am I the only one hacking VMS on Itanium? - VMS

This is a discussion on am I the only one hacking VMS on Itanium? - VMS ; Just curious. I've been trying to link an execlet and I've encountered not only a Macro compiler issue but now it looks like a linker problem. I can't be the only one with these issues... unless I'm the only hacker ...

+ Reply to Thread
Results 1 to 15 of 15

Thread: am I the only one hacking VMS on Itanium?

  1. am I the only one hacking VMS on Itanium?

    Just curious. I've been trying to link an execlet and I've encountered
    not only a Macro compiler issue but now it looks like a linker problem.
    I can't be the only one with these issues... unless I'm the only hacker
    still left on planet VMS.

    --
    VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)COM

    .... pejorative statements of opinion are entitled to constitutional protection
    no matter how extreme, vituperous, or vigorously expressed they may be. (NJSC)

    Copr. 2008 Brian Schenkenberger. Publication of _this_ usenet article outside
    of usenet _must_ include its contents in its entirety including this copyright
    notice, disclaimer and quotations.

  2. Re: am I the only one hacking VMS on Itanium?

    On Sep 16, 1:28 pm, VAXman- @SendSpamHere.ORG wrote:
    > Just curious. I've been trying to link an execlet and I've encountered
    > not only a Macro compiler issue but now it looks like a linker problem.
    > I can't be the only one with these issues... unless I'm the only hacker
    > still left on planet VMS.
    >
    > --
    > VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)COM
    >
    > ... pejorative statements of opinion are entitled to constitutional protection
    > no matter how extreme, vituperous, or vigorously expressed they may be. (NJSC)
    >
    > Copr. 2008 Brian Schenkenberger. Publication of _this_ usenet article outside
    > of usenet _must_ include its contents in its entirety including this copyright
    > notice, disclaimer and quotations.



    Brian,

    There are not all that many people who write execlets.For that matter,
    there are not that many of us who still use MACRO). What kind of issue
    are you encountering?

    - Bob Gezelter, http://www.rlgsc.com

  3. Re: am I the only one hacking VMS on Itanium?

    On 2008-09-16, VAXman- @SendSpamHere.ORG wrote:
    > Just curious. I've been trying to link an execlet and I've encountered
    > not only a Macro compiler issue but now it looks like a linker problem.
    > I can't be the only one with these issues... unless I'm the only hacker
    > still left on planet VMS.


    I've done a device driver, but it was in C. And I didn't figure out how
    to link it; I was busy, so one of my co-workers did. I haven't looked at
    the linker options file he came up with in detail, but a quick glance
    looked pretty similar to the Alpha one. I don't have it handy to look at
    just at the moment.
    --
    roger ivie
    rivie@ridgenet.net

  4. Re: am I the only one hacking VMS on Itanium?

    VAXman- @SendSpamHere.ORG writes:

    >Just curious. I've been trying to link an execlet and I've encountered
    >not only a Macro compiler issue but now it looks like a linker problem.
    >I can't be the only one with these issues... unless I'm the only hacker
    >still left on planet VMS.


    I'm doing Macro work on Itanium and we've found odd Macro problems, but
    the same problems exist on the Alpha macro compiler so I think we're the
    first to encounter them since the Alpha came out.

  5. Re: am I the only one hacking VMS on Itanium?

    In article , moroney@world.std.spaamtrap.com (Michael Moroney) writes:
    >VAXman- @SendSpamHere.ORG writes:
    >
    >>Just curious. I've been trying to link an execlet and I've encountered
    >>not only a Macro compiler issue but now it looks like a linker problem.
    >>I can't be the only one with these issues... unless I'm the only hacker
    >>still left on planet VMS.

    >
    >I'm doing Macro work on Itanium and we've found odd Macro problems, but
    >the same problems exist on the Alpha macro compiler so I think we're the
    >first to encounter them since the Alpha came out.


    I reported one recently to John Reagan... Here's a simple reproducer:

    .LIBRARY "SYS$LIBRARY:LIB.MLB"

    DECLARE_PSECT EXEC$NONPAGED_CODE
    ABC: .CALL_ENTRY
    MOVL DATA,R0
    RET

    ;;; DECLARE_PSECT EXEC$NONPAGED_CODE
    XYZ: .CALL_ENTRY
    MOVAL DATA,R0
    MOVL (R0),R0
    RET

    DECLARE_PSECT EXEC$NONPAGED_DATA
    DATA: .LONG 1

    .END

    Compile the above with /LIST/MACHINE and look at the PSECT attributes
    for EXEC$NONPAGED_LINKAGE in the listing. Then, remove the ";;;" com-
    menting out the second DECLARE_PSECT EXEC$NONPAGED_CODE. Compile with
    /LIST/MACHINE and look at the EXEC$NONPAGED_LINKAGE PSECT attributes!
    It's NOT the end of the world as the attributes can be rewritten in a
    linker option deck. It's just a nuisance.

