pyexiv2 fails on cifs shares served by win2k8-servers

[thoralf]

first off, i do not think that this is pyexiv's fault, but i do think that this issue should be reported somewhere …

this code:

        meta_source = pyexiv2.metadata.ImageMetadata(in_fn)
        meta_source.read()
        meta_dest = pyexiv2.metadata.ImageMetadata(out_fn)
        meta_dest.read()
        meta_source.copy(meta_dest)
        meta_dest["Exif.Photo.PixelXDimension"] = dest_x
        meta_dest["Exif.Photo.PixelYDimension"] = dest_y
        print "hier …"
        meta_dest.write()
        print "… hängt es"

hangs at meta_dest.write() iff:
- out_fn points to a file on a cifs share mounted from a win2k8 server
- the jpeg file referenced by meta_dest is larger than ~100kb

meta_dest.write() does not hang if:
- meta_dest is bigger than 100k and resides on a cifs share served by a samba server or a local file system
- meta_dest is smaller than 100k and resides on either cifs share
- meta_dest is bigger than 100k, resides on a cifs share served by the infamous win2k8 server and the python process executing the script is being strace'd
- meta_dest is bigger than 100k, resides on a cifs share served by the infamous win2k8 server and the share is mounted with mount.cifs' "cache=none"-option

a look at self._image._writeMetadata() makes me think that the issue might rather be caused by exiv2 itself than by pyexiv2 - thanks for having a look anyway. and please let me know if there's anything i can do to help to resolve this issue.

the pyexiv2 version i am using is the one in the debian wheezy repositores (0.3.2-5 linked against libexiv2-12 0.23-1).

with kind regards,
t.

[Olivier Tilloy (osomon)]

Thanks for the detailed report. pyexiv2 is not actively maintained any longer, it has been deprecated in favour of https://wiki.gnome.org/Projects/gexiv2.

As you point out, it looks like the issue could be in exiv2 itself. You might want to file a bug with all those details at http://dev.exiv2.org/projects/exiv2/issues.

[thoralf (tschulze-7)]

hi olivier,

On 03/19/2015 08:06 AM, Olivier Tilloy wrote:
> As you point out, it looks like the issue could be in exiv2 itself. You
> might want to file a bug with all those details at
> http://dev.exiv2.org/projects/exiv2/issues.

thank you - i filed a bug report there.

with kind regards,
t.

[thoralf (tschulze-7)]

hi there,

it seems that neither pyexiv2 nor exiv2 were to blame, but rather the cifs implementation in the linux kernel. see http://dev.exiv2.org/issues/1043 for details.

with kind regards,
t.

[Olivier Tilloy (osomon)]

Thanks for the head-up!