git: a051ca72e281 - main - Introduce m_get3()

[John Baldwin]

On 8/18/21 1:52 AM, Kristof Provost wrote:
> The branch main has been updated by kp:
> 
> URL: https://cgit.FreeBSD.org/src/commit/?id=a051ca72e2815b9bbba1e422f5abf22bc2a01551
> 
> commit a051ca72e2815b9bbba1e422f5abf22bc2a01551
> Author:     Kristof Provost <kp at FreeBSD.org>
> AuthorDate: 2021-08-07 18:02:21 +0000
> Commit:     Kristof Provost <kp at FreeBSD.org>
> CommitDate: 2021-08-18 06:48:27 +0000
> 
>      Introduce m_get3()
>      
>      Introduce m_get3() which is similar to m_get2(), but can allocate up to
>      MJUM16BYTES bytes (m_get2() can only allocate up to MJUMPAGESIZE).
>      
>      This simplifies the bpf improvement in f13da24715.
>      
>      Suggested by:   glebius
>      Differential Revision:  https://reviews.freebsd.org/D31455

Hmm, if this looped the way m_getm2() does I could use this in
icl_cxgbei_conn_pdu_append_data() in sys/dev/cxgbe/cxgbei/icl_cxgbei.c
where the code currently allocates 16k jumbo's by hand and then uses
m_getm2() for the trailer.

One almost wants a variant of m_getm*() where you can give an explicit
"max cluster size" perhaps instead of having to know the magic
implicit sizes for m_getm2 vs m_getm3.

-- 
John Baldwin