Re: git: 29029b06a6f7 - main - routing: remove info argument from add/change_route_nhop().

From: Alexander V. Chernikov <melifaro_at_ipfw.ru>
Date: Wed, 03 Aug 2022 09:18:53 UTC

> On 2 Aug 2022, at 20:41, Cy Schubert <Cy.Schubert@cschubert.com> wrote:
> 
> In message <202208010853.2718rMqh090712@gitrepo.freebsd.org>, "Alexander V. 
> Che
> rnikov" writes:
>> The branch main has been updated by melifaro:
>> 
>> URL: https://cgit.FreeBSD.org/src/commit/?id=29029b06a6f7fd51c34c9db88cc98a66
>> 53ee5046
>> 
>> commit 29029b06a6f7fd51c34c9db88cc98a6653ee5046
>> Author: Alexander V. Chernikov <melifaro@FreeBSD.org>
>> AuthorDate: 2022-07-28 13:41:11 +0000
>> Commit: Alexander V. Chernikov <melifaro@FreeBSD.org>
>> CommitDate: 2022-08-01 07:41:07 +0000
>> 
>> routing: remove info argument from add/change_route_nhop().
>> 
>> Currently, rt_addrinfo(info) serves as a main "transport" moving
>> state between various functions inside the routing subsystem.
>> As all of the fields are filled in directly by the customers, it
>> is problematic to maintain consistency, resulting in repeated checks
>> inside many functions. Additionally, there are multiple ways of
>> specifying the same value (RTAX_IFP vs rti_ifp / rti_ifa) and so on.
>> With the upcoming nhop(9) kpi it is possible to store all of the
>> required state in the nexthops in the consistent fashion, reducing the
>> need to use "info" in the KPI calls.
>> Finally, rt_addrinfo structure format was derived from the rtsock wire
>> format, which is different from other kernel routing users or netlink.
>> 
>> This cleanup simplifies upcoming nhop(9) kpi and netlink introduction.
>> 
>> Reviewed by: zlei.huang@gmail.com
>> Differential Revision: https://reviews.freebsd.org/D35972
>> MFC after: 2 weeks
> 
> Hi,
> 
> Just picking this commit:
> 
> One of these routing commits has caused a panic during setup of gif(4) to 
> Tunnelbroker.
Sorry for the breakage, should be fixed by 08bb0873ca88.
> 
> <118>Created clone interfaces: gif0 bridge0.
> <6>lo0: link state changed to UP
> <6>sk0: link state changed to DOWN
> <6>xl0: link state changed to DOWN
> <6>fxp0: link state changed to DOWN
> <6>nfe0: link state changed to DOWN
> <6>nfe1: link state changed to DOWN
> 
> 
> Fatal trap 12: page fault while in kernel mode
> cpuid = 1; apic id = 01
> fault virtual address = 0x70
> fault code = supervisor read data, page not present
> instruction pointer = 0x20:0xffffffff80806c66
> stack pointer = 0x28:0xfffffe00c43387b0
> frame pointer = 0x28:0xfffffe00c4338810
> code segment = base 0x0, limit 0xfffff, type 0x1b
> = DPL 0, pres 1, long 1, def32 0, gran 1
> processor eflags = interrupt enabled, resume, IOPL = 0
> current process = 513 (ifconfig)
> trap number = 12
> panic: page fault
> cpuid = 1
> time = 1659466745
> KDB: stack backtrace:
> db_trace_self_wrapper() at db_trace_self_wrapper+0x2b/frame 
> 0xfffffe00c4338570
> vpanic() at vpanic+0x151/frame 0xfffffe00c43385c0
> panic() at panic+0x43/frame 0xfffffe00c4338620
> trap_fatal() at trap_fatal+0x387/frame 0xfffffe00c4338680
> trap_pfault() at trap_pfault+0x4f/frame 0xfffffe00c43386e0
> calltrap() at calltrap+0x8/frame 0xfffffe00c43386e0
> --- trap 0xc, rip = 0xffffffff80806c66, rsp = 0xfffffe00c43387b0, rbp = 
> 0xfffffe
> 00c4338810 ---
> nhop_get_nhop() at nhop_get_nhop+0x26/frame 0xfffffe00c4338810
> nhop_create_from_info() at nhop_create_from_info+0x20c/frame 
> 0xfffffe00c4338860
