious_handler.rs"
 
 [[example]]
-name = "uffd_valid_handler"
-path = "examples/uffd/valid_handler.rs"
+name = "uffd_on_demand_handler"
+path = "examples/uffd/on_demand_handler.rs"
 
 [[example]]
 name = "uffd_fault_all_handler"

src/firecracker/examples/uffd/valid_handler.rs renamed to src/firecracker/examples/uffd/on_demand_handler.rs

@@ -92,7 +92,7 @@ def test_valid_handler(uvm_plain, snapshot):
     vm.spawn()
 
     # Spawn page fault handler process.
-    pf_handler = spawn_pf_handler(vm, uffd_handler("valid"), snapshot.mem)
+    pf_handler = spawn_pf_handler(vm, uffd_handler("on_demand"), snapshot.mem)
 
     vm.restore_from_snapshot(snapshot, resume=True, uffd_path=pf_handler.socket_path)
 

tests/integration_tests/functional/test_uffd.py

@@ -93,7 +93,7 @@ def test_hugetlbfs_snapshot(microvm_factory, guest_kernel_linux_5_10, rootfs):
     vm.spawn()
 
     # Spawn page fault handler process.
-    pf_handler = spawn_pf_handler(vm, uffd_handler("valid"), snapshot.mem)
+    pf_handler = spawn_pf_handler(vm, uffd_handler("on_demand"), snapshot.mem)
 
     vm.restore_from_snapshot(snapshot, resume=True, uffd_path=pf_handler.socket_path)
 
@@ -135,7 +135,7 @@ def test_hugetlbfs_diff_snapshot(microvm_factory, uvm_plain):
     vm.spawn()
 
     # Spawn page fault handler process.
-    pf_handler = spawn_pf_handler(vm, uffd_handler("valid"), snapshot_merged.mem)
+    pf_handler = spawn_pf_handler(vm, uffd_handler("on_demand"), snapshot_merged.mem)
 
     vm.restore_from_snapshot(
         snapshot_merged, resume=True, uffd_path=pf_handler.socket_path

tests/integration_tests/performance/test_huge_pages.py

@@ -141,7 +141,7 @@ def test_restore_latency(
 # wakes up, because it gets faulted in on the first page fault. In this scenario, we are not measuring UFFD
 # latencies, but KVM latencies of setting up missing EPT entries.
 @pytest.mark.nonci
-@pytest.mark.parametrize("uffd_handler", [None, "valid", "fault_all"])
+@pytest.mark.parametrize("uffd_handler", [None, "on_demand", "fault_all"])
 @pytest.mark.parametrize("huge_pages", HugePagesConfig)
 def test_post_restore_latency(
     microvm_factory, rootfs, guest_kernel_linux_5_10, metrics, uffd_handler, huge_pages