    Another is on Itanium. I can't get the linker to resolve a reference
    to MMG$ALLOC_PFN_MAP_SVA even though it's listed in SYS$BASE_IMAGE.MAP.
    This is putting a real damper on my progressing forward with a project.
    This one is NOT a nuisance.

    --
    VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)COM

    .... pejorative statements of opinion are entitled to constitutional protection
    no matter how extreme, vituperous, or vigorously expressed they may be. (NJSC)

    Copr. 2008 Brian Schenkenberger. Publication of _this_ usenet article outside
    of usenet _must_ include its contents in its entirety including this copyright
    notice, disclaimer and quotations.

  6. Re: am I the only one hacking VMS on Itanium?


    I have done one execlet which works on alpha and itanium. It is
    written in C,
    See http://eisner.encompasserve.org/~miller/mbmon010.zip

    I guess you need someone with access to Itanium VMS listings to see
    how execlects are linked on that platform.

  7. Re: am I the only one hacking VMS on Itanium?

    In article <3c9768ff-c88a-4a91-8de1-e049e143a825@25g2000hsx.googlegroups.com>, IanMiller writes:
    >
    >I have done one execlet which works on alpha and itanium. It is
    >written in C,
    >See http://eisner.encompasserve.org/~miller/mbmon010.zip
    >
    > I guess you need someone with access to Itanium VMS listings to see
    >how execlects are linked on that platform.


    NOT!!!

    I have myriad working execlets and drivers that I've linked. I'm perfectly
    fine with linking these. It turns out that the system routine in question
    has been elided from VMS on Itanium but the reference to it still listed in
    the SYS$BASE_IMAGE map.

    --
    VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)COM

    .... pejorative statements of opinion are entitled to constitutional protection
    no matter how extreme, vituperous, or vigorously expressed they may be. (NJSC)

    Copr. 2008 Brian Schenkenberger. Publication of _this_ usenet article outside
    of usenet _must_ include its contents in its entirety including this copyright
    notice, disclaimer and quotations.

  8. Re: am I the only one hacking VMS on Itanium?


    "Michael Moroney" wrote in message
    news:gap2vs$utd$1@pcls6.std.com...

    > I'm doing Macro work on Itanium and we've found odd Macro problems, but
    > the same problems exist on the Alpha macro compiler so I think we're the
    > first to encounter them since the Alpha came out.


    Do I know about them? I'm about to do some other Macro work/fixes so while
    I'm in there...

    John



  9. Re: am I the only one hacking VMS on Itanium?

    "John Reagan" writes:


    >"Michael Moroney" wrote in message
    >news:gap2vs$utd$1@pcls6.std.com...


    >> I'm doing Macro work on Itanium and we've found odd Macro problems, but
    >> the same problems exist on the Alpha macro compiler so I think we're the
    >> first to encounter them since the Alpha came out.


    >Do I know about them? I'm about to do some other Macro work/fixes so while
    >I'm in there...


    Yes. It's all that absolute addressing junk from a while ago.

  10. Re: am I the only one hacking VMS on Itanium?


    "Michael Moroney" wrote in message
    news:gar6gp$t53$1@pcls4.std.com...
    > Yes. It's all that absolute addressing junk from a while ago.


    Oh that, I have it mostly fixed but I need a little more testing before I
    check it in.

    For those of you keeping track at home, Macro (on both Alpha and I64) will
    generate incorrect code for certain absolute address modes.

    For instance,

    cvtbl ^x104,r0

    should fetch the byte at absolute address ^x104. However, the ^x104 is
    truncated to a byte and location 4 is fetched.

    John



  11. Re: am I the only one hacking VMS on Itanium?

    "John Reagan" writes:


    >"Michael Moroney" wrote in message
    >news:gar6gp$t53$1@pcls4.std.com...
    >> Yes. It's all that absolute addressing junk from a while ago.


    >Oh that, I have it mostly fixed but I need a little more testing before I
    >check it in.


