Re: Weird PUFFS reclaim activity on netbsd-6

[manu%netbsd.org@localhost (Emmanuel Dreyfus)]

Emmanuel Dreyfus <manu%netbsd.org@localhost> wrote:

> And here is an operation on 0xbb99d760, which is a reclaimed node

A posible explanation: perfuse tries to reuse nodes. If the same node is
looked up twice and not reclaimed in between, perfuse return the same
node, like here:

reqid: 251910, opclass 2, optype: PUFFS_VN_LOOKUP, cookie: 0xbb99d690,
    aux: 0xbb91902c, auxlen: 1192, pid: 9960, lwpid: 1
    puffs_cn: "makeinfo", len 8 op LOOKUP (flags 0x84054)
    since previous call: 0.000202
    new 0xbb99d760, type 0x2, size 0x2048, dev 0x817370b
(...)
reqid: 251936, opclass 2, optype: PUFFS_VN_LOOKUP, cookie: 0xbb99d690,
    aux: 0xbb91902c, auxlen: 1192, pid: 9960, lwpid: 1
    puffs_cn: "makeinfo", len 8 op LOOKUP (flags 0x84054)
    since previous call: 0.000199
    new 0xbb99d760, type 0x2, size 0x2048, dev 0x817370b

The kernel did not send a reclaim on 0xbb99d760 before doing the second
lookup. Is it legal to return the same node for both lookup? Rr should I
return a duplicate one (which looks rather inefficient)?

-- 
Emmanuel Dreyfus
http://hcpnet.free.fr/pubz
manu%netbsd.org@localhost