This patch implements initial backend support for a -mtune CPU controlled by a "tune-cpu" function attribute. If the attribute is not present X86 will use the resolved CPU from target-cpu attribute or command line.
This patch adds MC layer support a tune CPU. Each CPU now has two sets of features stored in their GenSubtargetInfo.inc tables . These features lists are passed separately to the Processor and ProcessorModel classes in tablegen. The tune list defaults to an empty list to avoid changes to non-X86. This annoyingly increases the size of static tables on all target as we now store 24 more bytes per CPU. I haven't quantified the overall impact, but I can if we're concerned.
One new test is added to X86 to show a few tuning features with mismatched tune-cpu and target-cpu/target-feature attributes to demonstrate independent control. Another new test is added to demonstrate that the scheduler model follows the tune CPU.
I have not added a -mtune to llc/opt or MC layer command line yet. With no attributes we'll just use the -mcpu for both. MC layer tools will always follow the normal CPU for tuning.
Differential Revision: https://reviews.llvm.org/D85165
In the ARM backend, for historical reasons we have only some targets
using Machine Scheduling. The rest use the old list scheduler as they
are using itinaries and the list scheduler seems to produce better code
(and not crash running out of register on v6m codes). So whether to use
the MIScheduler or not is checked at runtime from the subtarget
features.
This is fine, except for post-ra scheduling. Whether to use the old
post-ra list scheduler or the post-ra machine schedule is decided as the
pass manager is set up, in arms case from a newly constructed subtarget.
Under some situations, like LTO, this won't include the correct cpu so
can pick the wrong option. This can have a surprising effect on
performance.
To fix that, this patch overrides targetSchedulesPostRAScheduling and
addPreSched2 in the ARM backend, adding _both_ post-ra schedulers and
picking at runtime which to execute. To pick between the two I've had to
add a enablePostRAMachineScheduler() method that normally returns
enableMachineScheduler() && enablePostRAScheduler(), which can be
overridden to enable just one of PostRAMachineScheduler vs
PostRAScheduler.
Thanks to David Penry for the identifying this problem.
Differential Revision: https://reviews.llvm.org/D69775
These arrays are both keyed by CPU name and go into the same tablegenerated file. Merge them so we only need to store keys once.
This also removes a weird space saving quirk where we used the ProcDesc.size() to create to build an ArrayRef for ProcSched.
Differential Revision: https://reviews.llvm.org/D58939
llvm-svn: 355431
The description for CPUs was just the CPU name wrapped with "Select the " and " processor". We can just do that directly in the help printer instead of making a separate version in the binary for each CPU.
Also remove the Value field that isn't needed and was always 0.
Differential Revision: https://reviews.llvm.org/D58938
llvm-svn: 355429
This patch removes hidden codegen flag -print-schedule effectively reverting the
logic originally committed as r300311
(https://llvm.org/viewvc/llvm-project?view=revision&revision=300311).
Flag -print-schedule was originally introduced by r300311 to address PR32216
(https://bugs.llvm.org/show_bug.cgi?id=32216). That bug was about adding "Better
testing of schedule model instruction latencies/throughputs".
These days, we can use llvm-mca to test scheduling models. So there is no longer
a need for flag -print-schedule in LLVM. The main use case for PR32216 is
now addressed by llvm-mca.
Flag -print-schedule is mainly used for debugging purposes, and it is only
actually used by x86 specific tests. We already have extensive (latency and
throughput) tests under "test/tools/llvm-mca" for X86 processor models. That
means, most (if not all) existing -print-schedule tests for X86 are redundant.
When flag -print-schedule was first added to LLVM, several files had to be
modified; a few APIs gained new arguments (see for example method
MCAsmStreamer::EmitInstruction), and MCSubtargetInfo/TargetSubtargetInfo gained
a couple of getSchedInfoStr() methods.
Method getSchedInfoStr() had to originally work for both MCInst and
MachineInstr. The original implmentation of getSchedInfoStr() introduced a
subtle layering violation (reported as PR37160 and then fixed/worked-around by
r330615).
