[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH v2 10/11] target/arm: Decode aa32 armv8.3 3-same
From: |
Richard Henderson |
Subject: |
Re: [Qemu-devel] [PATCH v2 10/11] target/arm: Decode aa32 armv8.3 3-same |
Date: |
Mon, 15 Jan 2018 11:10:04 -0800 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.5.2 |
On 01/15/2018 10:46 AM, Peter Maydell wrote:
> This doesn't seem to line up with the Arm ARM decode. Your
> pattern and mask give
> op0 = 0x, op1 = 100, op2 = 0 and also bit 8 = 0.
> The ARM has 3reg-same decoded with
> op0 = 0x, op1 = 1x0, op2 = x
>
> (and some insns in the 3reg-same group have bit 8 == 1, like
> VSDOT and VUDOT.)
Ah, more v8.2 instructions that I wasn't even looking at...
> How are you proposing to do the Thumb decoding? Try to share
> some of the 3same-vs-2reg+scalar decode part, or just have
> them both do a similar kind of decode and call the
> disas_neon_insn_cp8_3same/cp8_index functions?
Hmm. I thought this was working via the "translate into the equivalent ARM
encoding" path. But it couldn't possibly be doing so, since
disas_neon_insn_cp8_3same is not a subroutine of disas_neon_data_insn.
I guess I'll have to verify that RISU is testing what I thought it was...
r~