maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/AArch64/GlobalISel/arm64-fallback.ll

261 lines
12 KiB
LLVM
Raw Normal View History

[AArch64][GlobalISel] Enable GlobalISel at -O0 by default Tests updated to explicitly use fast-isel at -O0 instead of implicitly. This change also allows an explicit -fast-isel option to override an implicitly enabled global-isel. Otherwise -fast-isel would have no effect at -O0. Differential Revision: https://reviews.llvm.org/D41362 llvm-svn: 321655
2018-01-03 00:30:47 +08:00
; RUN: not llc -O0 -global-isel -global-isel-abort=1 -verify-machineinstrs %s -o - 2>&1 | FileCheck %s --check-prefix=ERROR
[TargetPassConfig] Add a hook to tell whether GlobalISel should warm on fallback. Thanks to this patch, we know have a way to easly see if GlobalISel failed. llvm-svn: 280273
2016-09-01 02:43:04 +08:00
; RUN: llc -O0 -global-isel -global-isel-abort=0 -verify-machineinstrs %s -o - 2>&1 | FileCheck %s --check-prefix=FALLBACK
[GlobalISel] Emit opt remarks on isel fallbacks. Having more fine-grained information on the specific construct that caused us to fallback is valuable for large-scale data collection. We still have the fallback warning, that's also used for FastISel. We still need to remove the fallback warning, and teach FastISel to also emit remarks (it currently has a combination of the warning, stats, and debug prints: the remarks could unify all three). The abort-on-fallback path could also be better handled using remarks: one could imagine a "-Rpass-error", analoguous to "-Werror", which would promote missed/failed remarks to errors. It's not clear whether that would be useful for other remarks though, so we're not there yet. llvm-svn: 296013
2017-02-24 05:05:42 +08:00
; RUN: llc -O0 -global-isel -global-isel-abort=2 -pass-remarks-missed='gisel*' -verify-machineinstrs %s -o %t.out 2> %t.err
GlobalISel: improve translation fallback for constants. Returning 0 (NoReg) from getOrCreateVReg leads to unexpected situations later in the translation. It's better to return a valid (if undefined) register and let the rest of the instruction carry on as planned. llvm-svn: 288709
2016-12-06 05:40:33 +08:00
; RUN: FileCheck %s --check-prefix=FALLBACK-WITH-REPORT-OUT < %t.out
; RUN: FileCheck %s --check-prefix=FALLBACK-WITH-REPORT-ERR < %t.err
[GlobalISel] Disable GISel for big endian. This is due to PR26161 needing to be resolved before we can fix big endian bugs like PR35359. The work to split aggregates into smaller LLTs instead of using one large scalar will take some time, so in the mean time we'll fall back to SDAG. Some ARM BE tests xfailed for now as a result. Differential Revision: https://reviews.llvm.org/D40789 llvm-svn: 320388
2017-12-12 00:58:29 +08:00
; RUN: not llc -global-isel -mtriple aarch64_be %s -o - 2>&1 | FileCheck %s --check-prefix=BIG-ENDIAN
[GlobalISel] Add a fallback path to SDISel. When global-isel fails on a MachineFunction MF, MF will be cleaned up and given to SDISel. Thanks to this fallback, we can already perform correctness test even if we support only a small portion of the functions in a test. llvm-svn: 279891
2016-08-27 08:18:31 +08:00
; This file checks that the fallback path to selection dag works.
; The test is fragile in the sense that it must be updated to expose
; something that fails with global-isel.
; When we cannot produce a test case anymore, that means we can remove
; the fallback path.
target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
[GlobalISel] Get the AArch64 tests to work on Linux Mostly this just means changing the triple from aarch64-apple-ios to the generic aarch64--. Only one test needs more significant changes, but GlobalISel already does the right thing so it's ok to just change the checks. Differential Revision: https://reviews.llvm.org/D25532 llvm-svn: 284223
2016-10-14 18:19:40 +08:00
target triple = "aarch64--"
[GlobalISel] Add a fallback path to SDISel. When global-isel fails on a MachineFunction MF, MF will be cleaned up and given to SDISel. Thanks to this fallback, we can already perform correctness test even if we support only a small portion of the functions in a test. llvm-svn: 279891
2016-08-27 08:18:31 +08:00
[GlobalISel] Disable GISel for big endian. This is due to PR26161 needing to be resolved before we can fix big endian bugs like PR35359. The work to split aggregates into smaller LLTs instead of using one large scalar will take some time, so in the mean time we'll fall back to SDAG. Some ARM BE tests xfailed for now as a result. Differential Revision: https://reviews.llvm.org/D40789 llvm-svn: 320388
2017-12-12 00:58:29 +08:00
; BIG-ENDIAN: unable to translate in big endian mode
[GlobalISel] Get the AArch64 tests to work on Linux Mostly this just means changing the triple from aarch64-apple-ios to the generic aarch64--. Only one test needs more significant changes, but GlobalISel already does the right thing so it's ok to just change the checks. Differential Revision: https://reviews.llvm.org/D25532 llvm-svn: 284223
2016-10-14 18:19:40 +08:00
; We use __fixunstfti as the common denominator for __fixunstfti on Linux and
; ___fixunstfti on iOS
[GlobalISel] Emit opt remarks on isel fallbacks. Having more fine-grained information on the specific construct that caused us to fallback is valuable for large-scale data collection. We still have the fallback warning, that's also used for FastISel. We still need to remove the fallback warning, and teach FastISel to also emit remarks (it currently has a combination of the warning, stats, and debug prints: the remarks could unify all three). The abort-on-fallback path could also be better handled using remarks: one could imagine a "-Rpass-error", analoguous to "-Werror", which would promote missed/failed remarks to errors. It's not clear whether that would be useful for other remarks though, so we're not there yet. llvm-svn: 296013
2017-02-24 05:05:42 +08:00
; ERROR: unable to lower arguments: i128 (i128)* (in function: ABIi128)
[GlobalISel] Add a fallback path to SDISel. When global-isel fails on a MachineFunction MF, MF will be cleaned up and given to SDISel. Thanks to this fallback, we can already perform correctness test even if we support only a small portion of the functions in a test. llvm-svn: 279891
2016-08-27 08:18:31 +08:00
; FALLBACK: ldr q0,
[GlobalISel] Get the AArch64 tests to work on Linux Mostly this just means changing the triple from aarch64-apple-ios to the generic aarch64--. Only one test needs more significant changes, but GlobalISel already does the right thing so it's ok to just change the checks. Differential Revision: https://reviews.llvm.org/D25532 llvm-svn: 284223
2016-10-14 18:19:40 +08:00
; FALLBACK-NEXT: bl __fixunstfti
[TargetPassConfig] Add a hook to tell whether GlobalISel should warm on fallback. Thanks to this patch, we know have a way to easly see if GlobalISel failed. llvm-svn: 280273
2016-09-01 02:43:04 +08:00
;
[GlobalISel] Emit opt remarks on isel fallbacks. Having more fine-grained information on the specific construct that caused us to fallback is valuable for large-scale data collection. We still have the fallback warning, that's also used for FastISel. We still need to remove the fallback warning, and teach FastISel to also emit remarks (it currently has a combination of the warning, stats, and debug prints: the remarks could unify all three). The abort-on-fallback path could also be better handled using remarks: one could imagine a "-Rpass-error", analoguous to "-Werror", which would promote missed/failed remarks to errors. It's not clear whether that would be useful for other remarks though, so we're not there yet. llvm-svn: 296013
