Re: Realloc is blocking execution
måndag den 19 oktober 2009 klockan 11:59 skrev Goswin von Brederlow detta:
> Mike Hommey <mh@glandium.org> writes:
>
> > On Fri, Oct 16, 2009 at 12:02:26PM +0200, Bernhard R. Link wrote:
> >> * Mats Erik Andersson <mats.andersson@gisladisker.se> [091016 11:55]:
> >> > 4. The main process WM receives SIGHUP, and enters a signal handler.
> >> > The signal handler uses two calls: free_menuitems(), get_menuitems().
> >>
> >> If those calls call malloc or free or anything else this might be the
> >> problem. Memory allocation functions are not reentrant and due to
> >> threading support are easily blocking when used this way.
> >
> > Sadly, the manual page for these functions under linux doesn't seem to
> > say anything about that, contrary to other unices man pages. Maybe this
> > should be filed as a wishlist bug.
> >
> > Mike
> >
> >
> > --
> > To UNSUBSCRIBE, email to debian-mentors-REQUEST@lists.debian.org
> > with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
>
> man 7 signal
>
> Async-signal-safe functions
> A signal handling routine established by sigaction(2) or
> signal(2) must be very careful, since processing elsewhere may
> be interrupted at some arbitrary point in the execution of the
> program. POSIX has the concept of "safe function". If a
> signal interrupts the execution of an unsafe function, and
> handler calls an unsafe function, then the behavior of the
> program is undefined.
>
> POSIX.1-2004 (also known as POSIX.1-2001 Technical Corrigendum
> 2) requires an implementation to guarantee that the following
> functions can be safely called inside a signal handler:
>
> <long list follow>
>
> The list certainly does not contain malloc() or realloc(). So what you
> see is that the undefined behaviour changed from luckily working to
> blocking. The code does somthing wrong and now it bites you in the ass. :)
>
> I doubt there is anything to fix this other than fixing your code to
> get out of the signal handler as fast as possible and do the real work
> outside.
>
> MfG
> Goswin
>
>
> --
> To UNSUBSCRIBE, email to debian-mentors-REQUEST@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
>
Yes, I have already suggested this to the upstream author.
Thanks, for the pointer.
I just tried to compile and run the same source tar-ball
an a new system with Debian testing. I tried to keep the
number of packages down when doing a dist-upgrade from
Lenny, but the outcome was the sysrq-RIESUB both times!
And I did not even invoke any signal, so possibly the
installed system was incomplete to begin with.
Reply to: