> On 12 Jun 2026, at 15:42, Jakub Jelinek <[email protected]> wrote:
> 
> On Fri, Jun 12, 2026 at 03:37:46PM +0100, Iain Sandoe wrote:
>> Builds completed on aarch64 linux and (with devt patches) darwin,
>> thoughts?  Maybe LLONG_MAX is sufficiently portable?
>> or  OK for trunk?
> 
> This looks wrong.  __LONG_LONG_MAX__ is a GCC (and maybe Clang)
> implementation detail, so using it in sources which can be compiled by
> arbitrary other compilers (at least in theory) looks wrong.
> I think it should just use LLONG_MAX instead of LONG_LONG_MAX.

Yeah, I was half-way to doing that anyway - I think that this is more in line
with what the standards docs say than using HOST_WIDE_INT_MAX,

As attached, retested on aarch64 darwin and linux, OK for trunk?
thanks
Iain

Attachment: 0001-aarch64-Use-LLONG_MAX-instead-of-LONG_LONG_MAX.patch
Description: Binary data

> 
>> @@ -3794,7 +3794,7 @@ aarch64_expand_prefetch_range_builtin (tree exp, int 
>> fcode)
>>   non-negative integer.  However it is meaningless for
>>   values less than 2^15 or greater than 2^29.  */
>> uint64_t reuse_distance = require_const_argument (exp, argno++, 0,
>> -  LONG_LONG_MAX);
>> +  __LONG_LONG_MAX__);
>> 
>> if (seen_error ())
>>  return;
>> -- 
>> 2.50.1 (Apple Git-155)
> 
> Jakub
> 

Reply via email to