Re: [line start cache] Re: XEmacs 21.1.4 crash (relatively often) on Solaris 2.7 sparc

The answer is: yes, I can. I applied this patch to my copy of 21.1. I started xemacs -vanilla. When I attempted to edit the latest tkman Makefile, KA-BOOM! Here's the gdb backtrace. Lemme know if you want to see anything else. Fatal error: assertion failed, file /usr/local/src/xemacs-21.1-19990826/src/redisplay.c, line 6257, 0 Fatal error (6). #0 0x40246811 in kill () from /lib/libc.so.6 #1 0x80a3704 in fatal_error_signal () #2 0xbfffefd4 in ?? () #3 0x4024784f in abort () at ../sysdeps/generic/abort.c:83 #4 0x80a3649 in assert_failed () #5 0x8122165 in update_internal_cache_list () #6 0x8123086 in update_line_start_cache () #7 0x812108c in redisplay_window () #8 0x81211cb in redisplay_frame () #9 0x812131c in redisplay_device () #10 0x8121539 in redisplay_without_hooks () #11 0x8125fa2 in redisplay () #12 0x80c75ba in Fnext_event () #13 0x808df1e in Fcommand_loop_1 () #14 0x808e155 in command_loop_1 () #15 0x80ab276 in condition_case_1 () #16 0x808e2b7 in command_loop_2 () #17 0x80ab144 in internal_catch () #18 0x808dc2e in initial_command_loop () #19 0x80a2154 in sort_args () #20 0x80a3923 in main () Here's my installation info: uname -a: Linux mithril 2.2.11 #1 SMP Thu Aug 12 00:21:53 EDT 1999 i686 unknown ../../../src/xemacs-21.1.6/configure '--prefix=/usr/local/xemacs-21.1.6' '--with-gcc' '--site-includes=/usr/local/include' '--site-libraries=/usr/local/lib' '--site-runtime-libraries=/usr/local/lib' '--infopath=/usr/local/info' '--with-mule=no' '--compiler=/usr/local/egcs-snapshot/bin/gcc' '--cflags=-O3 -malign-double -pipe -march=pentiumpro -mcpu=pentiumpro -ffast-math -fno-exceptions' '--with-dialogs=no' '--package-path=/usr/local/xemacs-packages' That's gcc-2.95.1, BTW. - vin Jan> P.S. Can you trigger the abort in this patch? Jan> Ah, %$#＠%!#$. There is one ! too many in this patch. Jan> Here is what I meant Jan> Index: redisplay.c Jan> =================================================================== Jan> RCS file: /usr/CVSroot/XEmacs/xemacs-20/src/redisplay.c,v Jan> retrieving revision 1.57 Jan> diff -u -u -r1.57 redisplay.c Jan> --- redisplay.c 1999/05/12 14:29:13 1.57 Jan> +++ redisplay.c 1999/08/12 19:37:53 Jan> ＠＠ -6245,6 +6245,19 ＠＠ Jan> lsc.end = dl->end_bufpos; Jan> lsc.height = dl->ascent + dl->descent; Jan> + { Jan> + struct buffer *buf = XBUFFER(w->buffer); Jan> + /* Paranaoid line start cache check */ Jan> + if (! (lsc.start <= BUF_BEGV(buf) || Jan> + BUF_FETCH_CHAR(buf,lsc.start - 1) == (Emchar) '\n')) Jan> + { Jan> + fprintf(stderr,"Wrong stuff in line start cache!\n" Jan> + "Supposedly begins at: %10s", Jan> + (char *)BUF_BYTE_ADDRESS(buf,lsc.start)); Jan> + assert(0); Jan> + } Jan> + } Jan> + Jan> Dynarr_add (internal_cache, lsc); Jan> } Jan> } Jan> However there is no need to run that. For me it crashes displaying the Jan> splash screen. The reason: a wrapped line. Jan> (Sorry if this gets technical) Jan> The problem is that \ Jan> when a line is wrap \ Jan> ped like this Jan> Then there are 3 display lines and the second 2 don't actually start at Jan> a new line in the buffer, just in the display. Jan> Previous line_start_cache rediplay crashes have all been of the type Jan> where code actually assumed that the line_start_cache actually contained lines Jan> starting at "\n", and we have fixed the triggers for that. Jan> It seems now that we better check these assumptions. Jan> i'm too tired now to have a real look. Gunnar Do you remember where Jan> that problem occured, was it also at asssert(cur_elt>=0)? Jan> Jan Jan> BTW Hamish, does the reproduction recipe at the end of Jan> http://www.xemacs.org/list-archives/xemacs-beta/9903/msg00475.html Jan> "work" for you?