From owner-freebsd-stable@freebsd.org Sat Jul 22 20:12:37 2017 Return-Path: Delivered-To: freebsd-stable@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id B9855DAC3D1 for ; Sat, 22 Jul 2017 20:12:37 +0000 (UTC) (envelope-from truckman@FreeBSD.org) Received: from gw.catspoiler.org (unknown [IPv6:2602:304:b010:ef20::f2]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client CN "gw.catspoiler.org", Issuer "gw.catspoiler.org" (not verified)) by mx1.freebsd.org (Postfix) with ESMTPS id 6FD1E641BE for ; Sat, 22 Jul 2017 20:12:37 +0000 (UTC) (envelope-from truckman@FreeBSD.org) Received: from FreeBSD.org (mousie.catspoiler.org [192.168.101.2]) by gw.catspoiler.org (8.15.2/8.15.2) with ESMTP id v6MKCT95070706; Sat, 22 Jul 2017 13:12:33 -0700 (PDT) (envelope-from truckman@FreeBSD.org) Message-Id: <201707222012.v6MKCT95070706@gw.catspoiler.org> Date: Sat, 22 Jul 2017 13:12:29 -0700 (PDT) From: Don Lewis Subject: Re: stable/11 r321349 crashing immediately To: unp@ziemba.us cc: freebsd-stable@FreeBSD.org In-Reply-To: MIME-Version: 1.0 Content-Type: TEXT/plain; charset=us-ascii X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 22 Jul 2017 20:12:37 -0000 On 21 Jul, G. Paul Ziemba wrote: > truckman@freebsd.org (Don Lewis) writes: > >>On 21 Jul, G. Paul Ziemba wrote: >>> GENERIC kernel r321349 results in the following about a minute after >>> multiuser boot completes. >>> >>> What additional information should I provide to assist in debugging? >>> >>> Many thanks! >>> >>> [Extracted from /var/crash/core.txt.NNN] >>> >>> KDB: stack backtrace: >>> #0 0xffffffff810f6ed7 at kdb_backtrace+0xa7 >>> #1 0xffffffff810872a9 at vpanic+0x249 >>> #2 0xffffffff81087060 at vpanic+0 >>> #3 0xffffffff817d9aca at dblfault_handler+0x10a >>> #4 0xffffffff817ae93c at Xdblfault+0xac >>> #5 0xffffffff810cf76e at cpu_search_lowest+0x35e >>> #6 0xffffffff810cf76e at cpu_search_lowest+0x35e >>> #7 0xffffffff810d5b36 at sched_lowest+0x66 >>> #8 0xffffffff810d1d92 at sched_pickcpu+0x522 >>> #9 0xffffffff810d2b03 at sched_add+0xd3 >>> #10 0xffffffff8101df5c at intr_event_schedule_thread+0x18c >>> #11 0xffffffff8101ddb0 at swi_sched+0xa0 >>> #12 0xffffffff81261643 at netisr_queue_internal+0x1d3 >>> #13 0xffffffff81261212 at netisr_queue_src+0x92 >>> #14 0xffffffff81261677 at netisr_queue+0x27 >>> #15 0xffffffff8123da5a at if_simloop+0x20a >>> #16 0xffffffff8123d83b at looutput+0x22b >>> #17 0xffffffff8131c4c6 at ip_output+0x1aa6 >>> >>> doadump (textdump=1) at /usr/src/sys/kern/kern_shutdown.c:298 >>> 298 dumptid = curthread->td_tid; >>> (kgdb) #0 doadump (textdump=1) at /usr/src/sys/kern/kern_shutdown.c:298 >>> #1 0xffffffff810867e8 in kern_reboot (howto=260) >>> at /usr/src/sys/kern/kern_shutdown.c:366 >>> #2 0xffffffff810872ff in vpanic (fmt=0xffffffff81e5f7e0 "double fault", >>> ap=0xfffffe0839778ec0) at /usr/src/sys/kern/kern_shutdown.c:759 >>> #3 0xffffffff81087060 in panic (fmt=0xffffffff81e5f7e0 "double fault") >>> at /usr/src/sys/kern/kern_shutdown.c:690 >>> #4 0xffffffff817d9aca in dblfault_handler (frame=0xfffffe0839778f40) >>> at /usr/src/sys/amd64/amd64/trap.c:828 >>> #5 >>> #6 0xffffffff810cf422 in cpu_search_lowest ( >>> cg=0xffffffff826ccd98 , >>> low=>> ff8>) at /usr/src/sys/kern/sched_ule.c:782 >>> #7 0xffffffff810cf76e in cpu_search (cg=0xffffffff826cccb8 , >>> low=0xfffffe085cfa53b8, high=0x0, match=1) >>> at /usr/src/sys/kern/sched_ule.c:710 >>> #8 cpu_search_lowest (cg=0xffffffff826cccb8 , >>> low=0xfffffe085cfa53b8) at /usr/src/sys/kern/sched_ule.c:783 >>> #9 0xffffffff810cf76e in cpu_search (cg=0xffffffff826ccc80 , >>> low=0xfffffe085cfa5430, high=0x0, match=1) >>> at /usr/src/sys/kern/sched_ule.c:710 >>> #10 cpu_search_lowest (cg=0xffffffff826ccc80 , low=0xfffffe085cfa5430) >>> at /usr/src/sys/kern/sched_ule.c:783 >>> #11 0xffffffff810d5b36 in sched_lowest (cg=0xffffffff826ccc80 , >>> mask=..., pri=28, maxload=2147483647, prefer=4) >>> at /usr/src/sys/kern/sched_ule.c:815 >>> #12 0xffffffff810d1d92 in sched_pickcpu (td=0xfffff8000a3a9000, flags=4) >>> at /usr/src/sys/kern/sched_ule.c:1292 >>> #13 0xffffffff810d2b03 in sched_add (td=0xfffff8000a3a9000, flags=4) >>> at /usr/src/sys/kern/sched_ule.c:2447 >>> #14 0xffffffff8101df5c in intr_event_schedule_thread (ie=0xfffff80007e7ae00) >>> at /usr/src/sys/kern/kern_intr.c:917 >>> #15 0xffffffff8101ddb0 in swi_sched (cookie=0xfffff8000a386880, flags=0) >>> at /usr/src/sys/kern/kern_intr.c:1163 >>> #16 0xffffffff81261643 in netisr_queue_internal (proto=1, >>> m=0xfffff80026d00500, cpuid=0) at /usr/src/sys/net/netisr.c:1022 >>> #17 0xffffffff81261212 in netisr_queue_src (proto=1, source=0, >>> m=0xfffff80026d00500) at /usr/src/sys/net/netisr.c:1056 >>> #18 0xffffffff81261677 in netisr_queue (proto=1, m=0xfffff80026d00500) >>> at /usr/src/sys/net/netisr.c:1069 >>> #19 0xffffffff8123da5a in if_simloop (ifp=0xfffff800116eb000, >>> m=0xfffff80026d00500, af=2, hlen=0) at /usr/src/sys/net/if_loop.c:358 >>> #20 0xffffffff8123d83b in looutput (ifp=0xfffff800116eb000, >>> m=0xfffff80026d00500, dst=0xfffff80026ed6550, ro=0xfffff80026ed6530) >>> at /usr/src/sys/net/if_loop.c:265 >>> #21 0xffffffff8131c4c6 in ip_output (m=0xfffff80026d00500, opt=0x0, >>> ro=0xfffff80026ed6530, flags=0, imo=0x0, inp=0xfffff80026ed63a0) >>> at /usr/src/sys/netinet/ip_output.c:655 >>> #22 0xffffffff8142e1c7 in tcp_output (tp=0xfffff80026eb2820) >>> at /usr/src/sys/netinet/tcp_output.c:1447 >>> #23 0xffffffff81447700 in tcp_usr_send (so=0xfffff80011ec2360, flags=0, >>> m=0xfffff80026d14d00, nam=0x0, control=0x0, td=0xfffff80063ba1000) >>> at /usr/src/sys/netinet/tcp_usrreq.c:967 >>> #24 0xffffffff811776f1 in sosend_generic (so=0xfffff80011ec2360, addr=0x0, >>> uio=0x0, top=0xfffff80026d14d00, control=0x0, flags=0, >>> td=0xfffff80063ba1000) at /usr/src/sys/kern/uipc_socket.c:1360 >>> #25 0xffffffff811779bd in sosend (so=0xfffff80011ec2360, addr=0x0, uio=0x0, >>> top=0xfffff80026d14d00, control=0x0, flags=0, td=0xfffff80063ba1000) >>> at /usr/src/sys/kern/uipc_socket.c:1405 >>> #26 0xffffffff815276a2 in clnt_vc_call (cl=0xfffff80063ca0980, >>> ext=0xfffffe085cfa6e38, proc=4, args=0xfffff80026c3bc00, >>> resultsp=0xfffffe085cfa7110, utimeout=...) >>> at /usr/src/sys/rpc/clnt_vc.c:413 >>> #27 0xffffffff8152391c in clnt_reconnect_call (cl=0xfffff80063ca0c00, >>> ext=0xfffffe085cfa6e38, proc=4, args=0xfffff80026c3bc00, >>> resultsp=0xfffffe085cfa7110, utimeout=...) >>> at /usr/src/sys/rpc/clnt_rc.c:271 >>> #28 0xffffffff80e75628 in newnfs_request (nd=0xfffffe085cfa7110, >>> nmp=0xfffff80007e79c00, clp=0x0, nrp=0xfffff80007e79d28, >>> vp=0xfffff80011d9b588, td=0xfffff80063ba1000, cred=0xfffff800118c0100, >>> prog=100003, vers=3, retsum=0x0, toplevel=1, xidp=0x0, dssep=0x0) >>> at /usr/src/sys/fs/nfs/nfs_commonkrpc.c:760 >>> #29 0xffffffff80ee87f1 in nfscl_request (nd=0xfffffe085cfa7110, >>> vp=0xfffff80011d9b588, p=0xfffff80063ba1000, cred=0xfffff800118c0100, >>> stuff=0x0) at /usr/src/sys/fs/nfsclient/nfs_clport.c:952 >>> #30 0xffffffff80ea865c in nfsrpc_accessrpc (vp=0xfffff80011d9b588, mode=63, >>> cred=0xfffff800118c0100, p=0xfffff80063ba1000, nap=0xfffffe085cfa72e0, >>> attrflagp=0xfffffe085cfa73c0, rmodep=0xfffffe085cfa73b4, stuff=0x0) >>> at /usr/src/sys/fs/nfsclient/nfs_clrpcops.c:243 >>> #31 0xffffffff80ed9ec9 in nfs34_access_otw (vp=0xfffff80011d9b588, wmode=63, >>> td=0xfffff80063ba1000, cred=0xfffff800118c0100, >>> retmode=0xfffffe085cfa7540) at /usr/src/sys/fs/nfsclient/nfs_clvnops.c:283 >>> #32 0xffffffff80ecfb64 in nfs_access (ap=0xfffffe085cfa75f8) >>> at /usr/src/sys/fs/nfsclient/nfs_clvnops.c:426 >>> #33 0xffffffff81a539d4 in VOP_ACCESS_APV ( >>> vop=0xffffffff822ff8b8 , a=0xfffffe085cfa75f8) >>> at vnode_if.c:601 >>> #34 0xffffffff80eda726 in VOP_ACCESS (vp=0xfffff80011d9b588, accmode=64, >>> cred=0xfffff800118c0100, td=0xfffff80063ba1000) at ./vnode_if.h:254 >>> #35 0xffffffff80ecb925 in nfs_lookup (ap=0xfffffe085cfa7cf8) >>> at /usr/src/sys/fs/nfsclient/nfs_clvnops.c:1064 >>> #36 0xffffffff81a52a44 in VOP_LOOKUP_APV ( >>> vop=0xffffffff822ff8b8 , a=0xfffffe085cfa7cf8) >>> at vnode_if.c:127 >>> #37 0xffffffff811c6aad in VOP_LOOKUP (dvp=0xfffff80011d9b588, >>> vpp=0xfffffe085cfa8708, cnp=0xfffffe085cfa8730) at ./vnode_if.h:54 >>> #38 0xffffffff811c5b64 in lookup (ndp=0xfffffe085cfa86a8) >>> at /usr/src/sys/kern/vfs_lookup.c:886 >>> #39 0xffffffff811c4aa2 in namei (ndp=0xfffffe085cfa86a8) >>> at /usr/src/sys/kern/vfs_lookup.c:448 >>> #40 0xffffffff810050f0 in do_execve (td=0xfffff80063ba1000, >>> args=0xfffffe085cfa8838, mac_p=0x0) at /usr/src/sys/kern/kern_exec.c:446 >>> #41 0xffffffff810047fa in kern_execve (td=0xfffff80063ba1000, >>> args=0xfffffe085cfa8838, mac_p=0x0) at /usr/src/sys/kern/kern_exec.c:347 >>> #42 0xffffffff810041e2 in sys_execve (td=0xfffff80063ba1000, >>> uap=0xfffff80063ba1538) at /usr/src/sys/kern/kern_exec.c:221 >>> #43 0xffffffff817da5ed in syscallenter (td=0xfffff80063ba1000) >>> at /usr/src/sys/amd64/amd64/../../kern/subr_syscall.c:131 >>> #44 0xffffffff817d9d0b in amd64_syscall (td=0xfffff80063ba1000, traced=0) >>> at /usr/src/sys/amd64/amd64/trap.c:903 >>> #45 >>> #46 0x0000000800d5285a in ?? () >>> Backtrace stopped: Cannot access memory at address 0x7fffffffe7d8 >>> (kgdb) > >>The double fault is a pretty good indication that you overflowed the >>kernel stack. Having ~40 frames on the stack when the fault happened is >>consistent with that. > > First, thank you for this answer. I've been tearing my hair out for > several hours (crash, fsck, tweak, repeat). I haven't done intensive > work with this kernel so hope you'll entertain my simple questions > below. > >>It looks like you are trying to execute a program from an NFS file >>system that is exported by the same host. This isn't exactly optimal >>... > > Perhaps not optimal for the implementation, but I think it's a > common NFS scenario: define a set of NFS-provided paths for files > and use those path names on all hosts, regardless of whether they > happen to be serving the files in question or merely clients. > >>Your best bet for a quick workaround for the stack overflow would be to >>rebuild the kernel with a larger value of KSTACK_PAGES. You can find >>teh default in /usr/src/sys//conf/NOTES. > > By the way, my host is amd64, not i386. I don't know if that matters > to the larger discussion, but I'm not running in a resource-constrained > situation (i.e., the host has Xeon E3-1231 (4 cores, 8 threads) and > 32GB ram) > > It seems (from /usr/include/machine/param.h) that PAGE_SIZE is 2048, > so the default KSTACK_PAGES=4 yields a stack size of 8192. Is that > right? Page size is 4096. There is a read-only sysctl that shows the current value of kstack_pages: # sysctl kern.kstack_pages kern.kstack_pages: 4 It's interesting that you are running into this on amd64. Usually i386 is the problem child. Since the problem crops up in the scheduler, which is preparing to switch to another process (with a stack switch), I suspect that you aren't missing the limit by much. Try using the tunable to bump kstack_pages by one. I didn't see the tunable earlier since it is handled in a separate .c file from where KSTACK_PAGES and the kstack_pages sysctl appear in the source. > It's not clear to me how to determine the correct value of KSTACK_PAGES > from the stack trace above. If I merely subtract frame 44 from frame 0, > I get 817d9d0b - 810f6ed7 = 6e2e34 = 7,220,788 decimal. That can't be > right. I must be misunderstanding something. > >>It would probably be a good idea to compute the differences in the stack >>pointer values between adjacent stack frames to see of any of them are >>consuming an excessive amount of stack space. > > Hmm. I am assuming the first number for each frame is the stack pointer. > Yet, sometimes that value increases with call depth (see, e.g., when > if_simloop calls netisr_queue above, but also other places). What is > going on? Nope, it's the pc value at each call location. Point kgdb at the vmcore file and execute "i r rsp" (short for info registers) in each stack frame. When the stack overflows, I think there is a trap that causes a switch to an alternate stack for the double fault handler. > By the way, I have net.isr.dispatch=deferred (set earlier while I was > chasing this crash). > > thanks!