This patch adds codegen support for the saving/restoring
V8-V23 for functions specified with the aarch64_vector_pcs
calling convention attribute, as added in patch D51477.
Reviewers: t.p.northover, gberry, thegameg, rengolin, javed.absar, MatzeB
Reviewed By: thegameg
Differential Revision: https://reviews.llvm.org/D51479
llvm-svn: 342049
Since the outliner is a module pass, it doesn't get codegen size remarks like
the other codegen passes do. This adds size remarks *to* the outliner.
This is kind of a workaround, so it's peppered with FIXMEs; size remarks
really ought to not ever be handled by the pass itself. However, since the
outliner is the only "MachineModulePass", this works for now. Since the
entire purpose of the MachineOutliner is to produce code size savings, it
really ought to be included in codgen size remarks.
If we ever go ahead and make a MachineModulePass (say, something similar to
MachineFunctionPass), then all of this ought to be moved there.
llvm-svn: 342009
Summary:
Reserving registers x1-7 is used to support CONFIG_ARM64_LSE_ATOMICS in Linux kernel. This change adds support for reserving registers x1 through x7.
Reviewers: javed.absar, phosek, srhines, nickdesaulniers, efriedma
Reviewed By: nickdesaulniers, efriedma
Subscribers: niravd, jfb, manojgupta, nickdesaulniers, jyknight, efriedma, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D48580
llvm-svn: 341706
Summary:
I added a few ARM64 memset codegen tests in r341406 and r341493, and annotated
where the generated code was bad. This patch fixes the majority of the issues by
requesting that a 2xi64 vector be used for memset of 32 bytes and above.
The patch leaves the former request for f128 unchanged, despite f128
materialization being suboptimal: doing otherwise runs into other asserts in
isel and makes this patch too broad.
This patch hides the issue that was present in bzero_40_stack and bzero_72_stack
because the code now generates in a better order which doesn't have the store
offset issue. I'm not aware of that issue appearing elsewhere at the moment.
<rdar://problem/44157755>
Reviewers: t.p.northover, MatzeB, javed.absar
Subscribers: eraman, kristof.beyls, chrib, dexonsmith, llvm-commits
Differential Revision: https://reviews.llvm.org/D51706
llvm-svn: 341558
I'm looking at some codegen optimization in this area and want to make sure I understand the current codegen and don't regress it. This patch further expands the tests (which I already expanded in r341406) to capture more of the current code generation when it comes to stack-based small non-zero memset on arm64. This patch annotates some potential fixes.
llvm-svn: 341493
This was proposed as an IR transform in D49306, but it was not clearly justifiable as a canonicalization.
Here, we only do the transform when the target tells us that sqrt can be lowered with inline code.
This is the basic case. Some potential enhancements are in the TODO comments:
1. Generalize the transform for other exponents (allow more than 2 sqrt calcs if that's really cheaper).
2. If we have less fast-math-flags, generate code to avoid -0.0 and/or INF.
3. Allow the transform when optimizing/minimizing size (might require a target hook to get that right).
Note that by default, x86 converts single-precision sqrt calcs into sqrt reciprocal estimate with
refinement. That codegen is controlled by CPU attributes and can be manually overridden. We have plenty
of test coverage for that already, so I didn't bother to include extra testing for that here. AArch uses
its full-precision ops in all cases (not sure if that's the intended behavior or not, but that should
also be covered by existing tests).
Differential Revision: https://reviews.llvm.org/D51630
llvm-svn: 341481
Reland r341269. Use std::stable_sort when sorting constant condidates.
Reverting commit, r341365:
Revert r341269: [Constant Hoisting] Hoisting Constant GEP Expressions
One of the tests is failing 50% of the time when expensive checks are
enabled. Not sure how deep the problem is so just reverting while the
author can investigate so that the bots stop repeatedly failing and
blaming things incorrectly. Will respond with details on the original
commit.
Original commit, r341269:
[Constant Hoisting] Hoisting Constant GEP Expressions
Leverage existing logic in constant hoisting pass to transform constant GEP
expressions sharing the same base global variable. Multi-dimensional GEPs are
rewritten into single-dimensional GEPs.
https://reviews.llvm.org/D51396
Differential Revision: https://reviews.llvm.org/D51654
llvm-svn: 341417
I'm looking at some codegen optimization in this area and want to make sure I understand the current codegen and don't regress it. This patch simply expands the two existing tests to capture more of the current code generation when it comes to heap-based and stack-based small memset on arm64. The tested code is already pretty good, notably when it comes to using STP, FP stores, FP immediate generation, and folding one of the stores into a stack spill when possible. The uses of STUR could be improved, and some more pairing could occur. Straying from bzero patterns currently yield suboptimal code, and I expect a variety of small changes could make things way better.
llvm-svn: 341406
The runtime pseudo relocations can't handle the AArch64 format PC
relative addressing in adrp+add/ldr pairs. By using stubs, the potentially
dllimported addresses can be touched up by the runtime pseudo relocation
framework.
Differential Revision: https://reviews.llvm.org/D51452
llvm-svn: 341401
One of the tests is failing 50% of the time when expensive checks are
enabled. Not sure how deep the problem is so just reverting while the
author can investigate so that the bots stop repeatedly failing and
blaming things incorrectly. Will respond with details on the original
commit.
llvm-svn: 341365
For instructions that spill/fill to and from multiple frame-indices
in a single instruction, hasStoreToStackSlot and hasLoadFromStackSlot
should return an array of accesses, rather than just the first encounter
of such an access.
This better describes FI accesses for AArch64 (paired) LDP/STP
instructions.
Reviewers: t.p.northover, gberry, thegameg, rengolin, javed.absar, MatzeB
Reviewed By: MatzeB
Differential Revision: https://reviews.llvm.org/D51537
llvm-svn: 341301
Summary:
A follow-up for D49266 / rL337166 + D49497 / rL338044.
This is still the same pattern to check for the [lack of]
signed truncation, but in this case the constants and the predicate
are negated.
https://rise4fun.com/Alive/BDVhttps://rise4fun.com/Alive/n7Z
Reviewers: spatel, craig.topper, RKSimon, javed.absar, efriedma, dmgreen
Reviewed By: spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D51532
llvm-svn: 341287
Leverage existing logic in constant hoisting pass to transform constant GEP
expressions sharing the same base global variable. Multi-dimensional GEPs are
rewritten into single-dimensional GEPs.
Differential Revision: https://reviews.llvm.org/D51396
llvm-svn: 341269
Summary:
This is a continuation of https://reviews.llvm.org/D49727
Below the original text, current changes in the comments:
Currently, in line with GCC, when specifying reserved registers like sp or pc on an inline asm() clobber list, we don't always preserve the original value across the statement. And in general, overwriting reserved registers can have surprising results.
For example:
extern int bar(int[]);
int foo(int i) {
int a[i]; // VLA
asm volatile(
"mov r7, #1"
:
:
: "r7"
);
return 1 + bar(a);
}
Compiled for thumb, this gives:
$ clang --target=arm-arm-none-eabi -march=armv7a -c test.c -o - -S -O1 -mthumb
...
foo:
.fnstart
@ %bb.0: @ %entry
.save {r4, r5, r6, r7, lr}
push {r4, r5, r6, r7, lr}
.setfp r7, sp, #12
add r7, sp, #12
.pad #4
sub sp, #4
movs r1, #7
add.w r0, r1, r0, lsl #2
bic r0, r0, #7
sub.w r0, sp, r0
mov sp, r0
@APP
mov.w r7, #1
@NO_APP
bl bar
adds r0, #1
sub.w r4, r7, #12
mov sp, r4
pop {r4, r5, r6, r7, pc}
...
r7 is used as the frame pointer for thumb targets, and this function needs to restore the SP from the FP because of the variable-length stack allocation a. r7 is clobbered by the inline assembly (and r7 is included in the clobber list), but LLVM does not preserve the value of the frame pointer across the assembly block.
This type of behavior is similar to GCC's and has been discussed on the bugtracker: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=11807 . No consensus seemed to have been reached on the way forward. Clang behavior has briefly been discussed on the CFE mailing (starting here: http://lists.llvm.org/pipermail/cfe-dev/2018-July/058392.html). I've opted for following Eli Friedman's advice to print warnings when there are reserved registers on the clobber list so as not to diverge from GCC behavior for now.
The patch uses MachineRegisterInfo's target-specific knowledge of reserved registers, just before we convert the inline asm string in the AsmPrinter.
If we find a reserved register, we print a warning:
repro.c:6:7: warning: inline asm clobber list contains reserved registers: R7 [-Winline-asm]
"mov r7, #1"
^
Reviewers: efriedma, olista01, javed.absar
Reviewed By: efriedma
Subscribers: eraman, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D51165
llvm-svn: 341062
Providing that the load is known to be 4 byte aligned, we can optimise a
ldr(adr address) to just ldr address.
Differential Revision: https://reviews.llvm.org/D51030
llvm-svn: 341058
Hacker's Delight 10-17: when C is constant,
the result of X % C == 0 can be computed more cheaply
without actually calculating the remainder.
The motivation is discussed here:
https://bugs.llvm.org/show_bug.cgi?id=35479.
Patch by: hermord (Dmytro Shynkevych)!
For https://reviews.llvm.org/D50222
llvm-svn: 341047
Summary:
Global variables that are external and zero initialized are
supposed to be merged with global variables in the bss section
rather than the data section.
Reviewers: efriedma, rengolin, t.p.northover, javed.absar, asl, john.brawn, pcc
Reviewed By: efriedma
Subscribers: dmgreen, llvm-commits
Differential Revision: https://reviews.llvm.org/D51379
llvm-svn: 341008
https://reviews.llvm.org/D51197
Currently, IRTranslator (and GISel) seems to be arbitrarily picking
which overflow intrinsics get mapped into opcodes which either have a
carry as an input or not.
For intrinsics such as Intrinsic::uadd_with_overflow, translate it to an
opcode (G_UADDO) which doesn't have any carry inputs (similar to LLVM
IR).
This patch adds 4 missing opcodes for completeness - G_UADDO, G_USUBO,
G_SSUBE and G_SADDE.
llvm-svn: 340865
This solves the motivating case from:
https://bugs.llvm.org/show_bug.cgi?id=38527
If we are legalizing an FP vector op that maps to 1 of the LLVM intrinsics that mimic libm calls,
but we're going to end up with scalar libcalls for that vector type anyway, then we should unroll
the vector op into scalars before widening. This avoids libcalls because we've lost the knowledge
that some of the scalar elements are undef.
Differential Revision: https://reviews.llvm.org/D50791
llvm-svn: 340469
This adds the plumbing for the Tiny code model for the AArch64 backend. This,
instead of loading addresses through the normal ADRP;ADD pair used in the Small
model, uses a single ADR. The 21 bit range of an ADR means that the code and
its statically defined symbols need to be within 1MB of each other.
This makes it mostly interesting for embedded applications where we want to fit
as much as we can in as small a space as possible.
Differential Revision: https://reviews.llvm.org/D49673
llvm-svn: 340397
This reverts commit 7debc334e6421bb5251ef8f18e97166dfc7dd787.
I missed updating legalizer-info-validation.mir as I had assertions
turned off in my build and that specific test requires asserts. Fixed it
now.
llvm-svn: 340197
- Generate pointer authentication instructions
- The functions instrumented depend on function attribtues:
all (all functions instrumentent)
non-leaf (only those that spill LR)
none
- Function epilogues sign the LR before spilling to the stack and authenticate
the LR once restored
- If the target is v8.3a or greater than can use the combined authenticate and
return instruction
Differential revision: https://reviews.llvm.org/D49793
llvm-svn: 340018
https://reviews.llvm.org/D50401
Add opcodes for llvm.intrinsic.trunc, round, and update the IRTranslator
for the same.
Reviewed by: dsanders.
llvm-svn: 339977
There is no way in the universe, that doing a full-width division in
software will be faster than doing overflowing multiplication in
software in the first place, especially given that this same full-width
multiplication needs to be done anyway.
This patch replaces the previous implementation with a direct lowering
into an overflowing multiplication algorithm based on half-width
operations.
Correctness of the algorithm was verified by exhaustively checking the
output of this algorithm for overflowing multiplication of 16 bit
integers against an obviously correct widening multiplication. Baring
any oversights introduced by porting the algorithm to DAG, confidence in
correctness of this algorithm is extremely high.
Following table shows the change in both t = runtime and s = space. The
change is expressed as a multiplier of original, so anything under 1 is
“better” and anything above 1 is worse.
+-------+-----------+-----------+-------------+-------------+
| Arch | u64*u64 t | u64*u64 s | u128*u128 t | u128*u128 s |
+-------+-----------+-----------+-------------+-------------+
| X64 | - | - | ~0.5 | ~0.64 |
| i686 | ~0.5 | ~0.6666 | ~0.05 | ~0.9 |
| armv7 | - | ~0.75 | - | ~1.4 |
+-------+-----------+-----------+-------------+-------------+
Performance numbers have been collected by running overflowing
multiplication in a loop under `perf` on two x86_64 (one Intel Haswell,
other AMD Ryzen) based machines. Size numbers have been collected by
looking at the size of function containing an overflowing multiply in
a loop.
All in all, it can be seen that both performance and size has improved
except in the case of armv7 where code size has regressed for 128-bit
multiply. u128*u128 overflowing multiply on 32-bit platforms seem to
benefit from this change a lot, taking only 5% of the time compared to
original algorithm to calculate the same thing.
The final benefit of this change is that LLVM is now capable of lowering
the overflowing unsigned multiply for integers of any bit-width as long
as the target is capable of lowering regular multiplication for the same
bit-width. Previously, 128-bit overflowing multiply was the widest
possible.
Patch by Simonas Kazlauskas!
Differential Revision: https://reviews.llvm.org/D50310
llvm-svn: 339922
These correspond to the x86 tests added with rL339790 / rL339791, but I widened
the non-fsin tests to v3f32 to show the problem because AArch supports v2f32 ops.
llvm-svn: 339793
Similar to rL337966 - if the DAGCombiner's rotate matching was
working as expected, I don't think we'd see any test diffs here.
AArch only goes right, and PPC only goes left.
x86 has both, so no diffs there.
Differential Revision: https://reviews.llvm.org/D50091
llvm-svn: 339359
Summary:
Currently, in line with GCC, when specifying reserved registers like sp or pc on an inline asm() clobber list, we don't always preserve the original value across the statement. And in general, overwriting reserved registers can have surprising results.
For example:
```
extern int bar(int[]);
int foo(int i) {
int a[i]; // VLA
asm volatile(
"mov r7, #1"
:
:
: "r7"
);
return 1 + bar(a);
}
```
Compiled for thumb, this gives:
```
$ clang --target=arm-arm-none-eabi -march=armv7a -c test.c -o - -S -O1 -mthumb
...
foo:
.fnstart
@ %bb.0: @ %entry
.save {r4, r5, r6, r7, lr}
push {r4, r5, r6, r7, lr}
.setfp r7, sp, #12
add r7, sp, #12
.pad #4
sub sp, #4
movs r1, #7
add.w r0, r1, r0, lsl #2
bic r0, r0, #7
sub.w r0, sp, r0
mov sp, r0
@APP
mov.w r7, #1
@NO_APP
bl bar
adds r0, #1
sub.w r4, r7, #12
mov sp, r4
pop {r4, r5, r6, r7, pc}
...
```
r7 is used as the frame pointer for thumb targets, and this function needs to restore the SP from the FP because of the variable-length stack allocation a. r7 is clobbered by the inline assembly (and r7 is included in the clobber list), but LLVM does not preserve the value of the frame pointer across the assembly block.
This type of behavior is similar to GCC's and has been discussed on the bugtracker: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=11807 . No consensus seemed to have been reached on the way forward. Clang behavior has briefly been discussed on the CFE mailing (starting here: http://lists.llvm.org/pipermail/cfe-dev/2018-July/058392.html). I've opted for following Eli Friedman's advice to print warnings when there are reserved registers on the clobber list so as not to diverge from GCC behavior for now.
The patch uses MachineRegisterInfo's target-specific knowledge of reserved registers, just before we convert the inline asm string in the AsmPrinter.
If we find a reserved register, we print a warning:
```
repro.c:6:7: warning: inline asm clobber list contains reserved registers: R7 [-Winline-asm]
"mov r7, #1"
^
```
Reviewers: eli.friedman, olista01, javed.absar, efriedma
Reviewed By: efriedma
Subscribers: efriedma, eraman, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D49727
llvm-svn: 339257
Summary:
Ensure that NormalizedBuildVector returns a BUILD_VECTOR with operands of the
same type. This fixes an assertion failure in VerifySDNode.
Reviewers: SjoerdMeijer, t.p.northover, javed.absar
Reviewed By: SjoerdMeijer
Subscribers: kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D50202
llvm-svn: 339013
In expansion of FCOPYSIGN, the shift node is missing when the two
operands of FCOPYSIGN are of the same size. We should always generate
shift node (if the required shift bit is not zero) to put the sign
bit into the right position, regardless of the size of underlying
types.
Differential Revision: https://reviews.llvm.org/D49973
llvm-svn: 338665
The bug is visible in the constant-folded x86 tests. We can't use the
negated shift amount when the type is not power-of-2:
https://rise4fun.com/Alive/US1r
...so in that case, use the regular lowering that includes a select
to guard against a shift-by-bitwidth. This path is improved by only
calculating the modulo shift amount once now.
Also, improve the rotate (with power-of-2 size) lowering to use
a negate rather than subtract from bitwidth. This improves the
codegen whether we have a rotate instruction or not (although
we can still see that we're not matching to a legal rotate in
all cases).
llvm-svn: 338592
Previously we were just visiting the blocks in the function in IR order, which
is rather arbitrary. Therefore we wouldn't always visit defs before uses, but
the translation code relies on this assumption in some places.
Only codegen change seen in tests is an elision of a redundant copy.
Fixes PR38396
llvm-svn: 338476
Also refactors some existing code to materialize addresses for the large code
model so it can be shared between G_GLOBAL_VALUE and G_BLOCK_ADDR.
This implements PR36390.
Differential Revision: https://reviews.llvm.org/D49903
llvm-svn: 338337
This is exchanging a sub-of-1 with add-of-minus-1:
https://rise4fun.com/Alive/plKAH
This is another step towards improving select-of-constants codegen (see D48970).
x86 is the motivating target, and those diffs all appear to be wins. PPC and AArch64 look neutral.
I've limited this to early combining (!LegalOperations) in case a target wants to reverse it, but
I think canonicalizing to 'add' is more likely to produce further transforms because we have more
folds for 'add'.
Differential Revision: https://reviews.llvm.org/D49924
llvm-svn: 338317
This teaches the outliner to save LR to a register rather than the stack when
possible. This allows us to avoid bumping the stack in outlined functions in
some cases. By doing this, in a later patch, we can teach the outliner to do
something like this:
f1:
...
bl OUTLINED_FUNCTION
...
f2:
...
move LR's contents to a register
bl OUTLINED_FUNCTION
move the register's contents back
instead of falling back to saving LR in both cases.
llvm-svn: 338278
