All of lore.kernel.org
 help / color / mirror / Atom feed
From: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
To: Xueming Li <xuemingl@mellanox.com>, dev@dpdk.org
Subject: Re: [PATCH] eal/malloc: fix malloc cookie check
Date: Wed, 23 Aug 2017 14:50:54 +0100	[thread overview]
Message-ID: <a4d2d30d-4060-07f0-5f15-1f77d15fdd00@intel.com> (raw)
In-Reply-To: <20170823022926.169272-1-xuemingl@mellanox.com>

Hi Xueming,

I think I have understood what you are doing and why the change is 
required, but
I believe we can improve the commit message.

It might be even a good idea to split the patches:
1. RTE_MALLOC_DEBUG fix
2. malloc_elem_free fix


On 23/08/2017 03:29, Xueming Li wrote:
> From: xuemingl <xuemingl@mellanox.com>

'From: ..' not needed.

> DPDK uses it's own memory management, few regular memory profiler tool
> support DPDK now. Malloc cookie check provides limited memory corruption
> check, better than nothing. This patch fixes the following:
> 1. Replace broken generated configuration macro RTE_LIBRTE_MALLOC_DEBUG
> with RTE_MALLOC_DEBUG
> 2. Fix malloc size calculation when RTE_MALLOC_DEBUG cookie check
> enabled.

For the 2nd change, you should mention what is the problem.
The issue is that malloc_elem_free is clearing (setting to 0) the 
trailer cookie when RTE_MALLOC_DEBUG is enabled.

You could also add that you would trigger this issue if the elem being 
freed does not join the next elem.

