Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

adds semantics for the x86 SSE floating-point instructions #1466

Merged
merged 19 commits into from
Apr 15, 2022
Merged

adds semantics for the x86 SSE floating-point instructions #1466

merged 19 commits into from
Apr 15, 2022

Commits on Apr 13, 2022

  1. marks synthetic subroutines with proper attributes 

    All synthetic subroutines are marked with the synthetic attribute,
and special together with intrinsic functions are marked as
intrinsic.
    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    d922cc2 View commit details
    Browse the repository at this point in the history

  2. prevents C ABI pass from clobbering registers on intrinsic calls 

    a little bit hacky, but will make it better in the future
    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    820d22a View commit details
    Browse the repository at this point in the history

  3. do not correct the stack pointer on calls to intrinsic functions

    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    c84b5d3 View commit details
    Browse the repository at this point in the history

  4. adds the MXCSR register to the x86-64 environment

    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    c12dcc2 View commit details
    Browse the repository at this point in the history

  5. fixes a bug in the intrinisic primtive 

    the repetitive parameters were ignored
    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    07a0c25 View commit details
    Browse the repository at this point in the history

  6. fixes the IR contraction procedure again and enables unconditionally

    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    3813229 View commit details
    Browse the repository at this point in the history

  7. adds the semantics of the basic SSE floating-point operations

    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    2b19171 View commit details
    Browse the repository at this point in the history

  8. adds the implementation of some ieee754 intrinsics

    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    f5637d9 View commit details
    Browse the repository at this point in the history

  9. relaxes the contraction predicate

    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    b48f8e5 View commit details
    Browse the repository at this point in the history

  10. fixes the pcode floating-point lifter 

    we now are passing the tests!
    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    381afd6 View commit details
    Browse the repository at this point in the history

  11. makes the word-width primitive equal in both versions of Primus

    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    9d082bf View commit details
    Browse the repository at this point in the history

  12. adds the ieee754-is-nan primitive

    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    a3608e5 View commit details
    Browse the repository at this point in the history

  13. adds more SSE instruction, implements branchless comparison

    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    6f88230 View commit details
    Browse the repository at this point in the history

  14. allows the normalization procedure to fail 

    The Ghidra lifter generates code that is either invalid or our
interpretation of it is not valid, e.g., `int3` instructioin on
x86 is lifted as
```
CALL x;
RETURN y;
```

where according to the documentation both CALL and RETURN should be
treated initially as indirect jumps
    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    c3b0131 View commit details
    Browse the repository at this point in the history

  15. renames the system feature in Lisp context to the target-system 

    to prevent clashing with the Primus Lisp systems
    @ivg
    ivg committed Apr 13, 2022
    Configuration menu
    Copy the full SHA
    b78ae26 View commit details
    Browse the repository at this point in the history

Commits on Apr 14, 2022

  1. adds the Lisp.Semantics.context to control the Lisp definitions

    @ivg
    ivg committed Apr 14, 2022
    Configuration menu
    Copy the full SHA
    04f8df0 View commit details
    Browse the repository at this point in the history

  2. disables x86 FP intrinsic semantics when the legacy lifter is used 

    Otherwise, they both try to provide semantics that is conflicting.
    @ivg
    ivg committed Apr 14, 2022
    Configuration menu
    Copy the full SHA
    47cec64 View commit details
    Browse the repository at this point in the history

  3. adds the --x86-disable-floating-point-intrinsics as an escape hatch 

    This option makes it possible to still use the old BIL plugin feature
that translates unknown (to the BIL lifters) instructions into
intrinsic calls.
    @ivg
    ivg committed Apr 14, 2022
    Configuration menu
    Copy the full SHA
    dddcf04 View commit details
    Browse the repository at this point in the history

  4. updates the testsuite revision

    @ivg
    ivg committed Apr 14, 2022
    Configuration menu
    Copy the full SHA
    1b95ae4 View commit details
    Browse the repository at this point in the history