[LLD][ELF] Drop the string null terminator from the hash in splitStrings

[MaskRay]

Thanks for looking into the performance:)

This decreases work for xxhash and for x86-64 seems to make the code sequence 3 instructions shorter.

[andrewng]

Sorry, but I've missed some updates for LLD tests requiring zlib. Working on the fix now.

[thakis]

Looks like this breaks tests: http://45.33.8.238/linux/77072/step_11.txt

Please take a look and revert for now if it takes a while to fix.

[andrewng]

In D126484#3542066, @thakis wrote:

Looks like this breaks tests: http://45.33.8.238/linux/77072/step_11.txt

Please take a look and revert for now if it takes a while to fix.

Should already be fixed by commit a94d454390c6.

lld/ELF/InputSection.cpp

 void MergeInputSection::splitStrings(StringRef s, size_t entSize) {
   const bool live = !(flags & SHF_ALLOC) || !config->gcSections;
   const char *p = s.data(), *end = s.data() + s.size();
   if (!std::all_of(end - entSize, end, [](char c) { return c == 0; }))
     fatal(toString(this) + ": string is not null terminated");
   if (entSize == 1) {
     // Optimize the common case.
     do {
-      size_t size = strlen(p) + 1;
+      size_t size = strlen(p);
       pieces.emplace_back(p - s.begin(), xxHash64(StringRef(p, size)), live);
-      p += size;
+      p += size + 1;
     } while (p != end);
   } else {
     do {
-      size_t size = findNull(StringRef(p, end - p), entSize) + entSize;
+      size_t size = findNull(StringRef(p, end - p), entSize);
       pieces.emplace_back(p - s.begin(), xxHash64(StringRef(p, size)), live);
-      p += size;
+      p += size + entSize;
     } while (p != end);
   }
 }
 
 // Split non-SHF_STRINGS section. Such section is a sequence of
 // fixed size records.
 void MergeInputSection::splitNonStrings(ArrayRef<uint8_t> data,
                                         size_t entSize) {

lld/test/ELF/comment-gc.s

 # REQUIRES: x86
 # RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %s -o %t.o
 # RUN: llvm-mc -filetype=obj -triple=x86_64-unknown-linux %p/Inputs/comment-gc.s -o %t2.o
 # RUN: ld.lld %t.o %t2.o -o %t1 --gc-sections -shared
 # RUN: llvm-objdump -s %t1 | FileCheck %s
 
 # CHECK:      Contents of section .comment:
-# CHECK-NEXT: foo..LLD 1.0.bar
+# CHECK-NEXT: .LLD 1.0.foo.bar
 
 .ident "foo"
 
 .globl _start
 _start:
   nop

lld/test/ELF/merge-string.s

 // NOTAIL-NEXT: Address: 0x20D
 // NOTAIL-NEXT: Offset: 0x20D
 // NOTAIL-NEXT: Size:    7
 // NOTAIL-NEXT: Link: 0
 // NOTAIL-NEXT: Info: 0
 // NOTAIL-NEXT: AddressAlignment: 1
 // NOTAIL-NEXT: EntrySize: 1
 // NOTAIL-NEXT: SectionData (
-// NOTAIL-NEXT:   0000: 62630061 626300                     |bc.abc.|
+// NOTAIL-NEXT:   0000: 61626300 626300                     |abc.bc.|
 // NOTAIL-NEXT: )
 
 // NOMERGE:      Name:    .rodata1
 // NOMERGE-NEXT: Type:    SHT_PROGBITS
 // NOMERGE-NEXT: Flags [
 // NOMERGE-NEXT:   SHF_ALLOC
 // NOMERGE-NEXT:   SHF_MERGE
 // NOMERGE-NEXT:   SHF_STRINGS