Re: lynx-dev Possible patch for lynx2.8.5dev.16

[Thomas Dickey]

On Sat, 20 Dec 2003, Doug Kaufman wrote:

> Call frame traceback EIPs:
>   0x002b498f free+119, file malloc.c
>   0x000be669 HTAnchor_delete+137, file HTAnchor.c, line 699
>   0x000be5ab deleteLinks+75, file HTAnchor.c, line 587
>   0x000be7b1 HTAnchor_delete_links+65, file HTAnchor.c, line 735
>   0x00010fea HText_free+202, file GridText.c, line 1238
>   0x0001c9e0 HTuncache_current_document+96, file GridText.c, line 8636
>   0x00037eb6 handle_PREV_DOC+534, file LYMainLoop.c, line 4057
>   0x0003ad17 mainloop+3063, file LYMainLoop.c, line 7097
>   0x0002a3ab main+3739, file LYMain.c, line 2075
>   0x002b31a8 __crt1_startup+176, file crt1.c
>
>
> Does the following patch make sense as an attempt to fix this? I can't
> test, since I can't produce the crash again.

no - HTAnchor_delete() checks its parameter for null.   There are a few
free's in that function, but they're past that check.

>
> --- lynx2.8.5dev.16/lynx2-8-5/www/Library/Implementation/HTAnchor.c.ori       
> 2003-06-01 17:16:28.000000000 -0800
> +++ lynx2.8.5dev.16/lynx2-8-5/www/Library/Implementation/HTAnchor.c   
> 2003-12-20 18:27:34.000000000 -0800
> @@ -582,7 +582,7 @@
>        *  Test here to avoid calling overhead.
>        *  Don't delete if document is loaded or being loaded.
>        */
> -     if ((me->parent != parent) && !parent->underway &&
> +     if (parent && (me->parent != parent) && !parent->underway &&
>           (!parent->info || !parent->info->document)) {
>           HTAnchor_delete(parent);
>       }
>
> This lynx binary was built with patches 16a, 16b, 16c6, and 16c7a-iz.

I'll be home all of next week - would like to focus on lynx & complete
dev.17 (keep getting distracted by other more urgent bugs).

-- 
Thomas E. Dickey
http://invisible-island.net
ftp://invisible-island.net

; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden