kern/50976: panic: kmem_poison_check (sys___kevent50)

[tnn%NetBSD.org@localhost]

>Number:         50976
>Category:       kern
>Synopsis:       panic: kmem_poison_check (sys___kevent50)
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Mar 18 02:35:00 +0000 2016
>Originator:     Tobias Nygren
>Release:        current
>Organization:
>Environment:
NetBSD 7.99.26 amd64 (in a xen 4.6 HVM hosted on Linux)
>Description:
My pkgsrc-wip NetBSD/amd64 bulk build regularly hits this while running ghc. Regular bulk builds are unaffected, only ghc stuff from pkgsrc-wip triggers this.

panic: kmem_poison_check: 0xfffffe813c6752f8: 0x90 != 0x3c

fatal breakpoint trap in supervisor mode
trap type 1 code 0 rip ffffffff80114e75 cs 8 rflags 246 cr2 7f7ff0b98000 ilevel 0 rsp fffffe810e109a30
curlwp 0xfffffe81ece18700 pid 7065.2 lowest kstack 0xfffffe810e1062c0
Stopped in pid 7065.2 (ghc) at  netbsd:breakpoint+0x5:  leave
db{1}> bt
breakpoint() at netbsd:breakpoint+0x5
vpanic() at netbsd:vpanic+0x1be
panic() at netbsd:panic+0x43
kmem_intr_alloc() at netbsd:kmem_intr_alloc+0x312
kmem_zalloc() at netbsd:kmem_zalloc+0x56
kqueue_register() at netbsd:kqueue_register+0x2b
kevent1() at netbsd:kevent1+0x144
sys___kevent50() at netbsd:sys___kevent50+0x38
syscall() at netbsd:syscall+0xe7
--- syscall (number 435) ---
7f7ff643d5fa:
db{1}>
>How-To-Repeat:
run ghc stuff from pkgsrc-wip (trying to narrow this down)

>Fix:
This bug started to appear after updating from kernel 7.99.21 to kernel 7.99.26.