[v6] eal: use madvise to exclude unmapped memory from being dumped

  Currently, even though memory is mapped with PROT_NONE, this does not
cause it to be excluded from core dumps. This is counter-productive,
because in a lot of cases, this memory will go unused (e.g. when the
memory subsystem preallocates VA space but hasn't yet mapped physical
pages into it).

Use `madvise()` call with MADV_DONTDUMP parameter to exclude the
unmapped memory from being dumped.

Signed-off-by: Li Feng <fengli@smartx.com>
---
v6:
- use RTE_DONTDUMP in right place.

v5:
- fix a code style.

V4:
- simplify the code;
- remove the undo code in V3;

V3:
- add support for freebsd.
- when free_seg is called, mark the memory MADV_DONTDUMP.
- when alloc_seg is called, mark the memory MADV_DODUMP.

 lib/librte_eal/common/eal_common_memory.c | 17 +++++++++++++++++
 lib/librte_eal/linux/eal_memalloc.c       |  3 +++
 2 files changed, 20 insertions(+)
  

On 24-Apr-20 2:19 PM, Li Feng wrote:
> Currently, even though memory is mapped with PROT_NONE, this does not
> cause it to be excluded from core dumps. This is counter-productive,
> because in a lot of cases, this memory will go unused (e.g. when the
> memory subsystem preallocates VA space but hasn't yet mapped physical
> pages into it).
> 
> Use `madvise()` call with MADV_DONTDUMP parameter to exclude the
> unmapped memory from being dumped.
> 
> Signed-off-by: Li Feng <fengli@smartx.com>
> ---
> v6:
> - use RTE_DONTDUMP in right place.
> 
> v5:
> - fix a code style.
> 
> V4:
> - simplify the code;
> - remove the undo code in V3;
> 
> V3:
> - add support for freebsd.
> - when free_seg is called, mark the memory MADV_DONTDUMP.
> - when alloc_seg is called, mark the memory MADV_DODUMP.
> 
>   lib/librte_eal/common/eal_common_memory.c | 17 +++++++++++++++++
>   lib/librte_eal/linux/eal_memalloc.c       |  3 +++
>   2 files changed, 20 insertions(+)
> 
> diff --git a/lib/librte_eal/common/eal_common_memory.c b/lib/librte_eal/common/eal_common_memory.c
> index cc7d54e0c..cd20672ce 100644
> --- a/lib/librte_eal/common/eal_common_memory.c
> +++ b/lib/librte_eal/common/eal_common_memory.c
> @@ -40,6 +40,14 @@
>   static void *next_baseaddr;
>   static uint64_t system_page_sz;
>   
> +#ifdef RTE_EXEC_ENV_LINUX
> +#define RTE_DONTDUMP MADV_DONTDUMP
> +#elif RTE_EXEC_ENV_FREEBSD
> +#define RTE_DONTDUMP MADV_NOCORE
> +#else
> +#error "madvise doesn't support this OS"
> +#endif
> +
>   #define MAX_MMAP_WITH_DEFINED_ADDR_TRIES 5
>   void *
>   eal_get_virtual_area(void *requested_addr, size_t *size,
> @@ -179,6 +187,15 @@ eal_get_virtual_area(void *requested_addr, size_t *size,
>   			munmap(aligned_end, after_len);
>   	}
>   
> +	if (!unmap) {
> +		/*
> +		 * Exclude this pages from a core dump.
> +		 */
> +		if (madvise(aligned_addr, *size, RTE_DONTDUMP) != 0)
> +			RTE_LOG(DEBUG, EAL, "madvise failed: %s\n",
> +				strerror(errno));
> +	}
> +
>   	return aligned_addr;
>   }
>   
> diff --git a/lib/librte_eal/linux/eal_memalloc.c b/lib/librte_eal/linux/eal_memalloc.c
> index af6d0d023..f628e0991 100644
> --- a/lib/librte_eal/linux/eal_memalloc.c
> +++ b/lib/librte_eal/linux/eal_memalloc.c
> @@ -687,6 +687,9 @@ free_seg(struct rte_memseg *ms, struct hugepage_info *hi,
>   		return -1;
>   	}
>   
> +	if (madvise(ms->addr, ms->len, MADV_DONTDUMP) != 0)
> +		RTE_LOG(DEBUG, EAL, "madvise failed: %s\n", strerror(errno));
> +
>   	exit_early = false;
>   
>   	/* if we're using anonymous hugepages, nothing to be done */
> 

Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
  

On Fri, Apr 24, 2020 at 3:19 PM Li Feng <fengli@smartx.com> wrote:
>
> Currently, even though memory is mapped with PROT_NONE, this does not
> cause it to be excluded from core dumps. This is counter-productive,
> because in a lot of cases, this memory will go unused (e.g. when the
> memory subsystem preallocates VA space but hasn't yet mapped physical
> pages into it).
>
> Use `madvise()` call with MADV_DONTDUMP parameter to exclude the
> unmapped memory from being dumped.
>
> Signed-off-by: Li Feng <fengli@smartx.com>

Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>

Applied, thanks.