Revision 6e474083f3daf3a3546737f5d7d502ad12eb257c authored by Wei Xu on 01 December 2017, 10:10:36 UTC, committed by David S. Miller on 03 December 2017, 02:31:03 UTC
Matthew found a roughly 40% tcp throughput regression with commit c67df11f(vhost_net: try batch dequing from skb array) as discussed in the following thread: https://www.mail-archive.com/netdev@vger.kernel.org/msg187936.html Eventually we figured out that it was a skb leak in handle_rx() when sending packets to the VM. This usually happens when a guest can not drain out vq as fast as vhost fills in, afterwards it sets off the traffic jam and leaks skb(s) which occurs as no headcount to send on the vq from vhost side. This can be avoided by making sure we have got enough headcount before actually consuming a skb from the batched rx array while transmitting, which is simply done by moving checking the zero headcount a bit ahead. Signed-off-by: Wei Xu <wexu@redhat.com> Reported-by: Matthew Rosato <mjrosato@linux.vnet.ibm.com> Acked-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
1 parent fa935ca
fsl_pamu_domain.h
/*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License, version 2, as
* published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*
* Copyright (C) 2013 Freescale Semiconductor, Inc.
*
*/
#ifndef __FSL_PAMU_DOMAIN_H
#define __FSL_PAMU_DOMAIN_H
#include "fsl_pamu.h"
struct dma_window {
phys_addr_t paddr;
u64 size;
int valid;
int prot;
};
struct fsl_dma_domain {
/*
* Indicates the geometry size for the domain.
* This would be set when the geometry is
* configured for the domain.
*/
dma_addr_t geom_size;
/*
* Number of windows assocaited with this domain.
* During domain initialization, it is set to the
* the maximum number of subwindows allowed for a LIODN.
* Minimum value for this is 1 indicating a single PAMU
* window, without any sub windows. Value can be set/
* queried by set_attr/get_attr API for DOMAIN_ATTR_WINDOWS.
* Value can only be set once the geometry has been configured.
*/
u32 win_cnt;
/*
* win_arr contains information of the configured
* windows for a domain. This is allocated only
* when the number of windows for the domain are
* set.
*/
struct dma_window *win_arr;
/* list of devices associated with the domain */
struct list_head devices;
/* dma_domain states:
* mapped - A particular mapping has been created
* within the configured geometry.
* enabled - DMA has been enabled for the given
* domain. This translates to setting of the
* valid bit for the primary PAACE in the PAMU
* PAACT table. Domain geometry should be set and
* it must have a valid mapping before DMA can be
* enabled for it.
*
*/
int mapped;
int enabled;
/* stash_id obtained from the stash attribute details */
u32 stash_id;
struct pamu_stash_attribute dma_stash;
u32 snoop_id;
struct iommu_domain iommu_domain;
spinlock_t domain_lock;
};
/* domain-device relationship */
struct device_domain_info {
struct list_head link; /* link to domain siblings */
struct device *dev;
u32 liodn;
struct fsl_dma_domain *domain; /* pointer to domain */
};
#endif /* __FSL_PAMU_DOMAIN_H */
![swh spinner](/static/img/swh-spinner.gif)
Computing file changes ...