On Fri, Sep 18, 2020 at 08:06 AM, Benoit Ganne (bganne) wrote:
>
> I am trying to report the hardware interface queue depth.
As in, TX interface FIFO/Queue occupancy.
>
> You can use vnet_get_sup_hw_interface() for this, however HW queues are
> managed differently by each driver.
There is
Hi Mauricio,
> I am trying to report the hardware interface queue depth.
[...]
> Is there a way/function that maps vnet_buffer(b)->sw_if_index to it's
> appropriate hw_if_index???
You can use vnet_get_sup_hw_interface() for this, however HW queues are managed
differently by each driver.
See for
Hi all,
I am trying to report the hardware interface queue depth. I'm looking at
ip6_forward and it seems that the process to TX a packet is vlib_get_next_frame
... [do things with packet] ... vlib_validate_buffer_enqueue ...
vlib_put_next_frame.
Before I go down this rabbit hole of code
> Is there a way to determine who has cached and who is using a specific
> buffer?
> Is allocation of buffers bounded to a specific numa node? Or can they
> allocate across numa nodes? If so, how is this determined?
As mentioned by Damjan this would be too expensive in general. However if you
> On 4 Sep 2020, at 10:41, Solis JR, M. (Mauricio) via lists.fd.io
> wrote:
>
> Thank you, I'm beginning to understand now, then total = available + cached +
> used?
yes
>
> Is there a way to determine who has cached
What do you mean by “who have cached”?
Cached buffers are free
Thank you, I'm beginning to understand now, then total = available + cached +
used?
Is there a way to determine who has cached and who is using a specific buffer?
Is allocation of buffers bounded to a specific numa node? Or can they allocate
across numa nodes? If so, how is this determined?
> On 3 Sep 2020, at 18:24, Solis JR, M. (Mauricio) via lists.fd.io
> wrote:
>
> [Edited Message Follows]
>
> Are there any resources where I can read more about the VPP buffer
> architecture?
As Dave likes to say, “Use the force, read the source” :)
As I am mainly guilty for not having
> On 3 Sep 2020, at 15:27, Benoit Ganne (bganne) via lists.fd.io
> wrote:
>
>> I'm confused about what buffers-per-numa and buffer data-size are
>> referring to? If I'm trying to determine overall packet buffer
>> occupancy levels, should I use buffers-per-numa or data-size?
>
> A buffer in
[Edited Message Follows]
Are there any resources where I can read more about the VPP buffer architecture?
@Dave
In regards to identifying the number of total buffers, are you referring to the
vector size of vlib_buffer_pools_t in vlib_buffer_main?
Thanks,
Mauricio
-=-=-=-=-=-=-=-=-=-=-=-
Are there any resources where I can read more about the VPP buffer architecture?
Thanks,
Mauricio
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#17328): https://lists.fd.io/g/vpp-dev/message/17328
Mute This Topic:
To: mauricio.soli...@tno.nl; vpp-dev@lists.fd.io
Subject: Re: [vpp-dev] Buffer Occupancy Calculation #vnet #vpp
> I'm confused about what buffers-per-numa and buffer data-size are
> referring to? If I'm trying to determine overall packet buffer
> occupancy levels, should I use buffers-per-numa or
> I'm confused about what buffers-per-numa and buffer data-size are
> referring to? If I'm trying to determine overall packet buffer
> occupancy levels, should I use buffers-per-numa or data-size?
A buffer in VPP is composed of a 128-bytes header, a 128-bytes scratchpad (to
allow eg. tunneling
Hi,
I am looking to calculating the overall packet buffer occupancy. I've dug into
the buffer files and it looks like the main data structure holding buffer
information is "vlib_buffer_pool_t". This structure seems to take information
from the startup.conf "buffers {}" section. I'm confused
13 matches
Mail list logo