In retrospect, that new API could have been designed more optimally. We can
always query MCSchedModel to get the latency and throughput. More importantly,
the "sched-info" string should not have been generated by the subtarget.
Note, r317782 fixed an issue where "print-schedule" didn't work very well in the
presence of inline assembly. That commit is also reverted by this change.
Differential Revision: https://reviews.llvm.org/D57244
llvm-svn: 353043
This patch adds a new ReadAdvance definition named ReadInt2Fpu.
ReadInt2Fpu allows x86 scheduling models to accurately describe delays caused by
data transfers from the integer unit to the floating point unit.
ReadInt2Fpu currently defaults to a delay of zero cycles (i.e. no delay) for all
x86 models excluding BtVer2. That means, this patch is only a functional change
for the Jaguar cpu model only.
Tablegen definitions for instructions (V)PINSR* have been updated to account for
the new ReadInt2Fpu. That read is mapped to the the GPR input operand.
On Jaguar, int-to-fpu transfers are modeled as a +6cy delay. Before this patch,
that extra delay was added to the opcode latency. In practice, the insert opcode
only executes for 1cy. Most of the actual latency is actually contributed by the
so-called operand-latency. According to the AMD SOG for family 16h, (V)PINSR*
latency is defined by expression f+1, where f is defined as a forwarding delay
from the integer unit to the fpu.
When printing instruction latency from MCA (see InstructionInfoView.cpp) and LLC
(only when flag -print-schedule is speified), we now need to account for any
extra forwarding delays. We do this by checking if scheduling classes declare
any negative ReadAdvance entries. Quoting a code comment in TargetSchedule.td:
"A negative advance effectively increases latency, which may be used for
cross-domain stalls". When computing the instruction latency for the purpose of
our scheduling tests, we now add any extra delay to the formula. This avoids
regressing existing codegen and mca schedule tests. It comes with the cost of an
extra (but very simple) hook in MCSchedModel.
Differential Revision: https://reviews.llvm.org/D57056
llvm-svn: 351965
to reflect the new license.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351636
This is a fix for the problem arising in D47374 (PR37678):
https://bugs.llvm.org/show_bug.cgi?id=37678
We may not have throughput info because it's not specified in the model
or it's not available with variant scheduling, so assume that those
instructions can execute/complete at max-issue-width.
Differential Revision: https://reviews.llvm.org/D47723
llvm-svn: 334055
This patch extends the MCSchedModel API with new methods that can be used to
obtain the latency and reciprocal througput information for an MCInst.
Scheduling models have recently gained the ability to resolve variant scheduling
classes associated with MCInst objects. Before, models were only able to resolve
a variant scheduling class from a MachineInstr object.
This patch is mainly required by D47374 to avoid regressing a pair of x86
specific -print-schedule tests for btver2. Patch D47374 introduces a new variant
class to teach the btver scheduling model (x86 target) how to correctly compute
the latency profile for some zero-idioms using the new scheduling predicates.
The new methods added by this patch would be mainly used by llc when flag
-print-schedule is specified. In particular, tests that contain inline assembly
require that code is parsed at code emission stage into a sequence of MCInst.
That forces the print-schedule functionality to query the latency/rthroughput
information for MCInst instructions too. If we don't expose this new API, then
we lose "-print-schedule" test coverage as soon as variant scheduling classes
are added to the x86 models.
The tablegen SubtargetEmitter changes teaches how to query latency profile
information using a object that derives from TargetSubtargetInfo. Note that this
should really have been part of r333286. To avoid code duplication, the logic
that "resolves" variant scheduling classes for MCInst, has been moved to a
common place in MC. That logic is used by the "resolveVariantSchedClass" methods
redefined in override by the tablegen'd GenSubtargetInfo classes.
Differential Revision: https://reviews.llvm.org/D47536
llvm-svn: 333650
TargetSchedModel now always delegates to MCSchedModel the computation of
instruction latency and reciprocal throughput.
