Re: kern/44651 (mountd panics nullfs on large disk)

[Hauke Fath <hf%spg.tu-darmstadt.de@localhost>]

The following reply was made to PR kern/44651; it has been noted by GNATS.

From: Hauke Fath <hf%spg.tu-darmstadt.de@localhost>
To: gnats-bugs%NetBSD.org@localhost, kern-bug-people%NetBSD.org@localhost, gnats-admin%NetBSD.org@localhost
Cc: 
Subject: Re: kern/44651 (mountd panics nullfs on large disk)
Date: Fri, 22 Feb 2019 15:28:35 +0100

 FTR, this bug just reared its ugly head again... on the same machine, 
 running netbsd-7 from November 2018, re-purposed as nfs server for 
 sources, and Radmind server.
 
 I fscked the three partitions, dumped them to a sata disk, doubled the 
 size of the raid 10 array, and restored. Booting, I was greeted with
 
 NetBSD 7.2_STABLE (MONOLITHIC) #1: Wed Nov  7 12:01:16 CET 2018
  
 hf@Hochstuhl:/var/obj/netbsd-builds/7/amd64/sys/arch/amd64/compile/MONOLITHIC
 total memory = 4094 MB
 avail memory = 3956 MB
 
 [...]
 
 Starting sendmail.
 Starting smmsp.
 Starting radmind.
 Starting inetd.
 Starting cron.
 Fri Feb 22 11:04:47 CET 2019
 ifree: dev = 0xa804, ino = 7935387, fs = /u
 panic: ifree: freeing free inode
 cpu0: Begin traceback...
 vpanic() at netbsd:vpanic+0x13c
 snprintf() at netbsd:snprintf
 ffs_mapsearch() at netbsd:ffs_mapsearch
 ffs_freefile() at netbsd:ffs_freefile+0xfd
 ffs_reclaim() at netbsd:ffs_reclaim+0x15a
 VOP_RECLAIM() at netbsd:VOP_RECLAIM+0x2f
 vclean() at netbsd:vclean+0xbd
 vrelel() at netbsd:vrelel+0x636
 ufs_fhtovp() at netbsd:ufs_fhtovp+0x54
 ffs_fhtovp() at netbsd:ffs_fhtovp+0x4c
 VFS_FHTOVP() at netbsd:VFS_FHTOVP+0x1c
 layerfs_fhtovp() at netbsd:layerfs_fhtovp+0x21
 VFS_FHTOVP() at netbsd:VFS_FHTOVP+0x1c
 nfsrv_fhtovp() at netbsd:nfsrv_fffffhtovp+0x68
 nfs_namei() at netbsd:nfs_namei+0x154
 nfsrv_lookup() at netbsd:nfsrv_lookup+0x264
 do_nfssvc() at netbsd:do_nfssvc+0x347
 syscall() at netbsd:syscall+0x9a
 --- syscall (number 155) ---
 7f7ff703cb3a:
 cpu0: End traceback...
 
 dumping to dev 168,2 (offset=8391975, size=1048155):
 dump area improper
 
 
 uvm_fault(0xfffffe812e5048a8, 0x0, 2) -> e
 fatal page fault in supervisor mode
 trap type 6 code 2 rip ffffffff8061e38f cs 8 rflags 10286 cr2 84 ilevel 
 8 rsp fffffe8040963e00
 curlwp 0xfffffe812e4c3320 pid 279.1 lowest kstack 0xfffffe80409602c0
 rebooting...
 
 
 Just like before, fsck(8) did not find anything wrong with /u, which is 
 a 1000 GB partition now, about half full. The panic was reproducible.
 
 /etc/fstab null mounts are:
 
 /u/sources/netbsd-5/src         /public/netbsd-5        null    ro,hidden
 /u/sources/netbsd-6/src         /public/netbsd-6        null    ro,hidden
 /u/sources/netbsd-7/src         /public/netbsd-7        null    ro,hidden
 /u/sources/netbsd-8/src         /public/netbsd-8        null    ro,hidden
 /u/sources/netbsd-developer/src /public/netbsd-developer null   ro,hidden
 /u/sources/pkgsrc               /public/pkgsrc          null    ro,hidden
 #
 /u/packages/distfiles           /public/pkg-distfiles   null    rw,hidden
 #
 /u/sources/freebsd-11/src       /public/freebsd-11      null    ro,hidden
 
 % du -sh /public/*
 2.9G    /public/freebsd-11
 4.0K    /public/netbsd-5
 2.7G    /public/netbsd-6
 3.2G    /public/netbsd-7
 3.2G    /public/netbsd-8
 4.3G    /public/netbsd-developer
 10G     /public/pkg-distfiles
 2.2G    /public/pkgsrc
 %
 
 After disabling the largest of the null mounts (/public/pkg-distfiles) 
 the machine booted fine.
 
 For lack of a better idea, I upgraded to -8, and the panic does not 
 occur there.