Sdb unit test for sorting big lists

[GovanifY]

For now this unit test triggers a crash, related to radareorg/radare2#7079

[radare]

test_sdb.c:54:16: warning: incompatible integer to pointer conversion passing 'int' to parameter of type 'const char *' [-Wint-conversion]
                sdb_set (db, i-1, i, 0);
                             ^~~
../../src/sdb.h:146:31: note: passing argument to parameter 'key' here
int sdb_set(Sdb*, const char *key, const char *data, ut32 cas);
                              ^
test_sdb.c:54:21: warning: incompatible integer to pointer conversion passing 'int' to parameter of type 'const char *' [-Wint-conversion]
                sdb_set (db, i-1, i, 0);
                                  ^
../../src/sdb.h:146:48: note: passing argument to parameter 'data' here
int sdb_set(Sdb*, const char *key, const char *data, ut32 cas);
                                               ^
2 warnings generated.

[radare]

thats why it segfaults :P

[radare]

fixed your test case and its not crashing at all this is not a valid test for your issue

[radare]

also you are not checking if the resulting list is sorted or not

[GovanifY]

I thougt we didn't had in that case as long as it didn't crashed. Argh back to reproducing then, sorry >_>

[radare]

i am sorting 10 million items in this SdbList without any stack exhaustion or crash problem

[radare]

diff --git a/test/unit/test_sdb.c b/test/unit/test_sdb.c
index 95f29f6..4d76430 100644
--- a/test/unit/test_sdb.c
+++ b/test/unit/test_sdb.c
@@ -50,11 +50,12 @@ static int __cmp_asc(const void *a, const void *b) {
 bool test_sdb_list_big(void) {
        Sdb *db = sdb_new (NULL, NULL, false);
        int i;
-       for ( i=1; i < 100000; i++) {
-               sdb_set (db, i-1, i, 0);
+       for ( i=0; i < 10000; i++) {
+               sdb_num_set (db, sdb_fmt (0, "%d", i), i + 1, 0);
        }
        SdbList *list = sdb_foreach_list (db, true);
-       ls_sort(list, __cmp_asc);
+       ls_sort (list, __cmp_asc);
+// TODO: verify if its sorted
        sdb_free (db);
        mu_end;
 }

[GovanifY]

Patch applied, any idea on how to reproduce the issue then?

[GovanifY]

Also confirm it works here too even though that was to be expected: test_sdb_list_big OK

[GovanifY]

Got a segfault at 10M here, will try to reproduce at lower count and add here. Tooked a lot of time tho

[radare]

10 million items. The hash is a table of lists, maybe some stats to see how many items are in each bucket may help to identify the issue

…

On 25 Mar 2017, at 22:30, GovanifY ***@***.***> wrote: Got a segfault at 10M here, will try to reproduce at lower count and add here. Tooked a lot of time tho — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

[radare]

But maybe we have reached a limit in mmap

…

On 25 Mar 2017, at 22:30, GovanifY ***@***.***> wrote: Got a segfault at 10M here, will try to reproduce at lower count and add here. Tooked a lot of time tho — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

[GovanifY]

Test is pretty slow in its current form but works™

[GovanifY]

There is a chance, for mmap, I'll check that later, thanks for pointing out

[alvarofe]

Here you are sorting twice since sdb_foreach_list with the true flag already does the same operation as below

[alvarofe]

Working on this btw

[alvarofe]

Fixed in master. I lowered the amount to sort to 500000 just to save time when testing.

Test again please and tell me.