Messages in this thread |  | | | Date | Thu, 8 Mar 2012 15:05:05 -0800 | | From | Andrew Morton <> | | Subject | Re: [PATCH 2/2 v2] core dump: add VM_NODUMP, MADV_NODUMP, MADV_CLEAR_NODUMP |
| |
On Thu, 8 Mar 2012 13:44:17 -0800 (PST) Roland McGrath <roland@hack.frob.com> wrote:
> I'm vaguely dubious about the desireability of using this new feature. > (I would hate to try to debug a qemu bug from a dump where half the > interesting state was intentionally hidden from me.)
yup. But the dumpability is fully controllable by userspace: one can easily envisage large vma's full of data which just isn't interesting for a dump. Plus if you're a developer who is looking at a dump and important data isn't there, you can often delete one line, rebuild and run the test again.
> But I don't see > any problem with the implementation. > > As to the interface, I think the bits are OK but it would be more > in keeping with the existing MADV_* names to call the new values > MADV_DONTDUMP and MADV_DODUMP.
Yup. I queued a patch to do that.
|  |