On 6/23/20 7:41 PM, Ferruh Yigit wrote:
> Function 'rte_eth_dma_zone_reserve()' returns an existing memzone based
> on name match, but other requested attributes are discarded.
> This may cause driver using a memzone with wrong size or alignment.
>
> Verify size, alignment and socket_id for matched memzone, and do not use
> memzone if any one of the attributes are not justified.
>
> Reported-by: Renata Saiakhova <renata.saiakh...@ekinops.com>
> Signed-off-by: Ferruh Yigit <ferruh.yi...@intel.com>
Few minor notes below, other than that:
Reviewed-by: Andrew Rybchenko <arybche...@solarflare.com>
> ---
> Cc: Anatoly Burakov <anatoly.bura...@intel.com>
> ---
> lib/librte_ethdev/rte_ethdev.c | 12 +++++++++++-
> 1 file changed, 11 insertions(+), 1 deletion(-)
>
> diff --git a/lib/librte_ethdev/rte_ethdev.c b/lib/librte_ethdev/rte_ethdev.c
> index 8e10a6fc3..0fe84aadc 100644
> --- a/lib/librte_ethdev/rte_ethdev.c
> +++ b/lib/librte_ethdev/rte_ethdev.c
> @@ -4202,8 +4202,18 @@ rte_eth_dma_zone_reserve(const struct rte_eth_dev
> *dev, const char *ring_name,
> }
>
> mz = rte_memzone_lookup(z_name);
> - if (mz)
> + if (mz) {
> + if ((socket_id != SOCKET_ID_ANY && socket_id != mz->socket_id)
> ||
> + size > mz->len ||
> + (uintptr_t)mz->addr & (align - 1)) {
IMHO, it would be a bit more readable if the result compared
with 0. Up to you.
> + RTE_ETHDEV_LOG(ERR,
> + "memzone %s does not justify the requested
> attributes\n",
> + mz->name);
> + return NULL;
Initially I thought it is a bad behaviour and memzone should
be freed and reallocated, but it is even worse (implicit free).
May be it is better to highlight it in the description.
> + }
> +
> return mz;
> + }
>
> return rte_memzone_reserve_aligned(z_name, size, socket_id,
> RTE_MEMZONE_IOVA_CONTIG, align);
>
