Daniel Stenberg wrote:
> On Mon, 9 May 2005, Ben Greear wrote:
>
>> Any particular reason we would want to be using GnuTLS instead of
>> OpenSSL?
>
>
> I added the GnuTLS support mainly for two reasons.

[snip]

Sounds good to me.

I have a bug report for you, based on curl 7.13.2, though I think I've seen
similar bugs on older releases, I just haven't tracked it down so thoroughly.

These are very repeatable for me, and happen when I'm shutting down the application.

Unfortunately, the app itself is not open source, nor is it a stand-alone
thing that I could send you for more advanced debugging.

I will help however I can, and could even give you ssh access to my test machine
if you wish...

Here is the backtrace of the core:

#0 0x0818b41c in fputc ()
#1 0x0807cc1b in dprintf_formatf (data=0x8283f98, stream=0x818b3d4 <fputc>,
     format=0x81e437c " %% Total %% Received %% Xferd Average Speed Time
   Time Time Current\n", ' ' <repeats 33 times>, "Dload Upload Total Spent Left Speed\n", ap_save=0xbfffe9c8 "\020\t") at mprintf.c:645
#2 0x0807e0c0 in curl_mfprintf (whereto=0x1dfddec0, format=0x8283f20 "")
     at mprintf.c:1152
#3 0x08088fd9 in Curl_pgrsUpdate (conn=0x0) at progress.c:260
#4 0x08080a87 in Curl_readwrite (conn=0x826abd0, done=0xbfffebd6 "")
     at transfer.c:1359
#5 0x08082b8f in Curl_perform (data=0x82625e0) at transfer.c:1583
#6 0x0804bfcf in HL4Endp::doReadWriteRound (this=0x82557a8, now=@0xbfffec88)
     at l4Endp.cc:113
#7 0x0804df12 in HL4Endp::doTrafficRound (this=0x82557a8, now=@0xbfffec88)
     at l4Endp.cc:224
#8 0x08057c75 in L4HelperMgr::doTrafficRound () at l4Manager.cc:287
#9 0x080494ce in do_big_while () at l4helper.cc:293
#10 0x080488cc in main (argc=15, argv=0xbffff0c4) at l4helper.cc:172
(gdb)

in progress.c, line: 261, it seems that data->set.err is a bad pointer
to no-where (not null, just out of memory bounds).

Though the backtrace shows conn=0x0 in frame 3, I don't think it can
be, otherwise it would crash earlier in progress.c.

I looked at 'f' in frame 0, and it has not iterated a single time (ie, f == data).
So, it's not a string formatting error or walking off the end of a string.

Is there anything I need to do to initialize the data->set.err file?

Thanks,
Ben

-- 
Ben Greear <greearb_at_candelatech.com>
Candela Technologies Inc  http://www.candelatech.com