Re: [PATCH v1 4/5] util: add qemu_get_host_physmem utility function

[Alex Bennée]

Richard Henderson <richard.henderson@linaro.org> writes:

> On 7/17/20 7:24 AM, Christian Ehrhardt wrote:
>>     > +size_t qemu_get_host_physmem(void)
>>     > +{
>>     > +#ifdef _SC_PHYS_PAGES
>>     > +    long pages = sysconf(_SC_PHYS_PAGES);
>>     > +    if (pages > 0) {
>>     > +        return pages * qemu_real_host_page_size;
>> 
>>     The Linux man page warns that this product may overflow so maybe you 
>> could
>>     return pages here.
>> 
>> 
>> The caller might be even less aware of that than this function - so maybe
>> better handle it here.
>> How about handling overflows and cutting it to MiB before returning?
>
> Indeed, the caller may be less aware, so we should handle it here.  But I 
> don't
> think truncating to MiB helps at all, because again, the caller has to handle
> overflow.
>
> Better, I think, to saturate the result to ~(size_t)0 and leave it at
> that.

So I went for:

  size_t qemu_get_host_physmem(void)
  {
  #ifdef _SC_PHYS_PAGES
      long pages = sysconf(_SC_PHYS_PAGES);
      if (pages > 0) {
          if (pages > SIZE_MAX / qemu_real_host_page_size) {
              return SIZE_MAX;
          } else {
              return pages * qemu_real_host_page_size;
          }
      }
  #endif
      return 0;
  }

apparently the first case of saturating integer arithmetic outside of
the instruction emulation in QEMU :-/

-- 
Alex Bennée