No functional change intended.
llvm-svn: 330099
The TargetSchedModel is always initialized using the TargetSubtargetInfo's
MCSchedModel and TargetInstrInfo, so we don't need to extract those and
pass 3 parameters to init().
Differential Revision: https://reviews.llvm.org/D44789
llvm-svn: 329540
I don't know how to expose this in a test. There are ARM / AArch64
sched classes that include zero latency instructions, but I'm not
seeing sched info printed for those targets. X86 will almost
certainly have these soon (see PR36671), but no model has
'let Latency = 0' currently.
llvm-svn: 327518
Summary:
First, we need to explain the core of the vulnerability. Note that this
is a very incomplete description, please see the Project Zero blog post
for details:
https://googleprojectzero.blogspot.com/2018/01/reading-privileged-memory-with-side.html
The basis for branch target injection is to direct speculative execution
of the processor to some "gadget" of executable code by poisoning the
prediction of indirect branches with the address of that gadget. The
gadget in turn contains an operation that provides a side channel for
reading data. Most commonly, this will look like a load of secret data
followed by a branch on the loaded value and then a load of some
predictable cache line. The attacker then uses timing of the processors
cache to determine which direction the branch took *in the speculative
execution*, and in turn what one bit of the loaded value was. Due to the
nature of these timing side channels and the branch predictor on Intel
processors, this allows an attacker to leak data only accessible to
a privileged domain (like the kernel) back into an unprivileged domain.
The goal is simple: avoid generating code which contains an indirect
branch that could have its prediction poisoned by an attacker. In many
cases, the compiler can simply use directed conditional branches and
a small search tree. LLVM already has support for lowering switches in
this way and the first step of this patch is to disable jump-table
lowering of switches and introduce a pass to rewrite explicit indirectbr
sequences into a switch over integers.
However, there is no fully general alternative to indirect calls. We
introduce a new construct we call a "retpoline" to implement indirect
calls in a non-speculatable way. It can be thought of loosely as
a trampoline for indirect calls which uses the RET instruction on x86.
Further, we arrange for a specific call->ret sequence which ensures the
processor predicts the return to go to a controlled, known location. The
retpoline then "smashes" the return address pushed onto the stack by the
call with the desired target of the original indirect call. The result
is a predicted return to the next instruction after a call (which can be
used to trap speculative execution within an infinite loop) and an
actual indirect branch to an arbitrary address.
On 64-bit x86 ABIs, this is especially easily done in the compiler by
using a guaranteed scratch register to pass the target into this device.
For 32-bit ABIs there isn't a guaranteed scratch register and so several
different retpoline variants are introduced to use a scratch register if
one is available in the calling convention and to otherwise use direct
stack push/pop sequences to pass the target address.
This "retpoline" mitigation is fully described in the following blog
post: https://support.google.com/faqs/answer/7625886
We also support a target feature that disables emission of the retpoline
thunk by the compiler to allow for custom thunks if users want them.
These are particularly useful in environments like kernels that
routinely do hot-patching on boot and want to hot-patch their thunk to
different code sequences. They can write this custom thunk and use
`-mretpoline-external-thunk` *in addition* to `-mretpoline`. In this
case, on x86-64 thu thunk names must be:
```
__llvm_external_retpoline_r11
```
or on 32-bit:
```
__llvm_external_retpoline_eax
__llvm_external_retpoline_ecx
__llvm_external_retpoline_edx
__llvm_external_retpoline_push
```
And the target of the retpoline is passed in the named register, or in
the case of the `push` suffix on the top of the stack via a `pushl`
instruction.
There is one other important source of indirect branches in x86 ELF
binaries: the PLT. These patches also include support for LLD to
generate PLT entries that perform a retpoline-style indirection.
The only other indirect branches remaining that we are aware of are from
precompiled runtimes (such as crt0.o and similar). The ones we have
found are not really attackable, and so we have not focused on them
here, but eventually these runtimes should also be replicated for
retpoline-ed configurations for completeness.
