Commit Graph

31 Commits

Author SHA1 Message Date
Rui Ueyama 96f0e0bfc9 ELF2: Improve Target::relocateOne().
relocateOne is a function to apply a relocation. Previously, that
function took a pointer to Elf_Rel or Elf_Rela in addition to other
information that can be derived from the relocation entry. This patch
simplifies the parameter list. The new parameters, P or SA, are used
in the ELF spec to describe each relocation. These names make
relocateOne look like a mechanical, direct translation of the ELF spec.

llvm-svn: 251090
2015-10-23 02:40:46 +00:00
George Rimar 648a2c37fb [ELF2] - Lazy relocation support for x86_64.
Target has supportsLazyRelocations() method which can switch lazy relocations on/off (currently all targets are OFF except x64 which is ON). So no any other targets are affected now.

Differential Revision: http://reviews.llvm.org/D13856?id=37726

llvm-svn: 250808
2015-10-20 08:54:27 +00:00
Hal Finkel 6f97c2bc00 [ELF2] getLocalRelTarget should handle R_PPC64_TOC directly
R_PPC64_TOC does not have an associated symbol, but does have a non-zero VA
that target-specific code must compute using some non-trivial rule. We
handled this as a special case in PPC64TargetInfo::relocateOne, where
we knew to write this special address, but that did not work when creating shared
libraries. The special TOC address needs to be the subject of a
R_PPC64_RELATIVE relocation, and so we also need to know how to encode this
special address in the addend of that relocation.

Thus, some target-specific logic is necessary when creating R_PPC64_RELATIVE as
well. To solve this problem, we teach getLocalRelTarget to handle R_PPC64_TOC
as a special case. This allows us to remove the special case in
PPC64TargetInfo::relocateOne (simplifying code there), and naturally allows the
existing logic to do the right thing when creating associated R_PPC64_RELATIVE
relocations for shared libraries.

llvm-svn: 250555
2015-10-16 21:55:40 +00:00
Hal Finkel 736c741752 [ELF2] Set a target-specific VAStart for PPC64
After some additional post-commit (post-revert) discussion and research, this
reverts, in part, r250205, so the ABI-recommended starting address can be used
on PPC64 (as is done by other linkers).

Also, this addresses the FIXME in ELF/Writer.cpp by making VAStart a
target-dependent property.

llvm-svn: 250378
2015-10-15 07:49:07 +00:00
Rui Ueyama efc23de4dd ELF2: Move Target concrete classes from .h to .cpp.
llvm-svn: 250330
2015-10-14 21:30:32 +00:00
Rafael Espindola 227556e1c7 Add support for a R_X86_64_32 referring to a plt.
This can show up with a non-PIC .o being linked into an executable that uses
shared libraries.

llvm-svn: 250300
2015-10-14 16:15:46 +00:00
Simon Atanasyan 9c2d788dda [ELF2][mips] Support both big and little endian MIPS 32-bit targets
- Make the `MipsTargetInfo` template class with `ELFType` argument. Use
  the argument to select an appropriate relocation type and read/write
  routines.
- Add template function `add32` to add-and-write relocation value in
  both big and little endian cases. Keep the `add32le` to reduce code
  changes.

Differential Revision: http://reviews.llvm.org/D13723

llvm-svn: 250297
2015-10-14 14:24:46 +00:00
Hal Finkel d962ef0bf8 Revert: "Re: [lld] r249752 - Revert: r249728 - Roll back r249726 and r249723 because they broke buildbots."
This has turned out to be unnecessary, and while some ability to set VAStart
will be needed at some point, this is not clearly the right direction.

llvm-svn: 250205
2015-10-13 18:50:28 +00:00
Rui Ueyama c58656c7c0 Revert r250169: "This patch implements basic variant of lazy loading for x86_x64 and for X86 targets."
With this patch LLD is not able to self-host on x86-64 Linux.

llvm-svn: 250182
2015-10-13 16:59:30 +00:00
George Rimar 9fd8fcb5a4 This patch implements basic variant of lazy loading for x86_x64 and for X86 targets.
What was done:
1) .got.plt section is created for functions that requires PLT. .got.plt has 3 predefined empty entries now that are required for dynamic linker.
Also other new items created are configured to have correct jump to PLT[N].
2) PLT section now has PLT[0] entry, also others ones are configured to support PLT->GOT(.got.plt) calls.
3) Implemented .rel[a].plt sections (based on patch http://reviews.llvm.org/D13569).
4) Fixed plt relocations types (based on patch http://reviews.llvm.org/D13589).

NOTES:
The .plt.got zero entry is still empty now. According to ELF specification it should hold the address of the dynamic structure, referenced with the symbol
_DYNAMIC. The _DYNAMIC entry points to the .dynamic section which contains information used by the ELF interpreter to setup the binary.

Differential Revision: http://reviews.llvm.org/D13651

llvm-svn: 250169
2015-10-13 16:09:55 +00:00
Rui Ueyama 9100439391 ELF2: Move createTarget to Target.cpp.
Now all Target<Arch> classes are used only in Target.cpp.
We can put them in an anonymous namespace. In order to avoid
merge conflict with other people's patches, I'll do that later.

llvm-svn: 250168
2015-10-13 16:08:15 +00:00
Rui Ueyama e7e3348aa3 ELF2: Remove dead code.
This code is completely a stub. We can resurrect when we start working on ARM.

llvm-svn: 250167
2015-10-13 16:03:39 +00:00
Hal Finkel 87bbd5ffd4 [ELF2] Allow PPC64 to add the TOC-restore after .plt-based relocations
Under the PPC64 ELF ABI, functions that might call into other modules (and,
thus, need to load a different TOC base value into %r2), need to restore the
old value after the call. The old value is saved by the .plt code, and the
caller only needs to include a nop instruction after the call, which the linker
will transform into a TOC restore if necessary.

