Some larger projects were loading quite slowly with the current LLDB on macOS and macOS simulator builds. I did some instrument traces and found 3 main culprits:
- a LLDB timer that was put into a function that was called too often
- a std::set that was keeping track of the address of symbols that were already added
- a unnamed function generator in ObjectFile that was going slow due to allocations
In order to see this in action I ran the latest LLDB on a large application with many frameworks using the following method:
(lldb) script import time; start_time = time.perf_counter()
(lldb) file Large.app
(lldb) script print(time.perf_counter() - start_time)
I first range "sudo purge" to clear the system file caches to simulate a cold startup of the debugger, followed by two iterations with warm file caches.
Prior to this fix I was seeing the following timings:
17.68 (cold)
14.56 (warm 1)
14.52 (warm 2)
After this fix I was seeing:
11.32 (cold)
8.43 (warm 1)
8.49 (warm 2)
Differential Revision: https://reviews.llvm.org/D103504
This patch adds support for matching gtest's ASSERT_THAT, EXPECT_THAT, ON_CALL and EXPECT_CALL macros.
Reviewed By: ymandel, hokein
Differential Revision: https://reviews.llvm.org/D103195
When LLVM and MLIR are built as subprojects (via add_subdirectory),
the CMake configuration that indicates where the MLIR libraries are is
not necessarily in the same cmake/ directory as LLVM's configuration.
This patch removes that assumption about where MLIRConfig.cmake is
located.
(As an additional none, the %llvm_lib_dir substitution was never
defined, and so find_package(MLIR) in the build was succeeding for
other reasons.)
Reviewed By: stephenneuendorffer
Differential Revision: https://reviews.llvm.org/D103276
When checking for type properties we usually want to strip all kind of type
sugar from the type. For example, sugar like Clang's ElaboratedType or typedefs
rarely influence the fundamental behaviour of a type such as its byte size.
However we always need to preserve type sugar for everything else as it does
matter for users that their variable of type `size_t` instead of `unsigned long`
for example.
This patch fixes one such bug when trying to use the SBValue API to dereference
a type.
Reviewed By: werat, shafik
Differential Revision: https://reviews.llvm.org/D103532
This is a follow-up to D103280 that eases the use restrictions,
so we can handle the motivating case from:
https://llvm.org/PR50055
The loop code is adapted from similar use checks in
ExtendUsesToFormExtLoad() and SliceUpLoad(). I did not see an
easier way to filter out non-chain uses of load values.
Differential Revision: https://reviews.llvm.org/D103462
In 07ef8e679621 and 3ed9f6ebdeeb, `__nbuf` started to diverge from the amount
of space that was actually needed for the buffer. For 32-bit longs for example,
we allocate a buffer that is one larger than needed. Moreover, it is no longer
clear exactly where the extra +1 or +2 comes from - they're just numbers pulled
from thin air. This PR cleans up how `__nbuf` is calculated, and adds comments
to further clarify where each part comes from.
Specifically, it corrects the underestimation of the max size buffer needed
that the above two commits had to compensate for. The root cause looks to be
the use of signed type parameters to numeric_limits<>::digits. Since digits
only counts non-sign bits, the calculation was acting as though (for a signed
64-bit type) the longest value we would print was 2^63 in octal. However,
printing in octal treats values as unsigned, so it is actually 2^64. Thus,
using unsigned types and changing the final +2 to a +1 is probably a better
option.
Reviewed By: #libc, ldionne, Mordante
Differential Revision: https://reviews.llvm.org/D103339
Also adds support for live_support sections, no_dead_strip sections,
.no_dead_strip symbols.
Chromium Framework 345MB unstripped -> 250MB stripped
(vs 290MB unstripped -> 236M stripped with ld64).
Doing dead stripping is a bit faster than not, because so much less
data needs to be processed:
% ministat lld_*
x lld_nostrip.txt
+ lld_strip.txt
N Min Max Median Avg Stddev
x 10 3.929414 4.07692 4.0269079 4.0089678 0.044214794
+ 10 3.8129408 3.9025559 3.8670411 3.8642573 0.024779651
Difference at 95.0% confidence
-0.144711 +/- 0.0336749
-3.60967% +/- 0.839989%
(Student's t, pooled s = 0.0358398)
This interacts with many parts of the linker. I tried to add test coverage
for all added `isLive()` checks, so that some test will fail if any of them
is removed. I checked that the test expectations for the most part match
ld64's behavior (except for live-support-iterations.s, see the comment
in the test). Interacts with:
- debug info
- export tries
- import opcodes
- flags like -exported_symbol(s_list)
- -U / dynamic_lookup
- mod_init_funcs, mod_term_funcs
- weak symbol handling
- unwind info
- stubs
- map files
- -sectcreate
- undefined, dylib, common, defined (both absolute and normal) symbols
It's possible it interacts with more features I didn't think of,
of course.
I also did some manual testing:
- check-llvm check-clang check-lld work with lld with this patch
as host linker and -dead_strip enabled
- Chromium still starts
- Chromium's base_unittests still pass, including unwind tests
Implemenation-wise, this is InputSection-based, so it'll work for
object files with .subsections_via_symbols (which includes all
object files generated by clang). I first based this on the COFF
implementation, but later realized that things are more similar to ELF.
I think it'd be good to refactor MarkLive.cpp to look more like the ELF
part at some point, but I'd like to get a working state checked in first.
Mechanical parts:
- Rename canOmitFromOutput to wasCoalesced (no behavior change)
since it really is for weak coalesced symbols
- Add noDeadStrip to Defined, corresponding to N_NO_DEAD_STRIP
(`.no_dead_strip` in asm)
Fixes PR49276.
Differential Revision: https://reviews.llvm.org/D103324
During Loop Strength Reduce, if the terminating condition for the loop
is not immediately adjacent to the terminating branch and it has more
than one use, a clone of the condition will be created just before the
terminating branch and will be used as the branch condition. Currently,
whether the instructions are "immediately adjacent" is determined by
checking whether the next instruction after the condition is the
terminating branch; this is incorrect however, as the presence of a
debug intrinsic between the two will result in a change to the output.
This is fixed by using getNextNonDebugInstruction() instead.
Differential Revision: https://reviews.llvm.org/D103033
I forgot to move the message() call around as requested in D103428
before committing that change. Move it now.
Also, improve the ordinal uniq'ing comment. I hadn't realized that the
distinct-but-identical files happen with --reproduce and not in general.
No behavior change.
Differential Revision: https://reviews.llvm.org/D103522
Transfer the swiftasync attribute to the resume partial function according to
suspend.async specification. It's first argument denotes which argument is the
async context.
rdar://71499498
Differential Revision: https://reviews.llvm.org/D103285
This attribute applies to a using declaration, and permits importing a
declaration without knowing if that declaration exists. This is useful
for libc++ C wrapper headers that re-export declarations in std::, in
cases where the base C library doesn't provide all declarations.
This attribute was proposed in http://lists.llvm.org/pipermail/cfe-dev/2020-June/066038.html.
rdar://69313357
Differential Revision: https://reviews.llvm.org/D90188
This allows us to differentiate symbols from the system (e.g. system
includes or sysroot) differently than symbols defined in the user's
project, which can be used by editors to display them differently.
This is currently based on `FileCharacteristic`, but we can
consider alternatives such as `Sysroot` and file paths in the future.
Differential Revision: https://reviews.llvm.org/D101554
Add getDemandedBits method for uses so we can query demanded bits for each use. This can help getting better use information. For example, for the code below
define i32 @test_use(i32 %a) {
%1 = and i32 %a, -256
%2 = or i32 %1, 1
%3 = trunc i32 %2 to i8 (didn't optimize this to 1 for illustration purpose)
... some use of %3
ret %2
}
if we look at the demanded bit of %2 (which is all 32 bits because of the return), we would conclude that %a is used regardless of how its return is used. However, if we look at each use separately, we will see that the demanded bit of %2 in trunc only uses the lower 8 bits of %a which is redefined, therefore %a's usage depends on how the function return is used.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D97074
This patch uses the calculated maximum scalable VFs to build VPlans,
cost them and select a suitable scalable VF.
Reviewed By: paulwalker-arm
Differential Revision: https://reviews.llvm.org/D98722
We have been seeing this test fail intermittently on our
2 stage AArch64 bot.
As far back as https://lab.llvm.org/buildbot/#/builders/53/builds/2694
Likely due to a lack of resources at certain times on the
shared machine. Up the time limit to give us some more room.
(this limit only applies to the watchdog thread, so if the
test passes then it won't take 20s)
A recent patch:
https://reviews.llvm.org/rGe0921655b1ff8d4ba7c14be59252fe05b705920e
changed clangs AIX bitfield handling to use 4-byte bitfield containers,
matching XLs behavior. This change triggers static assert failures when
bootstrapping. Change the macro we check to enable bitfield packing on
AIX to `__clang__` which is defined by both xlclang and clang.
Differential Revision: https://reviews.llvm.org/D103474
llvm::getLoadStoreType was added recently and has the same implementation
as 'getMemInstValueType' in LoopVectorize.cpp. Since there is no
value in having two implementations, this patch removes the custom LV
implementation in favor of the generic one defined in Instructions.h.
Since https://reviews.llvm.org/D102046 some tests have
been falling back to fast unwinding on our Thumb bot.
This fails because fast unwinding does not work on Thumb.
By adding the extra information we ensure this does not happen
during testing, but the built library can still fast unwind
as a last resort.
Since there are some situations it can work in, like if
eveything is built with clang. During testing we've got gcc
built system libs and clang built tests.
The same change was made for sanitizer-common in
https://reviews.llvm.org/D96337.
Reviewed By: zatrazz
Differential Revision: https://reviews.llvm.org/D103463
This is a pre-commit of a test case D99439 which is a patch that
updates @llvm.powi to handle different int sizes for the exponent.
Problem is that @llvm.powi is used as an IR construct that maps
to RT libcalls to __powi* functions, and those lib functions depend
on sizeof(int) to use correct type for the exponent.
The test cases show that we use i32 for the powi expenent, which
later would result in wrong type being used in libcalls (miscompile).
But there are also a couple of the negative test cases that show
that we rewrite into using powi when having a uitofp conversion
from i16, which would be wrong when doing the libcall as an
"unsigned int" isn't guaranteed to fit inside the "int" argument
in the called libcall function.
Differential Revision: https://reviews.llvm.org/D102919
Use RuntimeLibcalls to get a common way to pick correct RTLIB::POWI_*
libcall for a given value type.
This includes a small refactoring of ExpandFPLibCall and
ExpandArgFPLibCall in SelectionDAGLegalize to share a bit of code,
plus adding an ExpandFPLibCall version that can be called directly
when expanding FPOWI/STRICT_FPOWI to ensure that we actually use
the same RTLIB::Libcall when expanding the libcall as we used when
checking the legality of such a call by doing a getLibcallName check.
Differential Revision: https://reviews.llvm.org/D103050
The FPOWI DAG node is normally lowered to a libcall to one of the
RTLIB::POWI* runtime functions and the exponent should normally
have a type matching sizeof(int) when making the call. Thus,
type promotion of the exponent could lead to an FPOWI with a type
for the second operand that would be incorrect when doing the
libcall (a situation which would be hard to detect post-legalization
if we allow such FPOWI nodes).
This patch is changing DAGTypeLegalizer::PromoteIntOp_FPOWI to
do the rewrite into a libcall directly instead of promoting the
operand. This way we can check that the exponent is smaller than
sizeof(int) and we can let TargetLowering handle promotion as
part of making the libcall. It could be noticed here that makeLibCall
has some knowledge about targets such as 64-bit RISCV, for which the
libcall argument should be extended to a type larger than sizeof(int).
Differential Revision: https://reviews.llvm.org/D102950
When rewriting
powf(2.0, itofp(x)) -> ldexpf(1.0, x)
exp2(sitofp(x)) -> ldexp(1.0, sext(x))
exp2(uitofp(x)) -> ldexp(1.0, zext(x))
the wrong type was used for the second argument in the ldexp/ldexpf
libc call, for target architectures with 16 bit "int" type.
The transform incorrectly used a bitcasted function pointer with
a 32-bit argument when emitting the ldexp/ldexpf call for such
targets.
The fault is solved by using the correct function prototype
in the call, by asking TargetLibraryInfo about the size of "int".
TargetLibraryInfo by default derives the size of the int type by
assuming that it is 16 bits for 16-bit architectures, and
32 bits otherwise. If this isn't true for a target it should be
possible to override that default in the TargetLibraryInfo
initializer.
Differential Revision: https://reviews.llvm.org/D99438
RVV vectors must be aligned to their element types, so anything less is
unaligned.
For regular loads and stores, our custom-lowering of fixed-length
vectors meant that we opted out of LegalizeDAG's built-in unaligned
expansion. This patch adds that logic in to our custom lower function.
For masked intrinsics, we declare that anything unaligned is not legal,
leaving the ScalarizeMaskedMemIntrin pass to do the expansion for us.
Note that neither of these methods can handle the expansion of
scalable-vector memory ops, so those cases are left alone by this patch.
Scalable loads and stores already go through expansion by default but
hit an assertion, and scalable masked intrinsics will silently generate
incorrect code. It may be prudent to return an error in both of these
cases.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D102493
