plugins/amdgpu: Implement parallel restore

This patch implements the entire logic to enable the offloading of
buffer object content restoration.

The goal of this patch is to offload the buffer object content
restoration to the main CRIU process so that this restoration can occur
in parallel with other restoration logic (mainly the restoration of
memory state in the restore blob, which is time-consuming) to speed up
the restore phase. The restoration of buffer object content usually
takes a significant amount of time for GPU applications, so
parallelizing it with other operations can reduce the overall restore
time.

It has three parts: the first replaces the restoration of buffer objects
in the target process by sending a parallel restore command to the main
CRIU process; the second implements the POST_FORKING hook in the amdgpu
plugin to enable buffer object content restoration in the main CRIU
process; the third stops the parallel thread in the RESUME_DEVICES_LATE
hook.

This optimization only focuses on the single-process situation (common
case). In other scenarios, it will turn to the original method. This is
achieved with the new `parallel_disabled` flag.

Signed-off-by: Yanning Yang <[email protected]>

plugins/amdgpu/Makefile

@@ -27,7 +27,7 @@ endif
 criu-amdgpu.pb-c.c: criu-amdgpu.proto
 		protoc-c --proto_path=. --c_out=. criu-amdgpu.proto
 
-amdgpu_plugin.so: amdgpu_plugin.c amdgpu_plugin_drm.c amdgpu_plugin_topology.c amdgpu_plugin_util.c criu-amdgpu.pb-c.c
+amdgpu_plugin.so: amdgpu_plugin.c amdgpu_plugin_drm.c amdgpu_plugin_topology.c amdgpu_plugin_util.c criu-amdgpu.pb-c.c amdgpu_socket_utils.c
 	$(CC) $(PLUGIN_CFLAGS) $(shell $(COMPEL) includes) $^ -o $@ $(PLUGIN_INCLUDE) $(PLUGIN_LDFLAGS) $(LIBDRM_INC)
 
 amdgpu_plugin_clean: