Try to load kernel object file
Joao Barros yazm:
> On 6/11/06, Vadim Goncharov <email@example.com> wrote:
>> 11.06.06 @ 22:36 Joao Barros wrote:
>> Original message is at:
>> > I'm very interested in this, great work! :-)
>> > I can't load the kld on my Sun Sparc, I think I messed up ld yesterday
>> > trying to patch for a bug that show's in firefox and mozilla. It
>> > compiles, just doesn't run. As soon as I have it up and running I'll
>> > give you feedback.
>> Umm, that's a kernel module, it shouldn't have any relations with ld.
>> diagnostics has it said on failed load?
> ultra5# make
> Warning: Object directory not changed from original /root/ng_tag
> @ -> /usr/src/sys
> machine -> /usr/src/sys/sparc64/include
> touch opt_netgraph.h
> cc -O2 -pipe -g -fno-strict-aliasing -Werror -D_KERNEL -DKLD_MODULE
> -nostdinc -I- -I/root/ng_tag -I. -I@ -I@/contrib/altq -I@/../include
> -I/usr/include -finline-limit=15000 -fno-common -mcmodel=medlow
> -msoft-float -ffreestanding -Wall -Wredundant-decls -Wnested-externs
> -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Winline
> -Wcast-qual -fformat-extensions -std=c99 -c ng_tag.c
> ld -d -warn-common -r -d -o ng_tag.kld ng_tag.o
> touch export_syms
> awk -f /sys/conf/kmod_syms.awk ng_tag.kld export_syms | xargs -J%
> objcopy % ng_tag.kld
> ld -Bshareable -d -warn-common -o ng_tag.ko ng_tag.kld
> objcopy --strip-debug ng_tag.ko
> ultra5# kldload ./ng_tag.kld
> kldload: can't load ./ng_tag.kld: Exec format error
> ultra5# file ng_tag.kld
> ng_tag.kld: ELF 64-bit MSB relocatable, SPARC V9, version 1 (FreeBSD),
> not stripped
>> > Have you tested it with pf? If so can you give me some examples?
>> No, it wasn't tested with pf. The problem with pf is that pf compiles
>> the rules at the time, so exact tags representation can change each time
>> (for this reason ipfw tags were made incompatible with pf), and you must
>> that values to supply them to . However, if you find a method how to
>> obtain tag values info from in-kernel pf structures, you'll be able
>> to use
>> it with pf. It doesn't support well integration with netgraph, though.
>> Another option is to use ipfw - it supports pf's altq(4) shaping, if
>> is all you need.
>> > I'm particularly interested in this for doing packed shaping,
>> > on P2P.
>> Yes, I'm also looking for possibility of shaping, but I can't test (no
>> resources) it currently. Also, as it seems non-trivial on current ipfw
>> dynamic rules implementation, I don't know if shaping will work at all.
> I'm not a ipfw user, but if this were to be possible it would be very
> nice :-)