DPDK patches and discussions
 help / color / mirror / Atom feed
From: Maxime Coquelin <maxime.coquelin@redhat.com>
To: "Mattias Rönnblom" <mattias.ronnblom@ericsson.com>,
	dev@dpdk.org, tiwei.bie@intel.com, jfreimann@redhat.com,
	zhihong.wang@intel.com, bruce.richardson@intel.com,
	konstantin.ananyev@intel.com
Subject: Re: [dpdk-dev] [PATCH 3/5] vhost: do not inline unlikely fragmented buffers code
Date: Thu, 23 May 2019 16:30:57 +0200	[thread overview]
Message-ID: <25057af9-8d1c-5e97-b648-6d344724291e@redhat.com> (raw)
In-Reply-To: <2cfed6b0-8a7d-ec05-4b67-f40bd9409a8b@ericsson.com>

Hi Mattias,

On 5/21/19 9:43 PM, Mattias Rönnblom wrote:
> On 2019-05-17 14:22, Maxime Coquelin wrote:
>> Handling of fragmented virtio-net header and indirect descriptors
>> tables was implemented to fix CVE-2018-1059. It should not never
>> happen with healthy guests and so are already considered as
>> unlikely code path.
>>
>> This patch moves these bits into non-inline dedicated functions
>> to reduce the I-cache pressure.
>>
>> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
>> ---
>>   lib/librte_vhost/vhost.c      |  33 +++++++++++
>>   lib/librte_vhost/vhost.h      |  35 +-----------
>>   lib/librte_vhost/virtio_net.c | 102 +++++++++++++++++++---------------
>>   3 files changed, 91 insertions(+), 79 deletions(-)
>>
>> diff --git a/lib/librte_vhost/vhost.c b/lib/librte_vhost/vhost.c
>> index 4a54ad6bd1..8a4379bc13 100644
>> --- a/lib/librte_vhost/vhost.c
>> +++ b/lib/librte_vhost/vhost.c
>> @@ -201,6 +201,39 @@ __vhost_log_cache_write(struct virtio_net *dev, 
>> struct vhost_virtqueue *vq,
>>   }
>> +void *
>> +alloc_copy_ind_table(struct virtio_net *dev, struct vhost_virtqueue *vq,
> 
> This function should have a prefix.

This function is just moved from vhost.h to vhost.c, so not the purpose
of the patch.

But I agree your comment, I'll send a patch to add a prefix.

> 
>> +        uint64_t desc_addr, uint64_t desc_len)
>> +{
>> +    void *idesc;
>> +    uint64_t src, dst;
>> +    uint64_t len, remain = desc_len;
>> +
>> +    idesc = rte_malloc(__func__, desc_len, 0);
>> +    if (unlikely(!idesc))
> 
> if (idesc == NULL)

Ditto, that is not the purpose of the patch that is just moving the
function.

I agree this is not matching the coding rules specified in the
documentation, though.

> 
>> +        return NULL;
>> +
>> +    dst = (uint64_t)(uintptr_t)idesc;
>> +
>> +    while (remain) {
> remain > 0

Ditto.

>> +        len = remain;
>> +        src = vhost_iova_to_vva(dev, vq, desc_addr, &len,
>> +                VHOST_ACCESS_RO);
>> +        if (unlikely(!src || !len)) {
>> +            rte_free(idesc);
>> +            return NULL;
>> +        }
>> +
>> +        rte_memcpy((void *)(uintptr_t)dst, (void *)(uintptr_t)src, len);
> 
> Just for my understanding: what difference does that (uintptr_t) cast do?

This is required to build 32bits (-Werror=int-to-pointer-cast)

