Optimize memory handling (#1963)

* optimize ram block handling

Save the last element of the ram_list. This allows to
faster find where to add new elements when they are not
bigger then page size.

* save ram_list freed

this keeps the optimization for find_ram_offset() intact after snapshot
restore.

* cow only clear the tlb of affected pages

* update flatview when possible

Building each flatview new when the memory has changed is quite
expensive when many MemoryRegions are used. This is an issue when using
snapshots.

* update benchmark for new api

* save flatview in context

this avoids rebuilding the flatview when restore a context.

* init context flatview with zero

* address_space_dispatch_clear remove subpage with higher priority

* docutemnt the options for UC_CTL_CONTEXT_MODE

Specialy stress that with UC_CTL_CONTEXT_MEMORY it is not possible to
use the context with a different unicorn object.

include/qemu.h

@@ -43,6 +43,7 @@ typedef struct {
 typedef struct RAMList {
     bool freed;
     RAMBlock *mru_block;
+    RAMBlock *last_block;
     QLIST_HEAD(, RAMBlock) blocks;
 } RAMList;
 

include/uc_priv.h

@@ -108,6 +108,8 @@ typedef MemoryRegion *(*uc_memory_mapping_t)(struct uc_struct *, hwaddr addr);
 
 typedef void (*uc_memory_filter_t)(MemoryRegion *, int32_t);
 
+typedef bool (*uc_flatview_copy_t)(struct uc_struct *, FlatView *, FlatView *, bool);
+
 typedef void (*uc_readonly_mem_t)(MemoryRegion *mr, bool readonly);
 
 typedef int (*uc_cpus_init)(struct uc_struct *, const char *);
@@ -288,6 +290,7 @@ struct uc_struct {
     uc_args_uc_ram_size_ptr_t memory_map_ptr;
     uc_memory_mapping_t memory_mapping;
     uc_memory_filter_t memory_filter_subregions;
+    uc_flatview_copy_t flatview_copy;
     uc_mem_unmap_t memory_unmap;
     uc_mem_unmap_t memory_moveout;
     uc_mem_unmap_t memory_movein;
@@ -427,6 +430,9 @@ struct uc_context {
     uc_mode mode;        // the mode of this context
     uc_arch arch;        // the arch of this context
     int snapshot_level;  // the memory snapshot level to restore
+    bool ramblock_freed; // wheter there was a some ramblock freed
+    RAMBlock *last_block;// The last element of the ramblock list
+    FlatView *fv;        // The current flatview of the memory
     char data[0];        // context
 };
 

include/unicorn/unicorn.h

@@ -1018,6 +1018,16 @@ struct uc_tlb_entry {
     uc_prot perms;
 };
 
+/*
+ Variables to control which state should be stored in the context.
+ Defaults to UC_CTL_CONTEXT_CPU. The options are used in a bitfield
+ so to enable more then one content the binary or of the required
+ contents can be use.
+ The UC_CTL_CONTEXT_MEMORY stores some pointers to internal allocated
+ memory. Therefor it's not possible to use this context with another
+ unicorn object.
+*/
+
 typedef enum uc_context_content {
     UC_CTL_CONTEXT_CPU = 1,
     UC_CTL_CONTEXT_MEMORY = 2,