2017-02-24 05:05:42 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to lower arguments: i128 (i128)* (in function: ABIi128)
GlobalISel: improve translation fallback for constants. Returning 0 (NoReg) from getOrCreateVReg leads to unexpected situations later in the translation. It's better to return a valid (if undefined) register and let the rest of the instruction carry on as planned. llvm-svn: 288709
2016-12-06 05:40:33 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for ABIi128
; FALLBACK-WITH-REPORT-OUT-LABEL: ABIi128:
; FALLBACK-WITH-REPORT-OUT: ldr q0,
; FALLBACK-WITH-REPORT-OUT-NEXT: bl __fixunstfti
[GlobalISel] Add a fallback path to SDISel. When global-isel fails on a MachineFunction MF, MF will be cleaned up and given to SDISel. Thanks to this fallback, we can already perform correctness test even if we support only a small portion of the functions in a test. llvm-svn: 279891
2016-08-27 08:18:31 +08:00
define i128 @ABIi128(i128 %arg1) {
GlobalISel: improve translation fallback for constants. Returning 0 (NoReg) from getOrCreateVReg leads to unexpected situations later in the translation. It's better to return a valid (if undefined) register and let the rest of the instruction carry on as planned. llvm-svn: 288709
2016-12-06 05:40:33 +08:00
%farg1 = bitcast i128 %arg1 to fp128
[GlobalISel] Add a fallback path to SDISel. When global-isel fails on a MachineFunction MF, MF will be cleaned up and given to SDISel. Thanks to this fallback, we can already perform correctness test even if we support only a small portion of the functions in a test. llvm-svn: 279891
2016-08-27 08:18:31 +08:00
%res = fptoui fp128 %farg1 to i128
ret i128 %res
}
GlobalISel: improve translation fallback for constants. Returning 0 (NoReg) from getOrCreateVReg leads to unexpected situations later in the translation. It's better to return a valid (if undefined) register and let the rest of the instruction carry on as planned. llvm-svn: 288709
2016-12-06 05:40:33 +08:00
Revert r319691: [globalisel][tablegen] Split atomic load/store into separate opcode and enable for AArch64. Some concerns were raised with the direction. Revert while we discuss it and look into an alternative llvm-svn: 319739
2017-12-05 13:52:07 +08:00
; The key problem here is that we may fail to create an MBB referenced by a
; PHI. If so, we cannot complete the G_PHI and mustn't try or bad things
; happen.
[CodeGen] Use MIR syntax for MachineMemOperand printing Get rid of the "; mem:" suffix and use the one we use in MIR: ":: (load 2)". rdar://38163529 Differential Revision: https://reviews.llvm.org/D42377 llvm-svn: 327580
2018-03-15 05:52:13 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: cannot select: G_STORE %6:gpr(s32), %2:gpr(p0) :: (store seq_cst 4 into %ir.addr) (in function: pending_phis)
GlobalISel: avoid looking too closely at PHIs when we bail. The function used to finish off PHIs by adding the relevant basic blocks can fail if we're aborting and still don't actually have the needed MachineBasicBlocks. So avoid trying in that case. llvm-svn: 288727
2016-12-06 07:10:19 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for pending_phis
GlobalISel: stop the legalizer from trying to handle oddly-sized types. It'll almost immediately fail because it always tries to half/double the size until it finds a legal one. Unfortunately, this triggers an assertion preventing the DAG fallback from being possible. llvm-svn: 288834
2016-12-07 02:38:29 +08:00
; FALLBACK-WITH-REPORT-OUT-LABEL: pending_phis:
GlobalISel: avoid looking too closely at PHIs when we bail. The function used to finish off PHIs by adding the relevant basic blocks can fail if we're aborting and still don't actually have the needed MachineBasicBlocks. So avoid trying in that case. llvm-svn: 288727
2016-12-06 07:10:19 +08:00
define i32 @pending_phis(i1 %tst, i32 %val, i32* %addr) {
br i1 %tst, label %true, label %false
end:
%res = phi i32 [%val, %true], [42, %false]
ret i32 %res
true:
Revert r319691: [globalisel][tablegen] Split atomic load/store into separate opcode and enable for AArch64. Some concerns were raised with the direction. Revert while we discuss it and look into an alternative llvm-svn: 319739
2017-12-05 13:52:07 +08:00
store atomic i32 42, i32* %addr seq_cst, align 4
GlobalISel: avoid looking too closely at PHIs when we bail. The function used to finish off PHIs by adding the relevant basic blocks can fail if we're aborting and still don't actually have the needed MachineBasicBlocks. So avoid trying in that case. llvm-svn: 288727
2016-12-06 07:10:19 +08:00
br label %end
false:
br label %end
GlobalISel: improve translation fallback for constants. Returning 0 (NoReg) from getOrCreateVReg leads to unexpected situations later in the translation. It's better to return a valid (if undefined) register and let the rest of the instruction carry on as planned. llvm-svn: 288709
2016-12-06 05:40:33 +08:00
}
GlobalISel: stop the legalizer from trying to handle oddly-sized types. It'll almost immediately fail because it always tries to half/double the size until it finds a legal one. Unfortunately, this triggers an assertion preventing the DAG fallback from being possible. llvm-svn: 288834
2016-12-07 02:38:29 +08:00
Re-commit: [globalisel] Add a combiner helpers for extending loads and use them in a pre-legalize combiner for AArch64 Summary: Depends on D45541 Reviewers: ab, aditya_nandakumar, bogner, rtereshin, volkan, rovka, javed.absar, aemerson Subscribers: aemerson, rengolin, mgorny, javed.absar, kristof.beyls, llvm-commits Differential Revision: https://reviews.llvm.org/D45543 The previous commit failed portions of the test-suite on GreenDragon due to duplicate COPY instructions and iterator invalidation. Both issues have now been fixed. To assist with this, a helper (cloneVirtualRegister) has been added to MachineRegisterInfo that can be used to get another register that has the same type and class/bank as an existing one. llvm-svn: 343654
2018-10-03 10:12:17 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to legalize instruction: %2:_(s32) = G_ZEXTLOAD %1:_(p0) :: (load 3 from `i24* undef`, align 1) (in function: odd_type_load)
[GlobalISel] Fix assert failure when legalizing non-power-2 loads. Until we support extending loads properly we're going to fall back for these. We already handle stores in the same way, so this is just being consistent. llvm-svn: 324001
2018-02-02 04:47:03 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for odd_type_load
; FALLBACK-WITH-REPORT-OUT-LABEL: odd_type_load
define i32 @odd_type_load() {
entry:
%ld = load i24, i24* undef, align 1
%cst = zext i24 %ld to i32
ret i32 %cst
}
GlobalISel: handle G_SEQUENCE fallbacks gracefully. There were two problems: + AArch64 was reusing random data from its binary op tables, which is complete nonsense for G_SEQUENCE. + Even when AArch64 gave up and said it couldn't handle G_SEQUENCE, the generic code asserted. llvm-svn: 288836
2016-12-07 02:38:38 +08:00
; General legalizer inability to handle types whose size wasn't a power of 2.
