This CL pulls out the refactorings used to support the new MachOWriter
in a followup CL to allow them to be reviewed separately.
Rename Elf -> ElfWriter. Also rename model classes used by ElfWriter
for concepts that exist both in ELF and Mach-O to ElfX. For example,
the old ELF-specific SymbolTable is renamed to ElfSymbolTable.
Adds SharedObjectWriter to serve as a base class for both ElfWriter
and the upcoming MachOWriter.
Adds a new AbstractWriteStream that serves as a common superclass
of both BaseWriteStream and SharedObjectWriter::WriteStream and
allows the creation of fully delegating WriteStreams that do not
maintain a local buffer.
Abstract the old Elf::SymbolData class into
SharedObjectWriter::SymbolData, which stores an enum value as the type
of the symbol instead of storing the ELF encoding of the type.
Rename the DwarfElfStream (which actually wasn't ELF specific, as
all the ELF-specific DWARF information is handled by ElfWriter) to
DwarfSharedObjectStream and put it in a separate header file.
Rename Image::compiled_to_elf() to Image::compiled_to_shared_object()
and add a separate Image::compiled_to_elf() that checks for the ELF
magic value at the DSO base. Also add Image::shared_object_start()
and Image::build_id_start() to return pointers to the DSO base and
the build ID note, respectively.
Refactor Image::build_id() and Image::build_id_length() to check
compiled_for_elf() prior to decoding the data pointed to by
build_id_start() as an ELF note section.
Create an AOTSnapshotType enum to specific the snapshot writer to use
in CreateAppAOTSnapshot instead of using an as_elf boolean.
TEST=refactorings, so existing tests on ci
Change-Id: Ia3ab37a4dff93b6e00390b123753be5a51fbdaaa
Cq-Include-Trybots: luci.dart.try:vm-aot-linux-debug-x64-try,vm-aot-mac-release-arm64-try,vm-aot-dwarf-linux-product-x64-try,vm-linux-debug-x64-try,vm-mac-debug-arm64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/421301
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
Even if the main program disables executable stacks, dlopen'ing a library that doesn't itself disable executable stack will switch the stack to executable. (Presumably in the name of compatability with GNU nested functions.)
Dart does not need executable stacks, and executable stacks are an additional attack surface.
TEST=readelf
Bug: b/229648756
Change-Id: Ia8c234ebc6178a26528d37b1359a10dd42039a9b
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/241540
Reviewed-by: Tess Strickland <sstrickl@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Implements a backend targeting RV32GC and RV64GC, based on Linux standardizing around GC. The assembler is written to make it easy to disable usage of C, but because the sizes of some instruction sequences are compile-time constants, an additional build configuration would need to be defined to make use of it.
The assembler and disassembler cover every RV32/64GC instruction. The simulator covers all instructions except accessing CSRs and the floating point state accessible through such, include accrued exceptions and dynamic rounding mode.
Quirks:
- RISC-V is a compare-and-branch architecture, but some existing "architecture-independent" parts of the Dart compiler assume a condition code architecture. To avoid rewriting these parts, we use a peephole in the assembler to map to compare-and-branch. See Assembler::BranchIf. Luckily nothing depended on taking multiple branches on the same condition code set.
- There are no hardware overflow checks, so we must use Hacker's Delight style software checks. Often these are very cheap: if the sign of one operand is known, a single branch is needed.
- The ranges of RISC-V branches and jumps are such that we use 3 levels of generation for forward branches, instead of the 2 levels of near and far branches used on ARM[64]. Nearly all code is handled by the first two levels with 20-bits of range, with enormous regex matchers triggering the third level that uses aupic+jalr to get 32-bits of range.
- For PC-relative calls in AOT, we always generate auipc+jalr pairs with 32-bits of range, so we never generate trampolines.
- Only a subset of registers are available in some compressed instructions, so we assign the most popular uses to these registers. In particular, THR, TMP[2], CODE and PP. This has the effect of assigning CODE and PP to volatile registers in the C calling convention, whereas they are assigned preserved registers on the other architectures. As on ARM64, PP is untagged; this is so short indices can be accessed with a compressed instruction.
