helgrind 中的 libxml2 多线程错误

Question

我有以下非常简单的程序：

#include <stdlib.h>
#include <stdio.h>
#include <pthread.h>
#include <libxml/parser.h>
#include <libxml/catalog.h>
#include <libxml/tree.h>
#include <libxml/HTMLparser.h>

static const int kHTMLParseFlags = 
  HTML_PARSE_NOBLANKS | HTML_PARSE_NOERROR | HTML_PARSE_NOWARNING | HTML_PARSE_NONET;

void* test1(void* ptr)
{
    htmlDocPtr doc = htmlReadFile("http://www.google.com", NULL, kHTMLParseFlags);
    xmlFreeDoc(doc);
}

void* test2(void* ptr)
{
    htmlDocPtr doc = htmlReadFile("http://www.lenta.ru", NULL, kHTMLParseFlags);
    xmlFreeDoc(doc);
}
int main(void)
{
    xmlInitParser();
    xmlInitializeCatalog();
    pthread_t thread1, thread2; 
    pthread_create(&thread1, NULL,  &test1, NULL);
    pthread_create(&thread2, NULL,  &test2, NULL);
    pthread_join(thread1, NULL);
    pthread_join(thread2, NULL);
    xmlCatalogCleanup();
    xmlCleanupParser();
    return EXIT_SUCCESS;
}

当我 运行 它在 valgrind --tool=helgrind 时它显示了很多竞争条件（ns_makecanon，等等）。这是什么原因？似乎 libxml2 是线程安全的。我正在用 gcc -I/usr/include/libxml2 temp.c -lxml2 -pthread.

编译它

来自 valgrind 输出的样本：

==2276== Possible data race during read of size 4 at 0x5EC2020 by thread #3
==2276== Locks held: none
==2276==    at 0x4EE5117: xmlCatalogXMLResolve (catalog.c:1637)
==2276==    by 0x4EE4F15: xmlCatalogListXMLResolve (catalog.c:2064)
==2276==    by 0x4EE5FBE: xmlACatalogResolve (catalog.c:2875)
==2276==    by 0x4EA21EA: xmlResolveResourceFromCatalog (xmlIO.c:4000)
==2276==    by 0x4EA46C3: xmlDefaultExternalEntityLoader (xmlIO.c:4065)
==2276==    by 0x4EA452E: xmlLoadExternalEntity (xmlIO.c:4133)
==2276==    by 0x4EB9D88: htmlCreateFileParserCtxt (HTMLparser.c:6314)
==2276==    by 0x4EBA97C: htmlReadFile (HTMLparser.c:6796)
==2276==    by 0x400A18: test2(void*) (in /home/a.out)
==2276==    by 0x4C32DF6: ??? (in /usr/lib/valgrind/vgpreload_helgrind-amd64-linux.so)
==2276==    by 0x51AC6A9: start_thread (pthread_create.c:333)
==2276== 
==2276== This conflicts with a previous write of size 4 by thread #2
==2276== Locks held: none
==2276==    at 0x4EE5131: xmlCatalogXMLResolve (catalog.c:1643)
==2276==    by 0x4EE4F15: xmlCatalogListXMLResolve (catalog.c:2064)
==2276==    by 0x4EE5FBE: xmlACatalogResolve (catalog.c:2875)
==2276==    by 0x4EA21EA: xmlResolveResourceFromCatalog (xmlIO.c:4000)
==2276==    by 0x4EA46C3: xmlDefaultExternalEntityLoader (xmlIO.c:4065)
==2276==    by 0x4EA452E: xmlLoadExternalEntity (xmlIO.c:4133)
==2276==    by 0x4EB9D88: htmlCreateFileParserCtxt (HTMLparser.c:6314)
==2276==    by 0x4EBA97C: htmlReadFile (HTMLparser.c:6796)
==2276==  Address 0x5ec2020 is 64 bytes inside a block of size 80 alloc'd
==2276==    at 0x4C2CFEF: malloc (in /usr/lib/valgrind/vgpreload_helgrind-amd64-linux.so)
==2276==    by 0x4EE2D3F: xmlNewCatalogEntry (catalog.c:280)
==2276==    by 0x4EE3340: xmlParseXMLCatalogOneNode (catalog.c:1187)
==2276==    by 0x4EE3871: xmlParseXMLCatalogNode (catalog.c:1256)
==2276==    by 0x4EE3871: xmlParseXMLCatalogNodeList (catalog.c:1323)
==2276==    by 0x4EE4C3E: xmlParseXMLCatalogFile (catalog.c:1388)
==2276==    by 0x4EE4C3E: xmlFetchXMLCatalogFile (catalog.c:1453)
==2276==    by 0x4EE4FEF: xmlCatalogListXMLResolve (catalog.c:2061)
==2276==    by 0x4EE5FBE: xmlACatalogResolve (catalog.c:2875)
==2276==    by 0x4EA21EA: xmlResolveResourceFromCatalog (xmlIO.c:4000)
==2276==    by 0x4EA46C3: xmlDefaultExternalEntityLoader (xmlIO.c:4065)
==2276==    by 0x4EA452E: xmlLoadExternalEntity (xmlIO.c:4133)
==2276==    by 0x4EB9D88: htmlCreateFileParserCtxt (HTMLparser.c:6314)
==2276==    by 0x4EBA97C: htmlReadFile (HTMLparser.c:6796)
==2276==  Block was alloc'd by thread #2

Answer 1

这确实是 libxml2 中的一个错误。 following code in xmlCatalogXMLResolve 不是线程安全的（应该使用原子增量或锁）：

if (catal->depth > MAX_CATAL_DEPTH) {
    xmlCatalogErr(catal, NULL, XML_CATALOG_RECURSION,
        "Detected recursion in catalog %s\n",
        catal->name, NULL, NULL);
    return(NULL);
}
catal->depth++;

libxml2 文档指出 concurrent loading should be thread-safe. I'd suggest you file a bug and/or raise the issue on the mailing list。