[CodeGen] Use MIR syntax for MachineMemOperand printing Get rid of the "; mem:" suffix and use the one we use in MIR: ":: (load 2)". rdar://38163529 Differential Revision: https://reviews.llvm.org/D42377 llvm-svn: 327580
2018-03-15 05:52:13 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to legalize instruction: G_STORE %1:_(s42), %0:_(p0) :: (store 6 into %ir.addr, align 8) (in function: odd_type)
GlobalISel: stop the legalizer from trying to handle oddly-sized types. It'll almost immediately fail because it always tries to half/double the size until it finds a legal one. Unfortunately, this triggers an assertion preventing the DAG fallback from being possible. llvm-svn: 288834
2016-12-07 02:38:29 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for odd_type
; FALLBACK-WITH-REPORT-OUT-LABEL: odd_type:
define void @odd_type(i42* %addr) {
%val42 = load i42, i42* %addr
[GISel]: Rework legalization algorithm for better elimination of artifacts along with DCE Legalization Artifacts are all those insts that are there to make the type system happy. Currently, the target needs to say all combinations of extends and truncs are legal and there's no way of verifying that post legalization, we only have *truly* legal instructions. This patch changes roughly the legalization algorithm to process all illegal insts at one go, and then process all truncs/extends that were added to satisfy the type constraints separately trying to combine trivial cases until they converge. This has the added benefit that, the target legalizerinfo can only say which truncs and extends are okay and the artifact combiner would combine away other exts and truncs. Updated legalization algorithm to roughly the following pseudo code. WorkList Insts, Artifacts; collect_all_insts_and_artifacts(Insts, Artifacts); do { for (Inst in Insts) legalizeInstrStep(Inst, Insts, Artifacts); for (Artifact in Artifacts) tryCombineArtifact(Artifact, Insts, Artifacts); } while(!Insts.empty()); Also, wrote a simple wrapper equivalent to SetVector, except for erasing, it avoids moving all elements over by one and instead just nulls them out. llvm-svn: 318210
2017-11-15 06:42:19 +08:00
store i42 %val42, i42* %addr
GlobalISel: stop the legalizer from trying to handle oddly-sized types. It'll almost immediately fail because it always tries to half/double the size until it finds a legal one. Unfortunately, this triggers an assertion preventing the DAG fallback from being possible. llvm-svn: 288834
2016-12-07 02:38:29 +08:00
ret void
}
GlobalISel: handle G_SEQUENCE fallbacks gracefully. There were two problems: + AArch64 was reusing random data from its binary op tables, which is complete nonsense for G_SEQUENCE. + Even when AArch64 gave up and said it couldn't handle G_SEQUENCE, the generic code asserted. llvm-svn: 288836
2016-12-07 02:38:38 +08:00
[CodeGen] Use MIR syntax for MachineMemOperand printing Get rid of the "; mem:" suffix and use the one we use in MIR: ":: (load 2)". rdar://38163529 Differential Revision: https://reviews.llvm.org/D42377 llvm-svn: 327580
2018-03-15 05:52:13 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to legalize instruction: G_STORE %1:_(<7 x s32>), %0:_(p0) :: (store 28 into %ir.addr, align 32) (in function: odd_vector)
[GlobalISel] LegalizerInfo: Enable legalization of non-power-of-2 types Summary: Legalize only if the type is marked as Legal or Custom. If not, return Unsupported as LegalizerHelper is not able to handle non-power-of-2 types right now. Reviewers: qcolombet, aditya_nandakumar, dsanders, t.p.northover, kristof.beyls, javed.absar, ab Reviewed By: kristof.beyls, ab Subscribers: dberris, rovka, igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D31711 llvm-svn: 299929
2017-04-11 18:10:14 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for odd_vector
; FALLBACK-WITH-REPORT-OUT-LABEL: odd_vector:
define void @odd_vector(<7 x i32>* %addr) {
%vec = load <7 x i32>, <7 x i32>* %addr
[GISel]: Rework legalization algorithm for better elimination of artifacts along with DCE Legalization Artifacts are all those insts that are there to make the type system happy. Currently, the target needs to say all combinations of extends and truncs are legal and there's no way of verifying that post legalization, we only have *truly* legal instructions. This patch changes roughly the legalization algorithm to process all illegal insts at one go, and then process all truncs/extends that were added to satisfy the type constraints separately trying to combine trivial cases until they converge. This has the added benefit that, the target legalizerinfo can only say which truncs and extends are okay and the artifact combiner would combine away other exts and truncs. Updated legalization algorithm to roughly the following pseudo code. WorkList Insts, Artifacts; collect_all_insts_and_artifacts(Insts, Artifacts); do { for (Inst in Insts) legalizeInstrStep(Inst, Insts, Artifacts); for (Artifact in Artifacts) tryCombineArtifact(Artifact, Insts, Artifacts); } while(!Insts.empty()); Also, wrote a simple wrapper equivalent to SetVector, except for erasing, it avoids moving all elements over by one and instead just nulls them out. llvm-svn: 318210
2017-11-15 06:42:19 +08:00
store <7 x i32> %vec, <7 x i32>* %addr
[GlobalISel] LegalizerInfo: Enable legalization of non-power-of-2 types Summary: Legalize only if the type is marked as Legal or Custom. If not, return Unsupported as LegalizerHelper is not able to handle non-power-of-2 types right now. Reviewers: qcolombet, aditya_nandakumar, dsanders, t.p.northover, kristof.beyls, javed.absar, ab Reviewed By: kristof.beyls, ab Subscribers: dberris, rovka, igorb, llvm-commits Differential Revision: https://reviews.llvm.org/D31711 llvm-svn: 299929
2017-04-11 18:10:14 +08:00
ret void
}
GlobalISel: fall back gracefully when we can't map an operand's size. AArch64 was asserting when it was asked to provide a register-bank of a size it couldn't deal with (in this case an s128 IMPLICIT_DEF). But we want a robust fallback path so this isn't allowed. llvm-svn: 294248
2017-02-07 05:57:06 +08:00
; AArch64 was asserting instead of returning an invalid mapping for unknown
; sizes.
