Max mtu #246
Merged
Max mtu #246
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
rjarry
force-pushed
the
max-mtu
branch
3 times, most recently
from
5b0c6a6 to
e83d3e9
Compare
maxime-leroy
requested changes
Jun 27, 2025
rjarry
force-pushed
the
max-mtu
branch
4 times, most recently
from
7c1c790 to
081c683
Compare
This partially reverts commit 62071f3. This is too risky, the stack does not support multi segment in a safe way. Keep the high MTU test. Also, enabling scatter/gather lowers performance with some drivers: DRIVER TEST RX/TX PKT/S mlx5 ipv4_fwd scatter/gather 15.5M (-0.6%) mlx5 ipv4_fwd simple 15.6M ice ipv4_fwd scatter/gather 13.8M (-12.7%) ice ipv4_fwd simple 15.8M Signed-off-by: Maxime Leroy <[email protected]> Signed-off-by: Robin Jarry <[email protected]>
Add a new parameter to enforce the maximum MTU on ports. Use the provided value to determine the size of mbuf segments. By default, the maximum MTU is set to 1800 which should cover most standard use cases. Increase the amount of allocated memory in test mode to allow creating ports with bigger mbufs. NB: larger mbufs lowers performance with some drivers: DRIVER TEST MBUF_SIZE PERF mlx5 ipv4_fwd 2K 15.6M mlx5 ipv4_fwd 16K 15.5M (-0.6%) ice ipv4_fwd 2K 15.8M ice ipv4_fwd 16K 14.7M (-7.0%) Signed-off-by: Robin Jarry <[email protected]>
Only ports have a MTU value which is initialized from whatever hardware has. Ensure VLAN interfaces always have the same MTU as their parent port. Initialize IPIP interfaces at 1480 bytes (1500 minus an extra IP header). Signed-off-by: Robin Jarry <[email protected]>
Currently, we don't check if the packets transmitted will fit in the configured MTU of the output port. Before proceeding further, drop the packets if they are larger than the configured MTU. Some day, we will need to support fragmentation for IPv4 and sending an ICMP6 "packet too big" error for IPv6. Signed-off-by: Robin Jarry <[email protected]>
| const struct gr_iface conf = { | ||
| .flags = iface->flags, | ||
| .mtu = iface->mtu, | ||
| .mtu = iface->mtu ?: 1480, |
Collaborator
There was a problem hiding this comment.
In which case, iface->mtu is zero ?
Collaborator
Author
There was a problem hiding this comment.
Unless users have specified a value for mtu in the initial interface creation message, it will be 0. It was OK until now since we'll drop packets that are bigger than output interface MTU.
I chose to initialize it to 1480 which seems like a good default.
Collaborator
maxime-leroy
left a comment
maxime-leroy
left a comment
There was a problem hiding this comment.
ICMP too big must also sent for IPV4 with DF bit set.
We don't have any function to send packets ?
Collaborator
Author
We do but at the moment, we don't support neither fragmentation nor sending ICMP too big packets. This will be implemented later. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.