SUMMARY: Bug in _XData32 from /usr/lib/64/libX11.so.4 ?

Well, I wish I had good news, but I really don't. James Noyes responded 
that he had encountered a similar problem (he is trying to verify my 
results now with gdb himself). However, Casper Dik responded thusly:

> This bug is listed now as "will not fix"; it appears to be caused
> by under specification of the XChangePropery() (when the format 32
> is specified, "longs" are converted as the standards specify rather 
> than
> "32 bit" quantaties the coders often expect)
>
> <snip>
>
> It seems that the standard specified that "32 bit " means "long"
> there a property type of "32" should be acommpanied with a 64(!)
> bit value.

I think it is fairly debatable whose bug this is. However, it is 
abundantly clear that Sun isn't going to "fix" what they feel isn't 
broken. The only solution for this is to convince the Afterstep people 
(and others who have this same bug, notably blackbox and its ilk) to 
send XChangeProperty() the "right" parameters.

Original post is below.

Alex

On Sun, 23 Mar 2003 14:17:41 -0500
alex avriette <avriettea@speakeasy.net> wrote:
> I've been trying today to get AfterStep compiled 64-bit (with gcc
> 3.2.2).
>
> Itcompiled happily and linked against the 64-bit libs I had fed it
> (such as libjpeg and libpng). My LD_LIBRARY_PATH is set up correctly. I
> am getting a "Bus Error" (reported to the shell). gdb(1) reports this
> actually as the dreaded SIGSEGV:
>
> Program received signal SIGSEGV, Segmentation fault.
> 0xffffffff7e85bbd8 in _XData32 () from /usr/lib/64/libX11.so.4
> (gdb) bt
> #0  0xffffffff7e85bbd8 in _XData32 () from /usr/lib/64/libX11.so.4
> #1  0xffffffff7e82cad4 in XChangeProperty () from
> /usr/lib/64/libX11.so.4
> #2  0x0000000100032bc0 in set_gnome_proxy ()
> #3  0x0000000100010750 in main ()
>
> I notice there are two patches that fix this problem:
>
> 108376-42
> 108652-66
>
> However, 108376-42 contains this note:
>
>> (from 108376-33)
>> 4499671 Bus Error in _XData32 when running a 64-bit application (back
>> out the
>> fix)
>>
>> (from 108376-32)
>> 4499671 Bus Error in _XData32 when running a 64-bit application
>
> and 108652-66 mentions:
>
>> 4499671 Bus Error in _XData32 when running a 64-bit application
>
> which is apparently rolled into the patch from revisions 43 and 44 of
> the same patch.
>
> However, when attempting to apply patch 108652, I get this:
>
> Checking installed patches...
> One or more patch packages included in
> 108652-66 are not installed on this system.
>
> I get the same error from 108376. Going over their manifest, I see:
>
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwacx
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwdxm
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwfa
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwfnt
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwice
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwicx
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwinc
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwman
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwplt
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwplx
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwpmn
> drwxr-xr-x   4 alex     other        512 Mar  5 01:39 SUNWxwslb
>
> All of these packages are installed on my machine (Solaris 9 12/02,
> Ultra 30), and pkginfo shows me information for all of them. The files
> on my machine ( /usr/openwin/lib/libX11.so.4
> /usr/openwin/lib/sparcv9/libX11.so.4 ) seem to be newer than the ones
> included in the patch.
>
> Should I just assume that this bug is still outstanding (it has been
> lurking since Solaris 7 it would seem), and wait for another patch? If
> that is the case, is there somebody I can harass at Sun? For what it is
> worth, this should be easily reproducible with the afterstep source and
> gcc 3.2.2 with CC=gcc and CFLAGS="-m64 -mcpu=v9".
>
> I am aware of SFWastep being available on the companion CD. However,
> that afterstep, for whatever reason, is not compiled with xinerama
> support (although it is enabled by default in the distribution, go
> figure). A window manager that works but only on one screen isn't of
> particular use to me.
>
> I'll gladly summarize here. Thanks in advance.
>
> Alex
>
> sources:
> afterstep: ftp://ftp.afterstep.org/stable/AfterStep-1.8.11.tar.gz
> 108376:
> http://sunsolve.sun.com/pub-cgi/
> retrieve.pl?doc=fpatches%2F108376&zone_32=_XData32
> 108652:
> http://sunsolve.sun.com/pub-cgi/
> retrieve.pl?doc=fpatches%2F108652&zone_32=_XData32
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers