AArch64: Fix cfinv disassembly issues
authorTamar Christina <tamar.christina@arm.com>
Mon, 27 Jan 2020 10:40:02 +0000 (10:40 +0000)
committerTamar Christina <tamar.christina@arm.com>
Mon, 27 Jan 2020 10:55:41 +0000 (10:55 +0000)
commit7568c93bf95a518797dfb2987b04911164c14a36
tree10b4933b73b7a607b53f2f76f5fb298f67a8a018
parent168f8c6ba008bed9fcc2fe7098416f608e4419af
AArch64: Fix cfinv disassembly issues

This fixes the preferred disassembly for cfinv.  The Armv8.4-a instruction
overlaps with the possible encoding space for msr.  This because msr allows you
to use unallocated encoding space using the general sA_B_cC_cD_E form.

However when an encoding does become allocated then we need to ensure that it's
used as the preferred disassembly.  The problem with cfinv is that its mask has
all bits sets because it has no arguments.

This causes issues for the Alias resolver in gas as it uses the mask to build
alias graph.  In this case it can't do it since it thinks almost everything
would alias with cfinv.  So instead we can only fix this by moving cfinv before
msr.

gas/ChangeLog:

PR 25403
* testsuite/gas/aarch64/armv8_4-a.d: Add cfinv.
* testsuite/gas/aarch64/armv8_4-a.s: Likewise.

opcodes/ChangeLog:

PR 25403
* aarch64-tbl.h (struct aarch64_opcode): Re-order cfinv.
* aarch64-asm-2.c: Regenerate
* aarch64-dis-2.c: Likewise.
* aarch64-opc-2.c: Likewise.
gas/ChangeLog
gas/testsuite/gas/aarch64/armv8_4-a.d
gas/testsuite/gas/aarch64/armv8_4-a.s
opcodes/ChangeLog
opcodes/aarch64-asm-2.c
opcodes/aarch64-dis-2.c
opcodes/aarch64-opc-2.c
opcodes/aarch64-tbl.h
This page took 0.024303 seconds and 4 git commands to generate.