topic Re: Unmessage in Operating System - OpenVMS

Unmessage

David R. Lennon — Mon, 13 Feb 2012 21:31:32 GMT

Hi,

Has anyone successfully gotten the unmessage utility to work under Alpha or Integrity VMS? And willing to share their code?

I mean the unmessage program by Joe Meadows that was on DECUS tapes and on ftp sites. This program "de-compiles" system message .EXE files into their source .MSG files.

I found this thread:

http://h30499.www3.hp.com/t5/Languages-and-Scripting/Changing-the-output-of-the-TYPE-command/td-p/4718882

But I am unable to successfully compile the source code attached to that thread, as well..

Thanks,

Dave

Re: Unmessage

H.Becker — Mon, 13 Feb 2012 21:56:47 GMT

What's the problem compiling the sources?

In unmsg.h, did you remove the GNU C __attribute__?

$ cc/vers

HP C V7.1-015 on OpenVMS Alpha V8.3

$ sh symb cc

%DCL-W-UNDSYM, undefined symbol - check validity and spelling

$ cc unmsg

$ link unmsg

Re: Unmessage

David R. Lennon — Mon, 13 Feb 2012 22:02:37 GMT

Thanks for the reply. Yes, but I still get this:

$ cc/vers

HP C V7.3-010 on OpenVMS Alpha V8.3

$ cc unmsg

struct mscdef

%CC-W-DUPTYPESPEC, In this declaration, the same type specifier occurs more than once.

at line number 12 in file TS_DEVICE:[LENNON.UNMSG]UNMSG.H;2

if (midx->midx$b_sanity != MIDX$C_SANITY)

......^

%CC-W-INCOMPDEREF, In this statement, "midx" is a pointer to an incomplete struct or union and should not be used as the left operan

d of a member dereference.

at line number 69 in file TS_DEVICE:[LENNON.UNMSG]UNMSG.C;1

if (midx->midx$b_sanity != MIDX$C_SANITY)

......^

%CC-E-NEEDMEMBER, In this statement, "midx$b_sanity" is not a member of "midx".

at line number 69 in file TS_DEVICE:[LENNON.UNMSG]UNMSG.C;1

for (i = 1; i < (midx->midx$w_size / MIDX$C_LENGTH); ++i)

...................^

%CC-W-INCOMPDEREF, In this statement, "midx" is a pointer to an incomplete struct or union and should not be used as the left operan

d of a member dereference.

at line number 74 in file TS_DEVICE:[LENNON.UNMSG]UNMSG.C;1

for (i = 1; i < (midx->midx$w_size / MIDX$C_LENGTH); ++i)

...................^

%CC-E-NEEDMEMBER, In this statement, "midx$w_size" is not a member of "midx".

at line number 74 in file TS_DEVICE:[LENNON.UNMSG]UNMSG.C;1

for (i = 1; i < (midx->midx$w_size / MIDX$C_LENGTH); ++i)

.....................^

%CC-W-INCOMPDEREF, In this statement, "midx" is a pointer to an incomplete struct or union and should not be used as the left operan

d of a member dereference.

at line number 98 in file TS_DEVICE:[LENNON.UNMSG]UNMSG.C;1

for (i = 1; i < (midx->midx$w_size / MIDX$C_LENGTH); ++i)

.....................^

%CC-E-NEEDMEMBER, In this statement, "midx$w_size" is not a member of "midx".

at line number 98 in file TS_DEVICE:[LENNON.UNMSG]UNMSG.C;1

If I use the original unmessage.h header file, it does compile but running it against a message file results in no output.

Thanks,

Dave

Re: Unmessage

H.Becker — Mon, 13 Feb 2012 22:19:17 GMT

I have no problems if I use the original unmessage.h:

$ rena UNMESSAGE.H UNMSG

$ cc unmsg

How is midx defined? It should be struct midxdef *midx; in unmsg.c, in function decompile_msc. midxdef is defined in header files, unmsg.h and unmessage.h.

Re: Unmessage

David R. Lennon — Mon, 13 Feb 2012 22:26:34 GMT

It seemed like there was some more stuff in unmsg.h than there was in unmessage.h but yes, if I use that one it compiles, but when I try to use it, I don't get any output:

$ mcr []unmsg sys$message:usb$msg.exe

I added some debugging printfs and it is "failing" at line 65:

if (msc->msc$w_sanity != MSC$C_SANITY)

return 0;

Any ideas on this?

Thanks again,

Dave

Re: Unmessage

Hoff — Mon, 13 Feb 2012 22:45:17 GMT

Attached is some working code for unmessage, tested against OpenVMS Alpha V8.3.

$ cc/ver
HP C V7.3-010 on OpenVMS Alpha V8.3
$

This is a one-source-module configuration, with no (non-system) headers required.

A couple of minor source code changes were incorporated.

Tested against a V8.3 message file.

Ah, and HPSC just keeps blowing up with ITRC-style blank-pages when posting. This is a reposting. Hopefully. This time, with a zipped file, in place of a text file attachment. (edit: yep; the zip worked. It posted. If you can't unpack that, I'll post the text version else-host and will post a pointer here.)

Re: Unmessage

H.Becker — Mon, 13 Feb 2012 23:17:54 GMT

>>> It seemed like there was some more stuff in unmsg.h than there was in unmessage.h

Yes, the pragmas to make sure the struct members are no member aligned (and some other stuff to make it compile on Linux as well).

It works for me (on Alpha, I don' have access to an Integrity):

$ mcr []unmsg sys$message:usb$msg.exe

.FACILITY USB,1905 /SYSTEM

.SEVERITY WARNING

_07718008 "fatal error encountered; operation terminated" /IDENT=ABORT

.BASE 100

_07718320 "USB Invalid device handle" /IDENT=INVALID_DEVICE

_07718328 "USB Invalid interface handle" /IDENT=INVALID_INTERFA

_07718330 "USB Invalid pipe handle" /IDENT=INVALID_PIPE

...

Dump the message file, starting with VBN 2, there should be the 7B (the #define MIDX$C_SANITY 123) at offset 42.

[edit: the struct mscdef needs to have __nomember_alignment]