Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Wed, 23 Jun 2021 14:32:24 -0400
From: Rich Felker <dalias@...c.org>
To: Srinivasan J <srinidpdk@...il.com>
Cc: musl@...ts.openwall.com
Subject: Re: Dumping process heap from the core file

On Wed, Jun 23, 2021 at 09:21:49AM +0530, Srinivasan J wrote:
> Hi Rich,
>             Thank you for replying. Unfortunately, I need to debug a
> core which used musl-libc version v1.1.24 to get details about size of
> allocations in the heap. I see that mal.heap is no longer valid since
> 2015 [commit e3bc22f1eff87b8f029a6ab31f1a269d69e4b053].  So as you
> said the only way to walk the heap would be using a patched libc.so?
> Please do share any pointers.

Ah, in that case you could just start walking the headers
backward/forward from any known starting chunk.

Rich



> Reference:
> https://patchwork.criu.org/patch/10472/
> 
> Regards,
> Srini
> 
> On Sat, Jun 5, 2021 at 9:20 PM Rich Felker <dalias@...c.org> wrote:
> >
> > On Thu, Jun 03, 2021 at 07:39:54AM +0530, Srinivasan J wrote:
> > > Hi,
> > >     Is there an easy way to dump the heap from the process core which
> > > is using musl-libc? The process is using musl-libc version v1.1.24. I
> > > see that there is a _dump_heap() function, but that's not up to date.
> > > I am thinking about using _dump_heap() as reference. But any other
> > > suggestions are welcome.
> >
> > The mallocng heap structures are actually more walkable. There is
> > intentionally no public interface for doing this, but I have some
> > debug code left over from development that you could link into
> > static-linked programs with a particular known musl/mallocng version
> > or patch into libc.so for debugging purposes. I'll dig it up and post
> > it. Ping me if I forget about it.
> >
> > Rich

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.