MRXS 2.2 - test failures

[Agelos Pappas]

Hi Benjamin,

in the first image you can see what the slide looks like zoomed out. In 
the second, I zoomed one step in and the application crashed reporting 
an access violation in openslide_read_region.

I didn't notice any corrupted images or crashes using the java viewer 
under windows. Is it multithreaded? I've been hunting this down the past 
few days and I didn't find it at all easy to reproduce.

It happens when threads run in parallel and is more evident when the 
program executes quickly. I wrote a simple desktop application that 
spawns N threads, each of which instantiates a single openslide_t, all 
to a specific slide file.
Every thread then walks from the top zoom level to the bottom reading 
256x256 tiles. If N equals to the number of CPU cores your system has, 
so that the app runs quicker, the application crashes almost immediately.
If I perform a lock in openslide_read_region calls, the application runs 
fine, but slower since tiles are read in series.

I have only noticed the problem in MRXS slides, but I didn't test it 
excessively using other type of slides. It does happen on public slides 
available in openslide.org.

This is definitely a thread safety related bug, but I cannot tell for 
sure if it lies within my application or in OpenSlide. I do know that we 
only encountered this when we switched to the latest OpenSlide binaries. 
This of course might very well be because the older OpenSlide version 
was doing locks itself and this could be preventing the bug in our 
application to show.

I will try using libjpeg and tell you what I find.

Regards
Pappas Agelos

On 13/11/2012 2:42 πμ, Benjamin Gilbert wrote:
> On 11/11/2012 09:40 AM, Agelos Pappas wrote:
>> My question therefore is, does the current GitHub version of OpenSlide
>> pass the provided mirax test cases, as far as valgrind is concerned? In
>> my system it looks like it does not.
>> I am running Ubuntu 12.04 LTS 32-bit in VMWare.
> The mirax test passes on 64-bit Fedora 17.  However, some of the other
> tests fail, due to:
>
>       https://sourceforge.net/p/libjpeg-turbo/bugs/48/
>
> I added --track-origins=yes to the valgrind command line, ran the mirax
> tests on a 32-bit Ubuntu 12.04 system, and got backtraces similar to the
> ones in that bug.  I'm still tracking it down, though.
>
> - Could you post a screenshot of the corruption?
>
> - Are you seeing corruption on Linux or only on Windows?
>
> - On Windows, can you reproduce it using the OpenSlide Java viewer?  You
> can run the viewer by double-clicking openslide-win32\bin\openslide.jar.
>
> - Can you rebuild OpenSlide against libjpeg, rather than libjpeg-turbo,
> and see if the problem goes away?  (Install libjpeg into /usr/local,
> then "./configure CPPFLAGS=-I/usr/local/include LDFLAGS=-L/usr/local/lib".)
>
> - Are you using the public test slides?
>
> Thanks,
> --Benjamin Gilbert
> _______________________________________________
> openslide-users mailing list
> openslide-users at lists.andrew.cmu.edu
> https://lists.andrew.cmu.edu/mailman/listinfo/openslide-users
>
>
> -----
> No virus found in this message.
> Checked by AVG - www.avg.com
> Version: 2012.0.2221 / Virus Database: 2441/5390 - Release Date: 11/12/12
>
>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: slide1.jpg
Type: image/jpeg
Size: 74502 bytes
Desc: not available
Url : http://lists.andrew.cmu.edu/pipermail/openslide-users/attachments/20121113/46c39e90/attachment-0002.jpg 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: slide2.jpg
Type: image/jpeg
Size: 119797 bytes
Desc: not available
Url : http://lists.andrew.cmu.edu/pipermail/openslide-users/attachments/20121113/46c39e90/attachment-0003.jpg