mbufs & EXT_EXTREF

From: Benjamin Close <Benjamin.Close_at_clearchain.com>
Date: Sat, 05 May 2007 00:25:42 +0930
Hi All,
    I'm working on a driver and am trying to use external storage 
previously allocated attached to mbufs.
However, I'm finding the extracted example below seems to decrement the 
system mbuf usage count below zero, causing thing like vmstat -z to wrap 
negatively.

void myfreefunc( void *, void * )
{}
...

struct mbuf *m;
m = m_getcl(M_DONTWAIT, MT_HEADER, M_EXT);
                if (m == NULL) {
                        device_printf(sc->sc_dev,
                            "could not allocate rx mbuf\n");
                        error = ENOBUFS;
                        goto fail;
                }


  /* attach mybuffer to mbuf */
MEXTADD(m, mybuffer, MYBUFFERSIZE,myfreefunc, NULL,0,EXT_EXTREF);
if ((m->m_flags & M_EXT) == 0) {
                    m_freem(m);
                    m=NULL;
                    error = ENOBUFS;
                    goto fail;
                }

/* Free the cluster */
m_freem(m);

Output from vmstat -z after running a equivilant example (this is after 
a number of runs, but even one run decrements from 1 alarge number)


mbuf_packet:              256,        0,      321,      319,    
22923,        0
mbuf:                     256,        0, 18446744073709551553,      
323,    95191,        0
mbuf_cluster:            2048,    25600,      640,       12,      
640,        0
mbuf_jumbo_pagesize:     4096,        0,        1,       43,    
26396,        0
mbuf_jumbo_9k:           9216,        0,        0,        0,        
0,        0
mbuf_jumbo_16k:         16384,        0,        0,        0,        
0,        0
mbuf_ext_refcnt:            4,        0,        0,        0,        
0,        0

Originally I tried m_get as apposed to m_getcl but that panics the 
kernel in m_extadd+0x15
Can anyone shed some light on what I'm missing?

I did notice whilst trying to trace this down, that m_getcl adds to 
zone_pack whilst mb_free_ext (called from m_free) frees from zone_mbuf 
for type EXT_EXTREF and thought that was odd.

Cheers,
    Benjamin
Received on Fri May 04 2007 - 12:56:05 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:09 UTC