On 20/05/2013 12:37, Bram Moolenaar wrote:
Charles Peacech wrote:
On Mon, May 20, 2013 at 5:28 PM, Charles <[email protected]> wrote:
Hi,
In my gvim, the new regexp engine crash gvim for this regexp
0x02a60c72 "htmlSpecialChar "&#\=[0-9A-Za-z]\{1,8};""
The crash happens here
/*
* Allocate and initialize nfa_state_T.
*/
static nfa_state_T *
new_state(c, out, out1)
int c;
nfa_state_T *out;
nfa_state_T *out1;
{
nfa_state_T *s;
if (istate >= nstate)
return NULL;
s = &state_ptr[istate++];
s->c = c;
s->out = out;
s->out1 = out1; // <----- Access violation here, probably because
s points to foreign memory
s->id = istate;
s->lastlist = 0;
s->lastthread = NULL;
s->visits = 0;
s->negated = FALSE;
return s;
}
It seems to be that the cause is insufficient initial size here
line 232:
/* A reasonable estimation for size */
nstate_max = (STRLEN(expr) + 1) * NFA_POSTFIX_MULTIPLIER;
When it crashed, it's trying to access member no. 714 while the
currently allocated array is only 631.
Yes, increasing the size works around the problem.
However, there is a check for the pointer not to go beyond the end:
#define EMIT(c) do { \
if (post_ptr >= post_end) \
return FAIL; \
*post_ptr++ = c; \
} while (0)
For some reason that is not working. Ah, it's adding the byte size to
the int pointer, that's wrong. Patch coming up...
It is non-portable code. It is not defined know when the post increment
is done, it could be before or after the assignment of c. I assume the
intent is that the post increment is done after the assignment so it
should be broken into two as follows:
*post_ptr = c; \
post_ptr++; \
TTFN
Mike
--
Make your mark in this world or at least spray in each corner.
--
--
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php
---
You received this message because you are subscribed to the Google Groups "vim_dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
