On Sun, Sep 14, 2025 at 9:15 PM Zaphod Beeblebrox <[email protected]> wrote:
>
> OK.  Thanks.  rpcbind used to be running it wasn't.  Ran rpcbind -d manually 
> and then ran (same machine) rpcinfo localhost.  rpcbind coredumps.  GDB says:
>
> (gdb) bt
> #0  ?? () at /usr/src/lib/libc/amd64/string/strlen.S:95 from /lib/libc.so.7
> #1  0x00001d9c3c5273a8 in xdr_string (xdrs=xdrs@entry=0x3ba5fb06f808, 
> cpp=cpp@entry=0x3ba5fb047460, maxsize=maxsize@entry=9000)
>     at /usr/src/lib/libc/xdr/xdr.c:701
> #2  0x00001d9c3c51b841 in xdr_rpcb (xdrs=0x3ba5fb06f808, objp=0x3ba5fb047450) 
> at /usr/src/lib/libc/rpc/rpcb_prot.c:63
> #3  0x00001d9c3c528df2 in xdr_reference (xdrs=xdrs@entry=0x3ba5fb06f808, 
> pp=pp@entry=0x1d9418ba7140 <list_rbl>,
>     size=size@entry=40, proc=proc@entry=0x1d9c3c51b7f0 <xdr_rpcb>) at 
> /usr/src/lib/libc/xdr/xdr_reference.c:92
> #4  0x00001d9c3c51b943 in xdr_rpcblist_ptr (xdrs=0x3ba5fb06f808, 
> rp=0x1d9418ba7140 <list_rbl>)
>     at /usr/src/lib/libc/rpc/rpcb_prot.c:124
> #5  0x00001d9c3c5205e4 in svc_vc_reply (xprt=0x3ba5fb0208c0, 
> msg=0x1d9c39a3d618) at /usr/src/lib/libc/rpc/svc_vc.c:646
> #6  0x00001d9c3c51c7ee in svc_sendreply (xprt=0x0, xprt@entry=0x3ba5fb0208c0,
>     xdr_results=xdr_results@entry=0x1d9c3c51b870 <xdr_rpcblist_ptr>, 
> xdr_location=0x2328) at /usr/src/lib/libc/rpc/svc.c:368
> #7  0x00001d9418ba0f52 in rpcb_service_3 (rqstp=0x1d9c39a3d740, 
> transp=0x3ba5fb0208c0)
>     at /usr/src/usr.sbin/rpcbind/rpcb_svc.c:175
> #8  0x00001d9c3c51d08d in svc_getreq_common (fd=<optimized out>) at 
> /usr/src/lib/libc/rpc/svc.c:664
> #9  0x00001d9c3c51d118 in svc_getreq_poll (pfdp=pfdp@entry=0x1d9c39a3dcf0, 
> pollretval=1) at /usr/src/lib/libc/rpc/svc.c:732
> #10 0x00001d9418ba3069 in my_svc_run () at 
> /usr/src/usr.sbin/rpcbind/rpcb_svc_com.c:1167
> #11 0x00001d9418b9f835 in main (argc=<optimized out>, argv=0x1d9c39a40000) at 
> /usr/src/usr.sbin/rpcbind/rpcbind.c:286
Yea, its busted alright. I reproduced the crash.
It appears that the r_addr field has not been set to a string.
(My current guess is that commit d50fc4b broke it for the
netlink case.)

I'll poke away at it and post if/when I have a patch for testing.

If you just want to work around it, you could try a NFSv4 mount,
which doesn't use rpcbind.

rick

>
>
> On Sun, Sep 14, 2025 at 8:51 PM Rick Macklem <[email protected]> wrote:
>>
>> On Fri, Sep 12, 2025 at 7:55 PM Zaphod Beeblebrox <[email protected]> wrote:
>> >
>> > My overall goal is to have a FreeBSD workstation before Windows 10 dies.  
>> > That's coming up pretty quickly, so I've been testing.  Because FreeBSD's 
>> > proton support is very much dependant on 15-CURRENT, I upgraded both the 
>> > new workstation and my poudriere box to the 15-current about 2 weeks ago.
>> >
>> > Now... normally, when the workstation boots, half a dozen NFS mounts to 
>> > the poudriere server (also fileserver) occur.  This has been the config 
>> > since ... well... since back to FreeBSD 4 or so.
>> >
>> > My workstation's fstab has lines of the form:
>> >
>> > vr:/vr1/tmp /d/vr/tmp nfs rw,-3,-T,-l,-i,-b 0 0
>> >
>> > 'vr' being the hostname and so on.  On vr,
>> > [2:31:331]root@vr:~> zfs get sharenfs vr1/tmp
>> > NAME     PROPERTY  VALUE                    SOURCE
>> > vr1/tmp  sharenfs  -maproot=nobody ump run  local
>> >
>> > Now... for some number of upgrades, this spits out lines like "-i 
>> > deprecated, use -o intr" ... which it still does --- but then it hangs ... 
>> > for one NFS time out and it moves onto the next share.  No errors on 
>> > either the client or the server.  What's changed here?  Both machines were 
>> > working before the upgrade from 14.3 to 15-c.
>>
>> Those options are from the Jurassic period, but they should still work, 
>> afaik.
>>
>> I suspect your mount problem is something related to networking
>> (firewall or ipv4 vs ipv6 or ???). Note that NFSv3 does need rpcbind
>> to be working.
>>
>> rick

Reply via email to