[GlobalISel] Emit opt remarks on isel fallbacks. Having more fine-grained information on the specific construct that caused us to fallback is valuable for large-scale data collection. We still have the fallback warning, that's also used for FastISel. We still need to remove the fallback warning, and teach FastISel to also emit remarks (it currently has a combination of the warning, stats, and debug prints: the remarks could unify all three). The abort-on-fallback path could also be better handled using remarks: one could imagine a "-Rpass-error", analoguous to "-Werror", which would promote missed/failed remarks to errors. It's not clear whether that would be useful for other remarks though, so we're not there yet. llvm-svn: 296013
2017-02-24 05:05:42 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to translate instruction: ret: ' ret i128 undef' (in function: sequence_sizes)
GlobalISel: fall back gracefully when we can't map an operand's size. AArch64 was asserting when it was asked to provide a register-bank of a size it couldn't deal with (in this case an s128 IMPLICIT_DEF). But we want a robust fallback path so this isn't allowed. llvm-svn: 294248
2017-02-07 05:57:06 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for sequence_sizes
; FALLBACK-WITH-REPORT-LABEL: sequence_sizes:
define i128 @sequence_sizes([8 x i8] %in) {
ret i128 undef
}
GlobalISel: represent atomic loads & stores via the MachineMemOperand. Also make sure the AArch64 backend doesn't try to convert them into normal loads and stores. llvm-svn: 294993
2017-02-14 06:14:16 +08:00
Revert r319691: [globalisel][tablegen] Split atomic load/store into separate opcode and enable for AArch64. Some concerns were raised with the direction. Revert while we discuss it and look into an alternative llvm-svn: 319739
2017-12-05 13:52:07 +08:00
; Just to make sure we don't accidentally emit a normal load/store.
[CodeGen] Use MIR syntax for MachineMemOperand printing Get rid of the "; mem:" suffix and use the one we use in MIR: ":: (load 2)". rdar://38163529 Differential Revision: https://reviews.llvm.org/D42377 llvm-svn: 327580
2018-03-15 05:52:13 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: cannot select: %2:gpr(s64) = G_LOAD %0:gpr(p0) :: (load seq_cst 8 from %ir.addr) (in function: atomic_ops)
Revert r319691: [globalisel][tablegen] Split atomic load/store into separate opcode and enable for AArch64. Some concerns were raised with the direction. Revert while we discuss it and look into an alternative llvm-svn: 319739
2017-12-05 13:52:07 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for atomic_ops
; FALLBACK-WITH-REPORT-LABEL: atomic_ops:
define i64 @atomic_ops(i64* %addr) {
store atomic i64 0, i64* %addr unordered, align 8
%res = load atomic i64, i64* %addr seq_cst, align 8
ret i64 %res
}
[GlobalISel] Don't translate intrinsics with metadata parameters. Some intrinsics take metadata parameters. These all need custom handling of some form, and cannot possibly be lowered generically to G_INTRINSIC calls with vreg operands. Reject them, instead of hitting an assert later in getOrCreateVReg. llvm-svn: 297209
2017-03-08 04:53:09 +08:00
; Make sure we don't mess up metadata arguments.
declare void @llvm.write_register.i64(metadata, i64)
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to translate instruction: call: ' call void @llvm.write_register.i64(metadata !0, i64 0)' (in function: test_write_register_intrin)
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for test_write_register_intrin
; FALLBACK-WITH-REPORT-LABEL: test_write_register_intrin:
define void @test_write_register_intrin() {
call void @llvm.write_register.i64(metadata !{!"sp"}, i64 0)
ret void
}
[GlobalISel] Fallback when failing to translate invoke. We unintentionally stopped falling back in r293670. While there, change an unusual construct. llvm-svn: 297425
2017-03-10 08:25:35 +08:00
@_ZTIi = external global i8*
declare i32 @__gxx_personality_v0(...)
; Check that we fallback on invoke translation failures.
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to translate instruction: invoke: ' invoke void %callee(i128 0)
; FALLBACK-WITH-REPORT-NEXT: to label %continue unwind label %broken' (in function: invoke_weird_type)
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for invoke_weird_type
; FALLBACK-WITH-REPORT-OUT-LABEL: invoke_weird_type:
define void @invoke_weird_type(void(i128)* %callee) personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
invoke void %callee(i128 0)
to label %continue unwind label %broken
broken:
landingpad { i8*, i32 } catch i8* bitcast(i8** @_ZTIi to i8*)
ret void
continue:
ret void
}
GlobalISel: allow quad-precision values to be dumped. Otherwise the fallback path fails with an assertion on AAPCS AArch64 targets, when "long double" is encountered. llvm-svn: 298273
2017-03-21 00:52:08 +08:00
; Check that we fallback on invoke translation failures.
[CodeGen] Move printing MO_FPImmediate operands to MachineOperand::print Work towards the unification of MIR and debug output by refactoring the interfaces. llvm-svn: 321110
2017-12-20 05:47:00 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to legalize instruction: %0:_(s128) = G_FCONSTANT fp128 0xL00000000000000004000000000000000
GlobalISel: allow quad-precision values to be dumped. Otherwise the fallback path fails with an assertion on AAPCS AArch64 targets, when "long double" is encountered. llvm-svn: 298273
2017-03-21 00:52:08 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for test_quad_dump
; FALLBACK-WITH-REPORT-OUT-LABEL: test_quad_dump:
define fp128 @test_quad_dump() {
ret fp128 0xL00000000000000004000000000000000
}
[GlobalISel] Support vector-of-pointers in LLT This fixes PR32471. As comment 10 on that bug report highlights (https://bugs.llvm.org//show_bug.cgi?id=32471#c10), there are quite a few different defendable design tradeoffs that could be made, including not representing pointers at all in LLT. I decided to go for representing vector-of-pointer as a concept in LLT, while keeping the size of the LLT type 64 bits (this is an increase from 48 bits before). My rationale for keeping pointers explicit is that on some targets probably it's very handy to have the distinction between pointer and non-pointer (e.g. 68K has a different register bank for pointers IIRC). If we keep a scalar pointer, it probably is easiest to also have a vector-of-pointers to keep LLT relatively conceptually clean and orthogonal, while we don't have a very strong reason to break that orthogonality. Once we gain more experience on the use of LLT, we can of course reconsider this direction. Rejecting vector-of-pointer types in the IRTranslator is also an option to avoid the crash reported in PR32471, but that is only a very short-term solution; also needs quite a bit of code tweaks in places, and is probably fragile. Therefore I didn't consider this the best option. llvm-svn: 300664
2017-04-19 15:23:57 +08:00
[GlobalISel] Use the target preferred type for G_EXTRACT_VECTOR_ELT index. Allows for better imported pattern re-use. llvm-svn: 345265
2018-10-25 22:04:54 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to legalize instruction: %2:_(p0) = G_EXTRACT_VECTOR_ELT %0:_(<2 x p0>), %3:_(s64) (in function: vector_of_pointers_extractelement)
