[patch] [PATCH] Fix memory leaks

Bernhard Walle bwalle at suse.de
Thu Sep 11 11:23:47 EDT 2008


This memory leaks appear in valigrind run:

    makedumpfile Completed.
    ==6024== 
    ==6024== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 2 from 1)
    ==6024== malloc/free: in use at exit: 8,838 bytes in 5 blocks.
    ==6024== malloc/free: 80 allocs, 75 frees, 192,858 bytes allocated.
    ==6024== For counts of detected errors, rerun with: -v
    ==6024== searching for pointers to 5 not-freed blocks.
    ==6024== checked 70,864 bytes.
    ==6024== 
    ==6024== 
    ==6024== 22 bytes in 1 blocks are definitely lost in loss record 2 of 5
    ==6024==    at 0x4C2560E: malloc (in /usr/lib64/valgrind/amd64-linux/vgpreload_memcheck.so)
    ==6024==    by 0x416955: initial (makedumpfile.c:3318)
    ==6024==    by 0x418954: create_dumpfile (makedumpfile.c:6000)
    ==6024==    by 0x4191AA: main (makedumpfile.c:6299)
    ==6024== 
    ==6024== 
    ==6024== 4,120 bytes in 1 blocks are definitely lost in loss record 4 of 5
    ==6024==    at 0x4C2560E: malloc (in /usr/lib64/valgrind/amd64-linux/vgpreload_memcheck.so)
    ==6024==    by 0x409F72: prepare_dump_bitmap (makedumpfile.c:4338)
    ==6024==    by 0x4187D8: create_dump_bitmap (makedumpfile.c:4361)
    ==6024==    by 0x41886C: create_dumpfile (makedumpfile.c:6005)
    ==6024==    by 0x4191AA: main (makedumpfile.c:6299)
    ==6024== 
    ==6024== 
    ==6024== 4,120 bytes in 1 blocks are definitely lost in loss record 5 of 5
    ==6024==    at 0x4C2560E: malloc (in /usr/lib64/valgrind/amd64-linux/vgpreload_memcheck.so)
    ==6024==    by 0x409F51: prepare_dump_bitmap (makedumpfile.c:4333)
    ==6024==    by 0x4187D8: create_dump_bitmap (makedumpfile.c:4361)
    ==6024==    by 0x41886C: create_dumpfile (makedumpfile.c:6005)
    ==6024==    by 0x4191AA: main (makedumpfile.c:6299)
    ==6024== 
    ==6024== LEAK SUMMARY:
    ==6024==    definitely lost: 8,262 bytes in 3 blocks.
    ==6024==      possibly lost: 0 bytes in 0 blocks.
    ==6024==    still reachable: 576 bytes in 2 blocks.
    ==6024==         suppressed: 0 bytes in 0 blocks.
    ==6024== Reachable blocks (those to which a pointer was found) are not shown.
    ==6024== To see them, rerun with: --leak-check=full --show-reachable=yes

Although it's not strictly necessary to free memory at the end of program run since
it gets freed automatically, it still makes sense to have a clean valgrind run to be
able to check the program for "real" memory leaks with valgrind.


Signed-off-by: Bernhard Walle <bwalle at suse.de>

---
 makedumpfile.c |   12 ++++++++++++
 1 file changed, 12 insertions(+)

--- a/makedumpfile.c
+++ b/makedumpfile.c
@@ -5268,6 +5268,14 @@ close_dump_bitmap(void)
 		ERRMSG("Can't close the bitmap file(%s). %s\n",
 		    info->name_bitmap, strerror(errno));
 	free(info->name_bitmap);
+
+	/* free 1st bitmap */
+	free(info->bitmap1);
+	info->bitmap1 = NULL;
+
+	/* free 2nd bitmap */
+	free(info->bitmap2);
+	info->bitmap2 = NULL;
 }
 
 void
@@ -5332,6 +5340,10 @@ close_files_for_creating_dumpfile(void)
 	else if (info->dump_level > DL_EXCLUDE_ZERO)
 		close_kernel_file();
 
+	/* free name for vmcoreinfo */
+	free(info->name_vmcoreinfo);
+	info->name_vmcoreinfo = NULL;
+
 	close_dump_memory();
 
 	close_dump_file();



More information about the kexec mailing list