Re: malloc/free is signal-safe?

[Dimitrios Apostolou <jimis at gmx dot net>]

On Tue, 4 Jun 2013, Siddhesh Poyarekar wrote:
> On 4 June 2013 14:35, Dimitrios Apostolou <jimis@gmx.net> wrote:
> > according to the libc manual, malloc is signal-safe in glibc. However I got
>
> If it really says that then it is a bug in the manual.  malloc is not
> async-signal safe  Please quote the section of the manual that
> mentions it and file a bug report.
>
> > a deadlock when I sigTERM'd a process during malloc(), and it tried to log
> > with vasprintf() (which itself calls malloc).
> >
> > Am I missing something regarding libc's malloc() reentrancy? Here is the
> > backtrace I got:
>
> Maybe you confused it with thread-safe?  malloc is thread-safe, but
> not async-signal-safe.


It seems you are right, I can't find the excerpt in the current libc
manual! It seems google had directed me to some very old manual:


Quoting 
http://www.cs.utah.edu/dept/old/texinfo/glibc-manual-0.02/library_21.html

 On the GNU system, malloc and free are safe to use in signal handlers 
because it blocks signals. As a result, the library functions that 
allocate space for a result are also safe in signal handlers. The obstack 
allocation functions are safe as long as you don't use the same obstack 
both inside and outside of a signal handler.


Thanks, I'll remove all logging from the signal handler.
Dimitris