[GlobalISel] Support vector-of-pointers in LLT This fixes PR32471. As comment 10 on that bug report highlights (https://bugs.llvm.org//show_bug.cgi?id=32471#c10), there are quite a few different defendable design tradeoffs that could be made, including not representing pointers at all in LLT. I decided to go for representing vector-of-pointer as a concept in LLT, while keeping the size of the LLT type 64 bits (this is an increase from 48 bits before). My rationale for keeping pointers explicit is that on some targets probably it's very handy to have the distinction between pointer and non-pointer (e.g. 68K has a different register bank for pointers IIRC). If we keep a scalar pointer, it probably is easiest to also have a vector-of-pointers to keep LLT relatively conceptually clean and orthogonal, while we don't have a very strong reason to break that orthogonality. Once we gain more experience on the use of LLT, we can of course reconsider this direction. Rejecting vector-of-pointer types in the IRTranslator is also an option to avoid the crash reported in PR32471, but that is only a very short-term solution; also needs quite a bit of code tweaks in places, and is probably fragile. Therefore I didn't consider this the best option. llvm-svn: 300664
2017-04-19 15:23:57 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for vector_of_pointers_extractelement
; FALLBACK-WITH-REPORT-OUT-LABEL: vector_of_pointers_extractelement:
GlobalISel: add G_IMPLICIT_DEF instruction. It looks like there are two target-independent but not GISel instructions that need legalization, IMPLICIT_DEF and PHI. These are already anomalies since their operands have important LLTs attached, so to make things more uniform it seems like a good idea to add generic variants. Starting with G_IMPLICIT_DEF. llvm-svn: 306875
2017-07-01 04:27:36 +08:00
@var = global <2 x i16*> zeroinitializer
[GlobalISel] Support vector-of-pointers in LLT This fixes PR32471. As comment 10 on that bug report highlights (https://bugs.llvm.org//show_bug.cgi?id=32471#c10), there are quite a few different defendable design tradeoffs that could be made, including not representing pointers at all in LLT. I decided to go for representing vector-of-pointer as a concept in LLT, while keeping the size of the LLT type 64 bits (this is an increase from 48 bits before). My rationale for keeping pointers explicit is that on some targets probably it's very handy to have the distinction between pointer and non-pointer (e.g. 68K has a different register bank for pointers IIRC). If we keep a scalar pointer, it probably is easiest to also have a vector-of-pointers to keep LLT relatively conceptually clean and orthogonal, while we don't have a very strong reason to break that orthogonality. Once we gain more experience on the use of LLT, we can of course reconsider this direction. Rejecting vector-of-pointer types in the IRTranslator is also an option to avoid the crash reported in PR32471, but that is only a very short-term solution; also needs quite a bit of code tweaks in places, and is probably fragile. Therefore I didn't consider this the best option. llvm-svn: 300664
2017-04-19 15:23:57 +08:00
define void @vector_of_pointers_extractelement() {
GlobalISel: add G_IMPLICIT_DEF instruction. It looks like there are two target-independent but not GISel instructions that need legalization, IMPLICIT_DEF and PHI. These are already anomalies since their operands have important LLTs attached, so to make things more uniform it seems like a good idea to add generic variants. Starting with G_IMPLICIT_DEF. llvm-svn: 306875
2017-07-01 04:27:36 +08:00
br label %end
block:
%dummy = extractelement <2 x i16*> %vec, i32 0
[GISel]: Rework legalization algorithm for better elimination of artifacts along with DCE Legalization Artifacts are all those insts that are there to make the type system happy. Currently, the target needs to say all combinations of extends and truncs are legal and there's no way of verifying that post legalization, we only have *truly* legal instructions. This patch changes roughly the legalization algorithm to process all illegal insts at one go, and then process all truncs/extends that were added to satisfy the type constraints separately trying to combine trivial cases until they converge. This has the added benefit that, the target legalizerinfo can only say which truncs and extends are okay and the artifact combiner would combine away other exts and truncs. Updated legalization algorithm to roughly the following pseudo code. WorkList Insts, Artifacts; collect_all_insts_and_artifacts(Insts, Artifacts); do { for (Inst in Insts) legalizeInstrStep(Inst, Insts, Artifacts); for (Artifact in Artifacts) tryCombineArtifact(Artifact, Insts, Artifacts); } while(!Insts.empty()); Also, wrote a simple wrapper equivalent to SetVector, except for erasing, it avoids moving all elements over by one and instead just nulls them out. llvm-svn: 318210
2017-11-15 06:42:19 +08:00
store i16* %dummy, i16** undef
[GlobalISel] Support vector-of-pointers in LLT This fixes PR32471. As comment 10 on that bug report highlights (https://bugs.llvm.org//show_bug.cgi?id=32471#c10), there are quite a few different defendable design tradeoffs that could be made, including not representing pointers at all in LLT. I decided to go for representing vector-of-pointer as a concept in LLT, while keeping the size of the LLT type 64 bits (this is an increase from 48 bits before). My rationale for keeping pointers explicit is that on some targets probably it's very handy to have the distinction between pointer and non-pointer (e.g. 68K has a different register bank for pointers IIRC). If we keep a scalar pointer, it probably is easiest to also have a vector-of-pointers to keep LLT relatively conceptually clean and orthogonal, while we don't have a very strong reason to break that orthogonality. Once we gain more experience on the use of LLT, we can of course reconsider this direction. Rejecting vector-of-pointer types in the IRTranslator is also an option to avoid the crash reported in PR32471, but that is only a very short-term solution; also needs quite a bit of code tweaks in places, and is probably fragile. Therefore I didn't consider this the best option. llvm-svn: 300664
2017-04-19 15:23:57 +08:00
ret void
GlobalISel: add G_IMPLICIT_DEF instruction. It looks like there are two target-independent but not GISel instructions that need legalization, IMPLICIT_DEF and PHI. These are already anomalies since their operands have important LLTs attached, so to make things more uniform it seems like a good idea to add generic variants. Starting with G_IMPLICIT_DEF. llvm-svn: 306875
2017-07-01 04:27:36 +08:00
end:
%vec = load <2 x i16*>, <2 x i16*>* undef
br label %block
[GlobalISel] Support vector-of-pointers in LLT This fixes PR32471. As comment 10 on that bug report highlights (https://bugs.llvm.org//show_bug.cgi?id=32471#c10), there are quite a few different defendable design tradeoffs that could be made, including not representing pointers at all in LLT. I decided to go for representing vector-of-pointer as a concept in LLT, while keeping the size of the LLT type 64 bits (this is an increase from 48 bits before). My rationale for keeping pointers explicit is that on some targets probably it's very handy to have the distinction between pointer and non-pointer (e.g. 68K has a different register bank for pointers IIRC). If we keep a scalar pointer, it probably is easiest to also have a vector-of-pointers to keep LLT relatively conceptually clean and orthogonal, while we don't have a very strong reason to break that orthogonality. Once we gain more experience on the use of LLT, we can of course reconsider this direction. Rejecting vector-of-pointer types in the IRTranslator is also an option to avoid the crash reported in PR32471, but that is only a very short-term solution; also needs quite a bit of code tweaks in places, and is probably fragile. Therefore I didn't consider this the best option. llvm-svn: 300664
