如何修复Valgrind日志中的内存泄漏

Question

This is the first time i have seen a valgrind log , this log is provided by other users (i can't run valgrind , since it's not supported stlinux (sh4)

for learning how could fix memory leak with valgrind log , would highly appreciated if any expert , choose even one line of bellow's valgrind error log (and tell us , the valgrind error line he intends for fix and post their fix (patch of source code)

then , would learn how we could fix memory leak from valgrind log , and i would do all rest fixing myself (i just need a example fix)

here is the app trac browser (source code) : http://www.streamboard.tv/oscam/browser/trunk/?rev=5375

would highly appreciate if any expert would take a look at these log (i really need help) , here are 4 valgrind logs :

http://www.4shared.com/office/04seUumN/valgrind_ 2 .html

http://www.4shared.com/office/WYmfxICb/valgrindlog.html

http://www.4shared.com/office/WGwlKeUK/valgrind.html

http://www.4shared.com/office/mkX4FAzd/valgrind_ 1 .html

PS :

1. since i can't run valgrind myself , please choose a memory leak with largest memory loss (leaks) , then , i could monitor those improvement with my eyes (top command , ps -aux)

2. my concern for fixing memory leaks , are mostly limited to these modules (for me the priority (importance) is in their number sequence ( i mean , fixing a memory leak with module-datastruct-list.c is in top priority for me , ....)

   • module-datastruct-list.c
   • oscam-garbage.c
   • oscam.c
   • module-cccam.c
   • module-dvbapi.c

Thanks in advance

Answer 1

I used Valgrind a long time ago so I don't remember well how to use it, but I know it is fairly easy. Valgrind logs show every memory leak of the program, due to your code, but also due to code you call from libraries. For every leak, the function call stack is displayed, like this :

==5313== Invalid write of size 4
==5313==    at 0x8048A27: test_2() (valgrind-tests.cc:37)
==5313==    by 0x8048CDF: main (valgrind-tests.cc:134)
==5313==    by 0x215BBE: __libc_start_main (in /lib/libc-2.3.2.so)
==5313==    by 0x8048910: (within /home/newren/examples/valgrind-tests)
==5313==    Address 0x1B3E024 is 0 bytes inside a block of size 4 free'd
==5313==    at 0x5419C5: __builtin_delete (vg_replace_malloc.c:244)
==5313==    by 0x5419E3: operator delete(void*) (vg_replace_malloc.c:253)
==5313==    by 0x8048A20: test_2() (valgrind-tests.cc:36)
==5313==    by 0x8048CDF: main (valgrind-tests.cc:134)
==5313==    by 0x215BBE: __libc_start_main (in /lib/libc-2.3.2.so)
==5313==    by 0x8048910: (within /home/newren/examples/valgrind-tests)

This means you try to write 4 bytes in memory, but you don't have access to theses bytes. This issue is located at line 37 of "valgrind-tests.cc" in this example.

The main problem of valgrind is that, as I said earlier, it displays memory leaks or memory warnings (like forgotten pointers) from libraries you use. To clean your logs, you can write valgrind rule files called " suppression files ".

More informations here and here .