[XeTeX] Segmentation fault in xdvipdfmx with Alvi Nastaleeq font

[Mike Maxwell]

Jonathan Kew wrote:
> On 13 Feb 2010, at 23:50, Mike Maxwell wrote:
>> ...the 64-bit Linux version in
>> bin/x86-64-linux/ is the old v0.7.7 dating from 18 Sept 09.
> 
> Aha, that's interesting. But harmless. It means the 64-bit Linux
> binaries were not rebuilt when that last-minute xdvipdfmx change was
> made, but that's OK because it wouldn't have affected them (except
> for the version number).

OK, then back to the segmentation fault (I've changed the subject line, 
since this is apparently a different seg fault that the OP was having 
with the CharisSIL font).  I get a segmentation fault with the Alvi 
Nastaleeq font, but not with the Nafees Nastaleeq font.

With the Alvi font: I run xdvipdfmx (this 64-bit v0.7.7, under Linux) on 
the .xdv file.  It appears to get to the end of the document (I see the 
last page #), then it outputs several screensful of msgs like
     name=onrNo Unicode mapping available: GID=15645
oops, I guess that should be
     No Unicode mapping available: GID=160, name=glyph160
(the msgs are all run together, no newlines).

Then at the end it says
     Segmentation fault
and the PDF is invalid (it won't load).

If I run xdvipdfmx with the -v flag, I get many add'l msgs, like the ff 
(I'm using dashes to separate sxns of msgs):
----------
 >][4<Alvi Nastaleeq(Alvi Nastaleeq:Regular)@13.09pt<NATIVE-FONTMAP:Alvi 
Nastaleeq/H/65536/0/0>
pdf_font>> Input encoding "Identity-H" requires at least 2 bytes.
pdf_font>> The -m <00> option will be assumed for 
"/groups/opt/share/fonts/alvi_Nastaleeq_Lahori_shipped.ttf".
(CID:Alvi Nastaleeq)
pdf_font>> Type0 font 
"/groups/opt/share/fonts/alvi_Nastaleeq_Lahori_shipped.ttf" 
cmap_id=<Identity-H,0> opened at font_id=<Alvi Nastaleeq/H/65536/0/0,7>.
---------
** WARNING ** Object reference with key "table.2.4" is in use.
---------
otf_cmap>> Creating ToUnicode CMap for 
"/groups/opt/share/fonts/alvi_Nastaleeq_Lahori_shipped.ttf"...
No Unicode mapping available: GID=1, name=glyph1
---------
I tried the -vv setting, but don't see anything obviously enlightening, 
except that the "No Unicode mapping available" msgs are separated by 
newlines.

Should I make a minimal example, or is this a known problem?
-- 
    Mike Maxwell
    What good is a universe without somebody around to look at it?
    --Robert Dicke, Princeton physicist