[tarantool-patches] [PATCH] sql: fix lemon memory leak reported by ASAN

[tarantool-patches] [PATCH] sql: fix lemon memory leak reported by ASAN

[Alexander Turenko]

It catched by ASAN at build time (lemon is executed to generate
parse.[ch]), so tarantool couldn't be built with -DENABLE_ASAN=ON.
---

no issue
https://github.com/tarantool/tarantool/tree/Totktonada/fix-lemon-memleak

 extra/lemon.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/extra/lemon.c b/extra/lemon.c
index 66e425f68..1efaac9e6 100644
--- a/extra/lemon.c
+++ b/extra/lemon.c
@@ -598,8 +598,11 @@ struct acttab {
 
 /* Free all memory associated with the given acttab */
 void acttab_free(acttab *p){
-  free( p->aAction );
-  free( p->aLookahead );
+  assert(p);
+  if (p->aAction)
+    free( p->aAction );
+  if (p->aLookahead)
+    free( p->aLookahead );
   free( p );
 }
 
@@ -4255,6 +4258,7 @@ void ReportTable(
     }
   }
   fprintf(out, "};\n"); lineno++;
+  acttab_free(pActtab);
 
   /* Output the yy_shift_ofst[] table */
   n = lemp->nxstate;
-- 
2.20.1

[tarantool-patches] Re: [PATCH] sql: fix lemon memory leak reported by ASAN

[Kirill Yukhin]

Hello,
On 05 Jan 16:13, Alexander Turenko wrote:
> It catched by ASAN at build time (lemon is executed to generate
> parse.[ch]), so tarantool couldn't be built with -DENABLE_ASAN=ON.
> ---
> 
> no issue
> https://github.com/tarantool/tarantool/tree/Totktonada/fix-lemon-memleak
I've pushed your patch into 2.1 branch.

Should we propose a patch to SQLite?

--
Regards, Kirill Yukhin

[tarantool-patches] Re: [PATCH] sql: fix lemon memory leak reported by ASAN

[Alexander Turenko]

On Wed, Jan 09, 2019 at 10:20:15AM +0300, Kirill Yukhin wrote:
> Hello,
> On 05 Jan 16:13, Alexander Turenko wrote:
> > It catched by ASAN at build time (lemon is executed to generate
> > parse.[ch]), so tarantool couldn't be built with -DENABLE_ASAN=ON.
> > ---
> > 
> > no issue
> > https://github.com/tarantool/tarantool/tree/Totktonada/fix-lemon-memleak
> I've pushed your patch into 2.1 branch.
> 
> Should we propose a patch to SQLite?

It is not so important unless you are using ASAN and I think in SQLite
it only run in the maintainer mode.

SQLite requires to sign-off dedication document before contribute. The
patch I think is not enough reason to do that.

WBR, Alexander Turenko.

[tarantool-patches] Re: [PATCH] sql: fix lemon memory leak reported by ASAN

[n.pettik]

> On 9 Jan 2019, at 10:57, Alexander Turenko <alexander.turenko@tarantool.org> wrote:
> 
> On Wed, Jan 09, 2019 at 10:20:15AM +0300, Kirill Yukhin wrote:
>> Hello,
>> On 05 Jan 16:13, Alexander Turenko wrote:
>>> It catched by ASAN at build time (lemon is executed to generate
>>> parse.[ch]), so tarantool couldn't be built with -DENABLE_ASAN=ON.
>>> ---
>>> 
>>> no issue
>>> https://github.com/tarantool/tarantool/tree/Totktonada/fix-lemon-memleak
>> I've pushed your patch into 2.1 branch.
>> 
>> Should we propose a patch to SQLite?
> 
> It is not so important unless you are using ASAN and I think in SQLite
> it only run in the maintainer mode.
> 
> SQLite requires to sign-off dedication document before contribute. The
> patch I think is not enough reason to do that.

You can simply write to sqlite dev list about it:
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-dev