Strip tools cause a few symbols in .dynsym to have bad section index.
This update safely keeps such broken symbols intact.
Test Plan:
```
ninja check-bolt
```
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D130073
There should not be a end of child mark before DW_AT_ranges, removed it and fixed unit offset.
Reviewed By: ayermolo
Differential Revision: https://reviews.llvm.org/D130335
The latest perf tool can return non-empty buffer when executing
buildid-list command, even when perf.data was recorded with -B flag.
Some binaries will be listed without the ID, while others may have a
recorded ID. Allow invalid entires on the input, while checking the
valid ones for the match.
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D130223
First of all, `LLVM_TOOLS_INSTALL_DIR` put there breaks our NixOS
builds, because `LLVM_TOOLS_INSTALL_DIR` defined the same as
`CMAKE_INSTALL_BINDIR` becomes an *absolute* path, and then when
downstream projects try to install there too this breaks because our
builds always install to fresh directories for isolation's sake.
Second of all, note that `LLVM_TOOLS_INSTALL_DIR` stands out against the
other specially crafted `LLVM_CONFIG_*` variables substituted in
`llvm/cmake/modules/LLVMConfig.cmake.in`.
@beanz added it in d0e1c2a550 to fix a
dangling reference in `AddLLVM`, but I am suspicious of how this
variable doesn't follow the pattern.
Those other ones are carefully made to be build-time vs install-time
variables depending on which `LLVMConfig.cmake` is being generated, are
carefully made relative as appropriate, etc. etc. For my NixOS use-case
they are also fine because they are never used as downstream install
variables, only for reading not writing.
To avoid the problems I face, and restore symmetry, I deleted the
exported and arranged to have many `${project}_TOOLS_INSTALL_DIR`s.
`AddLLVM` now instead expects each project to define its own, and they
do so based on `CMAKE_INSTALL_BINDIR`. `LLVMConfig` still exports
`LLVM_TOOLS_BINARY_DIR` which is the location for the tools defined in
the usual way, matching the other remaining exported variables.
For the `AddLLVM` changes, I tried to copy the existing pattern of
internal vs non-internal or for LLVM vs for downstream function/macro
names, but it would good to confirm I did that correctly.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D117977
Some unit tests has incorrect DW_AT_type offset since they are manual crafted, fix them to the correct offset.
Reviewed By: Amir, ayermolo
Differential Revision: https://reviews.llvm.org/D129828
This patch adds a dedicated class to keep track of each function's
layout. It also lays the groundwork for splitting functions into
multiple fragments (as opposed to a strict hot/cold split).
Reviewed By: maksfb
Differential Revision: https://reviews.llvm.org/D129518
Add a test for `-icp-inline` knob, which ensures that ICP is only performed for
functions that can be subsequently inlined.
Reviewed By: rafauler
Differential Revision: https://reviews.llvm.org/D129803
We previously support split jump table, where some jump table entries
target different fragments of same function. In this fix, we provide
support for another type of intra-indirect transfer: landing pad.
When C++ exception handling is used, compiler emits .gcc_except_table
that describes the location of catch block (landing pad) for specific
range that potentially invokes a throw(). Normally landing pads reside
in the function, but with -fsplit-machine-functions, landing pads can
be moved to another fragment. The intuition is, landing pads are rarely
executed, so compiler can move them to .cold section.
This update will mark all fragments that have landing pad to another
fragment as non-simple, and later propagate non-simple to all related
fragments.
This update also includes one manual test case: split-landing-pad.s
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D128561
As we are moving towards support for multiple fragments, loops that
iterate over all basic blocks of a function, but do not depend on the
order of basic blocks in the final layout, should iterate over binary
functions directly, rather than the layout.
Eventually, all loops using the layout list should either iterate over
the function, or be aware of multiple layouts. This patch replaces
references to binary function's block layout with the binary function
itself where only little code changes are necessary.
Reviewed By: maksfb
Differential Revision: https://reviews.llvm.org/D129585
There are two assumptions regarding jump table:
(a) It is accessed by only one fragment, say, Parent
(b) All entries target instructions in Parent
For (a), BOLT stores jump table entries as relative offset to Parent.
For (b), BOLT treats jump table entries target somewhere out of Parent
as INVALID_OFFSET, including fragment of same split function.
In this update, we extend (a) and (b) to include fragment of same split
functinon. For (a), we store jump table entries in absolute offset
instead. In addition, jump table will store all fragments that access
it. A fragment uses this information to only create label for jump table
entries that target to that fragment.
For (b), using absolute offset allows jump table entries to target
fragments of same split function, i.e., extend support for split jump
table. This can be done using relocation (fragment start/size) and
fragment detection heuristics (e.g., using symbol name pattern for
non-stripped binaries).
For jump table targets that can only be reached by one fragment, we
mark them as local label; otherwise, they would be the secondary
function entry to the target fragment.
Test Plan
```
ninja check-bolt
```
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D128474
The gold linker veneers are written between functions without symbols,
so we to handle it specially in BOLT.
Vladislav Khmelevsky,
Advanced Software Technology Lab, Huawei
Differential Revision: https://reviews.llvm.org/D129260
Since we now have +all feature for AArch64 disassembler, we can use it
in BOLT and allow it to disassemble all ARM instructions supported by LLVM.
Reviewed by: rafauler
Differential Revision: https://reviews.llvm.org/D129139
Add -experimental-shrink-wrapping flag to control when we
want to move callee-saved registers even when addresses of the stack
frame are captured and used in pointer arithmetic, making it more
challenging to do alias analysis to prove that we do not access
optimized stack positions. This alias analysis is not yet implemented,
hence, it is experimental. In practice, though, no compiler would emit
code to do pointer arithmetic to access a saved callee-saved register
unless there is a memory bug or we are failing to identify a
callee-saved reg, so I'm not sure how useful it would be to formally
prove that.
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D126115
Change shrink-wrapping to try a priority list of save
positions, instead of trying the best one and giving up if it doesn't
work. This also increases coverage.
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D126114
Add the option to run -equalize-bb-counts before shrink
wrapping to avoid unnecessarily optimizing some CFGs where profile is
inaccurate but we can prove two blocks have the same frequency.
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D126113
Refactor isStackAccess() to reflect updates by D126116. Now we only
handle simple stack accesses and delegate the rest of the cases to
getMemDataSize.
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D126112
Change how function score is calculated and provide more
detailed statistics when reporting back frame optimizer and shrink
wrapping results. In this new statistics, we provide dynamic coverage
numbers. The main metric for shrink wrapping is the number of executed
stores that were saved because of shrink wrapping (push instructions
that were either entirely moved away from the hot block or converted
to a stack adjustment instruction). There is still a number of reduced
load instructions (pop) that we are not counting at the moment. Also
update alloc combiner to report dynamic numbers, as well as frame
optimizer.
For debugging purposes, we also include a list of top 10 functions
optimized by shrink wrapping. These changes are aimed at better
understanding the impact of shrink wrapping in a given binary.
We also remove an assertion in dataflow analysis to do not choke on
empty functions (which makes no sense).
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D126111
There is a post-processing in ext-tsp block reordering that merges some blocks
into chains. This allows to maintain the original block order in the absense of
profile data and can be beneficial for code size (when fallthroughs are merged).
In the earlier version we could merge hot and cold (with zero execution count)
chains, that later were split by SplitFunction.cpp (when split-all-cold=1). The
diff eliminates the redundant merging.
It is unlikely the change will affect the performance of a binary in a
measurable way, as it is mostly operates with cold basic blocks. However, after
the diff the impact of split-all-cold is almost negligible and we can avoid the
extra function splitting.
Measuring on the clang binary (negative is good, positive is a regression):
**clang12**
benchmark1: `0.0253`
benchmark2: `-0.1843`
benchmark3: `0.3234`
benchmark4: `0.0333`
**clang10**
benchmark1 `-0.2517`
benchmark2 `-0.3703`
benchmark3 `-0.1186`
benchmark4 `-0.3822`
**clang7**
benchmark1 `0.2526`
benchmark2 `0.0500`
benchmark3 `0.3024`
benchmark4 `-0.0489`
**Overall**: `-0.0671 ± 0.1172` (insignificant)
Reviewed By: maksfb
Differential Revision: https://reviews.llvm.org/D129397
Summary:
Introduce NeverAlign fragment type.
