Welcome to the Builder Academy

Question Debug help

More
19 Mar 2023 15:34 - 19 Mar 2023 15:45 #10260 by cunning
Debug help was created by cunning
I am having some issues with 2 main TBA functions ( i thnk this is just valgrind) but i figured i would ask. There is another that i may or may not get help with here.

valgrind --log-file="leak.txt" --leak-check=full --show-reachable=yes --track-origins=yes bin/circle_dev_w -q 6001
==848167== Memcheck, a memory error detector
==848167== Copyright (C) 2002-2017, and GNU GPL'd, by Julian Seward et al.
==848167== Using Valgrind-3.18.1 and LibVEX; rerun with -h for copyright info
==848167== Command: bin/circle_dev_w -q 6001
==848167== Parent PID: 840102
==848167== 
==848167== Conditional jump or move depends on uninitialised value(s)
==848167==    at 0x1B8B40: fread_string (db.c:3915)
==848167==    by 0x1C4FBA: load_config (db.c:5862)
==848167==    by 0x1268B2: main (comm.c:285)
==848167==  Uninitialised value was created by a stack allocation
==848167==    at 0x1B8B22: fread_string (db.c:3915)

Code:
char *fread_string(FILE *fl, const char *error) {   char buf[MAX_STRING_LENGTH] = {'\0'}, tmp[513] = {'\0'};   char *point = NULL;   int done = 0, length = 0, templength = 0;   *buf = '\0';   *tmp = '\0';      do   {     memset(tmp, '\0', 513);     if (!fgets(tmp, 512, fl))     {       log("SYSERR: fread_string: format error at or near %s", error);       exit(1);     }     /* If there is a '~', end the string; else put an "\r\n" over the '\n'. */     /* now only removes trailing ~'s -- Welcor */     point = strchr(tmp, '\0');     if (point == NULL)     {       log("SYSERR: freed_string: end of string not found (db.c)");       log("String: %s", tmp);       exit(1);     }    [b] for (point--; (*point == '\r' || *point == '\n' || point == 0); point--)   ============> 3915[/b]       ;     if (*point == '~')     {       *point = '\0';       done = 1;     }     else     {       *(++point) = '\r';       *(++point) = '\n';       *(++point) = '\0';     }

==848167== Conditional jump or move depends on uninitialised value(s)
==848167==    at 0x1BD9A0: fread_clean_string (db.c:3982)
==848167==    by 0x21E8C6: read_ibt (ibt.c:197)
==848167==    by 0x21F08A: load_ibt_file (ibt.c:279)
==848167==    by 0x1C6C71: boot_db (db.c:1122)
==848167==    by 0x126D48: init_game (comm.c:561)
==848167==    by 0x126D48: main (comm.c:395)
==848167==  Uninitialised value was created by a stack allocation
==848167==    at 0x1BD981: fread_clean_string (db.c:3982)


/* fread_clean_string is the same as fread_string, but skips preceding spaces */
Code:
char *fread_clean_string(FILE *fl, const char *error) {   char buf[MAX_STRING_LENGTH] = {'\0'}, tmp[513] = {'\0'};   char *point = NULL, c = '\0';   int done = 0, length = 0, templength = 0;   *buf = '\0';   *tmp = '\0';   do   {     if (feof(fl))     {       log("%s", "fread_clean_string: EOF encountered on read.");       return 0;     }     c = getc(fl);   } while (isspace(c));   ungetc(c, fl);   do   {     if (!fgets(tmp, 512, fl))     {       log("SYSERR: fread_clean_string: format error at or near %s", error);       exit(1);     }     /* If there is a '~', end the string; else put an "\r\n" over the '\n'. */     /* now only removes trailing ~'s -- Welcor */     point = strchr(tmp, '\0');  [b]  [/b] for (point--; (*point == '\r' || *point == '\n'); point--)  ==================>3982       ;     if (*point == '~')     {       *point = '\0';       done = 1;     }     else     {       *(++point) = '\r';       *(++point) = '\n';       *(++point) = '\0';     }

==848167== HEAP SUMMARY:
==848167==     in use at exit: 3,600,349 bytes in 2,677 blocks
==848167==   total heap usage: 2,109,239 allocs, 2,106,562 frees, 272,968,248 bytes allocated
==848167== 
==848167== 5 bytes in 1 blocks are still reachable in loss record 1 of 369
==848167==    at 0x4848899: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==848167==    by 0x4BEC60E: strdup (strdup.c:42)
==848167==    by 0x4AE7B56: ??? (in /usr/lib/x86_64-linux-gnu/libmariadb.so.3)
==848167==    by 0x4AEC1DD: ??? (in /usr/lib/x86_64-linux-gnu/libmariadb.so.3)
==848167==    by 0x4BDDF67: __pthread_once_slow (pthread_once.c:116)
==848167==    by 0x249B69: connect_to_mysql (mysql.c:62)
==848167==    by 0x1C68AB: boot_db (db.c:984)
==848167==    by 0x126D48: init_game (comm.c:561)
==848167==    by 0x126D48: main (comm.c:395)
Code:
void connect_to_mysql() {      if (mysql_library_init(0, NULL, NULL))        =================> 62   {     log("SYSERR: Unable to initialize MySQL library.");     exit(1);   }   if (!(conn = mysql_init(NULL)))   {     log("SYSERR: Unable to initialize MySQL connection.");     exit(1);   }   my_bool reconnect = 1;   mysql_options(conn, MYSQL_OPT_RECONNECT, &reconnect);   if (!mysql_real_connect(conn, MYSQL_SERVER, MYSQL_USER, MYSQL_PASSWD, MYSQL_DB, 0, NULL, 0))   ===========> 78 same as 62 debug   {     log("SYSERR: Unable to connect to MySQL: %s", mysql_error(conn));     exit(1);   }
Last edit: 19 Mar 2023 15:45 by cunning.

