Export of internal Abseil changes

--
ed829ac612f090375427c3488827c6e74deb2e3f by Derek Mauro <dmauro@google.com>:

Update latest GCC/Clang Linux tests to Bazel 5.0.0 and CMake 3.22.2

PiperOrigin-RevId: 429369775

--
76952303c4d942288c4e7657ffb5893cec54a132 by Martijn Vels <mvels@google.com>:

Optimize Cord::ChunkIterator now that CordRepConcat is removed

PiperOrigin-RevId: 429321455

--
dcd0d287793649aba9b98268c5783e449a34749f by Martijn Vels <mvels@google.com>:

Add IsDataEdge() and DataEdgeValue() helper functions.

This moves repetitive logic accessing data edges into its own header, and more strongly defines the notion of what a data edge is, enforcing the internal invariants. This will also be incorporated in optimized Cord iteration logic once CordRepConcat is totally removed from the Cord code.

PiperOrigin-RevId: 429307248

--
6a0903962155988085bf8656743fda9c4cdcba6c by Abseil Team <absl-team@google.com>:

Make it clear that the probability function given for the zipf distribution is unnormalized, i.e., sum(p(x) for x = 0..k) != 100%.

Quoting Section 7 of the paper cited in the comments, where this formula comes from (emphasis mine): "We will consider the two parameter generalization as defined in Dagpunar [1988] with the *unnormalized* probability function ..."

PiperOrigin-RevId: 429068258

--
3899ff6d444ba755148bc521a6ee031d9e9d4485 by Abseil Team <absl-team@google.com>:

Internal Changes

PiperOrigin-RevId: 428644856

--
319de702d2b537cbb76c4c71277ae89b349b162e by Benjamin Barenblat <bbaren@google.com>:

Support symbolization on PA-RISC

Null out supervisor bits in PA-RISC addresses before symbolizing, and
handle function descriptor tables correctly.

Change symbolize_test.cc to use 32-bit aligned addresses, allowing that
test to pass on PA-RISC.

PiperOrigin-RevId: 428590564
GitOrigin-RevId: ed829ac612f090375427c3488827c6e74deb2e3f
Change-Id: Ie01ff3b9365fd45e5a55f858038552679f3180d3

2 files changed, 43 insertions, 10 deletions

diff --git a/absl/debugging/symbolize_elf.inc b/absl/debugging/symbolize_elf.inc
index 3ff343d6..ddccd590 100644
--- a/absl/debugging/symbolize_elf.inc
+++ b/absl/debugging/symbolize_elf.inc
@@ -323,6 +323,7 @@ class Symbolizer {
                                            const ptrdiff_t relocation,
                                            char *out, int out_size,
                                            char *tmp_buf, int tmp_buf_size);
+  const char *GetUncachedSymbol(const void *pc);
 
   enum {
     SYMBOL_BUF_SIZE = 3072,
@@ -1333,13 +1334,7 @@ static bool MaybeInitializeObjFile(ObjFile *obj) {
 // they are called here as well.
 // To keep stack consumption low, we would like this function to not
 // get inlined.
-const char *Symbolizer::GetSymbol(const void *const pc) {
-  const char *entry = FindSymbolInCache(pc);
-  if (entry != nullptr) {
-    return entry;
-  }
-  symbol_buf_[0] = '\0';
-
+const char *Symbolizer::GetUncachedSymbol(const void *pc) {
   ObjFile *const obj = FindObjFile(pc, 1);
   ptrdiff_t relocation = 0;
   int fd = -1;
@@ -1427,6 +1422,42 @@ const char *Symbolizer::GetSymbol(const void *const pc) {
   return InsertSymbolInCache(pc, symbol_buf_);
 }
 
+const char *Symbolizer::GetSymbol(const void *pc) {
+  const char *entry = FindSymbolInCache(pc);
+  if (entry != nullptr) {
+    return entry;
+  }
+  symbol_buf_[0] = '\0';
+
+#ifdef __hppa__
+  {
+    // In some contexts (e.g., return addresses), PA-RISC uses the lowest two
+    // bits of the address to indicate the privilege level. Clear those bits
+    // before trying to symbolize.
+    const auto pc_bits = reinterpret_cast<uintptr_t>(pc);
+    const auto address = pc_bits & ~0x3;
+    entry = GetUncachedSymbol(reinterpret_cast<const void *>(address));
+    if (entry != nullptr) {
+      return entry;
+    }
+
+    // In some contexts, PA-RISC also uses bit 1 of the address to indicate that
+    // this is a cross-DSO function pointer. Such function pointers actually
+    // point to a procedure label, a struct whose first 32-bit (pointer) element
+    // actually points to the function text. With no symbol found for this
+    // address so far, try interpreting it as a cross-DSO function pointer and
+    // see how that goes.
+    if (pc_bits & 0x2) {
+      return GetUncachedSymbol(*reinterpret_cast<const void *const *>(address));
+    }
+
+    return nullptr;
+  }
+#else
+  return GetUncachedSymbol(pc);
+#endif
+}
+
 bool RemoveAllSymbolDecorators(void) {
   if (!g_decorators_mu.TryLock()) {
     // Someone else is using decorators. Get out.
diff --git a/absl/debugging/symbolize_test.cc b/absl/debugging/symbolize_test.cc
index c710a3da..a62fa35d 100644
--- a/absl/debugging/symbolize_test.cc
+++ b/absl/debugging/symbolize_test.cc
@@ -392,12 +392,14 @@ TEST(Symbolize, InstallAndRemoveSymbolDecorators) {
                 DummySymbolDecorator, &c_message),
             0);
 
-  char *address = reinterpret_cast<char *>(1);
-  EXPECT_STREQ("abc", TrySymbolize(address++));
+  // Use addresses 4 and 8 here to ensure that we always use valid addresses
+  // even on systems that require instructions to be 32-bit aligned.
+  char *address = reinterpret_cast<char *>(4);
+  EXPECT_STREQ("abc", TrySymbolize(address));
 
   EXPECT_TRUE(absl::debugging_internal::RemoveSymbolDecorator(ticket_b));
 
-  EXPECT_STREQ("ac", TrySymbolize(address++));
+  EXPECT_STREQ("ac", TrySymbolize(address + 4));
 
   // Cleanup: remove all remaining decorators so other stack traces don't
   // get mystery "ac" decoration.