Summary:
Previously, if the threshold was 2, we were willing to speculatively
execute 2 cheap instructions in both basic blocks (thus we were willing
to speculatively execute cost = 4), but weren't willing to speculate
when one BB had 3 instructions and other one had no instructions,
even thought that would have total cost of 3.
This looks inconsistent to me.
I don't think `cmov`-like instructions will start executing
until both of it's inputs are available: https://godbolt.org/z/zgHePf
So i don't see why the existing behavior is the correct one.
Also, let's add it's own `cl::opt` for this threshold,
with default=4, so it is not stricter than the previous threshold:
will allow to fold when there are 2 BB's each with cost=2.
And since the logic has changed, it will also allow to fold when
one BB has cost=3 and other cost=1, or there is only one BB with cost=4.
This is an alternative solution to D65148:
This fix is mainly motivated by `signbit-like-value-extension.ll` test.
That pattern comes up in JPEG decoding, see e.g.
`Figure F.12 – Extending the sign bit of a decoded value in V`
of `ITU T.81` (JPEG specification).
That branch is not predictable, and it is within the innermost loop,
so the fact that that pattern ends up being stuck with a branch
instead of `select` (i.e. `CMOV` for x86) is unlikely to be beneficial.
This has great results on the final assembly (vanilla test-suite + RawSpeed): (metric pass - D67240)
| metric | old | new | delta | % |
| x86-mi-counting.NumMachineFunctions | 37720 | 37721 | 1 | 0.00% |
| x86-mi-counting.NumMachineBasicBlocks | 773545 | 771181 | -2364 | -0.31% |
| x86-mi-counting.NumMachineInstructions | 7488843 | 7486442 | -2401 | -0.03% |
| x86-mi-counting.NumUncondBR | 135770 | 135543 | -227 | -0.17% |
| x86-mi-counting.NumCondBR | 423753 | 422187 | -1566 | -0.37% |
| x86-mi-counting.NumCMOV | 24815 | 25731 | 916 | 3.69% |
| x86-mi-counting.NumVecBlend | 17 | 17 | 0 | 0.00% |
We significantly decrease basic block count, notably decrease instruction count,
significantly decrease branch count and very significantly increase `cmov` count.
Performance-wise, unsurprisingly, this has great effect on
target RawSpeed benchmark. I'm seeing 5 **major** improvements:
```
Benchmark Time CPU Time Old Time New CPU Old CPU New
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Samsung/NX3000/_3184416.SRW/threads:8/process_time/real_time_pvalue 0.0000 0.0000 U Test, Repetitions: 49 vs 49
Samsung/NX3000/_3184416.SRW/threads:8/process_time/real_time_mean -0.3064 -0.3064 226.9913 157.4452 226.9800 157.4384
Samsung/NX3000/_3184416.SRW/threads:8/process_time/real_time_median -0.3057 -0.3057 226.8407 157.4926 226.8282 157.4828
Samsung/NX3000/_3184416.SRW/threads:8/process_time/real_time_stddev -0.4985 -0.4954 0.3051 0.1530 0.3040 0.1534
Kodak/DCS760C/86L57188.DCR/threads:8/process_time/real_time_pvalue 0.0000 0.0000 U Test, Repetitions: 49 vs 49
Kodak/DCS760C/86L57188.DCR/threads:8/process_time/real_time_mean -0.1747 -0.1747 80.4787 66.4227 80.4771 66.4146
Kodak/DCS760C/86L57188.DCR/threads:8/process_time/real_time_median -0.1742 -0.1743 80.4686 66.4542 80.4690 66.4436
Kodak/DCS760C/86L57188.DCR/threads:8/process_time/real_time_stddev +0.6089 +0.5797 0.0670 0.1078 0.0673 0.1062
Sony/DSLR-A230/DSC08026.ARW/threads:8/process_time/real_time_pvalue 0.0000 0.0000 U Test, Repetitions: 49 vs 49
Sony/DSLR-A230/DSC08026.ARW/threads:8/process_time/real_time_mean -0.1598 -0.1598 171.6996 144.2575 171.6915 144.2538
Sony/DSLR-A230/DSC08026.ARW/threads:8/process_time/real_time_median -0.1598 -0.1597 171.7109 144.2755 171.7018 144.2766
Sony/DSLR-A230/DSC08026.ARW/threads:8/process_time/real_time_stddev +0.4024 +0.3850 0.0847 0.1187 0.0848 0.1175
Canon/EOS 77D/IMG_4049.CR2/threads:8/process_time/real_time_pvalue 0.0000 0.0000 U Test, Repetitions: 49 vs 49
Canon/EOS 77D/IMG_4049.CR2/threads:8/process_time/real_time_mean -0.0550 -0.0551 280.3046 264.8800 280.3017 264.8559
Canon/EOS 77D/IMG_4049.CR2/threads:8/process_time/real_time_median -0.0554 -0.0554 280.2628 264.7360 280.2574 264.7297
Canon/EOS 77D/IMG_4049.CR2/threads:8/process_time/real_time_stddev +0.7005 +0.7041 0.2779 0.4725 0.2775 0.4729
Canon/EOS 5DS/2K4A9929.CR2/threads:8/process_time/real_time_pvalue 0.0000 0.0000 U Test, Repetitions: 49 vs 49
Canon/EOS 5DS/2K4A9929.CR2/threads:8/process_time/real_time_mean -0.0354 -0.0355 316.7396 305.5208 316.7342 305.4890
Canon/EOS 5DS/2K4A9929.CR2/threads:8/process_time/real_time_median -0.0354 -0.0356 316.6969 305.4798 316.6917 305.4324
Canon/EOS 5DS/2K4A9929.CR2/threads:8/process_time/real_time_stddev +0.0493 +0.0330 0.3562 0.3737 0.3563 0.3681
```
That being said, it's always best-effort, so there will likely
be cases where this worsens things.
Reviewers: efriedma, craig.topper, dmgreen, jmolloy, fhahn, Carrot, hfinkel, chandlerc
Reviewed By: jmolloy
Subscribers: xbolva00, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D67318
llvm-svn: 372009
Summary:
- Functions to compute highlighting kinds for things are separated from
the ones that add highlighting tokens.
This keeps each of them more focused on what they're doing: getting
locations and figuring out the kind of the entity, correspondingly.
- Less special cases in visitor for various nodes.
This change is an NFC.
Reviewers: hokein
Reviewed By: hokein
Subscribers: MaskRay, jkorous, arphaman, kadircet, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D67341
llvm-svn: 372008
Related folds were added in:
rL125734
...the code comment about register pressure is discussed in
more detail in:
https://bugs.llvm.org/show_bug.cgi?id=2698
But 10 years later, perf testing bzip2 with this change now
shows a slight (0.2% average) improvement on Haswell although
that's probably within test noise.
Given that this is IR canonicalization, we shouldn't be worried
about register pressure though; the backend should be able to
adjust for that as needed.
This is part of solving PR43310 the theoretically right way:
https://bugs.llvm.org/show_bug.cgi?id=43310
...ie, if we don't cripple basic transforms, then we won't
need to add special-case code to detect larger patterns.
rL371940 and rL371981 are related patches in this series.
llvm-svn: 372007
Lower CTTZ on MVE using VBRSR and VCLS which will reverse the bits and
count the leading zeros, equivalent to a count trailing zeros (CTTZ).
llvm-svn: 372000
The static analyzer is warning about potential null dereferences of dyn_cast<> results - in these cases we can safely use cast<> directly as we know that these cases should all be the correct type, which is why its working atm and anyway cast<> will assert if they aren't.
llvm-svn: 371998
Now that the vectorizer can do tail-folding (rL367592), and the ARM backend
understands MVE masked loads/stores (rL371932), it's time to add the MVE
tail-folding equivalent of the X86 tests that I added.
llvm-svn: 371996
Since MBB was split *before* MI, the MI(s) will reside in JoinMBB (MBB) at
the point of erasing them, so calling StartMBB->erase() is actually wrong,
although it is "working" by all appearances.
Review: Ulrich Weigand
llvm-svn: 371995
Summary:
This allows enabling useaa on the command-line and will allow enabling the
feature on a per-CPU basis where benchmarking shows improvements.
This is modelled after the ARM/AArch64 target.
Reviewers: RKSimon, andreadb, craig.topper
Subscribers: javed.absar, kristof.beyls, hiraditya, ychen, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D67266
llvm-svn: 371989
r371875 moved some functionality around to a Basic header file, but
didn't move its definitions as well. This patch moves some things
around so that shared library building can work.
llvm-svn: 371985
Previously we only had a stub document.
Reviewed by: serge-sans-paille, MaskRay
Differential Revision: https://reviews.llvm.org/D67555
llvm-svn: 371983
MVE has VPT instructions, which perform the duties of both a VCMP and a VPST in
a single instruction, performing the compare and starting the VPT block in one.
This teaches the MVEVPTBlockPass to fold them, searching back through the
basicblock for a valid VCMP and creating the VPT from its operands.
There are some changes to the VPT instructions to accommodate this, altering
the order of the operands to match the VCMP better, and changing P0 register
defs to be VPR defs, as is used in other places.
Differential Revision: https://reviews.llvm.org/D66577
llvm-svn: 371982
This fold and several others were added in:
rL125734 <https://reviews.llvm.org/rL125734>
...with no explanation for the one-use checks other than the code
comments about register pressure.
Given that this is IR canonicalization, we shouldn't be worried
about register pressure though; the backend should be able to
adjust for that as needed.
This is part of solving PR43310 the theoretically right way:
https://bugs.llvm.org/show_bug.cgi?id=43310
...ie, if we don't cripple basic transforms, then we won't
need to add special-case code to detect larger patterns.
rL371940 is a related patch in this series.
llvm-svn: 371981
This blob was written before match() existed, so it
could probably be reduced significantly.
But I suspect it isn't well tested, so tests would have
to be added to reduce risk from logic changes.
llvm-svn: 371978
Summary:
For a given cursor position, it returns ranges that are interesting to the user.
Currently the semantic ranges correspond to the nodes of the syntax trees.
Subscribers: mgorny, jkorous, arphaman, kadircet, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D67358
llvm-svn: 371976
The static analyzer is warning about a potential null dereference of the cast_or_null result, I've split the cast_or_null check from the ->getUnderlyingInstr() call to avoid this, but it appears that we weren't seeing any null pointers in the dumped bundles in the first place.
llvm-svn: 371975
The static analyzer is warning about potential null dereferences of dyn_cast<> results - in these cases we can safely use cast<> directly as we know that these cases should all be the correct type, which is why its working atm and anyway cast<> will assert if they aren't.
llvm-svn: 371973
Summary:
The renamelib uses a tricky way to calculate the end location by relying
on decl name, this is incorrect for the overloaded operator (the name is
"operator++" instead of "++"), which will cause out-of-file offset.
We also disable renaming operator symbol, this case is tricky, and
renamelib doesnt handle it properly.
Reviewers: ilya-biryukov
Subscribers: MaskRay, jkorous, arphaman, kadircet, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D67607
llvm-svn: 371971
Provide a missing initializer to get rid of warning provoking buildbot
failures.
error: missing field 'rel' initializer
[-Werror,-Wmissing-field-initializers]
llvm-svn: 371970
The signature of __builtin_rotateright64 was by misstake changed from
unsigned to signed in r360863, this patch will change it back to
unsigned as intended.
This fixes pr43309
Reviewers: efriedma, hans
Reviewed By: hans
Differential Revision: https://reviews.llvm.org/D67606
llvm-svn: 371969
Summary:
Adds the following inline asm constraints for SVE:
- Upl: One of the low eight SVE predicate registers, P0 to P7 inclusive
- Upa: SVE predicate register with full range, P0 to P15
Reviewers: t.p.northover, sdesmalen, rovka, momchil.velikov, cameron.mcinally, greened, rengolin
Reviewed By: rovka
Subscribers: javed.absar, tschuett, rkruppe, psnobl, cfe-commits, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D66524
llvm-svn: 371967
The --fix-cortex-a8 option implements a linker workaround for the
coretex-a8 erratum 657417. A summary of the erratum conditions is:
- A 32-bit Thumb-2 branch instruction B.w, Bcc.w, BL, BLX spans two
4KiB regions.
- The destination of the branch is to the first 4KiB region.
- The instruction before the branch is a 32-bit Thumb-2 non-branch
instruction.
The linker fix is to redirect the branch to a patch not in the first
4KiB region. The patch forwards the branch on to its target.
The cortex-a8, is an old CPU, with the first implementation of this
workaround in ld.bfd appearing in 2009. The cortex-a8 has been used in
early Android Phones and there are some critical applications that still
need to run on a cortex-a8 that have the erratum. The patch is applied
roughly 10 times on LLD and 20 on Clang when they are built with
--fix-cortex-a8 on an Arm system.
The formal erratum description is avaliable in the ARM Core Cortex-A8
(AT400/AT401) Errata Notice document. This is available from Arm on
request but it seems to be findable via a web search.
Differential Revision: https://reviews.llvm.org/D67284
llvm-svn: 371965
Summary:
Finds calls that add element to protobuf repeated field in a loop
without calling Reserve() before the loop. Calling Reserve() first can avoid
unnecessary memory reallocations.
A new option EnableProto is added to guard this feature.
Patch by Cong Liu!
Reviewers: gribozavr, alexfh, hokein, aaron.ballman
Reviewed By: hokein
Subscribers: lebedev.ri, xazax.hun, Eugene.Zelenko, cfe-commits
Tags: #clang, #clang-tools-extra
Differential Revision: https://reviews.llvm.org/D67135
llvm-svn: 371963
After our previous machinecombiner exercises (rL371321, rL371818, rL371833), we
were still missing a few FP16 FMA patterns.
Differential Revision: https://reviews.llvm.org/D67576
llvm-svn: 371960
SystemZExpandPseudo:s only job was to expand LOCRMux instructions into jump
sequences. This needs to be done if expandLOCRPseudo() or expandSELRPseudo()
fails to find a legal opcode (all registers "high" or "low"). This task has
now been moved to SystemZPostRewrite while removing the SystemZExpandPseudo
pass.
It is in fact preferred to expand these pseudos directly after register
allocation in SystemZPostRewrite since the hinted register combinations are
then not subject to later optimizations.
Review: Ulrich Weigand
https://reviews.llvm.org/D67432
llvm-svn: 371959
If there is no readonly section, we map:
* The ELF header at imageBase+maxPageSize
* Program headers at imageBase+maxPageSize+sizeof(Ehdr)
* The first section .text at imageBase+maxPageSize+sizeof(Ehdr)+sizeof(program headers)
Due to the interaction between Writer<ELFT>::fixSectionAlignments and
LinkerScript::allocateHeaders,
`alignDown(p_vaddr(R PT_LOAD)) = alignDown(p_vaddr(RX PT_LOAD))`.
The RX PT_LOAD will override the R PT_LOAD at runtime, which is not ideal:
```
// PHDR at 0x401034, should be 0x400034
PHDR 0x000034 0x00401034 0x00401034 0x000a0 0x000a0 R 0x4
// R PT_LOAD contains just Ehdr and program headers.
// At 0x401000, should be 0x400000
LOAD 0x000000 0x00401000 0x00401000 0x000d4 0x000d4 R 0x1000
LOAD 0x0000d4 0x004010d4 0x004010d4 0x00001 0x00001 R E 0x1000
```
* createPhdrs allocates the headers to the R PT_LOAD.
* fixSectionAlignments assigns `imageBase+maxPageSize+sizeof(Ehdr)+sizeof(program headers)` (formula: `alignTo(dot, maxPageSize) + dot % config->maxPageSize`) to addrExpr of .text
* allocateHeaders computes the minimum address among SHF_ALLOC sections, i.e. addr(.text)
* allocateHeaders sets address of ELF header to `addr(.text)-sizeof(Ehdr)-sizeof(program headers) = imageBase+maxPageSize`
The main observation is that when the SECTIONS command is not used, we
don't have to call allocateHeaders. This requires an assumption that
the presence of PT_PHDR and addresses of headers can be decided
regardless of address information.
This may seem natural because dot is not manipulated by a linker script.
The other thing is that we have to drop the special rule for -T<section>
in `getInitialDot`. If -Ttext is smaller than the image base, the headers
will not be allocated with the old behavior (allocateHeaders is called)
but always allocated with the new behavior.
The behavior change is not a problem. Whether and where headers are
allocated can vary among linkers, or ld.bfd across different versions
(--enable-separate-code or not). It is thus advised to use a linker
script with the PHDRS command to have a consistent behavior across
linkers. If PT_PHDR is needed, an explicit --image-base can be a simpler
alternative.
Differential Revision: https://reviews.llvm.org/D67325
llvm-svn: 371957
Roman Lebedev