if_vxlan: Update *m0 after a pullup

[markj]

Summary

vxlan_input()'s caller is supposed to free *m0 if it is non-NULL after
the function returns. vxlan_input() failed to free *m0 after the pullup
however, so if m_pullup() fails or we hit an error case, we'd free the
mbuf twice.

Reported by: Yuxiang Yang, Yizhou Zhao, Xuewei Feng, Qi Li, and Ke Xu from Tsinghua University using GLM5.1 from Z.ai

[zlei]

Oops ! My stupid mistake.

vxlan_input() failed to free *m0 after the pullup
however, so if m_pullup() fails or we hit an error case, we'd free the
mbuf twice.

The description is not accurate.

m_pullup() can end up three cases:

1. It fails. Then the original mbuf is freed, and *m0 is set to NULL. The caller will not free a NULL mbuf. Nothing bad will happen.
2. It succeeds, but no new mbuf allocated. In this case the m == *m0; Nothing bad will happen.
3. It succeeds with new mbuf allocated. In this case m != *m0, thus will leak memory and free a freed mbuf.

[markj]

In D56944#1304377, @zlei wrote:

Oops ! My stupid mistake.

vxlan_input() failed to free *m0 after the pullup
however, so if m_pullup() fails or we hit an error case, we'd free the
mbuf twice.

The description is not accurate.

m_pullup() can end up three cases:

1. It fails. Then the original mbuf is freed, and *m0 is set to NULL. The caller will not free a NULL mbuf. Nothing bad will happen.

Yes, my mistake.

2. It succeeds, but no new mbuf allocated. In this case the m == *m0; Nothing bad will happen.

Right.

3. It succeeds with new mbuf allocated. In this case m != *m0, thus will leak memory and free a freed mbuf.

If an error occurs, yes.

[zlei]

Emm, the if_vxlan(4) is not virtualized yet. @hrs attempted the first, me the second, and @kp the third. It is better to proceed to have tests for it.

[markj]

In D56944#1304393, @zlei wrote:

Emm, the if_vxlan(4) is not virtualized yet. @hrs attempted the first, me the second, and @kp the third. It is better to proceed to have tests for it.

Sorry, I don't quite follow what you're asking for?

[zlei]

In D56944#1304407, @markj wrote:

In D56944#1304393, @zlei wrote:

Emm, the if_vxlan(4) is not virtualized yet. @hrs attempted the first, me the second, and @kp the third. It is better to proceed to have tests for it.

Sorry, I don't quite follow what you're asking for?

No, not asking for anything for this revision / review.

the if_vxlan(4) is not virtualized yet

That is a missing feature. See D2868 , D48646 and D52851 .