Very well put, Antonio Sir.
On 2019-05-27 at 8:10 AM, Antonio Scuri <[email protected]> wrote: The only possibility for a leak there if there is an infinite loop in the "for". The "for" is ended only when "return" is called right after free. If you can show me an actual situation where there is a leak, then we will try to fix it. But adding a free after the "for" is not a fix. Best, Scuri Em seg, 27 de mai de 2019 às 08:42, Ranier VF <[email protected]> escreveu: Hi Scuri, Acccording to Coverity this is a real bug: static void iAttribParse(Ihandle *ih, const char* str) 1463{ 1464 char env_buffer[256]; 1465 char* name=NULL; 1466 char* value=NULL; 1467 char state = 'a'; /* get attribute */ 1468 int end = 0; 1469 1470 env_str = str; 1471 1. Condition true, taking true branch. 7. Condition true, taking true branch. 15. Condition true, taking true branch. 1472 for (;;) 1473 { 2. Switch case value 8. 8. Switch case value 5. 16. Switch case value 5. 1474 switch (iAttribToken(env_buffer)) 1475 { CID 210525: Missing break in switch (MISSING_BREAK) [select issue] 1476 case IUPLEX_TK_END: /* same as IUPLEX_TK_COMMA */ 1477 end = 1; 1478 case IUPLEX_TK_COMMA: 3. Condition name, taking false branch. 1479 if (name) 1480 { CID 210685: Explicit null dereferenced (FORWARD_NULL) [select issue] 1481 IupStoreAttribute(ih, name, value); 1482 free(name); 1483 } 4. Condition end, taking false branch. 1484 if (end) 1485 return; 1486 name = value = NULL; 1487 state = 'a'; 5. Breaking from switch. 1488 break; 1489 1490 case IUPLEX_TK_SET: 1491 state = 'v'; /* get value */ 1492 break; 1493 1494 case IUPLEX_TK_NAME: 9. Condition state == 97, taking true branch. 17. Condition state == 97, taking true branch. 1495 if (state == 'a') 10. alloc_fn: Storage is returned from allocation function iupStrDup. [show details] 11. var_assign: Assigning: name = storage returned from iupStrDup(env_buffer). 12. Falling through to end of if statement. CID 210676 (#1 of 1): Resource leak (RESOURCE_LEAK)18. overwrite_var: Overwriting name in name = iupStrDup(env_buffer) leaks the storage that name points to. 1496 name = iupStrDup(env_buffer); 1497 else 1498 value = env_buffer; 13. Breaking from switch. 1499 break; 1500 } 6. Jumping back to the beginning of the loop. 14. Jumping back to the beginning of the loop. 1501 } 1502} Following the "conditions", that can actually occur, var name leaks pointer. IHMO I really believe fix this. Best regards, Ranier Vilela ________________________________________ De: Antonio Scuri <[email protected]> Enviado: domingo, 26 de maio de 2019 22:08 Para: IUP discussion list. Assunto: Re: [Iup-users] CID 210676 (#1 of 1): Resource leak (RESOURCE_LEAK) The code you added is an unreachable code. The "for" never ends normally, only in the return. Em sáb, 25 de mai de 2019 às 08:28, Ranier VF <[email protected]<mailto:[email protected]>> escreveu: Hi, Fix RESOURCE_LEAK at iup_attrib.c --- ..\..\a\src\iup_attrib.c Sat Jan 26 16:02:13 2019 +++ iup_attrib.c Sat May 25 08:25:17 2019 @@ -1499,6 +1499,11 @@ break; } } + if (name) + { + IupStoreAttribute(ih, name, value); + free(name); + } } Ihandle* IupSetAttributes(Ihandle *ih, const char* str) Best regards. Ranier Vilela_______________________________________________ Iup-users mailing list [email protected]<mailto:[email protected]> https://lists.sourceforge.net/lists/listinfo/iup-users _______________________________________________ Iup-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/iup-users
_______________________________________________ Iup-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/iup-users
