Hi Günther and Gerda, it's my impression too that something is wrong with debian tested/unstable, also due to other recent problems. Anyway, yesterday's late-night bugfix indeed eliminated a double-free problem (today verified with valgrind), so Gerda, in case you are checking out from SF anonymously (that is, not as a developer), i would recommend to try again, since changes need some time to propagate inside the SF cvs.
all the best, Thomas
geiger wrote:
Hi Gerda and Thomas,
Depending on what libraries you are using, I have heard rumours about a change of the memory allocator in debians c++ lib. Nothing concrete, but, the problem might be something that is not related to flext or pool at all, but only a bug/incoherent memory allocation thing in the standard c++ libs.
It might be worth to try it on Debian stable or testing or on ubuntu (or any other distro) and see if the problems persist.
Guenter
On Wed, 14 Dec 2005, Gerda Strobl wrote:
Hi Thomas,
I am using the newest cvs-sourceforge pool version. I reinstalled it again-but this doesn't change-(and I do not have any old version stored)-now I just receive either double free or corruption (fasttop)(like yesterday) or glibc detected *** corrupted double-linked list (look attached new debug report).
I am sorry but I am not able to reproduce properly, how to make that error.
(gdb) run Starting program: /usr/local/bin/pd [Thread debugging using libthread_db enabled] [New Thread -1210583360 (LWP 5276)] *** glibc detected *** corrupted double-linked list: 0xb7eb5218
Program received signal SIGABRT, Aborted. [Switching to Thread -1210583360 (LWP 5276)] 0xb7da99e7 in raise () from /lib/tls/libc.so.6 (gdb) where #0 0xb7da99e7 in raise () from /lib/tls/libc.so.6 #1 0xb7dab31b in abort () from /lib/tls/libc.so.6 #2 0xb7de0365 in __fsetlocking () from /lib/tls/libc.so.6 #3 0xb7de650c in malloc_usable_size () from /lib/tls/libc.so.6 #4 0xb7de75ab in free () from /lib/tls/libc.so.6 #5 0xb7de8901 in calloc () from /lib/tls/libc.so.6 #6 0x080a2f55 in getbytes (nbytes=0) at m_memory.c:24 #7 0xb7d6c665 in flext_root_single::operator new () from /usr/local/lib/pd/extra/pool.pd_linux #8 0xb7d65246 in pooldir::Reset () from /usr/local/lib/pd/extra/pool.pd_linux #9 0xb7d654f0 in pooldir::pooldir () from /usr/local/lib/pd/extra/pool.pd_linux #10 0xb7d63093 in pooldata::pooldata () from /usr/local/lib/pd/extra/pool.pd_linux #11 0xb7d5cc18 in pool::SetPool () from /usr/local/lib/pd/extra/pool.pd_linux #12 0xb7d5cd12 in pool::Init () from /usr/local/lib/pd/extra/pool.pd_linux ---Type <return> to continue, or q <return> to quit---quit
lg Gerda
Thomas Grill wrote:
Hi Gerda, never mind, could you try out the pool cvs version? I think i found it.
thanks, Thomas
i was testing rrad.eros~.pd from Frank (rradical/effects)- clicking several times on the type-toggle-and changing several times the values from the number boxes- then I want to close the patch-I get the message if I want to close the patch (y/n)-I say yes-the error message appears on the shell and the patch is not closing. I also cant open a new patch (ctrl+N).
sorry, i don't have rradical here
Unfortunately I just found out that this doesn't depend on a specific patch-just opening pool-help.pd -typing on some of the connected message produces same error.
Could you give me some exact guidelines? You know, the problem is that here it doesn't crash - so trying around doesn't make much sense for me - it's just a waste of time.
thanks, Thomas
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list