Thanks,
Sergio

>  From real test, it IS very helpful to detect memory corruption.
>
> A better designed DPDK application should quit nicely with resoure
> releasing so that all allocated memory could be checked against cookie.
>
> Fixes: af75078 ("first public release")
> Cc: Sergio Gonzalez Monroy <sergio.gonzalez.monroy@intel.com>
>
> Signed-off-by: Xueming Li <xuemingl@mellanox.com>
> ---
>   lib/librte_eal/common/malloc_elem.c |  8 ++++----
>   lib/librte_eal/common/malloc_elem.h |  4 ++--
>   test/test/test_malloc.c             | 10 +++++-----
>   3 files changed, 11 insertions(+), 11 deletions(-)
>
> diff --git a/lib/librte_eal/common/malloc_elem.c b/lib/librte_eal/common/malloc_elem.c
> index 150769057..889dffd21 100644
> --- a/lib/librte_eal/common/malloc_elem.c
> +++ b/lib/librte_eal/common/malloc_elem.c
> @@ -275,14 +275,14 @@ malloc_elem_free(struct malloc_elem *elem)
>   		return -1;
>   
>   	rte_spinlock_lock(&(elem->heap->lock));
> -	size_t sz = elem->size - sizeof(*elem);
> +	size_t sz = elem->size - sizeof(*elem) - MALLOC_ELEM_TRAILER_LEN;
>   	uint8_t *ptr = (uint8_t *)&elem[1];
>   	struct malloc_elem *next = RTE_PTR_ADD(elem, elem->size);
>   	if (next->state == ELEM_FREE){
>   		/* remove from free list, join to this one */
>   		elem_free_list_remove(next);
>   		join_elem(elem, next);
> -		sz += sizeof(*elem);
> +		sz += (sizeof(*elem) + MALLOC_ELEM_TRAILER_LEN);
>   	}
>   
>   	/* check if previous element is free, if so join with it and return,
> @@ -291,8 +291,8 @@ malloc_elem_free(struct malloc_elem *elem)
>   	if (elem->prev != NULL && elem->prev->state == ELEM_FREE) {
>   		elem_free_list_remove(elem->prev);
>   		join_elem(elem->prev, elem);
> -		sz += sizeof(*elem);
> -		ptr -= sizeof(*elem);
> +		sz += (sizeof(*elem) + MALLOC_ELEM_TRAILER_LEN);
> +		ptr -= (sizeof(*elem) + MALLOC_ELEM_TRAILER_LEN);
>   		elem = elem->prev;
>   	}
>   	malloc_elem_free_list_insert(elem);
> diff --git a/lib/librte_eal/common/malloc_elem.h b/lib/librte_eal/common/malloc_elem.h
> index f04b2d1e4..ce39129d9 100644
> --- a/lib/librte_eal/common/malloc_elem.h
> +++ b/lib/librte_eal/common/malloc_elem.h
> @@ -53,13 +53,13 @@ struct malloc_elem {
>   	volatile enum elem_state state;
>   	uint32_t pad;
>   	size_t size;
> -#ifdef RTE_LIBRTE_MALLOC_DEBUG
> +#ifdef RTE_MALLOC_DEBUG
>   	uint64_t header_cookie;         /* Cookie marking start of data */
>   	                                /* trailer cookie at start + size */
>   #endif
>   } __rte_cache_aligned;
>   
> -#ifndef RTE_LIBRTE_MALLOC_DEBUG
> +#ifndef RTE_MALLOC_DEBUG
>   static const unsigned MALLOC_ELEM_TRAILER_LEN = 0;
>   
>   /* dummy function - just check if pointer is non-null */
> diff --git a/test/test/test_malloc.c b/test/test/test_malloc.c
> index 013fd4407..5558acda4 100644
> --- a/test/test/test_malloc.c
> +++ b/test/test/test_malloc.c
> @@ -108,7 +108,7 @@ test_align_overlap_per_lcore(__attribute__((unused)) void *arg)
>   		}
>   		for(j = 0; j < 1000 ; j++) {
>   			if( *(char *)p1 != 0) {
> -				printf("rte_zmalloc didn't zero"
> +				printf("rte_zmalloc didn't zero "
>   				       "the allocated memory\n");
>   				ret = -1;
>   			}
> @@ -180,7 +180,7 @@ test_reordered_free_per_lcore(__attribute__((unused)) void *arg)
>   		}
>   		for(j = 0; j < 1000 ; j++) {
>   			if( *(char *)p1 != 0) {
> -				printf("rte_zmalloc didn't zero"
> +				printf("rte_zmalloc didn't zero "
>   				       "the allocated memory\n");
>   				ret = -1;
>   			}
> @@ -293,7 +293,7 @@ test_multi_alloc_statistics(void)
>   	struct rte_malloc_socket_stats pre_stats, post_stats ,first_stats, second_stats;
>   	size_t size = 2048;
>   	int align = 1024;
> -#ifndef RTE_LIBRTE_MALLOC_DEBUG
> +#ifndef RTE_MALLOC_DEBUG
>   	int trailer_size = 0;
>   #else
>   	int trailer_size = RTE_CACHE_LINE_SIZE;
> @@ -623,7 +623,7 @@ test_rte_malloc_validate(void)
>   	const size_t request_size = 1024;
>   	size_t allocated_size;
>   	char *data_ptr = rte_malloc(NULL, request_size, RTE_CACHE_LINE_SIZE);
> -#ifdef RTE_LIBRTE_MALLOC_DEBUG
> +#ifdef RTE_MALLOC_DEBUG
>   	int retval;
>   	char *over_write_vals = NULL;
>   #endif
> @@ -645,7 +645,7 @@ test_rte_malloc_validate(void)
>   	if (allocated_size < request_size)
>   		err_return();
>   
> -#ifdef RTE_LIBRTE_MALLOC_DEBUG
> +#ifdef RTE_MALLOC_DEBUG
>   
>   	/****** change the header to be bad */
>   	char save_buf[64];

  reply	other threads:[~2017-08-23 13:50 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-08-23  2:29 [PATCH] eal/malloc: fix malloc cookie check Xueming Li
2017-08-23 13:50 ` Sergio Gonzalez Monroy [this message]
2017-09-08 14:50 ` [PATCH v2 1/2] eal/malloc: fix RTE malloc debug macro Xueming Li
2017-09-08 14:50   ` [PATCH v2 2/2] eal/malloc: fix RTE malloc element free Xueming Li
2017-09-08 16:03   ` [PATCH v2 1/2] eal/malloc: fix RTE malloc debug macro Stephen Hemminger
2017-09-09  7:33     ` Xueming(Steven) Li
2017-09-09  7:33 ` [PATCH v3 " Xueming Li
2017-09-09  7:33   ` [PATCH v3 2/2] eal/malloc: fix RTE malloc element free Xueming Li
2017-09-13 12:06     ` Sergio Gonzalez Monroy
2017-10-09 20:56       ` Thomas Monjalon
2017-09-13 12:06   ` [PATCH v3 1/2] eal/malloc: fix RTE malloc debug macro Sergio Gonzalez Monroy

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=a4d2d30d-4060-07f0-5f15-1f77d15fdd00@intel.com \
    --to=sergio.gonzalez.monroy@intel.com \
    --cc=dev@dpdk.org \
    --cc=xuemingl@mellanox.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.