On 10/7/2015 12:38 PM, Daniel Jakots wrote:
On Wed, 7 Oct 2015 12:18:32 -0700, Scott Vanderbilt
<li...@datagenic.com> wrote:
Might anyone have an idea where I can start to look for the problem?
semarie@ explained how to debug/report tame(2) problem :
https://marc.info/?l=openbsd-tech&m=144412493925465&w=2
The post from tech@ is elucidating, but the diff provided by semarie@ is
bgpd-specific. Sadly, I'm not knowledgeable enough to extrapolate from
that how to patch httpd. If anyone could kindly provide a patch, I will
gladly recompile and generate the core.
The syscall (72) maps to SYS_kevent 72.
Here is the (I believe) relevant extract from the kdump output:
10818 httpd CALL tame(0x10525251dde9,0)
10818 httpd STRU tame request="malloc cmsg"
10818 httpd RET tame 0
10818 httpd CALL kbind(0x7f7fffff4938,0x18,0x67b077efcdf46f05)
10818 httpd RET kbind 0
10818 httpd CALL kbind(0x7f7fffff4938,0x18,0x67b077efcdf46f05)
10818 httpd RET kbind 0
10818 httpd CALL kbind(0x7f7fffff4938,0x18,0x67b077efcdf46f05)
10818 httpd RET kbind 0
10818 httpd CALL clock_gettime(CLOCK_MONOTONIC,0x7f7fffff4920)
10818 httpd STRU struct timespec { 4436<"Dec 31 17:13:56
1969">.154449146 }
10818 httpd RET clock_gettime 0
10818 httpd CALL kevent(17,0x10553aeb3800,4,0x105536daf000,64,0)
10818 httpd PSIG SIGKILL SIG_DFL
30352 httpd RET kevent 2
Thank you.