[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Libunwind-devel] Commit 32202619c42f58a9dac6eee3b9a4c231f1ce394a is
From: |
Arun Sharma |
Subject: |
Re: [Libunwind-devel] Commit 32202619c42f58a9dac6eee3b9a4c231f1ce394a is misleading |
Date: |
Sat, 27 Oct 2012 20:01:57 -0700 |
On Fri, Oct 26, 2012 at 10:35 AM, Paul Pluzhnikov
<address@hidden> wrote:
> The hint to call unw_backtrace early in thread creation is not very useful
> to tcmalloc users, as they have little control over exactly when
> calloc is called during thread creation, or when tcmalloc decides to
> record a stack trace.
>
I went back and read the earlier messages on the topic:
http://lists.gnu.org/archive/html/libunwind-devel/2011-04/msg00030.html
Back then, you thought something like unwind_per_thread_once() could
work. What has changed since then?
IIRC - the calloc originates from within pthread_setspecific() and all
we need to do is force a call to setspecific() from a safe (guaranteed
outside of malloc) context. Are there other calls to calloc you've
discovered since then?
-Arun