[ELF] Change std::max<uint64_t> to uint32_t for section alignment

Summary:
We use `uint32_t SectionBase::Alignment` and `uint32_t
PhdrEntry::p_align` despite alignments being 64 bits in ELF64.
Fix the std::max template arguments accordingly.

The currently 160-byte InputSection will become 168 bytes if we make SectionBase::Alignment uint64_t.

Differential Revision: https://reviews.llvm.org/D61171

llvm-svn: 359268

Author: MaskRay

lld/ELF/InputSection.cpp

@@ -75,7 +75,7 @@ InputSectionBase::InputSectionBase(InputFile *File, uint64_t Flags,
 
   // The ELF spec states that a value of 0 means the section has
   // no alignment constraits.
-  uint32_t V = std::max<uint64_t>(Alignment, 1);
+  uint32_t V = std::max<uint32_t>(Alignment, 1);
   if (!isPowerOf2_64(V))
     fatal(toString(this) + ": sh_addralign is not a power of 2");
   this->Alignment = V;
@@ -253,7 +253,7 @@ void InputSectionBase::parseCompressedHeader() {
     }
 
     UncompressedSize = Hdr->ch_size;
-    Alignment = std::max<uint64_t>(Hdr->ch_addralign, 1);
+    Alignment = std::max<uint32_t>(Hdr->ch_addralign, 1);
     RawData = RawData.slice(sizeof(*Hdr));
     return;
   }
@@ -271,7 +271,7 @@ void InputSectionBase::parseCompressedHeader() {
   }
 
   UncompressedSize = Hdr->ch_size;
-  Alignment = std::max<uint64_t>(Hdr->ch_addralign, 1);
+  Alignment = std::max<uint32_t>(Hdr->ch_addralign, 1);
   RawData = RawData.slice(sizeof(*Hdr));
 }