Messages in this thread | | | From | Linus Torvalds <> | Date | Sat, 16 Feb 2013 11:51:03 -0800 | Subject | Re: [-rc7 regression] Buggy commit: "mm: use aligned zone start for pfn_to_bitidx calculation" |
| |
On Sat, Feb 16, 2013 at 11:38 AM, Yinghai Lu <yinghai@kernel.org> wrote: > > but you forgot to update setup_usemap() for SPARSEMEM
Heh. I tried desperately to find a config to test my patch in, because I couldn't see how to even disable SPARSEMEM for my normal x86-64 build. But then I *only* tested it for that non-SPARSEMEM case, expecting that to be what showed any problems. The fact that I didn't bother testing my normal config is a bit ironic.
But I'd still like verification that it actually fixes Ingo's issue. It *looks* like it should, and it would explain the potential for memory corruption, but perhaps Ingo's odd lock-up is due to something really subtle and unrelated to the actual allocation size.
Linus
| |