On 05/22/2018 09:31 AM, Peter Maydell wrote:
> Do we also want f16 inputs to mean uint32_t arguments?
Yes. We don't want the compiler assuming zeros (for non-x86 abis) when we
aren't providing them.
r~
On 22 May 2018 at 17:27, Richard Henderson wrote:
> On 05/22/2018 09:19 AM, Peter Maydell wrote:
>> But what we actually have is
>> target/arm/helper.c:float16 HELPER(recpe_f16)(float16 input, void *fpstp)
>> target/arm/helper.c:float16 HELPER(vfp_sltoh)(uint32_t x, uint32_t
>> shift, void *fpst)
On 05/22/2018 09:19 AM, Peter Maydell wrote:
> But what we actually have is
> target/arm/helper.c:float16 HELPER(recpe_f16)(float16 input, void *fpstp)
> target/arm/helper.c:float16 HELPER(vfp_sltoh)(uint32_t x, uint32_t
> shift, void *fpst)
>
> (and others that you can find by grepping for "float
On 22 May 2018 at 17:11, Richard Henderson wrote:
> On 05/22/2018 07:56 AM, Peter Maydell wrote:
> The *helpers* are supposed to be returning uint32_t. That should be enforced
> by the prototype generated from "f16" being aliased to "i32".
>
> The softfloat routines are supposed to be returning f
On 05/22/2018 07:56 AM, Peter Maydell wrote:
>>> However, for TCG helper functions, although we allow the helper
>>> macro to say that the return type is 'f16', in helper-head.h we
>>> seem to just alias f16 as i32, and in dh_sizemask() we only record
>>> "is this 64 bits" and "is this signed". Am
On 22 May 2018 at 15:55, Richard Henderson wrote:
> On 05/22/2018 07:50 AM, Peter Maydell wrote:
>> Hi; I was reading about the x86-64 calling convention and I found
>> out that apparently for returns of smaller-than-register types the
>> high parts may contain garbage:
>> https://stackoverflow.co
On 05/22/2018 07:50 AM, Peter Maydell wrote:
> Hi; I was reading about the x86-64 calling convention and I found
> out that apparently for returns of smaller-than-register types the
> high parts may contain garbage:
> https://stackoverflow.com/questions/36706721/is-a-sign-or-zero-extension-required
Hi; I was reading about the x86-64 calling convention and I found
out that apparently for returns of smaller-than-register types the
high parts may contain garbage:
https://stackoverflow.com/questions/36706721/is-a-sign-or-zero-extension-required-when-adding-a-32bit-offset-to-a-pointer-for/36760539
