gtk-gnutella-devel Digest, Vol 29, Issue 3
|
View:
New views
1 Messages
—
Rating Filter:
Alert me
|
 |
gtk-gnutella-devel Digest, Vol 29, Issue 3
by gtk-gnutella-devel-request
::
Rate this Message:
Send gtk-gnutella-devel mailing list submissions to
gtk-gnutella-devel@... To subscribe or unsubscribe via the World Wide Web, visit https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-devel or, via email, send a message with subject or body 'help' to gtk-gnutella-devel-request@... You can reach the person managing the list at gtk-gnutella-devel-owner@... When replying, please edit your Subject line so it is more specific than "Re: Contents of gtk-gnutella-devel digest..." Today's Topics: 1. Re: getpeername() failed warning: getpeername not being called. (Matthew Lye) 2. Re: getpeername() failed warning: getpeername not being called. (Christian Biere) 3. Re: getpeername() failed warning: getpeername not being called. (Christian Biere) 4. Re: getpeername() failed warning: getpeername not being called. (Matthew Lye) 5. Crash at startup (Hauke Hachmann) 6. Re: Crash at startup (Hauke Hachmann) 7. Re: Crash at startup (Christian Biere) 8. Re: Crash at startup (Christian Biere) 9. Re: Crash at startup (Hauke Hachmann) ---------------------------------------------------------------------- Message: 1 Date: Wed, 13 May 2009 12:14:15 -0400 From: Matthew Lye <mlye@...> Subject: Re: [gtk-gnutella-devel] getpeername() failed warning: getpeername not being called. To: gtk-gnutella-devel@... Message-ID: <47FE50C0-C091-447D-A567-2FDC10A07F9C@...> Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes On 13-May-09, at 1:46 AM, Christian Biere wrote: > Do you get this warning for every incoming connection or just for > some? Just for some, albeit fairly frequently. (That said, it's happening with every call to getpeername$UNIX2003.) I'd have treated it as more of a problem if it was flooding the console. I shouldn't be getting any IPv6 from anywhere except my local wireless nodes talking to each other. I'll see if I can get return values from accept(), or probably accept $UNIX2003()*, later today; also see if I can catch the warning occurring with a manageable number of incoming connections to scan through, and/or figure out how to get the socket (identical to fd?) number to show up alongside the incoming IP number. *be prepared for dumps of assembly language interspersed with sarcastic comments**. Mystery computer science theater UNIX2003. **No, not really. ------------------------------ Message: 2 Date: Wed, 13 May 2009 19:29:44 +0200 From: Christian Biere <christianbiere@...> Subject: Re: [gtk-gnutella-devel] getpeername() failed warning: getpeername not being called. To: gtk-gnutella-devel@... Message-ID: <20090513172944.GA25033@...> Content-Type: text/plain; charset=utf-8 Matthew Lye wrote: > On 13-May-09, at 1:46 AM, Christian Biere wrote: > > Do you get this warning for every incoming connection or just for > > some? > Just for some, albeit fairly frequently. (That said, it's happening > with every call to getpeername$UNIX2003.) I'd have treated it as more > of a problem if it was flooding the console. What's the frequency? About once per second, minute, hour? > I shouldn't be getting any IPv6 from anywhere except my local wireless > nodes talking to each other. If you have IPv6 connectivity, the following should display a valid IPv6 address: echo print local_ip6|gtk-gnutella --shell You can use netstat, fstat and maybe lsof to see all sockets, their addresses and the latter two also show which file descriptor they are corresponding to. That should also tell you whether there are any IPv6 connections. > I'll see if I can get return values from accept(), or probably accept > $UNIX2003()*, later today; also see if I can catch the warning > occurring with a manageable number of incoming connections to scan > through, and/or figure out how to get the socket (identical to fd?) > number to show up alongside the incoming IP number. I don't know what's the frequency of incoming connections for you. If it's not multiple per second, it shouldn't be too difficult to see which connections trigger this issue and the timestamps shown by these tools should help, too. One possibility for this issue could be an invalid definition of the socklen_t type. I found at least one post in which someone wrote that his GCC used long instead of int or vice-versa causing an ABI mismatch. This may go unnoticed if the upper bits are conveniently within the acceptable boundaries for valid parameters. As the variables are mostly auto-variables on the stack which is typically dirty, this isn't completely unlikely. Maybe it only fails when the respective bits are all zero by incident causing these function to fail respectively not fill-in the source address. So what you could try is this: In src/core/sockets.c, replace all occurences of socklen_t with SOCKLEN_T, then add one of the following lines at the head of the file, after #include "common.h", for example: #define SOCKLEN_T short #define SOCKLEN_T int #define SOCKLEN_T long Try compiling with each of these three and see whether it helps. You'll have to ignore the compiler warnings, of course. -- Christian ------------------------------ Message: 3 Date: Wed, 13 May 2009 21:15:43 +0200 From: Christian Biere <christianbiere@...> Subject: Re: [gtk-gnutella-devel] getpeername() failed warning: getpeername not being called. To: gtk-gnutella-devel@... Message-ID: <20090513191543.GC25033@...> Content-Type: text/plain; charset=utf-8 Matthew Lye wrote: > On 13-May-09, at 1:46 AM, Christian Biere wrote: > > Do you get this warning for every incoming connection or just for > > some? > Just for some, albeit fairly frequently. (That said, it's happening > with every call to getpeername$UNIX2003.) You could call getpeername() unconditonally after each successful accept() to ensure it's not just a problem with getpeername(). However, getpeername() wouldn't be called if accept() had provided an address as it should. -- Christian ------------------------------ Message: 4 Date: Thu, 14 May 2009 13:12:46 -0400 From: Matthew Lye <mlye@...