For kernels or other freestanding or fully static executables, the
compiler switch `-mretpoline` is sufficient to fully mitigate this
particular attack. For dynamic executables, you must compile *all*
libraries with `-mretpoline` and additionally link the dynamic
executable and all shared libraries with LLD and pass `-z retpolineplt`
(or use similar functionality from some other linker). We strongly
recommend also using `-z now` as non-lazy binding allows the
retpoline-mitigated PLT to be substantially smaller.
When manually apply similar transformations to `-mretpoline` to the
Linux kernel we observed very small performance hits to applications
running typical workloads, and relatively minor hits (approximately 2%)
even for extremely syscall-heavy applications. This is largely due to
the small number of indirect branches that occur in performance
sensitive paths of the kernel.
When using these patches on statically linked applications, especially
C++ applications, you should expect to see a much more dramatic
performance hit. For microbenchmarks that are switch, indirect-, or
virtual-call heavy we have seen overheads ranging from 10% to 50%.
However, real-world workloads exhibit substantially lower performance
impact. Notably, techniques such as PGO and ThinLTO dramatically reduce
the impact of hot indirect calls (by speculatively promoting them to
direct calls) and allow optimized search trees to be used to lower
switches. If you need to deploy these techniques in C++ applications, we
*strongly* recommend that you ensure all hot call targets are statically
linked (avoiding PLT indirection) and use both PGO and ThinLTO. Well
tuned servers using all of these techniques saw 5% - 10% overhead from
the use of retpoline.
We will add detailed documentation covering these components in
subsequent patches, but wanted to make the core functionality available
as soon as possible. Happy for more code review, but we'd really like to
get these patches landed and backported ASAP for obvious reasons. We're
planning to backport this to both 6.0 and 5.0 release streams and get
a 5.0 release with just this cherry picked ASAP for distros and vendors.
This patch is the work of a number of people over the past month: Eric, Reid,
Rui, and myself. I'm mailing it out as a single commit due to the time
sensitive nature of landing this and the need to backport it. Huge thanks to
everyone who helped out here, and everyone at Intel who helped out in
discussions about how to craft this. Also, credit goes to Paul Turner (at
Google, but not an LLVM contributor) for much of the underlying retpoline
design.
Reviewers: echristo, rnk, ruiu, craig.topper, DavidKreitzer
Subscribers: sanjoy, emaste, mcrosier, mgorny, mehdi_amini, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D41723
llvm-svn: 323155
Re-commit of r322200: The testcase shouldn't hit machineverifiers
anymore with r322917 in place.
Large callframes (calls with several hundreds or thousands or
parameters) could lead to situations in which the emergency spillslot is
out of range to be addressed relative to the stack pointer.
This commit forces the use of a frame pointer in the presence of large
callframes.
This commit does several things:
- Compute max callframe size at the end of instruction selection.
- Add mirFileLoaded target callback. Use it to compute the max callframe size
after loading a .mir file when the size wasn't specified in the file.
- Let TargetFrameLowering::hasFP() return true if there exists a
callframe > 255 bytes.
- Always place the emergency spillslot close to FP if we have a frame
pointer.
- Note that `useFPForScavengingIndex()` would previously return false
when a base pointer was available leading to the emergency spillslot
getting allocated late (that's the whole effect of this callback).
Which made no sense to me so I took this case out: Even though the
emergency spillslot is technically not referenced by FP in this case
we still want it allocated early.
Differential Revision: https://reviews.llvm.org/D40876
llvm-svn: 322919
Revert for now as the testcase is hitting a pre-existing verifier error
that manifest as a failure when expensive checks are enabled (or
-verify-machineinstrs) is used.
This reverts commit r322200.
llvm-svn: 322231
Large callframes (calls with several hundreds or thousands or
parameters) could lead to situations in which the emergency spillslot is
out of range to be addressed relative to the stack pointer.
This commit forces the use of a frame pointer in the presence of large
callframes.
This commit does several things:
- Compute max callframe size at the end of instruction selection.
- Add mirFileLoaded target callback. Use it to compute the max callframe size
after loading a .mir file when the size wasn't specified in the file.
- Let TargetFrameLowering::hasFP() return true if there exists a
callframe > 255 bytes.
