When I compile the following simple document with XeLaTeX:

\documentclass{article}

\usepackage{dsfont}

\begin{document}
$\mathds N \subset \mathds Z \subset \mathds Q \subset \mathds R \subset \mathds C$
\end{document}


Adobe Reader (now version 10.1.3 but this problem existed for some time) does not display the double stroke symbols correctly. Instead it just gives an error message:

Cannot extract the embedded font 'RAXAIQ+dsrom10'. Some characters may not display or print correctly.

and replaces those symbols with a dot:

It should look like this:

Other readers like FoxIt or Okular have no problems displaying the document correctly.

After reading egregs comment, I looked into dsrom10.pfb. The easiest way for me to inspect it, is using fontforge. First, fontforge reads the font without problems. But when I try to recreate the pfb file, I get an error for every glyph:

Missing Points at Extrema

I read that it is possible to correct these with fontforge, but then the dsrom10.pfb was automatically generated by pktrace. So even if that would work (maybe I'll try it), it cannot be the final solution.

-
True: I see this with Adobe Reader on both Windows and OS X, but get the correct result with Preview and TeXworks. However, I'm not sure what can be said: probably a bug in Reader. –  Joseph Wright Jul 14 '12 at 17:55
That said, the problem only occurs with XeTeX: compiling with pdfTeX or LuaTeX gives the correct result. Perhaps an issue in XeTeX. –  Joseph Wright Jul 14 '12 at 17:59
Either latex-dvips-ps2pdf or pdflatex can produce the correct result. But xelatex cannot. –  kiss my armpit Jul 14 '12 at 18:07
I got an error message when opening the xelatex's PDF output. –  kiss my armpit Jul 14 '12 at 18:10
@JosephWright I'm inclined towards a bug in dsrom10.pfb, but I've no particular reason. If I use xdv2pdf instead of xdvipdfmx I get some warnings, but the result is shown correctly –  egreg Jul 14 '12 at 20:28

Egreg's guess was correct. But I would not call it a bug; rather an incompatibility between dsrom10.pfb and Adobe Reader, as FontForge had no problems loading it. (While for the other pdf viewers I can not say – maybe they are just ignoring the issue.)
What I did, was to first apply the add extrema tool of FontForge and replace my version of dsrom10.pfb with the newly exported one. Et voilà. Adobe Reader does not issue an error any more and displays the symbols correctly.
As the pfb file was created from the MetaFont, this could of course not be the final solution. So after finding the tool pktrace — which is now called mftrace — I could just recreate all the pfb files of dsfont. It seems this current version of the tool does a good enough job, as those files worked without any further intervention.
I hope I can reach the package maintainer of dsfont to get the correct fonts published.