[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#917034: xterm crashes on certain Unicode characters when font was selected with -fa



* Thomas Dickey <dickey@his.com> [2018-12-30 23:14]:

> On Sun, Dec 30, 2018 at 10:15:53PM +0100, Sven Joachim wrote:
>> On 2018-12-30 18:26 +0100, Alexander Meyer wrote:
>>> So the segfault only happens with that fonts.conf.
>> 
>> Thanks, that explains why others did not see it - and possibly also why
>> I could still reproduce #916349 in xterm 340, as I have my own
>> fonts.conf, albeit a much shorter and simpler one than yours.
>> 
>> Alas, with your file (saved as /tmp/fonts.conf) xterm does not even start
>> here, although I have fonts-noto-mono and fonts-noto-color-emoji
>> installed:
> 
> If xterm sees the latter, it will not start.  That's a problem that can
> only be solved by modifying Xft.  Not xterm.
> 
> But the short term goal is to get xterm working as well as feasible with
> Xft before moving on to see what can be done to make Xft handle the
> color bitmap fonts.
> 
> I'm looking for details on how to reproduce the segmentation violation
> which was reported.
>  
>> ,----
>> | $ FONTCONFIG_FILE=/tmp/foo/fonts.conf xterm -fa 'Noto Mono'      
> 
> Back up a step: with that fonts.conf, I get no result for
> 	fc-match 'Noto Mono'
> 
> while without it, I do get a match.  A quick look at strace tells me
> that setting FONTCONFIG_FILE eliminates a lot of the normal configuration
> information, and probably isn't what you want to do in reproducing the
> problem.

For the record: I disabled/enabled my fonts.conf by moving it out of
~/.config/fontconfig/ and then back inside. I did not set FONTCONFIG_FILE.

Also, as described in my original report, the segfault is not limited to
"Noto Mono" but seems to happen with any other font, too. (I've randomly
tried a few from my fc-list.)

>> | xterm: cannot match normal font "Noto Mono"
>> | xterm: Selected font has no non-zero height for ISO-8859-1 encoding
> 
> ...and because there's no match, for _this_ case I get the same failure.
> 
> However, that's not a bug (in xterm), but a problem with configuration.
> 
>> So I am a bit at a loss here.  And I am definitely not a an expert on
>> fontconfig (or fonts in general) either.
> 
> I'd like to see the XFT_DEBUG information from the known broken scenario.
> If it tells me that Xft is still loading the color-noto font, there's
> probably nothing that I can do to address that in xterm.

Please find attached the gzipped output of

$ XFT_DEBUG=1023 xterm -fa 'Noto Mono'

with xterm 341. All output happens directly after startup, there's no
output after issuing the command that displays the character that causes
the segfault.

Best
Alex

Attachment: xterm-341-XFT_DEBUG_1023.log.gz
Description: application/gzip


Reply to: