Re: [PATCH v5] hashtable: introduce a small and naive hashtable

[Sasha Levin <levinsasha928@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1746 bytes --]

On 09/16/2012 01:08 AM, Matthias Diener wrote:
> Sasha Levin (levinsasha928 <at> gmail.com) wrote:
>> On Sat, Sep 15, 2012 at 5:14 PM, Mathieu Desnoyers
>> <mathieu.desnoyers <at> efficios.com> wrote:
>>> * Sasha Levin (levinsasha928 <at> gmail.com) wrote:
> [...]
>>>> +#define hash_init(hashtable)                                                 \
>>>> +({                                                                           \
>>>> +     int __i;                                                                \
>>>> +                                                                             \
>>>> +     for (__i = 0; __i < HASH_BITS(hashtable); __i++)                        \
>>>
>>> I think this fails to initialize the whole table. You'd need
>>>
>>>   HASH_BITS -> HASH_SIZE
>>
>> Right.
>>
>> Unfortunately it's pretty hard catching something like this :/
>>
>>> Which brings the following question: how did you test this code ? It
>>> would be nice to have a small test module along with this patchset that
>>> stress-tests this simple hash table in various configurations (on stack,
>>> in data, etc).
>>
>> I do two things:
>>
>> - A small userspace test (since this header works just fine from
>> userspace as well).
> 
> 
> It would be interesting to run some experiments with this hashtable in
> userspace.
> Could you post the test code here?

Sure, I've attached the test code. There are 2 things to remember it:

 1. The code looks like crap :) I've never intended it to be seen by others.
 2. It should be used in the context of "sanitized" kernel headers so it could
be included directly. I usually work in the directory of lkvm, and compile this
code using:

	gcc -Iinclude/ -I../../include/ -O0 -ggdb  hashtest.c



[-- Attachment #2: hashtest.c --]
[-- Type: text/x-csrc, Size: 1352 bytes --]

#define BITS_PER_LONG 64
#include <linux/kernel.h>
#include <linux/hashtable.h>
#include <kvm/util.h>
#include <linux/log2.h>
#include <assert.h>

struct my_data {
	int x;
	struct hlist_node node;
};

static DECLARE_HASHTABLE(test, 10);
static struct my_data *values[1000000];

struct my_data *get_data(int v)
{
	struct my_data *p;
	struct hlist_node *n;

	hash_for_each_possible(test, p, n, node, v)
		if (p->x == v)
			return p;

	return NULL;
}

bool verify(int i)
{
	if (values[i] == NULL && get_data(i))
		printf("No data, but found at %p (%d)\n", get_data(i), i);
	if (values[i])
		return get_data(i) ? true : false;
	else
		return get_data(i) ? false : true;
}

int main(void)
{
	int i;

	hash_init(test);

	printf("Empty? %d\n", hash_empty(test));

	while (1) {
	for (i = 0; i < ARRAY_SIZE(values); i++) {
		int r = rand() % 3;

		if (i % 10000 == 0) {
			printf("\r%d ...", i);
			fflush(stdout);
		}

		switch (r) {
		case 0:
			assert(verify(i));
			if (values[i])
				break;
			values[i] = malloc(sizeof(values[i]));
			values[i]->x = i;
			hash_add(test, &values[i]->node, i);
			assert(verify(i));
			break;
		case 1:
			assert(verify(i));

			break;
		case 2:
			assert(verify(i));
			if (values[i]) {
				hash_del(&values[i]->node);
				free(values[i]);
				values[i] = NULL;
			}
			assert(verify(i));
			break;
		}
	}
	}
	return 0;
}