VFS: iput: trying to free free inode
VFS: device 00:00, inode 0, mode=00000000
and then the file system for $PWD is "busy" and can't be unmounted anymore
(can happen for ext2 and msdos-fs too). sometimes there are strange numbers
in this message like this:
VFS: iput: trying to free free inode
VFS: device 08:11, inode 0, mode=00000000
VFS: iput: trying to free free inode
VFS: device 4f:50, inode 18446739675676151240, mode=063026710
VFS: iput: trying to free free inode
VFS: device 00:00, inode 10580162311059456, mode=00000002
VFS: iput: trying to free free inode
VFS: device 00:00, inode 18446739675724787536, mode=00000000
command line which triggers this is
# /usr/lib/gcc-lib/alpha-linux/2.7.2/cc1 cca00716.i -quiet -dumpbase slip.c -mno-fp-regs -O2 -Wall -Wstrict-prototypes -fomit-frame-pointer -o cca00716.s
/soft/linux/include/linux/skbuff.h: In function `sl_bump':
/soft/linux/include/linux/skbuff.h:392: warning: `tmp' might be used uninitialized in this function
IOT trap (core dumped)
VFS: iput: trying to free free inode
VFS: device 00:00, inode 18446739675724787536, mode=00000000
# ls -l cca00716.*
-rwxr-xr-x 1 root root 155642 Nov 8 22:14 cca00716.i (compressed 33k)
-rwxr-xr-x 1 root root 19901 Nov 8 22:19 cca00716.s
(cca00716.s isn't complete, ends at the end of slip_unesc).
this IOT is 100% reproducable on both machines with both 2.0.22 and 2.1.7 kernel
and I get this VFS errors on 2.1.7 too.
why does the IOT trap happen at all? from strace:
read(4, "rcount < sl->buffsize) {\n\t\t\t"..., 2048) = 2048
write(5, "\t.align 3\n\t.ent slip_unesc\ns"..., 1888) = 1888
read(4, "\"2:\tldq_u %1,0(%5)\\n\"\t\"\ti"..., 2048) = 2048
read(4, "\"\t\"\tlda %0, 3b-1b(%2)\\n\"\t"..., 2048) = 2048
sigprocmask(SIG_UNBLOCK, [ABRT]) = 0 (old mask [])
getxpid() = 290
kill(290, SIGABRT) = 0
--- SIGABRT (IOT trap) ---
+++ killed by SIGABRT +++
does anyone else get this IOT when compiling slip.c too ?
if not, please let me know when you'd like to get a copy of the *.i file
to try to reproduce this...
Harald
PS: this is the first time I compiled an 2.1.x kernel on AXP;
slip.c was no problem ever before up to 2.0.22
--
All SCSI disks will from now on ___ _____
be required to send an email notice 0--,| /OOOOOOO\
24 hours prior to complete hardware failure! <_/ / /OOOOOOOOOOO\
\ \/OOOOOOOOOOOOOOO\
\ OOOOOOOOOOOOOOOOO|//
Harald Koenig, \/\/\/\/\/\/\/\/\/
Inst.f.Theoret.Astrophysik // / \\ \
koenig@tat.physik.uni-tuebingen.de ^^^^^ ^^^^^