The intended usage of this fragment is to insert it before a pair of
macro-op fusion eligible instructions. NeverAlign fragment ensures that
the next fragment (first instruction in the pair) does not end at a
given alignment boundary by emitting a minimal size nop if necessary.
In effect, it ensures that a pair of macro-fusible instructions is not
split by a given alignment boundary, which is a precondition for
macro-op fusion in modern Intel Cores (64B = cache line size, see Intel
Architecture Optimization Reference Manual, 2.3.2.1 Legacy Decode
Pipeline: Macro-Fusion).
This patch introduces functionality used by BOLT when emitting code with
MacroFusion alignment already in place.
The use case is different from BoundaryAlign and instruction bundling:
- BoundaryAlign can be extended to perform the desired alignment for the
first instruction in the macro-op fusion pair (D101817). However, this
approach has higher overhead due to reliance on relaxation as
BoundaryAlign requires in the general case - see
https://reviews.llvm.org/D97982#2710638.
- Instruction bundling: the intent of NeverAlign fragment is to prevent
the first instruction in a pair ending at a given alignment boundary, by
inserting at most one minimum size nop. It's OK if either instruction
crosses the cache line. Padding both instructions using bundles to not
cross the alignment boundary would result in excessive padding. There's
no straightforward way to request instruction bundling to avoid a given
end alignment for the first instruction in the bundle.
LLVM: https://reviews.llvm.org/D97982
Manual rebase conflict history:
https://phabricator.intern.facebook.com/D30142613
Test Plan: sandcastle
Reviewers: #llvm-bolt
Subscribers: phabricatorlinter
Differential Revision: https://phabricator.intern.facebook.com/D31361547
`__bolt_instr_data_dump()` does not lock the hash tables when iterating
over them, so the iteration can happen concurrently with a modification
done in another thread, when the table is in an inconsistent state. This
also has been observed in practice, when it caused a segmentation fault.
We fix this by locking hash tables during iteration. This is done by taking
the lock in `forEachElement()`.
The only other site of iteration, `resetCounters()`, has been correctly
locking the table even before this patch. This patch removes its `Lock`
because the lock is now taken in the inner `forEachElement()`.
Reviewed By: maksfb, yota9
Differential Revision: https://reviews.llvm.org/D129089
Compiler can generate calls to some functions implicitly, even under
constraints of freestanding environment. Make sure these functions are
available in our runtime objects.
Fixes test failures on some systems after https://reviews.llvm.org/D128960.
Reviewed By: yota9
Differential Revision: https://reviews.llvm.org/D129168
Changed acquire implemetaion to __atomic_test_and_set() and release
to __atomic_clear() so it eliminates inline asm usage and is arch
independent.
Elvina Yakubova,
Advanced Software Technology Lab, Huawei
Reviewers: yota9, maksfb, rafauler
Differential Revision: https://reviews.llvm.org/D129162
If BOLT instrumentation runtime uses XMM registers, it can interfere
with the user program causing crashes and unexpected behavior. This
happens as the instrumentation code preserves general purpose registers
only.
Build BOLT instrumentation runtime with "-mno-sse".
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D128960
Added support for mixing monolithic DWARF5 with legacy DWARF, and monolithic legacy and DWARF5 split dwarf.
Reviewed By: maksfb
Differential Revision: https://reviews.llvm.org/D128232
Generate INSTRINFO_OPERAND_TYPE table in X86GenInstrInfo.inc.
This diff adds support for instructions that were previously reported as having
memory access size 0. It replaces the heuristic of looking at instruction
register width to determine memory access width by instead checking the memory
operand type using tablegen-provided tables.
Reviewed By: skan
Differential Revision: https://reviews.llvm.org/D126116
Don't dump dot CFG graph for functions that should not be printed.
Reviewed By: rafauler, maksfb
Differential Revision: https://reviews.llvm.org/D128699
When SplitFunctions pass adds a trampoline code for exception landing
pads (limited to shared objects), it may increase the size of the hot
fragment making it larger than the whole function pre-split. When this
happens, the pass reverts the splitting action by restoring the original
block order and marking all blocks hot.
However, if createEHTrampolines() added new blocks to the CFG and
modified invoke instructions, simply restoring the original block layout
will not suffice as the new CFG has more blocks.