>> +
>> +        remain -= len;
>> +        dst += len;
>> +        desc_addr += len;
>> +    }
>> +
>> +    return idesc;
>> +}
>> +
>>   void
>>   cleanup_vq(struct vhost_virtqueue *vq, int destroy)
>>   {
>> diff --git a/lib/librte_vhost/vhost.h b/lib/librte_vhost/vhost.h
>> index 3ab7b4950f..ab26454e1c 100644
>> --- a/lib/librte_vhost/vhost.h
>> +++ b/lib/librte_vhost/vhost.h
>> @@ -488,6 +488,8 @@ void vhost_backend_cleanup(struct virtio_net *dev);
>>   uint64_t __vhost_iova_to_vva(struct virtio_net *dev, struct 
>> vhost_virtqueue *vq,
>>               uint64_t iova, uint64_t *len, uint8_t perm);
>> +void *alloc_copy_ind_table(struct virtio_net *dev, struct 
>> vhost_virtqueue *vq,
>> +            uint64_t desc_addr, uint64_t desc_len);
>>   int vring_translate(struct virtio_net *dev, struct vhost_virtqueue 
>> *vq);
>>   void vring_invalidate(struct virtio_net *dev, struct vhost_virtqueue 
>> *vq);
>> @@ -601,39 +603,6 @@ vhost_vring_call_packed(struct virtio_net *dev, 
>> struct vhost_virtqueue *vq)
>>           eventfd_write(vq->callfd, (eventfd_t)1);
>>   }
>> -static __rte_always_inline void *
>> -alloc_copy_ind_table(struct virtio_net *dev, struct vhost_virtqueue *vq,
>> -        uint64_t desc_addr, uint64_t desc_len)
>> -{
>> -    void *idesc;
>> -    uint64_t src, dst;
>> -    uint64_t len, remain = desc_len;
>> -
>> -    idesc = rte_malloc(__func__, desc_len, 0);
>> -    if (unlikely(!idesc))
>> -        return 0;
>> -
>> -    dst = (uint64_t)(uintptr_t)idesc;
>> -
>> -    while (remain) {
>> -        len = remain;
>> -        src = vhost_iova_to_vva(dev, vq, desc_addr, &len,
>> -                VHOST_ACCESS_RO);
>> -        if (unlikely(!src || !len)) {
>> -            rte_free(idesc);
>> -            return 0;
>> -        }
>> -
>> -        rte_memcpy((void *)(uintptr_t)dst, (void *)(uintptr_t)src, len);
>> -
>> -        remain -= len;
>> -        dst += len;
>> -        desc_addr += len;
>> -    }
>> -
>> -    return idesc;
>> -}
>> -
>>   static __rte_always_inline void
>>   free_ind_table(void *idesc)
>>   {
>> diff --git a/lib/librte_vhost/virtio_net.c 
>> b/lib/librte_vhost/virtio_net.c
>> index 35ae4992c2..494dd9957e 100644
>> --- a/lib/librte_vhost/virtio_net.c
>> +++ b/lib/librte_vhost/virtio_net.c
>> @@ -610,6 +610,35 @@ reserve_avail_buf_packed(struct virtio_net *dev, 
>> struct vhost_virtqueue *vq,
>>       return 0;
>>   }
>> +static void
>> +copy_vnet_hdr_to_desc(struct virtio_net *dev, struct vhost_virtqueue 
>> *vq,
> 
> __rte_noinline? Or you don't care about this function being inlined or not?

Right, I'll add it here and there in next revision.

I'll try to send a patch to fix the kind of style issues you reported.
If you want to do it that would be great, just let me know.

Thanks,
Maxime

  reply	other threads:[~2019-05-23 14:31 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-05-17 12:22 [dpdk-dev] [PATCH 0/5] vhost: I-cache pressure optimizations Maxime Coquelin
2019-05-17 12:22 ` [dpdk-dev] [PATCH 1/5] vhost: un-inline dirty pages logging functions Maxime Coquelin
2019-05-17 12:22 ` [dpdk-dev] [PATCH 2/5] vhost: do not inline packed and split functions Maxime Coquelin
2019-05-17 13:00   ` David Marchand
2019-05-17 14:42     ` Maxime Coquelin
2019-05-17 12:22 ` [dpdk-dev] [PATCH 3/5] vhost: do not inline unlikely fragmented buffers code Maxime Coquelin
2019-05-17 12:57   ` Maxime Coquelin
2019-05-21 19:43   ` Mattias Rönnblom
2019-05-23 14:30     ` Maxime Coquelin [this message]
2019-05-23 15:17       ` Mattias Rönnblom
2019-05-23 17:40         ` Maxime Coquelin
2019-05-17 12:22 ` [dpdk-dev] [PATCH 4/5] vhost: simplify descriptor's buffer prefetching Maxime Coquelin
2019-05-17 12:22 ` [dpdk-dev] [PATCH 5/5] eal/x86: force inlining of all memcpy and mov helpers Maxime Coquelin
2019-05-17 13:04 ` [dpdk-dev] [PATCH 0/5] vhost: I-cache pressure optimizations David Marchand
2019-05-17 14:42   ` Maxime Coquelin

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=25057af9-8d1c-5e97-b648-6d344724291e@redhat.com \
    --to=maxime.coquelin@redhat.com \
    --cc=bruce.richardson@intel.com \
    --cc=dev@dpdk.org \
    --cc=jfreimann@redhat.com \
    --cc=konstantin.ananyev@intel.com \
    --cc=mattias.ronnblom@ericsson.com \
    --cc=tiwei.bie@intel.com \
    --cc=zhihong.wang@intel.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).