- There are no push or pop instructions, so combining pushes and pops is preferred so we can update SP once.
- The C calling convention has a strongly aligned stack, but unlike on ARM64 we don't need to use an alternate stack pointer. The author ensured language was added to the RISC-V psABI making the OS responsible for realigning the stack pointer for signal handlers, allowing Dart to leave the stack pointer misaligned from the C calling convention's point of view until a foreign call.
- We don't bother with the link register tracking done on ARM[64]. Instead we make use of an alternate link register to avoid inline spilling in the write barrier.
Unimplemented:
- non-trivial FFI cases
- Compressed pointers - No intention to implement.
- Unboxed SIMD - We might make use of the V extension registers when the V extension is ratified.
- BigInt intrinsics
TEST=existing tests for IL level, new tests for assembler/disassembler/simulator
Bug: https://github.com/dart-lang/sdk/issues/38587
Bug: https://github.com/dart-lang/sdk/issues/48164
Change-Id: I991d1df4be5bf55efec5371b767b332d37dfa3e0
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/217289
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Ryan Macnak <rmacnak@google.com>
Before, picking a symbol name for a Code object when generating
assembly required adding a placeholder entry in the Dwarf object,
because the Dwarf object separately used a SnapshotTextObjectNamer
to determine the appropriate symbol name to use in relocations.
Instead, now the ImageWriter subclasses are entirely responsible for
creating the symbol name for a given Code object, and that name is
passed when adding the Code object to the Dwarf object.
For Elf output, we eagerly create symbols for sections and Code objects
again in the appropriate symbol table, and so looking up the start of a
given section or Code object can be done simply by asking for the
corresponding symbol's address, removing special case methods and other
workarounds.
We also clean up the generated static and dynamic symbol tables by
only generating local symbols for Code objects and BSS sections (these
are local in assembly output also), and only adding global symbols to
the dynamic symbol table.
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-debug-x64-try,vm-kernel-precomp-linux-debug-simarm_x64-try,vm-kernel-precomp-mac-release-simarm64-try,vm-kernel-mac-debug-x64-try,vm-kernel-win-debug-x64-try,vm-kernel-precomp-win-release-x64-try
Change-Id: Ie40c0c836681e98f345483136bb6860e5588ef30
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/166002
Reviewed-by: Martin Kustermann <kustermann@google.com>
For proper crashpad integration, we need to generate a build ID, as the
build ID generated by crashpad if there is not one will be a simple XOR
of the first text page, which rarely changes for Dart snapshots.
Assembly snapshots already have a build ID included by the assembler, so
we currently only do this for ELF snapshots.
Currently the build ID is a 128-bit hash value that is four separate
32-bit hash values concatenated together. Those hash values come from
the contents of the VM and isolate .text and .rodata sections.
This change also contains work to separate out the concepts of sections
and segments in the ELF builder. Now, consecutive allocated sections
with the same write and execute flags are combined into a single PT_LOAD
segment when possible, which reduces the padding needed to ensure that
segments start on page boundaries in ELF snapshots.
Bug: https://github.com/dart-lang/sdk/issues/42020
Change-Id: I42a837dae665a3902d881b8d151b49ede87d6c67
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-release-x64-try,vm-kernel-precomp-linux-product-x64-try,vm-kernel-precomp-linux-debug-x64-try,vm-kernel-precomp-linux-release-simarm_x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/150625
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Introduce an internal StreamingWriteStream wrapper and use that instead.
Also add more ELF-defined constants and use those appropriately.
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-release-simarm_x64-try,vm-kernel-precomp-linux-product-x64-try
Change-Id: I2bd34ebac60e789aea82c8f064c59351e2c55dd3
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/149291
Reviewed-by: Clement Skau <cskau@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
Removes an unnecessary change to the names of type testing stubs that
allowed name collisions to happen.
Old commit message:
The flag can be used when creating AOT snapshots. The resulting file can be
used with package:vm/dwarf/convert.dart to convert DWARF-based stack traces
to stack traces with function, file, and line number information.
