On Tue, 7 Sep 2004, Al Danial wrote:

>On my 64 bit Linux system (Opteron, Gentoo 2004.1 amd_64), the v3.0.6
>code builds fine but seg faults during 'make test' in one of the blob tests.
>Today I built the code with debugging enabled and used gdb to get a
>traceback.  It failed at line 411 of src/util.c, in a strlen() call.
>
>    407   nByte = 0;
>    408   va_start(ap, pz);
>    409   while( (z = va_arg(ap, const char*))!=0 ){
>    410     n = va_arg(ap, int);
>    411     if( n<=0 ) n = strlen(z);
>    412     nByte += n;
>    413   }
>    414   va_end(ap);
>
>I added printf's to check the contents of z and the code then segfaulted
>in one of my print statements.  Weird.  If anyone has debug suggestions
>I'll try them.        -- Al


As a complete stab in the dark, this could be a compiler fault. Richard,
do the reports occur with GCC, do you know?

My initial thoughts upon seeing va_arg et. al. is that type promotion to
(64 bit) long is occuring for variadic functions, and then the functions
are accessing the data with the wrong alignment or something like that.

Go up a frame, and try examining the arguments in gdb. Then go back down
and compare the values.

>
>
>On Tue, 7 Sep 2004 11:39:21 -0700 (PDT), Louis P. Santillan
><[EMAIL PROTECTED]> wrote:
>> Do the rumors apply to 64-bit Linux or do they apply
>> elsewhere?  I can give you access to either my 64-bit
>> Fedora Core 2 desktop (self-built) or Fedora Core 2
>> laptop (eMachines M6805) and I have access to an Apple
>> G5 (unfortunately I can't give you access to it).  I
>> remember trying to compile 2.8/3.0 on Fedora and I
>> believe they failed.  Fedora seems to have a screwy
>> /usr/include setup (also breaks Mozilla JavaScript
>> engine, SpiderMonkey).
>>
>> By chance, the M6805 is kind of up for sale (only
>> about 4 months old; looking to replace it with an
>> Athlon64 laptop with an nVidia graphics chip).  I'd
>> make it cheap for you Richard, or anybody else, if
>> you're interested.
>>
>> Louis
>>
>>
>>
>>
>> --- "D. Richard Hipp" <[EMAIL PROTECTED]> wrote:
>>
>> > I keep hearing rumors that SQLite is not working
>> > right on
>> > 64-bit unix.  But lacking access to any 64-bit
>> > hardware,
>> > there isn't much I can do about it.
>> >
>> > Does anybody out there have a 64-bit unix box that
>> > they
>> > can give me ssh access to so that I can test and
>> > debug
>> > SQLite?  Access can be intermittant (in other words,
>> > I
>> > call you up to let you know that I need to log on)
>> > and
>> > can come from a fixed IP address.
>> > --
>> > D. Richard Hipp -- [EMAIL PROTECTED] -- 704.948.4565
>> >
>> >
>>
>>
>> _______________________________
>> Do you Yahoo!?
>> Win 1 of 4,000 free domain names from Yahoo! Enter now.
>> http://promotions.yahoo.com/goldrush
>>
>

-- 
    /"\
    \ /    ASCII RIBBON CAMPAIGN - AGAINST HTML MAIL
     X                           - AGAINST MS ATTACHMENTS
    / \

Reply via email to