[dpdk-dev] [PATCH dpdk-dev v2] mempool: sort the rte_mempool_ops by name

[Olivier Matz]

Hi,

On Thu, Mar 05, 2020 at 04:20:40PM +0800, xiangxia.m.yue at gmail.com wrote:
> From: Tonghao Zhang <xiangxia.m.yue at gmail.com>
> 
> The order of mempool initiation affects mempool index in the
> rte_mempool_ops_table. For example, when building APPs with:
> 
> $ gcc -lrte_mempool_bucket -lrte_mempool_ring ...
> 
> The "bucket" mempool will be registered firstly, and its index
> in table is 0 while the index of "ring" mempool is 1. DPDK
> uses the mk/rte.app.mk to build APPs, and others, for example,
> Open vSwitch, use the libdpdk.a or libdpdk.so to build it.
> The mempool lib linked in dpdk and Open vSwitch is different.
> 
> The mempool can be used between primary and secondary process,
> such as dpdk-pdump and pdump-pmd/Open vSwitch(pdump enabled).
> There will be a crash because dpdk-pdump creates the "ring_mp_mc"
> ring which index in table is 0, but the index of "bucket" ring
> is 0 in Open vSwitch. If Open vSwitch use the index 0 to get
> mempool ops and malloc memory from mempool. The crash will occur:
> 
>     bucket_dequeue (access null and crash)
>     rte_mempool_get_ops (should get "ring_mp_mc",
>                          but get "bucket" mempool)
>     rte_mempool_ops_dequeue_bulk
>     ...
>     rte_pktmbuf_alloc
>     rte_pktmbuf_copy
>     pdump_copy
>     pdump_rx
>     rte_eth_rx_burst
> 
> To avoid the crash, there are some solution:
> * constructor priority: Different mempool uses different
>   priority in RTE_INIT, but it's not easy to maintain.
> 
> * change mk/rte.app.mk: Change the order in mk/rte.app.mk to
>   be same as libdpdk.a/libdpdk.so, but when adding a new mempool
>   driver in future, we must make sure the order.
> 
> * register mempool orderly: Sort the mempool when registering,
>   so the lib linked will not affect the index in mempool table.
> 
> Signed-off-by: Tonghao Zhang <xiangxia.m.yue at gmail.com>

Sorting the pool drivers certainly make things better than they
are today. However, there is still an issue as soon as the mempool
list is not the same on both sides.

I don't see any better solution anyway. Storing the ops pointers in the
shared memory won't work, since function addresses won't be the same in
the 2 processes.

Just one minor comment below.

> ---
> v2:
> 1. use the qsort to sort the mempool_ops.
> 2. tested: https://travis-ci.com/ovn-open-virtual-networks/dpdk-next-net/builds/151894026
> ---
>  lib/librte_mempool/rte_mempool_ops.c | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/lib/librte_mempool/rte_mempool_ops.c b/lib/librte_mempool/rte_mempool_ops.c
> index 22c5251..e9113cf 100644
> --- a/lib/librte_mempool/rte_mempool_ops.c
> +++ b/lib/librte_mempool/rte_mempool_ops.c
> @@ -17,6 +17,15 @@ struct rte_mempool_ops_table rte_mempool_ops_table = {
>  	.num_ops = 0
>  };
>  
> +static int
> +compare_mempool_ops(const void *a, const void *b)
> +{
> +	const struct rte_mempool_ops *m_a = a;
> +	const struct rte_mempool_ops *m_b = b;
> +
> +	return strcmp(m_a->name, m_b->name);
> +}
> +
>  /* add a new ops struct in rte_mempool_ops_table, return its index. */
>  int
>  rte_mempool_register_ops(const struct rte_mempool_ops *h)
> @@ -63,6 +72,11 @@ struct rte_mempool_ops_table rte_mempool_ops_table = {
>  	ops->get_info = h->get_info;
>  	ops->dequeue_contig_blocks = h->dequeue_contig_blocks;
>  
> +	/* sort the rte_mempool_ops by name. the order of the mempool
> +	 * lib initiation will not affect rte_mempool_ops index. */

initiation -> initialization

there is also a checkpatch comment:

WARNING:BLOCK_COMMENT_STYLE: Block comments use a trailing */ on a separate line
#153: FILE: lib/librte_mempool/rte_mempool_ops.c:76:
+	 * lib initiation will not affect rte_mempool_ops index. */


> +	qsort(rte_mempool_ops_table.ops, rte_mempool_ops_table.num_ops,
> +	      sizeof(rte_mempool_ops_table.ops[0]), compare_mempool_ops);
> +
>  	rte_spinlock_unlock(&rte_mempool_ops_table.sl);
>  
>  	return ops_index;
> -- 
> 1.8.3.1
> 

Then,
Acked-by: Olivier Matz <olivier.matz at 6wind.com>

Thanks!