Currently the saved file will be an ELF file with DWARF debugging information,
but this is subject to change in the future. To avoid being affected by any
changes in format, read the DWARF information from the file using
Dwarf.fromFile() in package:vm/dwarf/dwarf.dart.
Also adds --dwarf-stack-traces to the VM global flag list, so its value at
compilation will be read out of snapshots by the precompiled runtime.
Fixes https://github.com/dart-lang/sdk/issues/39512, which was due to
a missing compiler::target:: prefix for Instructions::HeaderSize() in
BlobImageWriter::WriteText().
Exposes the information suggested in
https://github.com/dart-lang/sdk/issues/39490 so that package:vm/dwarf
can appropriately convert absolute PC addresses to the correct virtual
address for a given DWARF line number program.
Bug: https://github.com/dart-lang/sdk/issues/35851
Change-Id: I6723449dceb0b89c054a1f1e70e7acd7749baf14
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-release-x64-try,vm-kernel-precomp-android-release-arm64-try,vm-kernel-precomp-mac-release-simarm64-try,vm-kernel-precomp-win-release-x64-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/128730
Reviewed-by: Martin Kustermann <kustermann@google.com>
REASON: the first commit in the chain contains a change that breaks internal
Flutter roll due to duplicated symbols for type testing stubs in assembly
snapshots.
Revert "[vm/compiler] Add --save-debugging-info flag to gen_snapshot."
This reverts commit c2b6c2e1db.
Revert "[vm] De-obfuscate function and file names in DWARF sections."
This reverts commit 4b8fd3c412.
Revert "[vm] Add docs for DWARF stack traces and related tools/libraries."
This reverts commit 5543e5ceb8.
Revert "[vm] Weaken dwarf_stack_trace expected stack trace."
This reverts commit 5824d3ed53.
Change-Id: I95ecd16fe493f61ae761dabfdd91a6ec0c2f0ca3
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/128776
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Siva Annamalai <asiva@google.com>
The flag can be used when creating AOT snapshots. The resulting file can be
used with package:vm/dwarf/convert.dart to convert DWARF-based stack traces
to stack traces with function, file, and line number information.
Currently the saved file will be an ELF file with DWARF debugging information,
but this is subject to change in the future. To avoid being affected by any
changes in format, read the DWARF information from the file using
Dwarf.fromFile() in package:vm/dwarf/dwarf.dart.
Also adds --dwarf-stack-traces to the VM global flag list, so its value at
compilation will be read out of snapshots by the precompiled runtime.
Fixes https://github.com/dart-lang/sdk/issues/39512, which was due to
a missing compiler::target:: prefix for Instructions::HeaderSize() in
BlobImageWriter::WriteText().
Exposes the information suggested in
https://github.com/dart-lang/sdk/issues/39490 so that package:vm/dwarf
can appropriately convert absolute PC addresses to the correct virtual
address for a given DWARF line number program.
Bug: https://github.com/dart-lang/sdk/issues/35851
Cq-Include-Trybots: luci.dart.try:vm-kernel-precomp-linux-debug-x64-try,vm-kernel-precomp-android-release-arm64-try,vm-kernel-precomp-mac-release-simarm_x64-try,vm-kernel-precomp-win-release-x64-try
Change-Id: I80d900fd9e5f1e2399ad3f2fd25c85131a7ea43c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/121857
Commit-Queue: Teagan Strickland <sstrickl@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Also switch some CQ bots using blobs to ELF. Once all embedders have migrated, we will remove blobs
support entirely.
Change-Id: Ie5e8c1187ad6c1af362b5715daafd3641bc8cc0e
Cq-Include-Trybots:luci.dart.try:vm-kernel-precomp-linux-debug-x64-try,vm-kernel-precomp-linux-release-simarm-try,vm-kernel-precomp-linux-release-simarm64-try,vm-kernel-precomp-bare-linux-release-simarm-try,vm-kernel-precomp-mac-debug-simarm_x64-try,vm-kernel-precomp-mac-release-simarm64-try,vm-kernel-precomp-win-release-x64-try,vm-kernel-precomp-android-release-arm-try
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/116620
Commit-Queue: Samir Jindel <sjindel@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Tess Strickland