Re: cygwin bughunt (Jip-hee!)
On Fri, Feb 18, 2005 at 11:29:38AM +0100, David Dindorp wrote: Christopher Faylor wrote: The test were performed with 1.5.10-3, as newer versions call upon me all sorts of other problems and thus can't be pushed to the failing box right now. Btw, I urge everyone to try the latest cygwin snapshot! This quote comes from a thread that suggested cygwin had mind control powers. If that was really true, you'd be using a snapshot by now. Will do. Has the problem been found that results in this error?: *** MapViewOfFileEx(0x188, in_h 0x188) failed, Win32 error 6 I don't know. You didn't produce a test case and I could never reproduce the problem. I did take a stab at fixing it and made some changes in that code so the error message will be different now, at least. 2) You can *try* a snapshot to see if it fixes your problem. Will do sort of implies that that was exactly the course of action I was contemplating :-). And asterisks around *try* implies that you should stop talking about it and start *doing* it. It is possible that this problem has been fixed by my recent attempts to correct the dreaded hyperthreading problem. However, if it hasn't been fixed, I'm not willing to debug problems in 1.5.10. It would be a fruitless endeavor. The code has changed in the last six months and, even if the problem was tracked down, we're not going to be releasing a new version of 1.5.10. If you want help, either use a snapshot or CVS. cgf -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: cygwin bughunt (Jip-hee!)
Christopher Faylor wrote: Ah, yes! You're the you don't want people to debug cygwin because you aren't spoon feeding me debugging information guy! That is nowhere near what was said. I said you should provide debugging versions of Cygwin, since large software packages are hell to build. I was wrong in the case of Cygwin and I admitted my mistake. I don't know what your problem is - but I'll bet it's hard to pronounce. You're the guy who insisted on dropping back to 1.5.10! I remember now! I'm not dropping back, it's a customer system, and it's been running 1.5.10 for some time now. In the future, please provide more context if you really want to be helped, especially after almost a month of silence. Apologies. A month was the time it took to get the debug version through testing and to the affected system. I thought the version number and a specific location in the source code (as provided) would be enough information to at least get a pointer from you on whether it's a problem you've seen before. What else do you need? The test were performed with 1.5.10-3, as newer versions call upon me all sorts of other problems and thus can't be pushed to the failing box right now. Btw, I urge everyone to try the latest cygwin snapshot! Will do. Has the problem been found that results in this error?: *** MapViewOfFileEx(0x188, in_h 0x188) failed, Win32 error 6 1) Regardless of whether the new versions call upon (you) all sorts of other problems, I doubt that anyone is going to spend any time tracking down problems in older versions. I'm not asking you to spend a lot of time on it. I just hope you could shed some light on the issue. Eg. have you seen hung Cygwin processes having to do with closing file descriptor 5 before? Any information on why it's closing file descriptor 5 (it's not used by the scripts) would be helpful too. 2) You can *try* a snapshot to see if it fixes your problem. Will do sort of implies that that was exactly the course of action I was contemplating :-). -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: cygwin bughunt (Jip-hee!)
Christopher Faylor wrote: Actually, we do. We provide the source code. It's easy to build. You are right; I was wrong. Building Cygwin is easy. (At least when it comes to newer versions :-p.) It even compiles under itself. *impressed*. It's been a few weeks, and I've tested with the debug version. Seems that it's hanging in some pipe related stuff, see GDB output further down. Other peculiarities noted: 1. When listing processes with PS, the hung process has the 'WINPID' in the cygwin PID column instead of the cyg PID. 2. The problem seems to only occur on one machine, which happens to be multi-processor (2*Xeon). At least it occurs very frequently on this box. I've seen a hang once on a UP machine. I tried to find more information about the process using Process Explorer, which seemingly triggers another strange condition - the CPU time of the process shot to 100%. It was looping in ntdll.dll!RtlConvertUiListToApiList+0x2fd. Here's backtraces from GDB (snipped a bit here and there). core dump 1 (gdb) info threads 3 process 3620 0x77e88785 in KERNEL32!GetModuleFileNameA () 2 process 4236 0x77f839eb in ntdll!ZwReadFile () * 1 process 3832 0x77f8376e in ntdll!ZwClose () (gdb) bt #0 0x77f8376e in ntdll!ZwClose () #1 0x77e87738 in KERNEL32!CloseHandle () #2 0x61073b16 in fhandler_pipe::close() (this=0x616d15f0) --- at ../../../../winsup/cygwin/pipe.cc:103 #3 0x6109ac34 in close (fd=5) at winsup/cygwin/cygheap.h:304 #4 0x6108db7f in _sigfe () at winsup/cygwin/cygserver.h:82 = core dump 2 (gdb) info threads 3 process 4496 0x77e88785 in KERNEL32!GetModuleFileNameA () 2 process 3796 0x77f839eb in ntdll!ZwReadFile () * 1 process 1528 0x77f8376e in ntdll!ZwClose () (gdb) bt #0 0x77f8376e in ntdll!ZwClose () #1 0x77e87738 in KERNEL32!CloseHandle () #2 0x61073b16 in fhandler_pipe::close() (this=0x616d0fd8) --- at ../../../../winsup/cygwin/pipe.cc:103 #3 0x6109ac34 in close (fd=5) at winsup/cygwin/cygheap.h:304 #4 0x6108db7f in _sigfe () at winsup/cygwin/cygserver.h:82 = core dump 3 (gdb) info threads 3 process 2424 0x77e88785 in KERNEL32!GetModuleFileNameA () 2 process 4568 0x77f839eb in ntdll!ZwReadFile () * 1 process 4540 0x77f8376e in ntdll!ZwClose () (gdb) bt #0 0x77f8376e in ntdll!ZwClose () #1 0x77e87738 in KERNEL32!CloseHandle () #2 0x61073b16 in fhandler_pipe::close() (this=0x616d15f0) --- at ../../../../winsup/cygwin/pipe.cc:103 #3 0x6109ac34 in close (fd=5) at winsup/cygwin/cygheap.h:304 #4 0x6108db7f in _sigfe () at winsup/cygwin/cygserver.h:82 = The test were performed with 1.5.10-3, as newer versions call upon me all sorts of other problems and thus can't be pushed to the failing box right now. Btw, I urge everyone to try the latest cygwin snapshot! Will do. Has the problem been found that results in this error?: *** MapViewOfFileEx(0x188, in_h 0x188) failed, Win32 error 6 -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Re: cygwin bughunt (Jip-hee!)
On Thu, Feb 17, 2005 at 02:23:42PM +0100, David Dindorp wrote: Christopher Faylor wrote: Actually, we do. We provide the source code. It's easy to build. You are right; I was wrong. Building Cygwin is easy. (At least when it comes to newer versions :-p.) It even compiles under itself. *impressed*. It's been a few weeks, and I've tested with the debug version. Er, who are you? **cgf checks archives Ah, yes! You're the you don't want people to debug cygwin because you aren't spoon feeding me debugging information guy! You're the guy who insisted on dropping back to 1.5.10! I remember now! In the future, please provide more context if you really want to be helped, especially after almost a month of silence. The test were performed with 1.5.10-3, as newer versions call upon me all sorts of other problems and thus can't be pushed to the failing box right now. Btw, I urge everyone to try the latest cygwin snapshot! Will do. Has the problem been found that results in this error?: *** MapViewOfFileEx(0x188, in_h 0x188) failed, Win32 error 6 1) Regardless of whether the new versions call upon (you) all sorts of other problems, I doubt that anyone is going to spend any time tracking down problems in older versions. 2) You can *try* a snapshot to see if it fixes your problem. cgf -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/