- Always place the emergency spillslot close to FP if we have a frame
pointer.
- Note that `useFPForScavengingIndex()` would previously return false
when a base pointer was available leading to the emergency spillslot
getting allocated late (that's the whole effect of this callback).
Which made no sense to me so I took this case out: Even though the
emergency spillslot is technically not referenced by FP in this case
we still want it allocated early.
Differential Revision: https://reviews.llvm.org/D40876
llvm-svn: 322200
All these headers already depend on CodeGen headers so moving them into
CodeGen fixes the layering (since CodeGen depends on Target, not the
other way around).
llvm-svn: 318490
This header includes CodeGen headers, and is not, itself, included by
any Target headers, so move it into CodeGen to match the layering of its
implementation.
llvm-svn: 317647
This fixes bugzilla 26810
https://bugs.llvm.org/show_bug.cgi?id=26810
This is intended to prevent sequences like:
movl %ebp, 8(%esp) # 4-byte Spill
movl %ecx, %ebp
movl %ebx, %ecx
movl %edi, %ebx
movl %edx, %edi
cltd
idivl %esi
movl %edi, %edx
movl %ebx, %edi
movl %ecx, %ebx
movl %ebp, %ecx
movl 16(%esp), %ebp # 4 - byte Reload
Such sequences are created in 2 scenarios:
Scenario #1:
vreg0 is evicted from physreg0 by vreg1
Evictee vreg0 is intended for region splitting with split candidate physreg0 (the reg vreg0 was evicted from)
Region splitting creates a local interval because of interference with the evictor vreg1 (normally region spliiting creates 2 interval, the "by reg" and "by stack" intervals. Local interval created when interference occurs.)
one of the split intervals ends up evicting vreg2 from physreg1
Evictee vreg2 is intended for region splitting with split candidate physreg1
one of the split intervals ends up evicting vreg3 from physreg2 etc.. until someone spills
Scenario #2
vreg0 is evicted from physreg0 by vreg1
vreg2 is evicted from physreg2 by vreg3 etc
Evictee vreg0 is intended for region splitting with split candidate physreg1
Region splitting creates a local interval because of interference with the evictor vreg1
one of the split intervals ends up evicting back original evictor vreg1 from physreg0 (the reg vreg0 was evicted from)
Another evictee vreg2 is intended for region splitting with split candidate physreg1
one of the split intervals ends up evicting vreg3 from physreg2 etc.. until someone spills
As compile time was a concern, I've added a flag to control weather we do cost calculations for local intervals we expect to be created (it's on by default for X86 target, off for the rest).
Differential Revision: https://reviews.llvm.org/D35816
Change-Id: Id9411ff7bbb845463d289ba2ae97737a1ee7cc39
llvm-svn: 316295
I did this a long time ago with a janky python script, but now
clang-format has built-in support for this. I fed clang-format every
line with a #include and let it re-sort things according to the precise
LLVM rules for include ordering baked into clang-format these days.
I've reverted a number of files where the results of sorting includes
isn't healthy. Either places where we have legacy code relying on
particular include ordering (where possible, I'll fix these separately)
or where we have particular formatting around #include lines that
I didn't want to disturb in this patch.
This patch is *entirely* mechanical. If you get merge conflicts or
anything, just ignore the changes in this patch and run clang-format
over your #include lines in the files.
Sorry for any noise here, but it is important to keep these things
stable. I was seeing an increasing number of patches with irrelevant
re-ordering of #include lines because clang-format was used. This patch
at least isolates that churn, makes it easy to skip when resolving
conflicts, and gets us to a clean baseline (again).
llvm-svn: 304787
TargetSubtargetInfo is filled with CodeGen specific interfaces nowadays
(getInstrInfo(), getFrameLowering(), getSelectionDAGInfo()) most of the
tuning flags like enablePostRAScheduler(), getAntiDepBreakMode(),
enableRALocalReassignment(), ... also do not seem to be universal enough
to make sense outside of CodeGen.
Differential Revision: https://reviews.llvm.org/D26948
llvm-svn: 287708
