[llvm-commits] CVS: llvm/lib/CodeGen/ELFWriter.cpp
Chris Lattner
sabre at nondot.org
Thu Jul 7 09:33:09 PDT 2005
On Thu, 7 Jul 2005, Reid Spencer wrote:
> I didn't realize you were going to implement this. I have a partial
> implementation in the same file name(s) that is based on libelf. I don't
Ouch, I'm sorry, I didn't know you were currently working on it :(
> see why we would want to re-invent this wheel, but I'm sure you have
> your reasons.
The short version is that using libelf doesn't solve the hard problems of
writing the ELF file: organizing the information coming out of the code
generator so that we can spit it out into the file.
It does solve the easy problem: handling the on disk format (which is
really really simple) but at a very high cost: it is LGPL, which puts some
serious restrictions on how we can build the software. It is also a
pretty big chunk of code to have to keep around in our source tree as
well.
BTW, I found the answer to your question about Sun's use of libelf. The
answer is that Sun is not using the LGPL libelf. The LGPL libelf is a
clone of the preexisting Sun implementation. This is why Sun can use
(their) libelf without any restrictions.
-Chris
> If you're opposed to libelf, then please remove it from
> the lib/CodeGen/ELF directory.
Done.
-Chris
> Reid.
>
> On Mon, 2005-06-27 at 01:29 -0500, Chris Lattner wrote:
>>
>> Changes in directory llvm/lib/CodeGen:
>>
>> ELFWriter.cpp added (r1.1)
>> ---
>> Log message:
>>
>> iniital checkin of ELFWriter implementation
>>
>> For now, the elf writer is only capable of emitting an empty elf file, with
>> a section table and a section table string table. This will be enhanced
>> in the future :)
>>
>>
>> ---
>> Diffs of the changes: (+230 -0)
>>
>> ELFWriter.cpp | 230 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>> 1 files changed, 230 insertions(+)
>>
>>
>> Index: llvm/lib/CodeGen/ELFWriter.cpp
>> diff -c /dev/null llvm/lib/CodeGen/ELFWriter.cpp:1.1
>> *** /dev/null Mon Jun 27 01:29:10 2005
>> --- llvm/lib/CodeGen/ELFWriter.cpp Mon Jun 27 01:29:00 2005
>> ***************
>> *** 0 ****
>> --- 1,230 ----
>> + //===-- ELFWriter.cpp - Target-independent ELF Writer code ----------------===//
>> + //
>> + // The LLVM Compiler Infrastructure
>> + //
>> + // This file was developed by the LLVM research group and is distributed under
>> + // the University of Illinois Open Source License. See LICENSE.TXT for details.
>> + //
>> + //===----------------------------------------------------------------------===//
>> + //
>> + // This file implements the target-independent ELF writer. This file writes out
>> + // the ELF file in the following order:
>> + //
>> + // #1. ELF Header
>> + // #2. '.data' section
>> + // #3. '.bss' section
>> + // ...
>> + // #X. '.shstrtab' section
>> + // #Y. Section Table
>> + //
>> + // The entries in the section table are laid out as:
>> + // #0. Null entry [required]
>> + // #1. ".data" entry - global variables with initializers. [ if needed ]
>> + // #2. ".bss" entry - global variables without initializers. [ if needed ]
>> + // #3. ".text" entry - the program code
>> + // ...
>> + // #N. ".shstrtab" entry - String table for the section names.
>> +
>> + //
>> + // NOTE: This code should eventually be extended to support 64-bit ELF (this
>> + // won't be hard), but we haven't done so yet!
>> + //
>> + //===----------------------------------------------------------------------===//
>> +
>> + #include "llvm/CodeGen/ELFWriter.h"
>> + #include "llvm/Module.h"
>> + #include "llvm/Target/TargetMachine.h"
>> + using namespace llvm;
>> +
>> + ELFWriter::ELFWriter(std::ostream &o, TargetMachine &tm) : O(o), TM(tm) {
>> + e_machine = 0; // e_machine defaults to 'No Machine'
>> + e_flags = 0; // e_flags defaults to 0, no flags.
>> +
>> + is64Bit = TM.getTargetData().getPointerSizeInBits() == 64;
>> + isLittleEndian = TM.getTargetData().isLittleEndian();
>> + }
>> +
>> + // doInitialization - Emit the file header and all of the global variables for
>> + // the module to the ELF file.
