Re: kern/54647: panic: ffs_newvnode: dup alloc ino=2782620

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Fri, Oct 25, 2019 at 09:05:01AM +0000, Kamil Rytarowski wrote:
> The following reply was made to PR kern/54647; it has been noted by GNATS.
> 
> From: Kamil Rytarowski <n54%gmx.com@localhost>
> To: gnats-bugs%netbsd.org@localhost
> Cc: 
> Subject: Re: kern/54647: panic: ffs_newvnode: dup alloc ino=2782620
> Date: Fri, 25 Oct 2019 11:01:27 +0200
> 
>  On 25.10.2019 06:50, lawrence_danna%apple.com@localhost wrote:
>  >> Number:         54647
>  >> Category:       kern
>  >> Synopsis:       panic: ffs_newvnode: dup alloc ino=3D2782620
>  >> Confidential:   no
>  >> Severity:       serious
>  >> Priority:       high
>  >> Responsible:    kern-bug-people
>  >> State:          open
>  >> Class:          sw-bug
>  >> Submitter-Id:   net
>  >> Arrival-Date:   Fri Oct 25 04:50:00 +0000 2019
>  >> Originator:     Lawrence D'Anna
>  >> Release:        8.1_STABLE
>  >> Organization:
>  > apple
>  >> Environment:
>  > NetBSD localhost 8.1_STABLE NetBSD 8.1_STABLE (GENERIC) #0: Thu Oct 17 1=
>  8:41:27 UTC 2019  mkrepro%mkrepro.NetBSD.org@localhost:/usr/src/sys/arch/amd64/compi=
>  le/GENERIC amd64
>  >
>  >> Description:
>  > netbsd# panic: ffs_newvnode: dup alloc ino=3D2782620 on /data: mode a1ed=
>  /a1ed gen 3835531f/3835531f size 0 blocks 8
>  > cpu0: Begin traceback...
>  > vpanic() at netbsd:vpanic+0x15d
>  > snprintf() at netbsd:snprintf
>  > ffs_newvnode() at netbsd:ffs_newvnode+0x530
>  > vcache_new() at netbsd:vcache_new+0x95
>  > ufs_makeinode() at netbsd:ufs_makeinode+0x38
>  > ufs_symlink() at netbsd:ufs_symlink+0x3a
>  > VOP_SYMLINK() at netbsd:VOP_SYMLINK+0x58
>  > do_sys_symlinkat.isra.4() at netbsd:do_sys_symlinkat.isra.4+0x219
>  > syscall() at netbsd:syscall+0x1ec
>  > --- syscall (number 57) ---
>  > 79d3d743e64a:
>  > cpu0: End traceback...
>  >
>  >> How-To-Repeat:
>  > I've got a VM I was using to build LLDB that panics like that every time=
>   I start the build back up again.
>  >> Fix:
>  >
>  
>  This is unfortunate that malformed filesystem images panic the kernel.
>  
>  I wrote a fuzzer for this purpose:
>  
>  http://netbsd.org/~kamil/rump/rump_pub_etfs_register_buffer.c
>  
>  It still needs man effort to fix the crashes.

This should be a mount option. I *want* a panic if the filesystem is
corrupted. No linux-like degraded mode please

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--