Question

I have a tailq with 100k+ entries of this struct:

struct entry {
char *file_name;
FILE *file;
TAILQ_ENTRY(entry) tailq;
};

The purpose is store thousands of file pointers for an application creates thousands of files and append stuffs to them.

On each increase of the tailq I have a foreach:

int c;
char temp[20];

struct entry *np;

TAILQ_FOREACH(np, &tailq_head[y], tailq) {
    if(strcmp(np->file_name, temp) == 0){
        c = 1;
        break;
    }
}

That searches for some temp name that already is on the tailq, if it isn’t in the tail then add id, else do not.

What can I do to improve performance? What is the faster structure that I can use? Should I calculate an integer hash to the temp variable to be compared in the foreach? Ideas?

Answer 1

Keeping an integer hash of the name in each entry will speed up the compare by a significant amount. It will also save one level of pointer indirection. But you’re still comparing against every single entry. If you store the entries in a structure that provides for efficient search without comparing against every single entry, like a hash table, the performance benefit would be even greater.