In addition to what Po-Hsu Lin and Daniel Axtens said, to have each of the kernels test-booted and checked for obvious regressions would be great and if in addition one of those can be tested for the reported problem being fixed, then this is sufficient. But mainly this procedure is to ensure we do not break things while trying to fix problems. And if there is some regression it takes time to isolate the change and revert it. Hence the urge to get reporter test as soon as possible.
In addition to what Po-Hsu Lin and Daniel Axtens said, to have each of the kernels test-booted and checked for obvious regressions would be great and if in addition one of those can be tested for the reported problem being fixed, then this is sufficient. But mainly this procedure is to ensure we do not break things while trying to fix problems. And if there is some regression it takes time to isolate the change and revert it. Hence the urge to get reporter test as soon as possible.