SUMMARY: rcp-ing to /dev/audio crash and Floppy om SLC

From: bernards@ecn.nl
Date: Tue Mar 12 1991 - 05:04:07 CST


Hi again Sun MGR's,
 
This is a brief summary about the rcp to a remote device causing the remote
system crashing and about the SLC and hooking up a floppy drive.
 
First , the most important issue: the crashing system caused by rcp-ing
an audio file to a remote Sparc.
 
Several "Hey , this happens to me too" replies , sounds as a kind of security
leak if someone does this to badge his collegue :-( ....
Others mentioned that rsh and dd buffers the data stream ond rcp does not.
I forgot to tell you about the SunOS version: It was the newest 4.1.1 version
in all cases. SDST60 kernels and no fancy stuff. In both cases however, there
was a PostScript laser attatched running transcript software.
At the moment we copied the audio file, there was nothing active on the lasers.
 
Jan mentioned:
 
>/dev/audio is a *device*, not an ordinary *file*. You try to copy a file on >t
   op of a device that the kernel may try to access.
>Doesn't sound too good to me.
 
Clyde mentioned:
 
>Since a lot of SunOS programs now use memory mapped file I/O, what
>probably happened was that the "cp" program at the recieving end of
>rcp opened /dev/audio and attempted to map the device into its
>address space with mmap and BOOM! I can't confirm this, but it seems
>a likely scenario.
>I guess that 'cat' works because neither the shell (doing
>the "> /dev/audio") nor dd attempts to use memory-mapped I/O.
 
Another Steve Mentioned:
 
> Rcp can handle device special files just fine. The problem here is a
>bug in either some of the STREAMS code or in the ftruncate system call; I
>ever got a chance to track that down further. The crash happens when rcp
>does a ftruncate on /dev/audio. (/Dev/audio is some sort of STREAMS-related
>file, as is /dev/nit, /dev/spx, and /dev/tcp.) I may even have sent a bug
>report to Sun about this; I don't remember offhand.
 
> I just rcped a file to /dev/ttyp0 just fine.
 
Kurt Mentioned:
 
>This is a bug in the kernel. The rcp "closes" the file with a call
>to ftruncate(), which the kernel then panic()'s on. You lose.
 
>I think it is fixed in 4.1.1 -- I know that it affect the machines under
>SunOS 4.0.3 and 4.1.
 
Unfortunately it is not fixed in 4.1.1 as it seems :-(
 
chucks@iplmail.orl.mmc.com chuck strickland
jan@eik.ii.uib.no Jan
clyde@sirius.cc.utexas.edu Clyde Hoover
pacacc!steve@sacto.west.sun.com Steve Riley
steve@umiacs.umd.edu Another Steve
synergy!kevin@sun.com Kevin Sheehan
xphoton!midi!miker@sunkist.west.sun.com
lidl@eng.umd.edu Kurt lidl ?
 
Now about the SLC Floppy drive:
 
Artecon does indeed sells a SCSI floppy Pizzabox with a 3 1/2" and a 5 1/4" FD.
included with ArtFD 1.2.2 driver software to access them as UNIX and DOS
floppies. Cost about US $1300,-
 
Steven Giuliano <steve@ibx.com> send me a summary I missed on SMGR's, Thanx
The Summary mentions several SCSI Floppy vendors and they are all pretty
expensive. ( > $900 )
 
Thanx all who replied to me
 
Greetings,
 
Marcel Bernards, UNIX & Net sysadm Netherlands Energy Research Foundation ECN
(and SURFnet IP/ICP), Phone: (+31 /0)2246 4579 Fax: (+31 /0)2246 1864
E-Mail: Bernards@ECN.NL, SnailMail: P.O. Box 1, 1755 ZG Petten
SCREAMNet : AAAAAARGHH!HUH?? : Disclaimer: "The AntiChrist is the Computer !"



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:11 CDT