[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [RFC PATCH v3 26/34] Hexagon (target/hexagon) macros referenced in i
From: |
Richard Henderson |
Subject: |
Re: [RFC PATCH v3 26/34] Hexagon (target/hexagon) macros referenced in instruction semantics |
Date: |
Sun, 30 Aug 2020 14:06:18 -0700 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.10.0 |
On 8/30/20 1:23 PM, Taylor Simpson wrote:
>> I think the merge stuff is a mistake. I think you can get the semantics that
>> you want with
>>
>> probe_read(ld_addr, ld_len)
>> qemu_st(st_value, st_addr)
>> qemu_ld(ld_value, ld_addr)
>>
>> In this way, all exceptions are recognized before the store is complete, the
>> normal memory operations handle any possible overlap.
>
> So, do this inside the helper? Or is there a way to generate TCG code?
I was thinking TCG code, where you can look at the packet before any code gen,
and decide whether or not this situation actually applies.
The probe is a simple helper call, for which the generic machinery exists
(probe_access, probe_write, probe_read). All you have to do is write the
wrapper.
The loads and stores are, well, normal loads and stores.
r~
[RFC PATCH v3 18/34] Hexagon (target/hexagon/imported) arch import - instruction semantics, Taylor Simpson, 2020/08/18
[RFC PATCH v3 19/34] Hexagon (target/hexagon/imported) arch import - instruction encoding, Taylor Simpson, 2020/08/18
[RFC PATCH v3 25/34] Hexagon (target/hexagon) macros to interface with the generator, Taylor Simpson, 2020/08/18