[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: build errors: 9172 Broken pipe and 9173 Segmentation fault
From: |
Andy Wingo |
Subject: |
Re: build errors: 9172 Broken pipe and 9173 Segmentation fault |
Date: |
Fri, 01 Apr 2011 11:08:10 +0200 |
User-agent: |
Gnus/5.13 (Gnus v5.13) Emacs/23.2 (gnu/linux) |
Hi,
On Tue 29 Mar 2011 19:07, Douglas Mencken <address@hidden> writes:
> <mark_weaver> dougmencken: the real problem seems to be that an
> infinite recursion results while handling a string decoding error. it
> is also curious why a decoding error is happening with uClibc that
> does not occur with glibc, but that's secondary.
> <mark_weaver> scm_from_locale_string calls scm_from_locale_stringn
> which calls scm_from_stringn (in your case UTF-8 presumably) which
> calls scm_decoding_error which calls scm_from_locale_string, ad
> infinitum.
> <mark_weaver> perhaps the decoding error handler is trying to print
> the erroneous string as part of its error message, or something.
> <mark_weaver> that would be my guess anyway.
I don't know what bug this is.
Your backtrace includes repeating lines line this:
#13263 0x480dc9cc in scm_from_locale_stringn (str=0x4812560c
.¢..±.¶.³..±.´.³..±.µ.µ..±.³.·..±.´.¶..±.¶.²..±.µ.·..±.µ.µ..±.³.·..±.¶.³..±.¶.´..±.¶.²..±.µ.±..±.µ.¶..±.´.·..±.µ.¶.¢,
len=<value optimized out>) at strings.c:1535
#13264 0x480dc9fc in scm_from_locale_string (str=<value optimized out>) at
strings.c:1529
#13265 0x480dca58 in scm_decoding_error (subr=<value optimized out>, err=38,
message=0x48125a54
.¢..±.µ.±..±.µ.¶..±.¶.°..±.¶.µ..±.¶.´..°.´.°..±.µ.´..±.µ.·..±.´.³..±.´.±..±.µ.´..±.´.µ..°.´.°..±.´.³..±.µ.·..±.µ.¶..±.¶.¶..±.´.µ..±.¶.²..±.¶.³..±.µ.±..±.µ.·..±.µ.¶..°.´.°..±.´.µ..±.¶.²..±.¶.²..±.µ.·..±.¶.².¢,
port=<value optimized out>) at strings.c:1438
#13266 0x480dcce4 in scm_from_stringn (str=0x4812560c
.¢..±.¶.³..±.´.³..±.µ.µ..±.³.·..±.´.¶..±.¶.²..±.µ.·..±.µ.µ..±.³.·..±.¶.³..±.¶.´..±.¶.²..±.µ.±..±.µ.¶..±.´.·..±.µ.¶.¢,
len=<value optimized out>, encoding=<value optimized out>, handler=<value
optimized out>) at strings.c:1494
But the only calls to scm_from_locale_string in scm_decoding_error are
on string literals:
scm_from_locale_string (subr),
scm_from_locale_string (message),
But the argument to scm_from_locale_stringn does not appear to be one of
the known strings that it could be.
In any case, it does seem more correct to use a fixed locale for these
strings, so I have changed it to use latin1 strings.
Regards,
Andy
--
http://wingolog.org/