> Subject: Re: [gtk-gnutella-devel] getpeername() failed warning: getpeername not being called. To: gtk-gnutella-devel@... Message-ID: <7F6255B6-0E1C-45D9-845C-4BC6F1FFA7B5@...> Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes On 13-May-09, at 3:15 PM, Christian Biere wrote: > Matthew Lye wrote: >> On 13-May-09, at 1:46 AM, Christian Biere wrote: >>> Do you get this warning for every incoming connection or just for >>> some? > >> Just for some, albeit fairly frequently. (That said, it's happening >> with every call to getpeername$UNIX2003.) > > You could call getpeername() unconditonally after each successful > accept() to ensure it's not just a problem with getpeername(). > However, > getpeername() wouldn't be called if accept() had provided an address > as it should. Under these circumstances, getpeername() does set a plausible address. It's setting 0.0.0.0 and failing only when called after accept(). Looking at the warnings in the code, there should be a different warning if getpeername() ever succeeds when accept() fails; this does not occur. I'll check the network stuff next session. ------------------------------ Message: 5 Date: Fri, 29 May 2009 01:47:34 +0200 From: Hauke Hachmann <haxe@...> Subject: [gtk-gnutella-devel] Crash at startup To: gtk-gnutella-devel@... Message-ID: <200905290147.34313.haxe@...> Content-Type: Text/Plain; charset="us-ascii" Hi list, I have a problem with latest svn revision 16799: I cannot even start the application. Directly after invokation I get this error: ** ERROR **: pmap already contains the new region aborting... CRASH (pid=14792) by SIGABRT Aborted Any ideas? Hauke ------------------------------ Message: 6 Date: Fri, 29 May 2009 15:53:34 +0200 From: Hauke Hachmann <haxe@...> Subject: Re: [gtk-gnutella-devel] Crash at startup To: Matthew Lye <mlye@...>, gtk-gnutella-devel@... Message-ID: <200905291553.34538.haxe@...> Content-Type: Text/Plain; charset="iso-8859-1" On Friday 29 May 2009, Matthew Lye wrote: > > I have a problem with latest svn revision 16799:... [...] > > That error message is from "lib/vmm.c", which was undergoing some > work as recently as revision 16799. > Just roll back to 16798 for the day if you need your client right > now. That did not work. The error must have been introduced earlier. I then took the time and bisected the error back to r16786. Everything before that works fine, everything starting from r16786 is broken. Going with r16785 for the moment. Bye, Hauke ------------------------------ Message: 7 Date: Fri, 29 May 2009 18:09:51 +0200 From: Christian Biere <christianbiere@...> Subject: Re: [gtk-gnutella-devel] Crash at startup To: gtk-gnutella-devel@... Message-ID: <20090529160951.GA21925@...> Content-Type: text/plain; charset=utf-8 Hauke Hachmann wrote: > On Friday 29 May 2009, Matthew Lye wrote: > > > I have a problem with latest svn revision 16799:... > [...] > > > > That error message is from "lib/vmm.c", which was undergoing some > > work as recently as revision 16799. > > Just roll back to 16798 for the day if you need your client right > > now. > > That did not work. The error must have been introduced earlier. I then > took the time and bisected the error back to r16786. Everything before > that works fine, everything starting from r16786 is broken. > > Going with r16785 for the moment. I can confirm this but I'm only seeing crashes when vmm_mmap() is used. This happens either by uploads or when loading the FAQ under Help->FAQ in the GUI. The crash doesn't occur immediately but a bit afterwards because there seems to be something screwed up with respect to tracking foreign regions. If it happens on *inux more frequently, it might be because /proc/self/maps is used then to find foreign regions. -- Christian ------------------------------ Message: 8 Date: Fri, 29 May 2009 19:37:05 +0200 From: Christian Biere <christianbiere@...> Subject: Re: [gtk-gnutella-devel] Crash at startup To: gtk-gnutella-devel@... Message-ID: <20090529173705.GA19545@...> Content-Type: text/plain; charset=utf-8 Hauke Hachmann wrote: > On Friday 29 May 2009, Matthew Lye wrote: > > > I have a problem with latest svn revision 16799:... > [...] > > > > That error message is from "lib/vmm.c", which was undergoing some > > work as recently as revision 16799. > > Just roll back to 16798 for the day if you need your client right > > now. > > That did not work. The error must have been introduced earlier. I then > took the time and bisected the error back to r16786. Everything before > that works fine, everything starting from r16786 is broken. This is probably fixed in SVN since r16807 now. -- Christian ------------------------------ Message: 9 Date: Fri, 29 May 2009 19:52:05 +0200 From: Hauke Hachmann <haxe@...> Subject: Re: [gtk-gnutella-devel] Crash at startup To: gtk-gnutella-devel@... Message-ID: <200905291952.05124.haxe@...> Content-Type: Text/Plain; charset="iso-8859-1" On Friday 29 May 2009, Christian Biere wrote: > This is probably fixed in SVN since r16807 now. Nope, the newest revision (16813) still crashes on startup. But with a different error message this time: FATAL: Assertion failure in vmm.c:2311: "amount != 0" h ------------------------------ ------------------------------------------------------------------------------ Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT is a gathering of tech-side developers & brand creativity professionals. Meet the minds behind Google Creative Lab, Visual Complexity, Processing, & iPhoneDevCamp as they present alongside digital heavyweights like Barbarian Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com ------------------------------ _______________________________________________ gtk-gnutella-devel mailing list gtk-gnutella-devel@... https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-devel End of gtk-gnutella-devel Digest, Vol 29, Issue 3 ************************************************* |
| Free embeddable forum powered by Nabble | Forum Help |
