i#4014 dr$sim phys: Handle page-spanning accesses (#5536)

Adds handling of page-spanning instruction fetch and data accesses.
For offline traces these are heuristics as we do not know the precise
sizes at tracing time.  Since an extra physical page value is not
harmful, the heuristics emit a second page if there is any decent
chance that it is reached.

Adds a data page span test by switching the drcacheoff.phys test to
use allasm_repstr and adding such an access there.  An instruction
cross is more difficult to synthesize.

Issue: #4014

clients/drcachesim/tests/allasm-repstr-basic-counts.templatex

@@ -11,11 +11,11 @@ Adios world!
 ---- <application exited with code 0> ----
 Basic counts tool results:
 Total counts:
-          95 total \(fetched\) instructions
-          23 total unique \(fetched\) instructions
+          98 total \(fetched\) instructions
+          26 total unique \(fetched\) instructions
            4 total non-fetched instructions
            0 total prefetches
-           5 total data loads
+           7 total data loads
            5 total data stores
            0 total icache flushes
            0 total dcache flushes
@@ -28,11 +28,11 @@ Total counts:
            0 total function return value markers
            4 total other markers
 Thread [0-9]* counts:
-          95 \(fetched\) instructions
-          23 unique \(fetched\) instructions
+          98 \(fetched\) instructions
+          26 unique \(fetched\) instructions
            4 non-fetched instructions
            0 prefetches
-           5 data loads
+           7 data loads
            5 data stores
            0 icache flushes
            0 dcache flushes

clients/drcachesim/tests/allasm_repstr.asm

@@ -53,6 +53,13 @@ _start:
         cld
         rep      movsb
 
+        // Test page-spanning accesses.
+        lea      rcx, page_str
+        // Somehow the GNU assembler 2.38 is adding the load size (4 here) to
+        // whatever displacement is listed (!!!), so these end up as -3 and -1.
+        mov      eax, DWORD [-7+rcx]
+        mov      eax, DWORD [-5+rcx]
+
         // Print a message in a loop for testing tracing windows.
         mov      ebx, 10          // Loop count.
 repeat:
@@ -76,3 +83,8 @@ hello_str:
         .string  "Hello world!\n"
 bye_str:
         .string  "Adios\n"
+        // Push .data onto a 2nd page to test page-spanning accesses.
+        // We assume 4K pages here.
+        .align   4096
+page_str:
+        .word    0

clients/drcachesim/tests/offline-allasm-repstr-basic-counts.templatex

@@ -10,11 +10,11 @@ Adios world!
 Adios world!
 Basic counts tool results:
 Total counts:
-          95 total \(fetched\) instructions
-          23 total unique \(fetched\) instructions
+          98 total \(fetched\) instructions
+          26 total unique \(fetched\) instructions
            4 total non-fetched instructions
            0 total prefetches
-           5 total data loads
+           7 total data loads
            5 total data stores
            0 total icache flushes
            0 total dcache flushes
@@ -27,11 +27,11 @@ Total counts:
            0 total function return value markers
            4 total other markers
 Thread [0-9]* counts:
-          95 \(fetched\) instructions
-          23 unique \(fetched\) instructions
+          98 \(fetched\) instructions
+          26 unique \(fetched\) instructions
            4 non-fetched instructions
            0 prefetches
-           5 data loads
+           7 data loads
            5 data stores
            0 icache flushes
            0 dcache flushes

clients/drcachesim/tests/offline-phys.templatex

@@ -1,19 +1,13 @@
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
-Hello world!
+Adios world!
+Adios world!
+Adios world!
+Adios world!
+Adios world!
+Adios world!
+Adios world!
+Adios world!
+Adios world!
+Adios world!
 Output format:
 <record#>: T<tid> <record details>
 ------------------------------------------------------------
@@ -25,6 +19,10 @@ Output format:
         6: T[0-9]+ <marker: tid [0-9]+ on core [0-9]+>
         7: T[0-9]+ <marker: physical address for following virtual: 0x[0-9a-f][0-9a-f]+>
         8: T[0-9]+ <marker: virtual address for prior physical: 0x[0-9a-f][0-9a-f]+>
-        9: T[0-9]+ <marker: timestamp [0-9]+>
-       10: T[0-9]+ <marker: tid [0-9]+ on core [0-9]+>
-       11: T[0-9]+ ifetch .*
+        9: T[0-9]+ <marker: physical address for following virtual: 0x[0-9a-f][0-9a-f]+>
+       10: T[0-9]+ <marker: virtual address for prior physical: 0x[0-9a-f][0-9a-f]+>
+       11: T[0-9]+ <marker: physical address for following virtual: 0x[0-9a-f][0-9a-f]+>
+       12: T[0-9]+ <marker: virtual address for prior physical: 0x[0-9a-f][0-9a-f]+>
+       13: T[0-9]+ <marker: timestamp [0-9]+>
+       14: T[0-9]+ <marker: tid [0-9]+ on core [0-9]+>
+       15: T[0-9]+ ifetch .*

clients/drcachesim/tests/offline-windows-asm.templatex

@@ -22,11 +22,11 @@ Adios world!
 Hit tracing window #5 limit: disabling tracing.
 Basic counts tool results:
 Total counts:
-          50 total \(fetched\) instructions
-          18 total unique \(fetched\) instructions
+          53 total \(fetched\) instructions
+          21 total unique \(fetched\) instructions
            4 total non-fetched instructions
            0 total prefetches
-           5 total data loads
+           7 total data loads
            5 total data stores
            0 total icache flushes
            0 total dcache flushes
@@ -40,11 +40,11 @@ Total counts:
           11 total other markers
 Total windows: 7
 Window #0:
-          12 window \(fetched\) instructions
-          12 window unique \(fetched\) instructions
+          15 window \(fetched\) instructions
+          15 window unique \(fetched\) instructions
            4 window non-fetched instructions
            0 window prefetches
-           5 window data loads
+           7 window data loads
            5 window data stores
            0 window icache flushes
            0 window dcache flushes
@@ -152,11 +152,11 @@ Window #6:
            0 window function return value markers
            1 window other markers
 Thread [0-9]* counts:
-          12 \(fetched\) instructions
-          12 unique \(fetched\) instructions
+          15 \(fetched\) instructions
+          15 unique \(fetched\) instructions
            4 non-fetched instructions
            0 prefetches
-           5 data loads
+           7 data loads
            5 data stores
            0 icache flushes
            0 dcache flushes

clients/drcachesim/tracer/physaddr.cpp

@@ -66,6 +66,9 @@ physaddr_t::physaddr_t()
     , fd_(-1)
     , v2p_(nullptr)
     , count_(0)
+    , num_hit_cache_(0)
+    , num_hit_table_(0)
+    , num_miss_(0)
 #endif
 {
 #ifdef LINUX
@@ -246,7 +249,7 @@ physaddr_t::virtual2physical(void *drcontext, addr_t virt, OUT addr_t *phys,
     NOTIFY(3, "v2p: %p => entry " HEX64_FORMAT_STRING " @ offs " INT64_FORMAT_STRING "\n",
            vpage, entry, offs);
     if (!TESTALL(PAGEMAP_VALID, entry) || TESTANY(PAGEMAP_SWAP, entry)) {
-        NOTIFY(1, "v2p failure: entry is invalid for %p\n", vpage);
+        NOTIFY(1, "v2p failure: entry %p is invalid for %p\n", vpage);
         return false;
     }
     addr_t ppage = (addr_t)((entry & PAGEMAP_PFN) << page_bits_);