]> xenbits.xensource.com Git - qemu-xen.git/commit
target/i386: fix operand size for VCOMI/VUCOMI instructions
authorPaolo Bonzini <pbonzini@redhat.com>
Tue, 9 May 2023 14:17:15 +0000 (16:17 +0200)
committerMichael Tokarev <mjt@tls.msk.ru>
Thu, 18 May 2023 17:46:47 +0000 (20:46 +0300)
commitdb8051ad59f416a91ad4121d3d11f52e2879c429
treebea9e74c8b66516a57ea2ddc85a7ef0cdcb61de1
parent1e029102e6b6b176d692944d9df94c6b46185912
target/i386: fix operand size for VCOMI/VUCOMI instructions

Compared to other SSE instructions, VUCOMISx and VCOMISx are different:
the single and double precision versions are distinguished through a
prefix, however they use no-prefix and 0x66 for SS and SD respectively.
Scalar values usually are associated with 0xF2 and 0xF3.

Because of these, they incorrectly perform a 128-bit memory load instead
of a 32- or 64-bit load.  Fix this by writing a custom decoding function.

I tested that the reproducer is fixed and the test-avx output does not
change.

Reported-by: Gabriele Svelto <gsvelto@mozilla.com>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1637
Fixes: f8d19eec0d53 ("target/i386: reimplement 0x0f 0x28-0x2f, add AVX", 2022-10-18)
Cc: qemu-stable@nongnu.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
(cherry picked from commit 2b55e479e6fcbb466585fd25077a50c32e10dc3a)
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>
target/i386/tcg/decode-new.c.inc