>> + bool ELFWriter::doInitialization(Module &M) {
>> + outbyte(0x7F); // EI_MAG0
>> + outbyte('E'); // EI_MAG1
>> + outbyte('L'); // EI_MAG2
>> + outbyte('F'); // EI_MAG3
>> + outbyte(is64Bit ? 2 : 1); // EI_CLASS
>> + outbyte(isLittleEndian ? 1 : 2); // EI_DATA
>> + outbyte(1); // EI_VERSION
>> + for (unsigned i = OutputBuffer.size(); i != 16; ++i)
>> + outbyte(0); // EI_PAD up to 16 bytes.
>> +
>> + // This should change for shared objects.
>> + outhalf(1); // e_type = ET_REL
>> + outhalf(e_machine); // e_machine = whatever the target wants
>> + outword(1); // e_version = 1
>> + outaddr(0); // e_entry = 0 -> no entry point in .o file
>> + outaddr(0); // e_phoff = 0 -> no program header for .o
>> +
>> + ELFHeader_e_shoff_Offset = OutputBuffer.size();
>> + outaddr(0); // e_shoff
>> + outword(e_flags); // e_flags = whatever the target wants
>> +
>> + assert(!is64Bit && "These sizes need to be adjusted for 64-bit!");
>> + outhalf(52); // e_ehsize = ELF header size
>> + outhalf(0); // e_phentsize = prog header entry size
>> + outhalf(0); // e_phnum = # prog header entries = 0
>> + outhalf(40); // e_shentsize = sect header entry size
>> +
>> +
>> + ELFHeader_e_shnum_Offset = OutputBuffer.size();
>> + outhalf(0); // e_shnum = # of section header ents
>> + ELFHeader_e_shstrndx_Offset = OutputBuffer.size();
>> + outhalf(0); // e_shstrndx = Section # of '.shstrtab'
>> +
>> + // Add the null section.
>> + SectionList.push_back(ELFSection());
>> +
>> + // Okay, the ELF header has been completed, emit the .data section next.
>> + ELFSection DataSection(".data", OutputBuffer.size());
>> + for (Module::global_iterator I = M.global_begin(), E = M.global_end();
>> + I != E; ++I)
>> + EmitDATASectionGlobal(I);
>> +
>> + // If the .data section is nonempty, add it to our list.
>> + if ((DataSection.Size = OutputBuffer.size()-DataSection.Offset)) {
>> + DataSection.Align = 4; // FIXME: Compute!
>> + SectionList.push_back(DataSection);
>> + }
>> +
>> + // Okay, emit the .bss section next.
>> + ELFSection BSSSection(".bss", OutputBuffer.size());
>> + for (Module::global_iterator I = M.global_begin(), E = M.global_end();
>> + I != E; ++I)
>> + EmitBSSSectionGlobal(I);
>> +
>> + // If the .bss section is nonempty, add it to our list.
>> + if ((BSSSection.Size = OutputBuffer.size()-BSSSection.Offset)) {
>> + BSSSection.Align = 4; // FIXME: Compute!
>> + SectionList.push_back(BSSSection);
>> + }
>> +
>> + return false;
>> + }
>> +
>> + // isCOMM - A global variable should be emitted to the common area if it is zero
>> + // initialized and has linkage that permits it to be merged with other globals.
>> + static bool isCOMM(GlobalVariable *GV) {
>> + return GV->getInitializer()->isNullValue() &&
>> + (GV->hasLinkOnceLinkage() || GV->hasInternalLinkage() ||
>> + GV->hasWeakLinkage());
>> + }
>> +
>> + // EmitDATASectionGlobal - Emit a global variable to the .data section if it
>> + // belongs there.
>> + void ELFWriter::EmitDATASectionGlobal(GlobalVariable *GV) {
>> + if (!GV->hasInitializer()) return;
>> +
>> + // Do not emit a symbol here if it should be emitted to the common area.
>> + if (isCOMM(GV)) return;
>> +
>> + EmitGlobal(GV);
>> + }
>> +
>> + void ELFWriter::EmitBSSSectionGlobal(GlobalVariable *GV) {
>> + if (!GV->hasInitializer()) return;
>> +
>> + // FIXME: We don't support BSS yet!