For proper backout of the split, modify the original layout by merging
in trampoline blocks immediately before their matching targets. As a
result, the number of blocks increases, but the number of instructions
and the function size remains the same as pre-split.
Add an assertion for the number of blocks when updating a function
layout.
Reviewed By: rafauler
Differential Revision: https://reviews.llvm.org/D128696
For test purposes, we want to split functions at a random split point
to be able to test different layouts without relying on the profile.
This patch introduces an option, that randomly chooses a split point
to partition blocks of a function into hot and cold regions.
Reviewed By: Amir, yota9
Differential Revision: https://reviews.llvm.org/D128773
This reverts commit 425dda76e9.
This commit is currently causing BOLT to crash in one of our
binaries and needs a bit more checking to make sure it is safe
to land.
The gold linker veneers are written between functions without symbols,
so we to handle it specially in BOLT.
Vladislav Khmelevsky,
Advanced Software Technology Lab, Huawei
Differential Revision: https://reviews.llvm.org/D128082
ICP peel for inline mode only makes sense for calls, not jump tables.
Plus, add a check that the Target BinaryFunction is found.
Reviewed By: rafauler
Differential Revision: https://reviews.llvm.org/D128404
The SplitFunctions pass does not distinguish between various splitting
modes anymore. This change updates the command line interface to
reflect this behavior by deprecating values passed to the
--split-function option.
Reviewed By: rafauler
Differential Revision: https://reviews.llvm.org/D128558
DWARF 5 added two new attributes DW_AT_call_pc and DW_AT_call_return_pc.
Adding support for them.
Reviewed By: maksfb
Differential Revision: https://reviews.llvm.org/D128526
Add functionality to allow splitting code with C++ exceptions in shared
libraries and PIEs. To overcome a limitation in exception ranges format,
for functions with fragments spanning multiple sections, add trampoline
landing pads in the same section as the corresponding throwing range.
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D127936
Allow cold fragment to get new address.
Our previous assumption is that a fragment (.cold) is only reached
through the main fragment of same function. In addition, .cold fragment
must be reached through either (a) direct transfer, or (b) split jump
table. For (a), we perform a simple fix-up. For (b), we currently mark
all relevant fragments as non-simple. Therefore, there is no need to
get new address for .cold fragment.
This is not always the case, as function entry can be rarely executed,
and is placed in .text.cold segment. Essentially we cannot tell which
the source-level function entry is based on hot and cold segments,
so we must treat each fragment a function on its own. Therfore, we
remove the assertion that a function entry cannot be cold fragment.
Test Plan:
```
ninja check-bolt
```
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D128111
Resolve a crash related to split functions
Due to split function optimization, a function can be divided to two
fragments, and both fragments can access same jump table. This
violates the assumption that a jump table can only have one parent
function, which causes a crash during instrumentation.
We want to support the case: different functions cannot access same
jump tables, but different fragments of same function can!
As all fragments are from same function, we point JT::Parent to one
specific fragment. Right now it is the first disassembled fragment, but
we can point it to the function's main fragment later.
Functions are disassembled sequentially. Previously, at the end of
processing a function, JT::OffsetEntries is cleared, so other fragment
can no longer reuse JT::OffsetEntries. To extend the support for split
function, we only clear JT::OffsetEntries after all functions are
disassembled.
Let say A.hot and A.cold access JT of three targets {X, Y, Z}, where
X and Y are in A.hot, and Z is in A.cold. Suppose that A.hot is
disassembled first, JT::OffsetEntries = {X',Y',INVALID_OFFSET}. When
A.cold is disassembled, it cannot reuse JT::OffsetEntries above due to
different fragment start. A simple solution:
A.hot = {X',Y',INVALID_OFFSET}
A.cold = {INVALID_OFFSET, INVALID_OFFSET, INVALID_OFFSET}
We update the assertion to allow different fragments of same function
to get the same JumpTable object.
Potential improvements:
A.hot = {X',Y',INVALID_OFFSET}
A.cold = {INVALID_OFFSET, INVALID_OFFSET, Z'}
The main issue is A.hot and A.cold have separate CFGs, thus jump table
targets are still constrained within fragment bounds.
Future improvements:
A.hot = {X, Y, Z}
A.cold = {X, Y, Z}
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D127924
Huan Nguyen