Please Log in or Create an account to join the conversation.

More
19 Mar 2023 23:16 #10265 by thomas
Replied by thomas on topic Debug help
The two first ones are false positives. The variables in question are initialized in the same function, actually.

I must admit I know little about the C mysql-library, but this also looks like a false positive.

Please Log in or Create an account to join the conversation.

More
19 Mar 2023 23:32 - 19 Mar 2023 23:35 #10267 by cunning
Replied by cunning on topic Debug help
I thought so myself, but i definately have one that needs looked into. 

==1124415== Invalid read of size 8
==1124415==    at 0x238C67: next_in_list (lists.c:548)
==1124415==    by 0x238C67: simple_list (lists.c:657)
==1124415==    by 0x238D97: free_list (lists.c:409)
==1124415==    by 0x126FB9: main (comm.c:419)
==1124415==  Address 0x5b78f48 is 8 bytes inside a block of size 24 free'd
==1124415==    at 0x484B27F: free (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==1124415==    by 0x238D8F: free_list (lists.c:410)
==1124415==    by 0x126FB9: main (comm.c:419)
==1124415==  Block was alloc'd at
==1124415==    at 0x484DA83: calloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==1124415==    by 0x2387BE: create_item (lists.c:370)
==1124415==    by 0x2387BE: add_to_list (lists.c:428)
==1124415==    by 0x238863: create_list (lists.c:359)
==1124415==    by 0x1C6A8D: boot_db (db.c:1019)
==1124415==    by 0x126D68: init_game (comm.c:561)
==1124415==    by 0x126D68: main (comm.c:395)
==1124415== 
==1124415== 
==1124415== HEAP SUMMARY:
==1124415==     in use at exit: 329,281 bytes in 1,028 blocks
==1124415==   total heap usage: 3,009,434 allocs, 3,008,406 frees, 262,207,244 bytes allocated

Code:
struct item_data *create_item(void) {   struct item_data *pNewItem = NULL;   CREATE(pNewItem, struct item_data, 1);   pNewItem->pNextItem = NULL;   pNewItem->pPrevItem = NULL;   pNewItem->pContent = NULL;   return (pNewItem); }
Code:
void free_list(struct list_data *pList) {   void *pContent = NULL;   simple_list(NULL);   if (pList && pList->iSize)     while ((pContent = simple_list(pList)))   ========================> line 409       remove_from_list(pContent, pList);                     /* Global List for debugging */   if (pList != global_lists)     remove_from_list(pList, global_lists);   free(pList); }
Code:
void *next_in_list(struct iterator_data *pIterator) {   void *pContent = NULL;   struct item_data *pTempItem = NULL;   if (pIterator->pList == NULL)   {     mudlog(NRM, ADMLVL_GOD, TRUE, "WARNING: Attempting to get content from iterator with NULL list.");     return NULL;   }   /* Cycle down the list */   pTempItem = pIterator->pItem->pNextItem;     ====================? line 548   pIterator->pItem = pTempItem;   /* Grab the content */   pContent = pIterator->pItem ? pIterator->pItem->pContent : NULL;   return (pContent); }
Last edit: 19 Mar 2023 23:35 by cunning.

Please Log in or Create an account to join the conversation.

More
20 Mar 2023 00:17 #10268 by cunning
Replied by cunning on topic Debug help
Just for the record, the mysql errors were me not closing the DB when we destroy_db() called. Leaving all those sql connections opened. It seriously reduced my error output by 700 lines.

Please Log in or Create an account to join the conversation.

More
27 Mar 2023 00:47 #10276 by soth
Replied by soth on topic Debug help
Hi,
I do not use c with mysql, but I do with c# and one thing I learned back in 2005 was make sure you close and dispose of those connections when you are done with them :) Even though the garbage collector is supposed to clean up I usually use obj.Dispose() anyway.

I am just now starting to try and use a debugger for linux myself.

Cheers

Please Log in or Create an account to join the conversation.

More
27 Mar 2023 00:51 #10278 by cunning
Replied by cunning on topic Debug help
Thank you for your response. I fixed that a few days ago by closing all sockets during destroy_db().

I just cannot solve the 3rd one I did above with lists.

Please Log in or Create an account to join the conversation.

Time to create page: 0.212 seconds