ether_sprintf() doesn't work about libuinet HOT 12 CLOSED

yup, it's in kvprintf().

However, the problem is that ether_sprintf() calls sprintf(), and that calls the libc sprintf rather than the kernel print functions:

Breakpoint 4, 0x0000000801c04f64 in snprintf () from /lib/libc.so.7
(gdb) bt
#0 0x0000000801c04f64 in snprintf () from /lib/libc.so.7
#1 0x0000000000449ba3 in ether_sprintf (ap=0x81b40f198 "") at /home/adrian/git/github/erikarn/libuinet/lib/libuinet/../../sys/net/if_ethersubr.c:1169
#2 0x000000000040d568 in uinet_pfil_in_hook_v4 (arg=, m=, ifp=0x49e0b9, dir=457240984, inp=0x4a2a24) at uinet_api.c:1245
#3 0x000000000044d3e5 in pfil_run_hooks (ph=, mp=0x7ffffebf5d80, ifp=0x802ced900, dir=1, inp=0x0) at /home/adrian/git/github/erikarn/libuinet/lib/libuinet/../../sys/net/pfil.c:82
#4 0x000000000046af45 in ip_input (m=0x800710200) at /home/adrian/git/github/erikarn/libuinet/lib/libuinet/../../sys/netinet/ip_input.c:513
#5 0x000000000044c4ec in netisr_dispatch_src (proto=, source=, m=0x800710200) at /home/adrian/git/github/erikarn/libuinet/lib/libuinet/../../sys/net/netisr.c:1013
#6 0x000000000044a9fe in ether_input_internal (ifp=, m=0x800710200, m=0x800710200, m=0x800710200, m=0x800710200, m=0x800710200, m=0x800710200, m=0x800710200, m=0x800710200, m=0x800710200, m=0x800710200, m=0x800710200, m=0x800710200,

from libuinet.

That is a FreeBSD kernel printf extension, but we are using the FreeBSD kernel printf.

from libuinet.

Ah, sprintf... let me look, probably not including the kernel version at the moment, but there's really not much reason not to.

from libuinet.

Yeah I'm giving it a shot atm.

from libuinet.

Interesting. If we include it in libuinet, even the userland code sees the symbol via the kernel definition, not the userland definition?

How'd you make printf() work right?

from libuinet.

That doesn't sound right, let me try to reproduce.

from libuinet.

Ok, I'll go commit something in my repo that you can pull. But yeah, the userland code is calling the kernel code. Different bug. :-P

from libuinet.

What's the behavior you are getting that leads you to think the userland code is seeing the kernel definition?

from libuinet.

I'll open a different bug for that. :)

from libuinet.

I've fixed this in my tree:

erikarn@d541f3a

would you mind just manually pulling that and merging it in?

from libuinet.

hah, seems you already fixed it! Woo!

from libuinet.

Yeah I just committed the repro version I made here. It pulled the balance of the routines in from subr_prf.c, and it didn't have the linkage issue you reported.

from libuinet.