iasecc: Avoid memory leaks

[popovec]

Thanks oss-fuzz

https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=65562

[popovec]

When I was looking at the code of the iasecc_delete_file() function, I found one thing, after deleting the file using SM, it does not update card->cache.current_ef. Can someone who knows details about the "iaescc" code confirm whether the following patch should be applied?

index b20123d65..7ec1fd575 100644
--- a/src/libopensc/card-iasecc.c
+++ b/src/libopensc/card-iasecc.c
@@ -1573,13 +1573,11 @@ iasecc_delete_file(struct sc_card *card, const struct sc_path *path)
                rv = sc_transmit_apdu(card, &apdu);
                LOG_TEST_RET(ctx, rv, "APDU transmit failed");
                rv = sc_check_sw(card, apdu.sw1, apdu.sw2);
-               LOG_TEST_RET(ctx, rv, "Delete file failed");
-
-               if (card->cache.valid) {
-                       sc_file_free(card->cache.current_ef);
-               }
-               card->cache.current_ef = NULL;
        }
+       LOG_TEST_RET(ctx, rv, "Delete file failed");
+       if (card->cache.valid)
+               sc_file_free(card->cache.current_ef);
+       card->cache.current_ef = NULL;
 
        LOG_FUNC_RETURN(ctx, rv);
 }

[Jakuje]

When I was looking at the code of the iasecc_delete_file() function, I found one thing, after deleting the file using SM, it does not update card->cache.current_ef. Can someone who knows details about the "iaescc" code confirm whether the following patch should be applied?

I would say yes, but I am not that much familiar with the iasecc. Maybe @vjardin ?

[vjardin]

I would say yes, but I am not that much familiar with the iasecc. Maybe @vjardin ?

I would need to run some tests with some IASECC cards, but I do not have time for some short terms ; can it wait a bit ?

[popovec]

Feel free to find time to test it, it's not urgent. It can be fixed in another PR.

[frankmorgner]

some of the functionality around current_ef/df has been removed in the last releases, because it was breaking use cases with concurrent access to the card. Also, speedup was very limited with respect to the complexity added. In the iasecc driver the handling of current_ef/df has not been removed, because it is deeply woven into the driver. Feel free to simplify that...

[Jakuje]

Thank you! Merging to move on with fuzzing. @popovec Please move the discussion regarding to the delete file either to new issue or draft PR so it wont get lost.