> rib_add_route() at rib_add_route+0x1e9/frame 0xfffffe00c43388e0
> rib_handle_ifaddr_info() at rib_handle_ifaddr_info+0xd7/frame 
> 0xfffffe00c4338950
> in6_update_ifa() at in6_update_ifa+0xb6d/frame 0xfffffe00c4338ad0
> in6_control() at in6_control+0x83c/frame 0xfffffe00c4338bc0
> ifioctl() at ifioctl+0x7bc/frame 0xfffffe00c4338cc0
> kern_ioctl() at kern_ioctl+0x26d/frame 0xfffffe00c4338d30
> sys_ioctl() at sys_ioctl+0x100/frame 0xfffffe00c4338e00
> amd64_syscall() at amd64_syscall+0x10c/frame 0xfffffe00c4338f30
> fast_syscall_common() at fast_syscall_common+0xf8/frame 0xfffffe00c4338f30
> --- syscall (54, FreeBSD ELF64, sys_ioctl), rip = 0x640bee2d97a, rsp = 
> 0x640bb37
> 3718, rbp = 0x640bb373760 ---
> Uptime: 22s
> Dumping 370 out of 5089 MB:..5%..13%..22%..31%..44%..52%..61%..74%..83%..91%
> 
> __curthread () at /opt/src/git-src/sys/amd64/include/pcpu_aux.h:59
> 59 __asm("movq %%gs:%P1,%0" : "=r" (td) : "n" (offsetof(struct 
> pcpu
> ,
> (kgdb) #0 __curthread () at /opt/src/git-src/sys/amd64/include/pcpu_aux.h:5
> 9
> #1 dump_savectx () at /opt/src/git-src/sys/kern/kern_shutdown.c:405
> #2 0xffffffff806be738 in dumpsys (di=0x0)
> at /opt/src/git-src/sys/x86/include/dump.h:87
> #3 doadump (textdump=1) at /opt/src/git-src/sys/kern/kern_shutdown.c:434
> #4 kern_reboot (howto=260) at /opt/src/git-src/sys/kern/kern_shutdown.c:541
> #5 0xffffffff806bec1e in vpanic (fmt=<optimized out>,
> ap=ap@entry=0xfffffe00c4338600)
> at /opt/src/git-src/sys/kern/kern_shutdown.c:979
> #6 0xffffffff806bea53 in panic (fmt=<unavailable>)
> at /opt/src/git-src/sys/kern/kern_shutdown.c:903
> #7 0xffffffff80a34bb7 in trap_fatal (frame=0xfffffe00c43386f0, eva=112)
> at /opt/src/git-src/sys/amd64/amd64/trap.c:942
> #8 0xffffffff80a34c0f in trap_pfault (frame=0xfffffe00c43386f0,
> usermode=false, signo=<optimized out>, ucode=<optimized out>)
> at /opt/src/git-src/sys/amd64/amd64/trap.c:761
> #9 <signal handler called>
> #10 0xffffffff80806c66 in get_aifp (nh=0xfffff8002291a200)
> at /opt/src/git-src/sys/net/route/nhop_ctl.c:138
> #11 nhop_get_nhop (nh=nh@entry=0xfffff8002291a200,
> perror=perror@entry=0xfffffe00c4338834)
> at /opt/src/git-src/sys/net/route/nhop_ctl.c:319
> #12 0xffffffff8080680c in nhop_create_from_info (
> rnh=rnh@entry=0xfffff80004924400, info=info@entry=0xfffffe00c4338a20,
> nh_ret=nh_ret@entry=0xfffffe00c4338890)
> at /opt/src/git-src/sys/net/route/nhop_ctl.c:299
> #13 0xffffffff8080c6a9 in create_rtentry (rnh=0xfffff80004924400,
> info=0xfffffe00c4338a20, prt=<optimized out>)
> at /opt/src/git-src/sys/net/route/route_ctl.c:661
> #14 add_route (rnh=0xfffff80004924400, info=0xfffffe00c4338a20,
> rc=0xfffffe00c43388f8) at /opt/src/git-src/sys/net/route/route_ctl.c:713
> #15 rib_add_route (fibnum=<optimized out>, info=0xfffffe00c4338a20,
> rc=0xfffffe00c43388f8) at /opt/src/git-src/sys/net/route/route_ctl.c:586
> #16 0xffffffff8080d4bf in rib_action (fibnum=579969536, 
> action=action@entry=1,
> info=0xfffffe00c4338834, rc=0xfffff8002291a280,
> rc@entry=0xfffffe00c43388f8)
> at /opt/src/git-src/sys/net/route/route_ctl.c:1257
> #17 0xffffffff8080f3d7 in rib_handle_ifaddr_one (fibnum=0, cmd=1,
> info=0xfffffe00c4338a20)
> at /opt/src/git-src/sys/net/route/route_ifaddrs.c:81
> #18 rib_handle_ifaddr_info (fibnum=579969536, cmd=cmd@entry=1, 
> info=0x100004,
> info@entry=0xfffffe00c4338a20)
> at /opt/src/git-src/sys/net/route/route_ifaddrs.c:107
> #19 0xffffffff8087001d in in6_handle_dstaddr_rtrequest (cmd=1,
> ia=0xfffff80022887480) at /opt/src/git-src/sys/netinet6/in6.c:1307
> #20 in6_notify_ifa (ifp=<optimized out>, ia=<optimized out>,
> ifra=<optimized out>, hostIsNew=<optimized out>)
> at /opt/src/git-src/sys/netinet6/in6.c:1518
> #21 in6_update_ifa_internal (ifp=0xfffff80022721800, 
> ifra=0xfffff8000799e400,
> ia=0xfffff80022887480, hostIsNew=<optimized out>, flags=<optimized out>)
> at /opt/src/git-src/sys/netinet6/in6.c:1196
> #22 in6_update_ifa (ifp=0xfffff80022721800, ifra=0xfffff8000799e400,
> ia=0xfffff80022887480, flags=flags@entry=0)
> at /opt/src/git-src/sys/netinet6/in6.c:899
> #23 0xffffffff8086effc in in6_control (so=<optimized out>,
> cmd=<optimized out>, data=<optimized out>, ifp=<optimized out>,
> td=<optimized out>) at /opt/src/git-src/sys/netinet6/in6.c:572
> #24 0xffffffff807e465c in ifioctl (so=0xfffff80016ffa000, cmd=2156423451,
> data=<optimized out>, td=0x0) at /opt/src/git-src/sys/net/if.c:3189
> #25 0xffffffff807310ad in fo_ioctl (fp=0xfffff8000775cc80,
> com=18446741877978007604, data=0x100004, active_cred=0xfffff8002291a280,
> td=0xfffffe00c44acac0) at /opt/src/git-src/sys/sys/file.h:365
> #26 kern_ioctl (td=0x0, td@entry=0xfffffe00c44acac0, fd=<optimized out>,
> com=18446741877978007604, com@entry=2156423451,
> data=0x100004 <error: Cannot access memory at address 0x100004>,
> data@entry=0xfffff8000799e400 "gif0")
> at /opt/src/git-src/sys/kern/sys_generic.c:803
> #27 0xffffffff80730d90 in sys_ioctl (td=0xfffffe00c44acac0,
> uap=0xfffffe00c44aceb0) at /opt/src/git-src/sys/kern/sys_generic.c:711
> #28 0xffffffff80a354ac in syscallenter (td=0xfffffe00c44acac0)
> at /opt/src/git-src/sys/amd64/amd64/../../kern/subr_syscall.c:189
> #29 amd64_syscall (td=0xfffffe00c44acac0, traced=0)
> at /opt/src/git-src/sys/amd64/amd64/trap.c:1187
> #30 <signal handler called>
> #31 0x00000640bee2d97a in ?? ()
> Backtrace stopped: Cannot access memory at address 0x640bb373718
> (kgdb)
> 
> 
> At the point of panic we see:
> 
> (kgdb) l
> 133		 * the packet back to ourselves, the ifp would be the loopback
> 134		 * interface. However, we'd rather know the interface associated
> 135		 * to the destination address (which should probably be one of
> 136		 * our own addresses).
> 137		 */
> 138		if ((nh->nh_ifp->if_flags & IFF_LOOPBACK) &&
> 139				nh->gw_sa.sa_family == AF_LINK) {
> 140			aifp = ifnet_byindex(nh->gwl_sa.sdl_index);
> 141			if (aifp == NULL) {
> 142				FIB_NH_LOG(LOG_WARNING, nh, "unable to get aifp for %s index %d",
> (kgdb) p nh
> $1 = (const struct nhop_object *) 0xfffff8002291a200
> (kgdb) p nh->nh_ifp
> $2 = (struct ifnet *) 0x0
> 
> 
> My relevant rc.conf lines are:
> 
> if kqcheckyesno use_tunnelbroker; then
> # tunnelbroker
> tunnel_broker="gif0"
> create_args_gif0="tunnel MY_IPv4_ADDR REMOTE_IPv4_ADDR"
> ifconfig_gif0_ipv6="inet6 MY_ASSIGNED_IPv6_TUNNEL_ENDPOINT_ADDR 
> prefixlen 128 mtu 1480"
> ipv6_defaultrouter="MY_IPv6_REMOTE_TUNNEL_ENDPOINT_ADDR"
> ipv6_gateway_enable="YES"
> rtsold_enable="YES" # Set to YES to enable an IPv6 
> router
> rtsold_flags="-F -m gif0" # Flags to an IPv6 router 
> solicitation
> else
> rtsold_enable="NO" # Set to YES to enable an IPv6 
> router
> tunnel_broker=''
> fi
> 
> 
> 
> -- 
> Cheers,
> Cy Schubert <Cy.Schubert@cschubert.com>
> FreeBSD UNIX: <cy@FreeBSD.org> Web: http://www.FreeBSD.org
> NTP: <cy@nwtime.org> Web: https://nwtime.org
> 
> 			e**(i*pi)+1=0