Valgrind检测PostgreSQL时出现错误

3

我正在尝试使用PostgreSQL更新/更改名为xxx的表中的两列- comment 和 id。虽然编译器没有显示任何错误,但是Valgrind给出了重大错误。代码如下:

int main()
{
    PGconn *conn;
        PGresult *res;
    int i=0,nFields=0,row=0,col=0;

    conn = PQconnectdb("dbname=test1 host=localhost user=postgres password=yyy");
    if(PQstatus(conn) == CONNECTION_BAD) 
    {
            fprintf(stderr, "Connection to database \"%s\" failed.\n", PQerrorMessage(conn));
            fprintf(stderr, "%s", PQerrorMessage(conn));
            exit_nicely(conn);
    }

    res = PQexec(conn, "IF COL_LENGTH('xxx','comment') IS NULL");
    if(res)
        res = PQexec(conn, "ALTER TABLE xxx ADD comment VARCHAR(500)");
    else
        res = PQexec(conn, "UPDATE TABLE xxx ADD comment VARCHAR(500)");

    res = PQexec(conn, "IF COL_LENGTH('xxx','id') IS NULL");
    if(res)
        res = PQexec(conn, "ALTER TABLE xxx ADD id VARCHAR(50)");
    else
        res = PQexec(conn, "UPDATE TABLE xxx ADD id VARCHAR(50)");

    res = PQexec(conn, "SELECT * FROM xxx");
    if((!res) || (PQresultStatus(res) != PGRES_TUPLES_OK))
    {
        fprintf(stderr, "SELECT command did not return tuples properly\n");
        PQclear(res);
    }

    PQclear(res);
    PQfinish(conn);

    return 0;
}       

Valgrind给出以下错误:

$ valgrind --track-origins=yes --leak-check=full ./output xaa

(注:该命令用于检测内存泄漏和跟踪内存来源)
    ==4525== Memcheck, a memory error detector
    ==4525== Copyright (C) 2002-2010, and GNU GPL'd, by Julian Seward et al.
    ==4525== Using Valgrind-3.6.1 and LibVEX; rerun with -h for copyright info
    ==4525== Command: ./Gwidd_uniprot_map2 xaa
    ==4525== 
    --4525-- ./Gwidd_uniprot_map2:
    --4525-- dSYM directory has wrong UUID; consider using --dsymutil=yes
    ==4525== 
    ==4525== HEAP SUMMARY:
    ==4525==     in use at exit: 262,994 bytes in 751 blocks
    ==4525==   total heap usage: 1,012 allocs, 261 frees, 345,158 bytes allocated
    ==4525== 
    ==4525== 2,248 (200 direct, 2,048 indirect) bytes in 1 blocks are definitely lost in loss record 400 of 414
    ==4525==    at 0x100011345: malloc (vg_replace_malloc.c:236)
    ==4525==    by 0x10007B7E0: PQmakeEmptyPGresult (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10008DF30: pqGetErrorNotice3 (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10008CC00: pqParseInput3 (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007ECD9: parseInput (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007EE3F: PQgetResult (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007F6F1: PQexecFinish (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007F1DC: PQexec (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10000199A: main (in ./Gwidd_uniprot_map2)
    ==4525== 
    ==4525== 2,248 (200 direct, 2,048 indirect) bytes in 1 blocks are definitely lost in loss record 401 of 414
    ==4525==    at 0x100011345: malloc (vg_replace_malloc.c:236)
    ==4525==    by 0x10007B7E0: PQmakeEmptyPGresult (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10008DF30: pqGetErrorNotice3 (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10008CC00: pqParseInput3 (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007ECD9: parseInput (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007EE3F: PQgetResult (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007F6F1: PQexecFinish (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007F1DC: PQexec (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x1000019B5: main (in ./Gwidd_uniprot_map2)
    ==4525== 
    ==4525== 2,248 (200 direct, 2,048 indirect) bytes in 1 blocks are definitely lost in loss record 402 of 414
    ==4525==    at 0x100011345: malloc (vg_replace_malloc.c:236)
    ==4525==    by 0x10007B7E0: PQmakeEmptyPGresult (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10008DF30: pqGetErrorNotice3 (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10008CC00: pqParseInput3 (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007ECD9: parseInput (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007EE3F: PQgetResult (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007F6F1: PQexecFinish (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007F1DC: PQexec (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x1000019DF: main (in ./Gwidd_uniprot_map2)
    ==4525== 
    ==4525== 2,248 (200 direct, 2,048 indirect) bytes in 1 blocks are definitely lost in loss record 403 of 414
    ==4525==    at 0x100011345: malloc (vg_replace_malloc.c:236)
    ==4525==    by 0x10007B7E0: PQmakeEmptyPGresult (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10008DF30: pqGetErrorNotice3 (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10008CC00: pqParseInput3 (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007ECD9: parseInput (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007EE3F: PQgetResult (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007F6F1: PQexecFinish (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x10007F1DC: PQexec (in /usr/lib/libpq.5.dylib)
    ==4525==    by 0x1000019FA: main (in ./Gwidd_uniprot_map2)
    ==4525== 
    ==4525== LEAK SUMMARY:
    ==4525==    definitely lost: 800 bytes in 4 blocks
    ==4525==    indirectly lost: 8,192 bytes in 4 blocks
    ==4525==      possibly lost: 0 bytes in 0 blocks
    ==4525==    still reachable: 254,002 bytes in 743 blocks
    ==4525==         suppressed: 0 bytes in 0 blocks
    ==4525== Reachable blocks (those to which a pointer was found) are not shown.
    ==4525== To see them, rerun with: --leak-check=full --show-reachable=yes
    ==4525== 
    ==4525== For counts of detected and suppressed errors, rerun with: -v
    ==4525== ERROR SUMMARY: 4 errors from 4 contexts (suppressed: 0 from 0)

