On 18/10/2021 12:01, Jan Beulich wrote:
> On 18.10.2021 12:08, Jane Malalane wrote:
>
>> /*
>> * Failure here with E2BIG indicates Xen is missing the bugfix to map
>> * resources larger than 32 frames.
>> */
>> if ( !res )
>> - return fail(" Fail: Map %d - %s\n", errno, strerror(errno));
>> + return fail(" Fail: Map grant table %d - %s\n", errno,
>> strerror(errno));
>>
>> + /* Put each gref at a unique offset in its frame. */
>> + for ( unsigned int i = 0; i < nr_frames; i++ )
>> + {
>> + unsigned int gref = i * (XC_PAGE_SIZE / sizeof(*gnttab)) + i;
>> +
>> + refs[i] = gref;
>> + domids[i] = domid;
>> +
>> + gnttab[gref].domid = 0;
>> + gnttab[gref].frame = gfn;
>> + gnttab[gref].flags = GTF_permit_access;
>> + }
> To make obvious that you're done with gnttab[], perhaps better unmap it
> here rather than at the bottom?
This is just test code. We could unmap it earlier, but that makes it
irritating if you ever need to insert printk()'s.
>> @@ -123,8 +162,25 @@ static void test_domain_configurations(void)
>>
>> printf(" Created d%u\n", domid);
>>
>> - test_gnttab(domid, t->create.max_grant_frames);
>> + rc = xc_domain_setmaxmem(xch, domid, -1);
> That's an unbelievably large upper bound that you set. Since you
> populate ...
>
>> + if ( rc )
>> + {
>> + fail(" Failed to set max memory for domain: %d - %s\n",
>> + errno, strerror(errno));
>> + goto test_done;
>> + }
>> +
>> + rc = xc_domain_populate_physmap_exact(xch, domid, ARRAY_SIZE(ram),
>> 0, 0, ram);
> ... only a single page, can't you get away with a much smaller value?
Yes, but again, this is test code.
Furthermore, there are other plans for further testing which would mean
1 wouldn't be appropriate here. All we want is "don't choke on limits
while we're performing testing".
~Andrew