>> + return;
>> +
>> + EmitGlobal(GV);
>> + }
>> +
>> + void ELFWriter::EmitGlobal(GlobalVariable *GV) {
>> + }
>> +
>> +
>> + bool ELFWriter::runOnMachineFunction(MachineFunction &MF) {
>> + return false;
>> + }
>> +
>> + /// doFinalization - Now that the module has been completely processed, emit
>> + /// the ELF file to 'O'.
>> + bool ELFWriter::doFinalization(Module &M) {
>> + // Emit the string table for the sections in the ELF file we have.
>> + EmitSectionTableStringTable();
>> +
>> + // Emit the .o file section table.
>> + EmitSectionTable();
>> +
>> + // Emit the .o file to the specified stream.
>> + O.write((char*)&OutputBuffer[0], OutputBuffer.size());
>> +
>> + // Free the output buffer.
>> + std::vector<unsigned char>().swap(OutputBuffer);
>> + return false;
>> + }
>> +
>> + /// EmitSectionTableStringTable - This method adds and emits a section for the
>> + /// ELF Section Table string table: the string table that holds all of the
>> + /// section names.
>> + void ELFWriter::EmitSectionTableStringTable() {
>> + // First step: add the section for the string table to the list of sections:
>> + SectionList.push_back(ELFSection(".shstrtab", OutputBuffer.size()));
>> + SectionList.back().Type = 3; // SHT_STRTAB
>> +
>> + // Now that we know which section number is the .shstrtab section, update the
>> + // e_shstrndx entry in the ELF header.
>> + fixhalf(SectionList.size()-1, ELFHeader_e_shstrndx_Offset);
>> +
>> + // Set the NameIdx of each section in the string table and emit the bytes for
>> + // the string table.
>> + unsigned Index = 0;
>> +
>> + for (unsigned i = 0, e = SectionList.size(); i != e; ++i) {
>> + // Set the index into the table. Note if we have lots of entries with
>> + // common suffixes, we could memoize them here if we cared.
>> + SectionList[i].NameIdx = Index;
>> +
>> + // Add the name to the output buffer, including the null terminator.
>> + OutputBuffer.insert(OutputBuffer.end(), SectionList[i].Name.begin(),
>> + SectionList[i].Name.end());
>> + // Add a null terminator.
>> + OutputBuffer.push_back(0);
>> +
>> + // Keep track of the number of bytes emitted to this section.
>> + Index += SectionList[i].Name.size()+1;
>> + }
>> +
>> + // Set the size of .shstrtab now that we know what it is.
>> + SectionList.back().Size = Index;
>> + }
>> +
>> + /// EmitSectionTable - Now that we have emitted the entire contents of the file
>> + /// (all of the sections), emit the section table which informs the reader where
>> + /// the boundaries are.
>> + void ELFWriter::EmitSectionTable() {
>> + // Now that all of the sections have been emitted, set the e_shnum entry in
>> + // the ELF header.
>> + fixhalf(SectionList.size(), ELFHeader_e_shnum_Offset);
>> +
>> + // Now that we know the offset in the file of the section table (which we emit
>> + // next), update the e_shoff address in the ELF header.
>> + fixaddr(OutputBuffer.size(), ELFHeader_e_shoff_Offset);
>> +
>> + // Emit all of the section table entries.
>> + for (unsigned i = 0, e = SectionList.size(); i != e; ++i) {
>> + const ELFSection &S = SectionList[i];
>> + outword(S.NameIdx); // sh_name - Symbol table name idx
>> + outword(S.Type); // sh_type - Section contents & semantics
>> + outword(S.Flags); // sh_flags - Section flags.
>> + outaddr(S.Addr); // sh_addr - The mem address this section appears in.
>> + outaddr(S.Offset); // sh_offset - The offset from the start of the file.
>> + outword(S.Size); // sh_size - The section size.
>> + outword(S.Link); // sh_link - Section header table index link.
>> + outword(S.Info); // sh_info - Auxillary information.
>> + outword(S.Align); // sh_addralign - Alignment of section.
>> + outword(S.EntSize); // sh_entsize - Size of each entry in the section.
>> + }
>> +
>> + // Release the memory allocated for the section list.
>> + std::vector<ELFSection>().swap(SectionList);
>> + }
>>
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://mail.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
-Chris
--
http://nondot.org/sabre/
http://llvm.cs.uiuc.edu/
More information about the llvm-commits
mailing list