请提供解决问题的建议。
以上代码的修改:
int main()
{
    PGconn *conn;
    PGresult *res;
    int i=0,nFields=0,row=0,col=0;

    conn = PQconnectdb("dbname=test1 host=localhost user=postgres password=madhurima");
    if(PQstatus(conn) == CONNECTION_BAD) 
    {
        fprintf(stderr, "Connection to database \"%s\" failed.\n", PQerrorMessage(conn));
        fprintf(stderr, "%s", PQerrorMessage(conn));
        exit_nicely(conn);
    }

    res = PQexec(conn, "IF COL_LENGTH('protein_sequence','comment') IS NULL");
    PQclear(res);
    if(res)
    {
        res = PQexec(conn, "ALTER TABLE protein_sequence ADD comment VARCHAR(500)");
        PQclear(res);
    }
    else
    {
        res = PQexec(conn, "UPDATE TABLE protein_sequence ADD comment VARCHAR(500)");
        PQclear(res);
    }
    res = PQexec(conn, "IF COL_LENGTH('protein_sequence','uniprotid') IS NULL");
    PQclear(res);
    if(res)
    {
        res = PQexec(conn, "ALTER TABLE protein_sequence ADD uniprotid VARCHAR(50)");
        PQclear(res);
    }
    else
    {
        res = PQexec(conn, "UPDATE TABLE protein_sequence ADD uniprotid VARCHAR(50)");
        PQclear(res);
    }

    res = PQexec(conn, "SELECT * FROM protein_sequence");
    PQclear(res);
    if((!res) || (PQresultStatus(res) != PGRES_TUPLES_OK))
    {
        fprintf(stderr, "SELECT command did not return tuples properly\n");
        PQclear(res);
    }

    PQclear(res);
    PQfinish(conn);

    return 0;
}   

来自Valgrind的新错误:

==9234== Memcheck, a memory error detector
==9234== Copyright (C) 2002-2010, and GNU GPL'd, by Julian Seward et al.
==9234== Using Valgrind-3.6.1 and LibVEX; rerun with -h for copyright info
==9234== Command: ./Gwidd_uniprot_map2 xaa
==9234== 
--9234-- ./Gwidd_uniprot_map2:
--9234-- dSYM directory has wrong UUID; consider using --dsymutil=yes
==9234== Invalid read of size 4
==9234==    at 0x1000804CF: PQresultStatus (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A3B: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dff78 is 40 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid read of size 4
==9234==    at 0x10007CCCD: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dffe8 is 152 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid read of size 8
==9234==    at 0x10007CCE2: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dffe0 is 144 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid read of size 8
==9234==    at 0x10007CD26: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013e0008 is 184 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid read of size 8
==9234==    at 0x10007CD3F: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dff60 is 16 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid write of size 8
==9234==    at 0x10007CD5D: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dff58 is 8 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid write of size 8
==9234==    at 0x10007CD69: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dff60 is 16 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid write of size 8
==9234==    at 0x10007CD75: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dff70 is 32 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid write of size 8
==9234==    at 0x10007CD81: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dfff8 is 168 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid write of size 8
==9234==    at 0x10007CD90: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dffe0 is 144 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid write of size 4
==9234==    at 0x10007CD9F: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dffe8 is 152 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== Invalid free() / delete / delete[]
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A72: main (in ./Gwidd_uniprot_map2)
==9234==  Address 0x1013dff50 is 0 bytes inside a block of size 200 free'd
==9234==    at 0x100010E9F: free (vg_replace_malloc.c:366)
==9234==    by 0x10007CDB4: PQclear (in /usr/lib/libpq.5.dylib)
==9234==    by 0x100001A2B: main (in ./Gwidd_uniprot_map2)
==9234== 
==9234== 
==9234== HEAP SUMMARY:
==9234==     in use at exit: 254,002 bytes in 743 blocks
==9234==   total heap usage: 1,012 allocs, 270 frees, 345,158 bytes allocated
==9234== 
==9234== LEAK SUMMARY:
==9234==    definitely lost: 0 bytes in 0 blocks
==9234==    indirectly lost: 0 bytes in 0 blocks
==9234==      possibly lost: 0 bytes in 0 blocks
==9234==    still reachable: 254,002 bytes in 743 blocks
==9234==         suppressed: 0 bytes in 0 blocks
==9234== Reachable blocks (those to which a pointer was found) are not shown.
==9234== To see them, rerun with: --leak-check=full --show-reachable=yes
==9234== 
==9234== For counts of detected and suppressed errors, rerun with: -v
==9234== ERROR SUMMARY: 12 errors from 12 contexts (suppressed: 0 from 0)
1个回答

1
考虑以下来自你的代码中的片段:
    res = PQexec(conn, "IF COL_LENGTH('xxx','comment') IS NULL");
    if(res)
        res = PQexec(conn, "ALTER TABLE xxx ADD comment VARCHAR(500)");
    else
        res = PQexec(conn, "UPDATE TABLE xxx ADD comment VARCHAR(500)");

这里有几个问题:
  1. 你从未在第一个PQexec的结果上调用PQclear,这导致了内存泄漏。请记住,res保存了指向动态分配的返回结构的指针。这就是valgrind抱怨的原因。

  2. 你没有正确检查第一条查询的结果。仅仅检查PGexec的返回值不为空并不能说明执行命令的结果,它只确认libpq能够分配足够的内存来返回结果结构。

  3. 你也没有检查错误。你应该调用PQresultStatus来确定查询是否成功执行。

  4. 如果你检查错误,我怀疑你会发现你尝试执行的SQL存在问题。SQL中没有IF语句。PL/pgSQL中有一个,但那是在存储过程的上下文中。

有关正确调用的更多详细信息,请参阅libpq文档。


谢谢您的建议。我添加了PQclear,解决了内存泄漏的问题,但带来了新的错误:大小为4/8的无效写入,我在Valgrind文档中找到了这个错误,它指的是内存堆溢出。请给予建议。 - user3258515
我没有看到你现在使用的代码,所以很难说。请在你的问题中发布修改后的源代码(放在底部,保留原始代码,这样其他人阅读时就能理解历史记录)。 - harmic
只是一个提醒:如果表中不存在id和comment列,我想要添加它们。如果它们已经存在,则只需更新这些列。你认为程序中有逻辑错误吗? - user3258515

网页内容由stack overflow 提供, 点击上面的
可以查看英文原文,
原文链接