It was a buffer overflow! In $GEMPAK/source/prorgrams/gui/nmap2/nmap_spfw.c,
the dimension of the local string variable, "tmpStr1" in the spfw_formatConf
function needs to be increased from 10. I chose 20. Recompile and voila, spf
files load properly!
I've attached a corrected file (from the 6.8.0 distribution).
Michael, if it tests out ok, please roll it out for future versions ...
--Kevin
_____________________________________________
Kevin Tyle, Systems Administrator
Dept. of Atmospheric & Environmental Sciences
University at Albany
Earth Science 235, 1400 Washington Avenue
Albany, NY 12222
Email: ktyle@xxxxxxxxxx
Phone: 518-442-4578
_____________________________________________
________________________________________
From: James Murakami [tenki@xxxxxxxxxxxxxx]
Sent: Monday, August 19, 2013 3:32 PM
To: Tyle, Kevin R
Cc: Stonie R. Cooper; Unidata gembud
Subject: Re: [gembud] Nmap2 problem on Mac OS X 10.6.8
Hi Kevin,
Not the most convenient way to go, but your suggestion of using the
command line option for restoring a spf file works. Thank you.
Michael James said he'd investigate the matter, but your email sounds
like it won't be a simple corrective solution.
James
----------------------------------------------
James Murakami
Staff Meteorologist/Student Affairs
Department of Atmospheric and Oceanic Sciences
University of California, Los Angeles
405 Hilgard Ave.
Los Angeles, CA 90095-1565
e-mail: tenki@xxxxxxxxxxxxxx
telephone: 310-825-2418
Fax: 310-206-5219
----------------------------------------------
On 08/19/2013 11:58 AM, Tyle, Kevin R wrote:
> I can confirm James' observations. The problem looks like it may be
> difficult to trace ... I can get the program to abort at line 812 in
> $GEMPAK/source/programs/gui/nmap2/nmap_spfw.c:
>
> if( ! (spftext = XtMalloc((Cardinal)flen))) {
>
> However, inserting some printf statements, as well as some other innocuous
> code changes, cause the error to appear further along in the code ... which
> likely means that the actual error lies elsewhere. Since the Abort error
> often is tied to a stack violation on OSX, I tried recompiling with the
> -fno-stack-protector option, which corrects NMAP2 aborts on Ubuntu, but this
> did not help. I suspect that there is some insidious buffer overflow hiding
> in the NMAP2 code base. Maybe compiling with -Wall may turn up something.
>
> In the meantime, James, you can still load your .spf files in NMAP2 on OSX by
> using the command line option to load spf files ... e.g., "nmap2 -as
> sample.spf".
>
> I also can confirm the problem with loading vg files on OSX.
>
> Cheers,
>
> Kevin
>
> _____________________________________________
> Kevin Tyle, Systems Administrator
> Dept. of Atmospheric & Environmental Sciences
> University at Albany
> Earth Science 235, 1400 Washington Avenue
> Albany, NY 12222
> Email: ktyle@xxxxxxxxxx
> Phone: 518-442-4578
> _____________________________________________
>
>
> -----Original Message-----
> From: gembud-bounces@xxxxxxxxxxxxxxxx
> [mailto:gembud-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Stonie R. Cooper
> Sent: Wednesday, August 14, 2013 7:52 PM
> To: James Murakami
> Cc: Unidata gembud
> Subject: Re: [gembud] Nmap2 problem on Mac OS X 10.6.8
>
> Shoot. That is ugly. I don't even have access to a modern Mac OS to try to
> debug. You would need to use a debugger with the "-g" flag at compile time,
> or add some stdout code with fflush() calls to try to find the issue.
>
> On 08/14/2013 11:54 AM, James Murakami wrote:
>> Hi Stonie,
>>
>> Thanks for the suggestion, but that didn't help. I edited
>> $GEMTBL/nmap/spf.nmap and created a test directory for the ~/Documents
>> folder. However, Nmap2 still crashed upon trying to restore the test
>> setting(tried all sorts of variations).
>> I tried running Gempak under bash(I'm a tcsh user by default), but the
>> results were the same.
>>
>> James
>>
>> ----------------------------------------------
>> James Murakami
>> Staff Meteorologist/Student Affairs
>> Department of Atmospheric and Oceanic Sciences University of
>> California, Los Angeles
>> 405 Hilgard Ave.
>> Los Angeles, CA 90095-1565
>>
>>
>> e-mail: tenki@xxxxxxxxxxxxxx
>> telephone: 310-825-2418
>> Fax: 310-206-5219
>> ----------------------------------------------
>>
>> On 08/13/2013 04:51 PM, Stonie R. Cooper wrote:
>>> James - when I had last played with this, the issue was pathing.
>>>
>>> But this was pre 10.6.x - so I am only guessing.
>>>
>>> When I saved a data setting, it stored it local to the path from
>>> which I ran the nmap2 session, usually just my home directory.
>>>
>>> But the Mac OS bash shell always wanted to read it from ~/Documents.
>>>
>>> See if that helps.
>>>
>>> Stonie
>>>
>>> On 08/13/2013 05:24 PM, James Murakami wrote:
>>>> Hi,
>>>>
>>>> I recently compiled GEMPAK 6.8.0 on a mini-Mac(OS X version 10.6.8).
>>>> It seemed to compile fine, but one problem I ran into regards one
>>>> functionality of Nmap2. The "Restore Data Settings" button(within
>>>> the Data Selection Window) doesn't work. I created a map with radar
>>>> imagery and surface observations superimposed. I clicked on the
>>>> "Save Data Settings" button. Later, when I tried to bring up the
>>>> same setting, the program crashes. This doesn't occur with Gempak on
>>>> my Linux(CentOS) machine. Another quirk I found is that Nmap2 nor
>>>> the program, gpmap, will plot a vgf file(only the map background
>>>> shows up). Again, it's not a problem on my Linux machine(using the same
>>>> data files).
>>>>
>>>> Have others noticed these problems, and is there a solution for it?
>>>> The reason I'm using a Mac in this case it that it'll be a
>>>> replacement for a Department display for the public. The current
>>>> min-mac uses binary version of GEMPAK 5.11.1. I need to upgrade
>>>> GEMPAK to be able to display gini satellite imagery(just copying the
>>>> latest satellite configuration tables don't work for gini imagery).
>>>> The closet space behind the display case is just large enough to
>>>> accommodate a mini-mac.
>>>>
>>>> James
>>>>
>>> .
>>>
>>
> _______________________________________________
> gembud mailing list
> gembud@xxxxxxxxxxxxxxxx
> For list information or to unsubscribe, visit:
> http://www.unidata.ucar.edu/mailing_lists/
>
>
> .
>
Attachment:
nmap_spfw.c
Description: nmap_spfw.c
gembud archives: