arm64: lift sxtl, sxtl2, sshll, sshll2#5461
Closed
yrp604 wants to merge 1 commit into
Closed
Conversation
|
|
galenbwill
added
Type: Enhancement
Arch: ARM64
Contributor
|
Closing PR because superseded by e3b91f9 |
galenbwill
added a commit
that referenced
this pull request
Jul 8, 2024
…, improved, and/or fixed Merged #5461: Author: yrp <yrp604@protonmail.com> Date: Sat May 25 21:00:26 2024 -0700 arm64: lift sxtl, sxtl2, sshll, sshll2 Partial list of detailed changes squashed into this commit (see https://github.com/Vector35/binaryninja-api/tree/arm64_improving_intrinsics for detailed commit history): * add lifting for sshll/sxtl * reverted neon_intrinsics.cpp to restore scvtf intrinsics * lifted sxtl/2, sshll, ushll, sshl, sshr, ushl, ushr, and changed the lifting of uxtl/2 to be consistent with sxtl/2 * reformatted arm64test.py and added tests for sxtl/2, sshll, ushll, sshl, sshr, ushl, ushr, and uxtl/2 fix scvtf (unroll because no intrinsic) and fsub (missing register assignment) half-precision vector cases * added preferIntrinsics setting to arm64 * added lifting for movn * fixed incorrect int/float conversions for FMOV, made half-precision immediates survive the lift to M/HLIL * fix missing break in SCVT; optimize MOVK * improved preferIntrinsics * fixed bad lifting introduced for movn * fixed bad settings definition for preferIntrinsics * added intrinsic definition for DUP from general register * added direct lifting of scalar version of FADDP, and fixed intrinsics for vector version * added direct lifting of scalar version of FABD, and fixed intrinsics for vector version * fixes to test_gen.py: gets the correct encoding instead of sometimes getting fooled by the mnemonic * fixed lifting of UCVTF; reviewed/fixed all intrinsics through SQXTUN * reviewed/fixed remaining intrinsics after SQXTUN * added lifting for FNMUL * WIP intrinsics improvements * WIP intrinsics improvements 2 * WIP intrinsics improvements: FCVT*_asisdmisc_R * added B.AL, B.NV, CASP* * direct lifting of scalar FSQRT instruction * SETREG now elides setting of targeting zero registers * fixed test_gen.py to correctly regenerate arm64test.py * unroll vector MOV operations, USHL no longer uses intrinsic for scalars * updated existing tests in arm64test.py for latest lifting changes * fixed CASH* and CASB* incorrectly accessing temp register in comparison (resulting in comparing to NOP) * lifting all variants of TBL as intrinsic * fixes/improvements to test_gen.py * lifting all variants of TBX as intrinsic * added tests for CAS*, UMUL*, UADD*, FABD, FABS, FADDP, FMAX, FMAXNM, FMIN, FMINNM, FNEG, FNMUL, FCMEQ, FCMGE, FCMGT, FMLA, FMLS * added tests for all aliases of SBFM
galenbwill
added a commit
that referenced
this pull request
Jul 8, 2024
…, improved, and/or fixed Merged #5461: Author: yrp <yrp604@protonmail.com> Date: Sat May 25 21:00:26 2024 -0700 arm64: lift sxtl, sxtl2, sshll, sshll2 Partial list of detailed changes squashed into this commit (see https://github.com/Vector35/binaryninja-api/tree/arm64_improving_intrinsics for detailed commit history): * add lifting for sshll/sxtl * reverted neon_intrinsics.cpp to restore scvtf intrinsics * lifted sxtl/2, sshll, ushll, sshl, sshr, ushl, ushr, and changed the lifting of uxtl/2 to be consistent with sxtl/2 * reformatted arm64test.py and added tests for sxtl/2, sshll, ushll, sshl, sshr, ushl, ushr, and uxtl/2 fix scvtf (unroll because no intrinsic) and fsub (missing register assignment) half-precision vector cases * added preferIntrinsics setting to arm64 * added lifting for movn * fixed incorrect int/float conversions for FMOV, made half-precision immediates survive the lift to M/HLIL * fix missing break in SCVT; optimize MOVK * improved preferIntrinsics * fixed bad lifting introduced for movn * fixed bad settings definition for preferIntrinsics * added intrinsic definition for DUP from general register * added direct lifting of scalar version of FADDP, and fixed intrinsics for vector version * added direct lifting of scalar version of FABD, and fixed intrinsics for vector version * fixes to test_gen.py: gets the correct encoding instead of sometimes getting fooled by the mnemonic * fixed lifting of UCVTF; reviewed/fixed all intrinsics through SQXTUN * reviewed/fixed remaining intrinsics after SQXTUN * added lifting for FNMUL * WIP intrinsics improvements * WIP intrinsics improvements 2 * WIP intrinsics improvements: FCVT*_asisdmisc_R * added B.AL, B.NV, CASP* * direct lifting of scalar FSQRT instruction * SETREG now elides setting of targeting zero registers * fixed test_gen.py to correctly regenerate arm64test.py * unroll vector MOV operations, USHL no longer uses intrinsic for scalars * updated existing tests in arm64test.py for latest lifting changes * fixed CASH* and CASB* incorrectly accessing temp register in comparison (resulting in comparing to NOP) * lifting all variants of TBL as intrinsic * fixes/improvements to test_gen.py * lifting all variants of TBX as intrinsic * added tests for CAS*, UMUL*, UADD*, FABD, FABS, FADDP, FMAX, FMAXNM, FMIN, FMINNM, FNEG, FNMUL, FCMEQ, FCMGE, FCMGT, FMLA, FMLS * added tests for all aliases of SBFM
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Partially fixes #5417
Also superficial re-ordering of the tests to make it easier to add new tests.
With both PRs targeting #5417 we get this LLIL:
and this HLIL:
Finally, it looks like there are some other unlifted SHL variants that would probably follow the same template. Let me know if you've come across any of these and I can take a look.