2017-04-19 15:23:57 +08:00
}
[GlobalISel][IRTranslator] Use RPO traversal when visiting blocks to translate. 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
2018-08-01 10:17:42 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to legalize instruction: G_STORE %2:_(<2 x p0>), %1:_(p0) :: (store 16 into `<2 x i16*>* undef`) (in function: vector_of_pointers_insertelement)
[GlobalISel] Support vector-of-pointers in LLT This fixes PR32471. As comment 10 on that bug report highlights (https://bugs.llvm.org//show_bug.cgi?id=32471#c10), there are quite a few different defendable design tradeoffs that could be made, including not representing pointers at all in LLT. I decided to go for representing vector-of-pointer as a concept in LLT, while keeping the size of the LLT type 64 bits (this is an increase from 48 bits before). My rationale for keeping pointers explicit is that on some targets probably it's very handy to have the distinction between pointer and non-pointer (e.g. 68K has a different register bank for pointers IIRC). If we keep a scalar pointer, it probably is easiest to also have a vector-of-pointers to keep LLT relatively conceptually clean and orthogonal, while we don't have a very strong reason to break that orthogonality. Once we gain more experience on the use of LLT, we can of course reconsider this direction. Rejecting vector-of-pointer types in the IRTranslator is also an option to avoid the crash reported in PR32471, but that is only a very short-term solution; also needs quite a bit of code tweaks in places, and is probably fragile. Therefore I didn't consider this the best option. llvm-svn: 300664
2017-04-19 15:23:57 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for vector_of_pointers_insertelement
; FALLBACK-WITH-REPORT-OUT-LABEL: vector_of_pointers_insertelement:
define void @vector_of_pointers_insertelement() {
GlobalISel: add G_IMPLICIT_DEF instruction. It looks like there are two target-independent but not GISel instructions that need legalization, IMPLICIT_DEF and PHI. These are already anomalies since their operands have important LLTs attached, so to make things more uniform it seems like a good idea to add generic variants. Starting with G_IMPLICIT_DEF. llvm-svn: 306875
2017-07-01 04:27:36 +08:00
br label %end
block:
%dummy = insertelement <2 x i16*> %vec, i16* null, i32 0
[GISel]: Rework legalization algorithm for better elimination of artifacts along with DCE Legalization Artifacts are all those insts that are there to make the type system happy. Currently, the target needs to say all combinations of extends and truncs are legal and there's no way of verifying that post legalization, we only have *truly* legal instructions. This patch changes roughly the legalization algorithm to process all illegal insts at one go, and then process all truncs/extends that were added to satisfy the type constraints separately trying to combine trivial cases until they converge. This has the added benefit that, the target legalizerinfo can only say which truncs and extends are okay and the artifact combiner would combine away other exts and truncs. Updated legalization algorithm to roughly the following pseudo code. WorkList Insts, Artifacts; collect_all_insts_and_artifacts(Insts, Artifacts); do { for (Inst in Insts) legalizeInstrStep(Inst, Insts, Artifacts); for (Artifact in Artifacts) tryCombineArtifact(Artifact, Insts, Artifacts); } while(!Insts.empty()); Also, wrote a simple wrapper equivalent to SetVector, except for erasing, it avoids moving all elements over by one and instead just nulls them out. llvm-svn: 318210
2017-11-15 06:42:19 +08:00
store <2 x i16*> %dummy, <2 x i16*>* undef
[GlobalISel] Support vector-of-pointers in LLT This fixes PR32471. As comment 10 on that bug report highlights (https://bugs.llvm.org//show_bug.cgi?id=32471#c10), there are quite a few different defendable design tradeoffs that could be made, including not representing pointers at all in LLT. I decided to go for representing vector-of-pointer as a concept in LLT, while keeping the size of the LLT type 64 bits (this is an increase from 48 bits before). My rationale for keeping pointers explicit is that on some targets probably it's very handy to have the distinction between pointer and non-pointer (e.g. 68K has a different register bank for pointers IIRC). If we keep a scalar pointer, it probably is easiest to also have a vector-of-pointers to keep LLT relatively conceptually clean and orthogonal, while we don't have a very strong reason to break that orthogonality. Once we gain more experience on the use of LLT, we can of course reconsider this direction. Rejecting vector-of-pointer types in the IRTranslator is also an option to avoid the crash reported in PR32471, but that is only a very short-term solution; also needs quite a bit of code tweaks in places, and is probably fragile. Therefore I didn't consider this the best option. llvm-svn: 300664
2017-04-19 15:23:57 +08:00
ret void
GlobalISel: add G_IMPLICIT_DEF instruction. It looks like there are two target-independent but not GISel instructions that need legalization, IMPLICIT_DEF and PHI. These are already anomalies since their operands have important LLTs attached, so to make things more uniform it seems like a good idea to add generic variants. Starting with G_IMPLICIT_DEF. llvm-svn: 306875
2017-07-01 04:27:36 +08:00
end:
%vec = load <2 x i16*>, <2 x i16*>* undef
br label %block
[GlobalISel] Support vector-of-pointers in LLT This fixes PR32471. As comment 10 on that bug report highlights (https://bugs.llvm.org//show_bug.cgi?id=32471#c10), there are quite a few different defendable design tradeoffs that could be made, including not representing pointers at all in LLT. I decided to go for representing vector-of-pointer as a concept in LLT, while keeping the size of the LLT type 64 bits (this is an increase from 48 bits before). My rationale for keeping pointers explicit is that on some targets probably it's very handy to have the distinction between pointer and non-pointer (e.g. 68K has a different register bank for pointers IIRC). If we keep a scalar pointer, it probably is easiest to also have a vector-of-pointers to keep LLT relatively conceptually clean and orthogonal, while we don't have a very strong reason to break that orthogonality. Once we gain more experience on the use of LLT, we can of course reconsider this direction. Rejecting vector-of-pointer types in the IRTranslator is also an option to avoid the crash reported in PR32471, but that is only a very short-term solution; also needs quite a bit of code tweaks in places, and is probably fragile. Therefore I didn't consider this the best option. llvm-svn: 300664
2017-04-19 15:23:57 +08:00
}
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
[CodeGen] Use MIR syntax for MachineMemOperand printing Get rid of the "; mem:" suffix and use the one we use in MIR: ":: (load 2)". rdar://38163529 Differential Revision: https://reviews.llvm.org/D42377 llvm-svn: 327580
2018-03-15 05:52:13 +08:00
; FALLBACK-WITH-REPORT-ERR remark: <unknown>:0:0: unable to legalize instruction: G_STORE %3, %4 :: (store 12 into `i96* undef`, align 16) (in function: nonpow2_add_narrowing)
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for nonpow2_add_narrowing
; FALLBACK-WITH-REPORT-OUT-LABEL: nonpow2_add_narrowing:
define void @nonpow2_add_narrowing() {
%a = add i128 undef, undef
%b = trunc i128 %a to i96
%dummy = add i96 %b, %b
[GISel]: Rework legalization algorithm for better elimination of artifacts along with DCE Legalization Artifacts are all those insts that are there to make the type system happy. Currently, the target needs to say all combinations of extends and truncs are legal and there's no way of verifying that post legalization, we only have *truly* legal instructions. This patch changes roughly the legalization algorithm to process all illegal insts at one go, and then process all truncs/extends that were added to satisfy the type constraints separately trying to combine trivial cases until they converge. This has the added benefit that, the target legalizerinfo can only say which truncs and extends are okay and the artifact combiner would combine away other exts and truncs. Updated legalization algorithm to roughly the following pseudo code. WorkList Insts, Artifacts; collect_all_insts_and_artifacts(Insts, Artifacts); do { for (Inst in Insts) legalizeInstrStep(Inst, Insts, Artifacts); for (Artifact in Artifacts) tryCombineArtifact(Artifact, Insts, Artifacts); } while(!Insts.empty()); Also, wrote a simple wrapper equivalent to SetVector, except for erasing, it avoids moving all elements over by one and instead just nulls them out. llvm-svn: 318210
2017-11-15 06:42:19 +08:00
store i96 %dummy, i96* undef
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
ret void
}
[CodeGen] Use MIR syntax for MachineMemOperand printing Get rid of the "; mem:" suffix and use the one we use in MIR: ":: (load 2)". rdar://38163529 Differential Revision: https://reviews.llvm.org/D42377 llvm-svn: 327580
2018-03-15 05:52:13 +08:00
; FALLBACK-WITH-REPORT-ERR remark: <unknown>:0:0: unable to legalize instruction: G_STORE %3, %4 :: (store 12 into `i96* undef`, align 16) (in function: nonpow2_add_narrowing)
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for nonpow2_or_narrowing
; FALLBACK-WITH-REPORT-OUT-LABEL: nonpow2_or_narrowing:
define void @nonpow2_or_narrowing() {
%a = add i128 undef, undef
%b = trunc i128 %a to i96
%dummy = or i96 %b, %b
[GISel]: Rework legalization algorithm for better elimination of artifacts along with DCE Legalization Artifacts are all those insts that are there to make the type system happy. Currently, the target needs to say all combinations of extends and truncs are legal and there's no way of verifying that post legalization, we only have *truly* legal instructions. This patch changes roughly the legalization algorithm to process all illegal insts at one go, and then process all truncs/extends that were added to satisfy the type constraints separately trying to combine trivial cases until they converge. This has the added benefit that, the target legalizerinfo can only say which truncs and extends are okay and the artifact combiner would combine away other exts and truncs. Updated legalization algorithm to roughly the following pseudo code. WorkList Insts, Artifacts; collect_all_insts_and_artifacts(Insts, Artifacts); do { for (Inst in Insts) legalizeInstrStep(Inst, Insts, Artifacts); for (Artifact in Artifacts) tryCombineArtifact(Artifact, Insts, Artifacts); } while(!Insts.empty()); Also, wrote a simple wrapper equivalent to SetVector, except for erasing, it avoids moving all elements over by one and instead just nulls them out. llvm-svn: 318210
2017-11-15 06:42:19 +08:00
store i96 %dummy, i96* undef
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
ret void
}
[CodeGen] Use MIR syntax for MachineMemOperand printing Get rid of the "; mem:" suffix and use the one we use in MIR: ":: (load 2)". rdar://38163529 Differential Revision: https://reviews.llvm.org/D42377 llvm-svn: 327580
2018-03-15 05:52:13 +08:00
; FALLBACK-WITH-REPORT-ERR remark: <unknown>:0:0: unable to legalize instruction: G_STORE %0, %1 :: (store 12 into `i96* undef`, align 16) (in function: nonpow2_load_narrowing)
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for nonpow2_load_narrowing
; FALLBACK-WITH-REPORT-OUT-LABEL: nonpow2_load_narrowing:
define void @nonpow2_load_narrowing() {
%dummy = load i96, i96* undef
[GISel]: Rework legalization algorithm for better elimination of artifacts along with DCE Legalization Artifacts are all those insts that are there to make the type system happy. Currently, the target needs to say all combinations of extends and truncs are legal and there's no way of verifying that post legalization, we only have *truly* legal instructions. This patch changes roughly the legalization algorithm to process all illegal insts at one go, and then process all truncs/extends that were added to satisfy the type constraints separately trying to combine trivial cases until they converge. This has the added benefit that, the target legalizerinfo can only say which truncs and extends are okay and the artifact combiner would combine away other exts and truncs. Updated legalization algorithm to roughly the following pseudo code. WorkList Insts, Artifacts; collect_all_insts_and_artifacts(Insts, Artifacts); do { for (Inst in Insts) legalizeInstrStep(Inst, Insts, Artifacts); for (Artifact in Artifacts) tryCombineArtifact(Artifact, Insts, Artifacts); } while(!Insts.empty()); Also, wrote a simple wrapper equivalent to SetVector, except for erasing, it avoids moving all elements over by one and instead just nulls them out. llvm-svn: 318210
2017-11-15 06:42:19 +08:00
store i96 %dummy, i96* undef
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
ret void
}
GlobalISel: Consolidate load/store legalization The fewerElementsVectors implementation for load/stores handles the scalar reduction case just as well, so drop the redundant code in narrowScalar. This also introduces support for narrowing irregular size breakdowns for scalars. llvm-svn: 353125
2019-02-05 08:26:12 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to legalize instruction: %4:_(s64) = G_EXTRACT %3:_(s96), 0 (in function: nonpow2_store_narrowing)
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for nonpow2_store_narrowing
; FALLBACK-WITH-REPORT-OUT-LABEL: nonpow2_store_narrowing:
define void @nonpow2_store_narrowing(i96* %c) {
%a = add i128 undef, undef
%b = trunc i128 %a to i96
store i96 %b, i96* %c
ret void
}
GlobalISel: Consolidate load/store legalization The fewerElementsVectors implementation for load/stores handles the scalar reduction case just as well, so drop the redundant code in narrowScalar. This also introduces support for narrowing irregular size breakdowns for scalars. llvm-svn: 353125
2019-02-05 08:26:12 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to legalize instruction: %0:_(s96) = G_CONSTANT i96 0 (in function: nonpow2_constant_narrowing)
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for nonpow2_constant_narrowing
; FALLBACK-WITH-REPORT-OUT-LABEL: nonpow2_constant_narrowing:
define void @nonpow2_constant_narrowing() {
store i96 0, i96* undef
ret void
}
; Currently can't handle vector lengths that aren't an exact multiple of
; natively supported vector lengths. Test that the fall-back works for those.
[CodeGen] Use MIR syntax for MachineMemOperand printing Get rid of the "; mem:" suffix and use the one we use in MIR: ":: (load 2)". rdar://38163529 Differential Revision: https://reviews.llvm.org/D42377 llvm-svn: 327580
2018-03-15 05:52:13 +08:00
; FALLBACK-WITH-REPORT-ERR-G_IMPLICIT_DEF-LEGALIZABLE: (FIXME: this is what is expected once we can legalize non-pow-of-2 G_IMPLICIT_DEF) remark: <unknown>:0:0: unable to legalize instruction: %1:_(<7 x s64>) = G_ADD %0, %0 (in function: nonpow2_vector_add_fewerelements
[CodeGen] Print RegClasses on MI in verbose mode r322086 removed the trailing information describing reg classes for each register. This patch adds printing reg classes next to every register when individual operands/instructions/basic blocks are printed. In the case of dumping MIR or printing a full function, by default don't print it. Differential Revision: https://reviews.llvm.org/D42239 llvm-svn: 322867
2018-01-19 01:59:06 +08:00
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to legalize instruction: %2:_(s64) = G_EXTRACT_VECTOR_ELT %1:_(<7 x s64>), %3:_(s64) (in function: nonpow2_vector_add_fewerelements)
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for nonpow2_vector_add_fewerelements
; FALLBACK-WITH-REPORT-OUT-LABEL: nonpow2_vector_add_fewerelements:
define void @nonpow2_vector_add_fewerelements() {
%dummy = add <7 x i64> undef, undef
[GISel]: Rework legalization algorithm for better elimination of artifacts along with DCE Legalization Artifacts are all those insts that are there to make the type system happy. Currently, the target needs to say all combinations of extends and truncs are legal and there's no way of verifying that post legalization, we only have *truly* legal instructions. This patch changes roughly the legalization algorithm to process all illegal insts at one go, and then process all truncs/extends that were added to satisfy the type constraints separately trying to combine trivial cases until they converge. This has the added benefit that, the target legalizerinfo can only say which truncs and extends are okay and the artifact combiner would combine away other exts and truncs. Updated legalization algorithm to roughly the following pseudo code. WorkList Insts, Artifacts; collect_all_insts_and_artifacts(Insts, Artifacts); do { for (Inst in Insts) legalizeInstrStep(Inst, Insts, Artifacts); for (Artifact in Artifacts) tryCombineArtifact(Artifact, Insts, Artifacts); } while(!Insts.empty()); Also, wrote a simple wrapper equivalent to SetVector, except for erasing, it avoids moving all elements over by one and instead just nulls them out. llvm-svn: 318210
2017-11-15 06:42:19 +08:00
%ex = extractelement <7 x i64> %dummy, i64 0
store i64 %ex, i64* undef
[GlobalISel] Enable legalizing non-power-of-2 sized types. This changes the interface of how targets describe how to legalize, see the below description. 1. Interface for targets to describe how to legalize. In GlobalISel, the API in the LegalizerInfo class is the main interface for targets to specify which types are legal for which operations, and what to do to turn illegal type/operation combinations into legal ones. For each operation the type sizes that can be legalized without having to change the size of the type are specified with a call to setAction. This isn't different to how GlobalISel worked before. For example, for a target that supports 32 and 64 bit adds natively: for (auto Ty : {s32, s64}) setAction({G_ADD, 0, s32}, Legal); or for a target that needs a library call for a 32 bit division: setAction({G_SDIV, s32}, Libcall); The main conceptual change to the LegalizerInfo API, is in specifying how to legalize the type sizes for which a change of size is needed. For example, in the above example, how to specify how all types from i1 to i8388607 (apart from s32 and s64 which are legal) need to be legalized and expressed in terms of operations on the available legal sizes (again, i32 and i64 in this case). Before, the implementation only allowed specifying power-of-2-sized types (e.g. setAction({G_ADD, 0, s128}, NarrowScalar). A worse limitation was that if you'd wanted to specify how to legalize all the sized types as allowed by the LLVM-IR LangRef, i1 to i8388607, you'd have to call setAction 8388607-3 times and probably would need a lot of memory to store all of these specifications. Instead, the legalization actions that need to change the size of the type are specified now using a "SizeChangeStrategy". For example: setLegalizeScalarToDifferentSizeStrategy( G_ADD, 0, widenToLargerAndNarrowToLargest); This example indicates that for type sizes for which there is a larger size that can be legalized towards, do it by Widening the size. For example, G_ADD on s17 will be legalized by first doing WidenScalar to make it s32, after which it's legal. The "NarrowToLargest" indicates what to do if there is no larger size that can be legalized towards. E.g. G_ADD on s92 will be legalized by doing NarrowScalar to s64. Another example, taken from the ARM backend is: for (unsigned Op : {G_SDIV, G_UDIV}) { setLegalizeScalarToDifferentSizeStrategy(Op, 0, widenToLargerTypesUnsupportedOtherwise); if (ST.hasDivideInARMMode()) setAction({Op, s32}, Legal); else setAction({Op, s32}, Libcall); } For this example, G_SDIV on s8, on a target without a divide instruction, would be legalized by first doing action (WidenScalar, s32), followed by (Libcall, s32). The same principle is also followed for when the number of vector lanes on vector data types need to be changed, e.g.: setAction({G_ADD, LLT::vector(8, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(16, 8)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(8, 16)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(2, 32)}, LegalizerInfo::Legal); setAction({G_ADD, LLT::vector(4, 32)}, LegalizerInfo::Legal); setLegalizeVectorElementToDifferentSizeStrategy( G_ADD, 0, widenToLargerTypesUnsupportedOtherwise); As currently implemented here, vector types are legalized by first making the vector element size legal, followed by then making the number of lanes legal. The strategy to follow in the first step is set by a call to setLegalizeVectorElementToDifferentSizeStrategy, see example above. The strategy followed in the second step "moreToWiderTypesAndLessToWidest" (see code for its definition), indicating that vectors are widened to more elements so they map to natively supported vector widths, or when there isn't a legal wider vector, split the vector to map it to the widest vector supported. Therefore, for the above specification, some example legalizations are: * getAction({G_ADD, LLT::vector(3, 3)}) returns {WidenScalar, LLT::vector(3, 8)} * getAction({G_ADD, LLT::vector(3, 8)}) then returns {MoreElements, LLT::vector(8, 8)} * getAction({G_ADD, LLT::vector(20, 8)}) returns {FewerElements, LLT::vector(16, 8)} 2. Key implementation aspects. How to legalize a specific (operation, type index, size) tuple is represented by mapping intervals of integers representing a range of size types to an action to take, e.g.: setScalarAction({G_ADD, LLT:scalar(1)}, {{1, WidenScalar}, // bit sizes [ 1, 31[ {32, Legal}, // bit sizes [32, 33[ {33, WidenScalar}, // bit sizes [33, 64[ {64, Legal}, // bit sizes [64, 65[ {65, NarrowScalar} // bit sizes [65, +inf[ }); Please note that most of the code to do the actual lowering of non-power-of-2 sized types is currently missing, this is just trying to make it possible for targets to specify what is legal, and how non-legal types should be legalized. Probably quite a bit of further work is needed in the actual legalizing and the other passes in GlobalISel to support non-power-of-2 sized types. I hope the documentation in LegalizerInfo.h and the examples provided in the various {Target}LegalizerInfo.cpp and LegalizerInfoTest.cpp explains well enough how this is meant to be used. This drops the need for LLT::{half,double}...Size(). Differential Revision: https://reviews.llvm.org/D30529 llvm-svn: 317560
2017-11-07 18:34:34 +08:00
ret void
}
[GlobalISel] Fall back to SDISel for swifterror/swiftself attributes. We don't currently support these, fall back until we do. llvm-svn: 337994
2018-07-26 09:25:58 +08:00
%swift_error = type {i64, i8}
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to lower arguments due to swifterror/swiftself: void (%swift_error**)* (in function: swifterror_param)
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for swifterror_param
define void @swifterror_param(%swift_error** swifterror %error_ptr_ref) {
ret void
}
; FALLBACK-WITH-REPORT-ERR: remark: <unknown>:0:0: unable to translate instruction: alloca: ' %error_ptr_ref = alloca swifterror %swift_error*' (in function: swifterror_alloca)
; FALLBACK-WITH-REPORT-ERR: warning: Instruction selection used fallback path for swifterror_alloca
; We can't currently test the call parameters being swifterror because the value
; must come from a swifterror alloca or parameter, at which point we already
; fallback. As long as those cases work however we should be fine.
define void @swifterror_alloca(i8* %error_ref) {
entry:
%error_ptr_ref = alloca swifterror %swift_error*
store %swift_error* null, %swift_error** %error_ptr_ref
call void @swifterror_param(%swift_error** swifterror %error_ptr_ref)
ret void
}