In order to do this the relocation handler needs two things:

 1. It needs to know whether the call instruction it is modifying is targeting
    a .plt stub that will load a new TOC base value (necessitating a restore after
    the call).

 2. It needs to know where the buffer ends, so that it does not accidentally
    run off the end of the buffer when looking for the 'nop' instruction after the
    call.

Given these two pieces of information, we can insert the restore instruction in
place of the following nop when necessary.

llvm-svn: 250110
2015-10-12 21:19:18 +00:00
Hal Finkel be0823dcc4 [ELF2] Implement PPC64TargetInfo::isRelRelative
This is essentially pattern-matching against the x86 target, and generates the
analogous PPC64 relocation.

llvm-svn: 250102
2015-10-12 20:58:52 +00:00
Rui Ueyama 0cef68913e ELF2: Handle default entry name in the driver.
llvm-svn: 250048
2015-10-12 15:23:54 +00:00
Hal Finkel e3c2626aa4 [ELF2] Make the system page size a target-dependent property
The required page alignment is different on different targets. On PowerPC, for
example, we need 64K pages (the loader won't set different permissions on a
finer granularity than that). I've set the existing targets to what I believe
to be the correct values, and have updated the regression tests accordingly.

llvm-svn: 249760
2015-10-08 22:23:54 +00:00
Hal Finkel 6c2a3b8368 [ELF2] Make the .plt entry size target dependent
The size of a .plt entry is different on different targets (it is,
specifically, much larger than 8 on all PPC ABIs). There is no functional
change here (later patches to create .plt entries for PPC64 will depend on this
change).

llvm-svn: 249756
2015-10-08 21:51:31 +00:00
Hal Finkel 4729064a1c Revert: r249728 - Roll back r249726 and r249723 because they broke buildbots.
Reapply r249726 (and r249723), hopefully with the correct test fixups this time.

Original commit message:

Address a FIXME in ELF/Writer.cpp: Make VAStart a target-dependent property.
I've set the values for the existing targets to what I believe to be the
correct values, and updated the regression tests accordingly.

llvm-svn: 249752
2015-10-08 21:25:04 +00:00
Rui Ueyama af21d92320 ELF2: Do not pass GotVA because it's accessible as Out<ELFT>::Got->getVA().
llvm-svn: 249729
2015-10-08 20:06:07 +00:00
Rui Ueyama 7114f0597b Roll back r249726 and r249723 because they broke buildbots.
llvm-svn: 249728
2015-10-08 20:00:02 +00:00
Hal Finkel eb84ce8bd1 [ELF2] Make VAStart a target-dependent property
Address a FIXME in ELF/Writer.cpp: Make VAStart a target-dependent property.
I've set the values for the existing targets to what I believe to be the
correct values, and updated the regression tests accordingly.

llvm-svn: 249723
2015-10-08 19:36:39 +00:00
Rafael Espindola ae24400424 Create R_X86_64_RELATIVE when needed.
The dynamic relocation code needs refactoring, but it is probably better
to do it with this test passing.

llvm-svn: 249340
2015-10-05 19:30:12 +00:00
Rafael Espindola 3ef3a4c9ff Start adding support for static programs using dynamic libraries.
This is just enough for a hello world using a dynamic glibc.

llvm-svn: 248854
2015-09-29 23:22:16 +00:00
Rafael Espindola 8acb95c7ab Add support for R_386_GOT32.
llvm-svn: 248799
2015-09-29 14:42:37 +00:00
Rafael Espindola 6d7fcdb518 Add R_386_GOTPC support.
llvm-svn: 248793
2015-09-29 13:36:32 +00:00
Simon Atanasyan 49829a1bfe [ELF2] Add initial MIPS support
Besides a trivial MIPS support the patch introduces new TargetInfo class
member getDefaultEntry() to override default name of the entry symbol.
MIPS uses __start for that.

Differential Revision:  http://reviews.llvm.org/D13227

llvm-svn: 248779
2015-09-29 05:34:03 +00:00
Davide Italiano cde9336717 [ELF2] Add initial AArch64 support.
Differential Revision:	http://reviews.llvm.org/D13167

llvm-svn: 248641
2015-09-26 00:32:04 +00:00
Rafael Espindola 7f07442bb6 Move the last remaining hard coded relocations to Target.
Unfortunately the i386 and x86_64 relocation have the same numerical value
and it is a probably a bit much to add got support for another architecture
just to test this.

llvm-svn: 248326
2015-09-22 21:35:51 +00:00
Rafael Espindola 1d6063e804 Stub out ARM and PPC targets so that we can use a switch to create them.
llvm-svn: 248324
2015-09-22 21:24:52 +00:00
Rafael Espindola c40108858d Move relocation processing to Target.
I will add a couple of ppc64 relocs in the next patches.

llvm-svn: 248319
2015-09-22 20:54:08 +00:00
Rafael Espindola 01205f79a4 Start adding target abstractions.
This is just enough to get PLT working on 32 bit x86.

The idea behind using a virtual interface is that it should be easy to
convert any of the functions to template parameters if any turns out to be
performance critical.

llvm-svn: 248308
2015-09-22 18:19:46 +00:00