Re: Segmentation fault on 2.0.5
On 13.08.2019 17:45, Maria Jan Matejka wrote:
define myvar = false;
function something () {
if myvar then {
return false;
}
return true;
}
protocol device DEVICE {
}
Could this be because of the new filter code or something else?
Hello
Thanks for bugreport. Yes, it is the new filter code, most likely
it is a bug in handling of statically eliminable (constant-false)
if statements without else branch.
Similar issue also happens for empty blocks:
if true then { }
What about the attached patch?
Maria
Hi,
I can confirm that the patch works for my setup.
Thank you,
Radu
Re: Segmentation fault on 2.0.5
>> define myvar = false;
>>
>> function something () {
>> if myvar then {
>> return false;
>> }
>>
>> return true;
>> }
>>
>> protocol device DEVICE {
>>
>> }
>>
>> Could this be because of the new filter code or something else?
>
> Hello
>
> Thanks for bugreport. Yes, it is the new filter code, most likely
> it is a bug in handling of statically eliminable (constant-false)
> if statements without else branch.
>
> Similar issue also happens for empty blocks:
>
> if true then { }
>
What about the attached patch?
Maria
diff --git a/filter/config.Y b/filter/config.Y
index a67a72a8..340053ba 100644
--- a/filter/config.Y
+++ b/filter/config.Y
@@ -446,7 +446,7 @@ CF_KEYWORDS(FUNCTION, PRINT, PRINTN, UNSET, RETURN,
%nonassoc THEN
%nonassoc ELSE
-%type cmds_int
+%type cmds_int cmd_prep
%type term block cmd cmds constant constructor print_list var_list function_call symbol_value bgp_path_expr bgp_path bgp_path_tail
%type dynamic_attr
%type static_attr
@@ -624,17 +624,25 @@ cmds: /* EMPTY */ { $$ = NULL; }
| cmds_int { $$ = $1.begin; }
;
-cmds_int: cmd {
+cmd_prep: cmd {
$$.begin = $$.end = $1;
- while ($$.end->next)
-$$.end = $$.end->next;
+ if ($1)
+while ($$.end->next)
+ $$.end = $$.end->next;
+}
+ ;
+
+cmds_int: cmd_prep
+ | cmds_int cmd_prep {
+ if (!$1.begin)
+$$ = $2;
+ else if (!$2.begin)
+$$ = $1;
+ else {
+$$.begin = $1.begin;
+$$.end = $2.end;
+$1.end->next = $2.begin;
}
- | cmds_int cmd {
- $$.begin = $1.begin;
- $1.end->next = $2;
- $$.end = $2;
- while ($$.end->next)
-$$.end = $$.end->next;
}
;
diff --git a/filter/test.conf b/filter/test.conf
index 09a4a88a..b1342819 100644
--- a/filter/test.conf
+++ b/filter/test.conf
@@ -1178,6 +1178,10 @@ bt_test_suite(t_include, "Testing including another config file");
function t_if_else()
int i;
{
+ /* Empty blocks regression test */
+ if true then {}
+ else {}
+
if true then
bt_assert(true);
@@ -1187,6 +1191,10 @@ int i;
bt_assert(true);
else
bt_assert(false);
+
+ /* Empty blocks regression test */
+ if true then {}
+ else {}
}
bt_test_suite(t_if_else, "Testing if-else statement");
smime.p7s
Description: S/MIME Cryptographic Signature
Re: Segmentation fault on 2.0.5
On Sat, Aug 10, 2019 at 02:22:18PM +0300, Radu Anghel wrote:
> Hello,
>
> I recently upgraded BIRD from 2.0.4 to 2.0.5 and with the same (pretty
> large) .conf I got Segmentation fault.
>
> Until today I didn't have time to debug it, but here it goes:
>
> With this simple .conf I get seg fault on 2.0.5 while 2.0.4 works as
> expected. (if I define myvar = true it doesn't crash on 2.0.5)
>
> log syslog all;
>
> define myvar = false;
>
> function something () {
> if myvar then {
> return false;
> }
>
> return true;
> }
>
> protocol device DEVICE {
>
> }
>
> Could this be because of the new filter code or something else?
Hello
Thanks for bugreport. Yes, it is the new filter code, most likely
it is a bug in handling of statically eliminable (constant-false)
if statements without else branch.
Similar issue also happens for empty blocks:
if true then { }
--
Elen sila lumenn' omentielvo
Ondrej 'Santiago' Zajicek (email: santi...@crfreenet.org)
OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net)
"To err is human -- to blame it on a computer is even more so."
Segmentation fault on 2.0.5
Hello,
I recently upgraded BIRD from 2.0.4 to 2.0.5 and with the same (pretty
large) .conf I got Segmentation fault.
Until today I didn't have time to debug it, but here it goes:
With this simple .conf I get seg fault on 2.0.5 while 2.0.4 works as
expected. (if I define myvar = true it doesn't crash on 2.0.5)
log syslog all;
define myvar = false;
function something () {
if myvar then {
return false;
}
return true;
}
protocol device DEVICE {
}
Could this be because of the new filter code or something else?
Radu
