Re: "Invalid Media Format" error

From: John Laird (nospam_at_laird-towers.org.uk)
Date: 06/16/04 

Date: Wed, 16 Jun 2004 14:12:00 +0100

On Wed, 16 Jun 2004 11:58:38 +0000 (UTC), John F
<john@SeeSigForAddress.invalid.com> wrote:

>Thanks very much for explanation, John. At least now I know what's
>wrong ( a netbsd guy describes the same problem and similar fix at
>http://mail-index.netbsd.org/netbsd-bugs/2001/08/29/0003.html ), and
>even how to fix it. But I do wish Everhart had taken Dachtera's
>follow-up suggestion and automated the procedure. If your notes
>documenting typos are in machine-readable form, could I trouble you
>once again to post them? I think I can pretty much follow Glenn's
>outlined procedure, but I won't know what I'm doing. Thanks again,

You're not going to catch me being overly-critical of Glenn as he has been
very generous with his time in dealing with this and other issues I have
raised, where he has been far better informed than me ;-)

For some reason, I wrote up what I did. Amazing. This is an addendum to
Glenn's original article, then. You will need a binary editor or suitable
patch utility. These may vary in appearance, use left-right or right-left,
but the principles should be the same. Note that I patched version 2 of
sys$dkdriver - version 1, as shipped with VMS6.2, had the same data
structures but apparently lacked code elsewhere in the TCQ area, and version
2 was part of a SCSI ECO kit.

Sadly, Alpha VMS lacks PATCH... If you really know your stuff, you could
use a text editor. EDT will preserve file characteristics if used
carefully, but you'll need to "shorten" the file record length (to 128
bytes, say) and restore it afterwards with SET FILE/ATTR (although I think
the VMS image activator doesn't care, but it'll be easier to check your
patches with DIFF if you do).

Note this may be a futile exercise if you intend to use the disk as a
bootable system disk. There may well be a level of boot driver that won't
play ball.

---------------------------------------------------------------------------

Glenn's description quotes the wrong hex longword value preceding the "GENE"
entry. It is not 9203000 but 36070000 (offset 220). However, his "patch"
command does modify the right byte. In fact, the correct value of 36070000
has clearly already been patched !

Note that his vfe dump format is reversed from VMS DUMP format, in that it
puts low-byte first and dumps longwords left-to-right.

Although I wished to patch an Alpha driver, I also have a Vaxstation and
found it easier to use VMS PATCH. Here is the journal file from my
editing session. (I had previously used SEARCH/NUM/FORM=NONULL to identify
the correct block, and the DUMP to find the "GENERIC" entry as a byte offset
within that block, seemingly byte 00CC within block 81).

Comments have been added, and for authenticity the journal file even shows
an erroneous command...

-----------------------------------------------------------------------------

PATCH Version 6.2 5-MAY-1994

IMAGE FILE BEING PATCHED: "SYS$SYSDEVICE:[JOHN]SYS$DKDRIVER.EXE;2"
JOURNAL FILE: "SYS$SYSDEVICE:[JOHN]SYS$DKDRIVER.JNL;1"
DATE/TIME OF PATCH: 12-JAN-2001 15:46:24.47

%PATCH-I-NOGBL, some or all global symbols not accessible
%PATCH-I-NOLCL, image does not contain local symbols
PATCH>set rad dec ! DEBUG command not recognised :-)
%PATCH-W-SYNTAX, command syntax error at or near ' rad dec'
PATCH>eval 80*200
00010000
PATCH>exa 0100cc
000100CC: 00000336 ! Note VFE shows this as 36030000
PATCH>exa
000100D0: 454E4547
PATCH>ex/asc .
000100D0: 'GENE' ! Here's the GENERIC disk entry
PATCH>exa 0100cc
000100CC: 00000336 ! Flags in preceding longword
PATCH>ex/byte 100cc
000100CC: 36
PATCH>ex/byte ! And this is the byte we are interested in
000100CD: 03
PATCH>rep/byte
LOC> .
OLD> 03
OLD> EXIT
NEW> 07
NEW> EXIT
old: 000100CD: 03
new: 000100CD: 07
PATCH>update
%PATCH-I-WRTFIL, updating image file SYS$SYSDEVICE:[JOHN]SYS$DKDRIVER.EXE;3
PATCH>exit 

-- 
Just when I make ends meet, someone moves one end. 
Mail john rather than nospam...

Relevant Pages

  • Re: "Invalid Media Format" error
    ... John Laird wrote: ... Sadly, Alpha VMS lacks PATCH... ... Note that his vfe dump format is reversed from VMS DUMP format, ...
    (comp.sys.dec)
  • Re: VMS listed first on HP Web site! VMS mentioned 17 times on the same
    ... > impressed by the VMS presence. ... And, ooh look, it appears that the Windows update process is flawed too. ... Windows Update Patch Process Faulty, ... The problem is that a system may have the registry keys associated ...
    (comp.os.vms)
  • Re: OT: security flaw in Solaris and Trusted Solaris
    ... >> well Andrew, VMS would have handled this a little differently ... ... The Solaris sadmin deamon is more closely comparable to VMS' ... Unless the wording in that patch is extremely clear then ...
    (comp.os.vms)
  • Re: BACKUP causing alpha node to crash
    ... nomenclature and hiearchy of patch names. ... kits "sequential" as you suggest would be to give all the patch kits for a ... the VMS build. ... UPDATE kits have been used for quite a few releases now. ...
    (comp.os.vms)
  • Re: The GNU Ada compiler
    ... >> I presume you mean something other than the standard VMS utilities ... >> PATCH and DIFFERENCES. ... You get them for VMS as well via GNV. ... Unix diff is more like the VMS utility, though one of its features is ...
    (comp.lang.ada)
Quantcast