    >For those of you keeping track at home, Macro (on both Alpha and I64) will
    >generate incorrect code for certain absolute address modes.


    >For instance,


    >cvtbl ^x104,r0


    >should fetch the byte at absolute address ^x104. However, the ^x104 is
    >truncated to a byte and location 4 is fetched.


    Did you get all of them? It seems for a while HP was giving us a trial
    version that fixed one mode or one instruction or something every two weeks
    or so until my boss told them to fix all of them before bugging us.

    Our workaround was things like:
    cvtbl ^x104[r31],r0

  12. Re: am I the only one hacking VMS on Itanium?


    "Michael Moroney" wrote in message
    news:garkms$e7i$3@pcls4.std.com...
    > Did you get all of them? It seems for a while HP was giving us a trial
    > version that fixed one mode or one instruction or something every two
    > weeks
    > or so until my boss told them to fix all of them before bugging us.
    >
    > Our workaround was things like:
    > cvtbl ^x104[r31],r0


    I hope so. I fixed more after I sent that trial compiler to you.

    John



  13. Re: am I the only one hacking VMS on Itanium?

    In this case listings would have shown that the linkage of that routine was
    eliminated for Itanium.

    There are more people working on Itanium. I've done my fair share of drivers,
    execlets, uwss and other kernel stuff on all platforms. But listings are
    really necessary for that sort of work.

    Jur.


    VAXman- @SendSpamHere.ORG wrote, On 17-9-2008 12:45:
    > In article <3c9768ff-c88a-4a91-8de1-e049e143a825@25g2000hsx.googlegroups.com>, IanMiller writes:
    >> I have done one execlet which works on alpha and itanium. It is
    >> written in C,
    >> See http://eisner.encompasserve.org/~miller/mbmon010.zip
    >>
    >> I guess you need someone with access to Itanium VMS listings to see
    >> how execlects are linked on that platform.

    >
    > NOT!!!
    >
    > I have myriad working execlets and drivers that I've linked. I'm perfectly
    > fine with linking these. It turns out that the system routine in question
    > has been elided from VMS on Itanium but the reference to it still listed in
    > the SYS$BASE_IMAGE map.
    >


  14. Re: am I the only one hacking VMS on Itanium?

    In article <48d1edab$0$183$e4fe514c@news.xs4all.nl>, Jur van der Burg <"lddriver at digiater dot nl"> writes:
    >In this case listings would have shown that the linkage of that routine was
    >eliminated for Itanium.
    >
    >There are more people working on Itanium. I've done my fair share of drivers,
    >execlets, uwss and other kernel stuff on all platforms. But listings are
    >really necessary for that sort of work.


    Well, I don't know about that. I've decided to learn this stuff the old
    fashioned way by looking at all of the executive with SDA. It takes some
    time to trace down argument usage, etc., but I've managed thus far to get
    all sorts of things ported without listings!

    I've been munging page tables, mapping memory and, yes, even intercepting
    executive routines and system services (change mode and mode of caller);
    all without listings! Sad that Garrett E. Brown could believe that I was
    dumber than a rock when it came to porting VMS code.

    I really should have looked more carefully at the MMG$ALLOC_PFN_MAP_SVA
    in the SYS$BASE_IMAGE.MAP file. Chalk that one up to Itanium instruction
    blindness -- similar to snow-blindness. I spent two days and figured
    out a replacement -- after several crashes. They're fun to pour through
    though.

    --
    VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)COM

    .... pejorative statements of opinion are entitled to constitutional protection
    no matter how extreme, vituperous, or vigorously expressed they may be. (NJSC)

    Copr. 2008 Brian Schenkenberger. Publication of _this_ usenet article outside
    of usenet _must_ include its contents in its entirety including this copyright
    notice, disclaimer and quotations.

  15. Re: am I the only one hacking VMS on Itanium?

    "John Reagan" writes:


    >"Michael Moroney" wrote in message
    >news:gap2vs$utd$1@pcls6.std.com...


    >> I'm doing Macro work on Itanium and we've found odd Macro problems, but
    >> the same problems exist on the Alpha macro compiler so I think we're the
    >> first to encounter them since the Alpha came out.


    >Do I know about them? I'm about to do some other Macro work/fixes so while
    >I'm in there...


    I just found something else that causes Itanium Macro to toss its cookies
    (accvio). I'll narrow it down to a reproducer and get it to you when I
    get a chance, but it doesn't seem serious (has to do with a wrong/missing
    macro definition)


+ Reply to Thread