VMD-L Mailing List

From: yjcoshc_at_gmail.com
Date: Thu Jun 01 2017 - 09:58:42 CDT

Hello,

   Thanks for fixing the issue! I can build VMD in my own machine. If
you are busy now you can send me a patch. I will test it.

Thanks again,

yjcoshc

在 2017年06月01日 05:33, John Stone 写道:
> Hi,
> I have found and fixed the issue that caused the crash that
> you observe, but I don't quite understand what the issue is with the file.
> It seems that some of the timesteps in the file you provided don't contain
> coordinates, and that upsets the VMD gromacs plugin. I have made the VMD
> Gromacs plugin tolerate such files, but I'm puzzled why you would want a
> trajectory that periodically has a sequence of frame without any coordinates.
> The next test versions of VMD we make available on the web site will contain
> the new plugin. If you would like to try out a new version soon, send me a
> note in about a week (I'm away from the office traveling until then anyway)
> and I can see about getting you a build. During the time while I'm away
> a bunch of our build machines are down due to building construction projects
> underway here, so I can't do VMD builds remotely.
>
> Cheers,
> John Stone
> vmd_at_ks.uiuc.edu
>
> On Sat, Apr 29, 2017 at 07:43:53PM +0800, yjcoshc wrote:
>> Hi,
>>
>> Any ideas? I managed to generate a small(36M) trr file that could
>> crash the VMD. You can download it at:
>>
>> http://chinfo.nankai.edu.cn/files/vmdtrr/traj.trr
>>
>> The gmx check result is:
>>
>> Command line:
>> gmx check -f ./traj.trr -s1 ./eq.tpr
>>
>> Reading file ./eq.tpr, VERSION 2016.3 (single precision)
>> Checking file ./traj.trr
>> trr version: GMX_trn_file (single precision)
>> Reading frame 0 time 0.000
>> # Atoms 43044
>> Last frame 50 time 100.000
>>
>>
>> Item #frames Timestep (ps)
>> Step 51 2
>> Time 51 2
>> Lambda 51 2
>> Coords 11 10
>> Velocities 11 10
>> Forces 51 2
>> Box 51 2
>>
>> The gromacs running tpr file is at
>> http://chinfo.nankai.edu.cn/files/vmdtrr/eq.tpr. I ran it 50000
>> steps.
>>
>> Thanks,
>>
>> yjcoshc
>>
>>
>> ??? 2017???03???03??? 04:25, John Stone ??????:
>>> Hi,
>>> Looking at the source code it seems that this is an unusual .trr file.
>>> Would it be possible for you to post this file somewhere where
>>> I could download it? That would allow me to debug the problem in more
>>> detail. It is conceivable that this file triggers an extremely unusual
>>> path through the code, which could explain the segfault you observe, but
>>> I need access to the input file to be certain. If you can post the file
>>> somewhere, I should be able to fix it and provide an updated plugin for
>>> VMD 1.9.3 that will address this issue.
>>>
>>> Cheers,
>>> John Stone
>>> vmd_at_ks.uiuc.edu
>>>
>>> On Wed, Mar 01, 2017 at 12:11:21PM +0800, yjcoshc_at_gmail.com wrote:
>>>> Hi,
>>>>
>>>> I ran gmx check on my trr file:
>>>>
>>>> Checking file ./traj.trr
>>>> trr version: GMX_trn_file (single precision)
>>>> Reading frame 0 time 0.000
>>>> # Atoms 2422
>>>> Last frame 50000 time 50000.000
>>>> WARNING: Incomplete header: nr 50001 time 50000
>>>>
>>>>
>>>> Item #frames Timestep (ps)
>>>> Step 50001 1
>>>> Time 50001 1
>>>> Lambda 50001 1
>>>> Coords 5001 10
>>>> Velocities 5001 10
>>>> Forces 50001 1
>>>> Box 50001 1
>>>>
>>>> The machine has 32G memory and 29G available.
>>>>
>>>> Thanks,
>>>>
>>>> yjcoshc
>>>>
>>>>
>>>> ??? 2017???03???01??? 02:57, John Stone ??????:
>>>>> Hi,
>>>>> How much memory does your machine have? Are you certain that
>>>>> you have enough physical memory for the trajectory to be loaded
>>>>> within VMD? The backtrace shows it dies in an optimized memmove()
>>>>> call which indicates to me that it ran beyond a buffer boundary,
>>>>> perhaps because a previous memory allocation had failed. Ignoring
>>>>> the on-disk size of your TRR file, how many atoms are in the structure
>>>>> and how many frames are you trying to load into RAM?
>>>>>
>>>>> Cheers,
>>>>> John Stone
>>>>> vmd_at_ks.uiuc.edu
>>>>>
>>>>> On Sat, Feb 25, 2017 at 10:54:27AM +0800, yjcoshc_at_gmail.com wrote:
>>>>>> Hello everyone,
>>>>>>
>>>>>> I am using VMD 1.9.3 and trying to load a trr trajectory from
>>>>>> GROMACS 2016.1 in linux(opensuse tumbleweed). The VMD crashes with
>>>>>> Segmentation fault. I have tried attaching the gdb to it and print
>>>>>> the backtrace as follow:
>>>>>>
>>>>>> Thread 1 "vmd_LINUXAMD64" received signal SIGSEGV, Segmentation fault.
>>>>>> 0x00007fd1e406372f in __memmove_avx_unaligned_erms () from /lib64/libc.so.6
>>>>>> (gdb) backtrace
>>>>>> #0 0x00007fd1e406372f in __memmove_avx_unaligned_erms () from
>>>>>> /lib64/libc.so.6
>>>>>> #1 0x00000000007a24e6 in read_trr_timestep(void*, int,
>>>>>> molfile_timestep_t*) ()
>>>>>> #2 0x0000000000685181 in MolFilePlugin::next(Molecule*) ()
>>>>>> #3 0x00000000005eecd4 in CoorPluginData::next(Molecule*) ()
>>>>>> #4 0x00000000006755b0 in Molecule::get_new_frames() ()
>>>>>> #5 0x0000000000676879 in Molecule::prepare() ()
>>>>>> #6 0x00000000005f4d4a in Displayable::draw_prepare() ()
>>>>>> #7 0x00000000006d0e83 in Scene::prepare() ()
>>>>>> #8 0x00000000006e37f2 in VMDApp::VMDupdate(int) [clone .part.106] ()
>>>>>> #9 0x000000000049ea32 in main ()
>>>>>>
>>>>>> It seems a bug in the molfile plugin because I have failed to load
>>>>>> the trajectory in the latest pymol(also segfault).
>>>>>>
>>>>>> The trr file is 1.6G. Both the text version and the GUI version of
>>>>>> vmd in linux crash but the windows version doesn't have the problem.
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> yjcoshc