Strange bug: cp -r to msdos disk seg faults under 1.3.32

david@cs.uow.edu.au
Wed, 11 Oct 1995 10:56:22 +1000 (EST)


My system will reliably cause cp to seg fault and leave /dev/fd1
busy/unmountable when I do the following under 1.3.32:

# mount /dev/fd1 /mnt # 1.44 drive with msdos floppy
# cp -r /usr/src/linux/fs/msdos /mnt
^^^^- an example - other directories do the same.

If I repeat using 1.2.13 there is no seg fault and no problem unmounting /mnt.
If I just cp files rather than using -r to copy directories it is fine.
After the fault I rebooted into dos and ran scandisk on the floppy with no
errors found.

I am running Slackware 2.3 from the Aug 95 InfoMagic set. Both kernels are
identically configured using a.out and no modules. I have 2 hds on a holtek
eide vlb card with the 2nd port disabled, fd-950 scsi card configured but
no devices connected at the moment.

The seg fault info is:

general protection: 0000
CPU: 0
EIP: 0010:00151a52
EFLAGS: 00010202
eax: 00002de0 ebx: 0 ecx: 0149bfb4 edx: 014dcee0
esi: ffffffff es: 0018 ebp: 1 esp: 01508df4
ds: 0018 es: 0018 fs: 002b gs: 002b ss: 0018
Process cp (pid: 80, process nr: 3, stackpage=01508000)
Stack: 0 2bf 1e2310 2bd 149c600 0 149c600 149bfb4
151b93 1e2310 2de 0 0 ffffffff 1508edc 1508ed8
1fa 0 0 1e2310 151c0c 1e2310 1fa 0
Call Trace: 151b93 151c0c 151e23 14dca6 14de4b 14dea3 12b785
120410 120542 10a829 1b002b
Code: 89 06 8b 7c 24 18 0f b7 6c 1f 1a 83 7c 24 38 00 75 18 8b 4c

Can anyone reproduce this problem or suggest what might have changed?
If necessary I could go back and build an earlier 1.3 kernel to try and locate
where the change occurs.

-- 
David Wilson	Dept CompSci Uni Wollongong Australia	david@cs.uow.edu.au