Re: xboard sends SIGINT to gnuchess, and gnuchess exits

[Simon Waters]

Michael Wang wrote:
> 
> Correction on
> > (3) xboard-4.2.5 and gnuchess-5.03beta runs ok on Solaris 8 x86.
> 
> It depends how I move. If I (white, computer black) move horse
> to the inner position, and then another horse also to the inner
> position (towards the center), then I got the following:
> 
> 14881:  gettimeofday(0x08087134)                        = 0
> 14881:  write(1, " 7 .   - 1 2   6 0 0   6".., 39)      = 39
> 14881:  write(1, " 1 .   . . .   e 7 e 5\n".., 29)      = 29
> 14881:  read(0, 0x08F4730C, 5120)       (sleeping...)
> 14881:      Received signal #2, SIGINT, in read() [caught]
> 14881:        siginfo: SIGINT pid=14880 uid=100
> 14881:  read(0, 0x08F4730C, 5120)                       Err#4 EINTR
> 14881:  sigaction(SIGINT, 0x08047658, 0x080476B4)       = 0
> 14881:  setcontext(0x08047744)
> 14881:  read(0, " t i m e   2 9 3 2 3\n o".., 5120)     = 27
> 14881:  write(1, " 2 .   g 1 f 3\n", 8)                 = 8
> 14881:      Incurred fault #6, FLTBOUNDS  %pc = 0x080708AA
> 14881:        siginfo: SIGSEGV SEGV_MAPERR addr=0x08F74646
> 14881:      Received signal #11, SIGSEGV [default]
> 14881:        siginfo: SIGSEGV SEGV_MAPERR addr=0x08F74646
> 14881:          *** process killed ***

Unless Tim is inspired (he's a much more experienced developer
than I) I assume we have fixed the SIGINT problem, and can spare
him more e-mail.

Stuart and I both get bug-gnu-chess e-mail - so we don't need
copies.

I don't have reason to suspect xboard in any of this. Solaris in
places but not xboard.

You are running with "xboard" argument to gnuchess now? i.e.
-fcp "gnuchess xboard"

> > (1) gnuchess-5.03beta compiles ok, and runs ok on Solaris 8 x86.
> > (2) gnuchess-5.03beta compiles ok, but does NOT run ok on Solaris 8 Sparc.
> >     ./gnuchess does not respond to any of the test command: "post", "depth 
> > 8",
> >     "go", "quit" on Solaris 8 Sparc.

> > (3) xboard-4.2.5 and gnuchess-5.03beta runs ok on Solaris 8 x86.
xboard is fine.

I (well lclint) found a few errors in 5.03beta which might
explain the SIGSEV, handling of book, and a few C format
modifiers were of the wrong type, which seems to be more of a
problem when you clear out unused variables.

The SPARC problem I suspect is the ENDIANNESS test in util.c. I
have a half rewritten "configure.in" file for 5.03beta that puts
in the right test (i.e. GNU standard) for ENDIANNESS (It's only
a one line change, but needs more work to actually do anything).

Stuart is going to check these problems on his SUN - cheers. If
he finds the CVS version is okay I'll package it up as a later
beta release. 

> > (4) xboard-4.2.5 and gnuchess-5.02 does NOT runs ok on Solaris 8 x86.
> >     same error as on Solaris 8 Sparc, shown below:

Okay that one we know about, and have fixed in 5.03beta.

-- 
Are you using the Internet to best effect ? www.eighth-layer.com
Tel: +44(0)1395 232769      ICQ: 116952768
Moderated discussion of teleworking at news:uk.business.telework