XML解析树内存泄漏及大文件内存不足问题求助（附Valgrind日志）

我编写了一款将XML解析为树结构的XMLParser，功能运行正常，但存在内存占用过高的问题——解析4GB大小的XML文件时会因内存不足无法执行。当前的处理流程是先将XML文件转换为char*，再进行解析，并通过malloc为每个Node结构体分配内存。我不清楚何时需要调用free()释放这些节点，甚至不确定是否需要释放，担心释放操作会破坏树结构，同时还收到了Valgrind的内存泄漏报错，具体日志如下：

==3002== HEAP SUMMARY:
==3002==     in use at exit: 8,995 bytes in 423 blocks
==3002==   total heap usage: 427 allocs, 137 frees, 24,691 bytes allocated
==3002== 
==3002== 19 bytes in 1 blocks are indirectly lost in loss record 1 of 8
==3002==    at 0x4848899: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==3002==    by 0x491460E: strdup (strdup.c:42)
==3002==    by 0x109B9D: parse (main.c:223)
==3002==    by 0x10A06B: main (main.c:333)
==3002== 
==3002== 48 bytes in 1 blocks are indirectly lost in loss record 2 of 8
==3002==    at 0x484DA83: calloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==3002==    by 0x10934B: new_node (main.c:9)
==3002==    by 0x10A054: main (main.c:332)
==3002== 
==3002== 206 bytes in 32 blocks are definitely lost in loss record 3 of 8
==3002==    at 0x4848899: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==3002==    by 0x491460E: strdup (strdup.c:42)
==3002==    by 0x1096E9: parse (main.c:108)
==3002==    by 0x10A06B: main (main.c:333)
==3002== 
==3002== 551 bytes in 101 blocks are indirectly lost in loss record 4 of 8
==3002==    at 0x4848899: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==3002==    by 0x491460E: strdup (strdup.c:42)
==3002==    by 0x1096E9: parse (main.c:108)
==3002==    by 0x10A06B: main (main.c:333)
==3002== 
==3002== 608 bytes in 19 blocks are indirectly lost in loss record 5 of 8
==3002==    at 0x484DA83: calloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==3002==    by 0x109648: add_attribute (main.c:85)
==3002==    by 0x109DC4: parse (main.c:275)
==3002==    by 0x10A06B: main (main.c:333)
==3002== 
==3002== 1,083 bytes in 134 blocks are indirectly lost in loss record 6 of 8
==3002==    at 0x4848899: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==3002==    by 0x491460E: strdup (strdup.c:42)
==3002==    by 0x109D13: parse (main.c:255)
==3002==    by 0x10A06B: main (main.c:333)
==3002== 
==3002== 6,432 bytes in 134 blocks are indirectly lost in loss record 7 of 8
==3002==    at 0x484DA83: calloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==3002==    by 0x10934B: new_node (main.c:9)
==3002==    by 0x109C3A: parse (main.c:238)
==3002==    by 0x10A06B: main (main.c:333)
==3002== 
==3002== 8,789 (48 direct, 8,741 indirect) bytes in 1 blocks are definitely lost in loss record 8 of 8
==3002==    at 0x484DA83: calloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==3002==    by 0x10934B: new_node (main.c:9)
==3002==    by 0x109B7E: parse (main.c:221)
==3002==    by 0x10A06B: main (main.c:333)
==3002== 
==3002== LEAK SUMMARY:
==3002==    definitely lost: 254 bytes in 33 blocks
==3002==    indirectly lost: 8,741 bytes in 390 blocks
==3002==      possibly lost: 0 bytes in 0 blocks
==3002==    still reachable: 0 bytes in 0 blocks
==3002==         suppressed: 0 bytes in 0 blocks
==3002== 
==3002== For lists of detected and suppressed errors, rerun with: -s

想知道问题的根源，是因为分配了不必要的内存，还是内存释放时机错误？

内容的提问来源于stack exchange，提问作者Ali