Re: [PATCH 1/4] coredump: add an interface to control the core dump routine

[Pavel Machek <pavel@ucw.cz>]

Hi!

> I have discussed with my colleagues why you say "ugly" against my
> procfs interface, then I noticed I may have misunderstood what you said.
> Is the reason for saying "ugly" two interfaces, i.e. preexisting ulimit
> (get/setrlimit) and my proc entry, exist to control core file size? 

Yes.

> > Plus, what you are doing can be done in userspace using google
> > coredumper.
> 
> I think that the needs differ between userland core dumper user and
> in-kernel core dumper user.  Pros and cons also differ.
> 
> Some of people (such as system admins, distro vendors, etc) need
> highly reliable core dumper because they don't want to experience
> same failures again and they don't hope that another failure is
> caused by core dumping.  Userland core dumper is useful because
> it is relatively easy to be customized, but its reliability highly
> depends on the application programs.

Fix userland core dumper to be reliable, then.

> If the stack for signal handlers is not set up carefully, if the
> data used by userland core dumper has been destroyed, if
> coredump_omit_anon_shared flag has been overwritten by bad data,
> or if the address of functions have been destroyed, the userland
> core dumper may fail to dump.  So in-kernel solutoin is required
> by enterprise users.

It should be possible to dump from separate process.

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html