Bochs X86 Emulator - Solaris

This is a discussion on Bochs X86 Emulator - Solaris ; Hi All, I am trying to work with the bochs developers on a bug I have open with them, in that I cannot get Bochs to compile under solaris 10 x86. I have opened: https://sourceforge.net/tracker/inde...80&atid=112580 Would anyone be kind enough ...

+ Reply to Thread
Results 1 to 6 of 6

Thread: Bochs X86 Emulator

  1. Bochs X86 Emulator

    Hi All,

    I am trying to work with the bochs developers on a bug I have
    open with them, in that I cannot get Bochs to compile under
    solaris 10 x86.

    I have opened:

    https://sourceforge.net/tracker/inde...80&atid=112580

    Would anyone be kind enough to download the
    source from http://bochs.sourceforge.net and attempt a plain old
    ../configure and make/gmake

    I am using gcc.

    The reason I am asking is that one of the maintainers apparently
    tested on a Solaris x86 machine and claims there is no problem.

    I sure as hell cant get it to build. And I have been able to build bochs
    on a variety of systems in the past (linux/bsd) sucessfully, just a no
    go with Solaris.

    Could those who are willing, test for me, and if its broken, append a
    comment to the bug report I have open with them? or if a sucess, just
    post here.

    Appreciate anyone willing to participate/help

    Regards
    Alex

  2. Re: Bochs X86 Emulator

    Alex wrote:
    > Hi All,
    >
    > I am trying to work with the bochs developers on a bug I have
    > open with them, in that I cannot get Bochs to compile under
    > solaris 10 x86.
    >
    > I have opened:
    >
    > https://sourceforge.net/tracker/inde...80&atid=112580
    >
    >
    > Would anyone be kind enough to download the
    > source from http://bochs.sourceforge.net and attempt a plain old
    > ./configure and make/gmake
    >
    > I am using gcc.
    >
    > The reason I am asking is that one of the maintainers apparently
    > tested on a Solaris x86 machine and claims there is no problem.
    >
    > I sure as hell cant get it to build. And I have been able to build bochs
    > on a variety of systems in the past (linux/bsd) sucessfully, just a no
    > go with Solaris.
    >
    > Could those who are willing, test for me, and if its broken, append a
    > comment to the bug report I have open with them? or if a sucess, just
    > post here.
    >
    > Appreciate anyone willing to participate/help
    >
    > Regards
    > Alex


    Tried under Solaris 10 u0 (sparc).
    g++ --version is 2.95.3

    iodev/hdimage.cc did not compile because of void * passing, fixed types
    and got through compiler.

    Link failed because -lsocket and -lnsl were not included. Once I hacked
    the make file then linked and ran up.



    Tried under Solaris 10 u2 (x86)
    g++ --version is 3.4.3

    Get the same error as you

    gmake[1]: Entering directory `/import/orac/home/jon/bochs-2.3/cpu'
    g++ -c -I.. -I./.. -I../instrument/stubs -I./../instrument/stubs
    -D_XOPEN_SOURCE_EXTENDED=1 -D__EXTENSIONS__ -D_FILE_OFFSET_BITS=64
    -D_LARGE_FILES init.cc -o init.o
    In file included from cpu.h:38,
    from init.cc:30:
    .../disasm/disasm.h:365: error: expected unqualified-id before numeric
    constant
    .../disasm/disasm.h:371: error: expected unqualified-id before numeric
    constant

    Have not posted this to SF, but you can do that.

    John.

  3. Re: Bochs X86 Emulator

    John Doe wrote:
    > Tried under Solaris 10 u0 (sparc).
    > g++ --version is 2.95.3
    >
    > iodev/hdimage.cc did not compile because of void * passing, fixed types
    > and got through compiler.
    >
    > Link failed because -lsocket and -lnsl were not included. Once I hacked
    > the make file then linked and ran up.
    >
    >
    >
    > Tried under Solaris 10 u2 (x86)
    > g++ --version is 3.4.3
    >
    > Get the same error as you
    >
    > gmake[1]: Entering directory `/import/orac/home/jon/bochs-2.3/cpu'
    > g++ -c -I.. -I./.. -I../instrument/stubs -I./../instrument/stubs
    > -D_XOPEN_SOURCE_EXTENDED=1 -D__EXTENSIONS__ -D_FILE_OFFSET_BITS=64
    > -D_LARGE_FILES init.cc -o init.o
    > In file included from cpu.h:38,
    > from init.cc:30:
    > ../disasm/disasm.h:365: error: expected unqualified-id before numeric
    > constant
    > ../disasm/disasm.h:371: error: expected unqualified-id before numeric
    > constant
    >
    > Have not posted this to SF, but you can do that.
    >
    > John.


    Thanks for taking the time to try John,
    Appreciate it :-)

    Alex

  4. Re: Bochs X86 Emulator


    John Doe wrote:
    > Alex wrote:
    > > Hi All,
    > >
    > > I am trying to work with the bochs developers on a bug I have
    > > open with them, in that I cannot get Bochs to compile under
    > > solaris 10 x86.
    > >
    > > I have opened:
    > >
    > > https://sourceforge.net/tracker/inde...80&atid=112580
    > >
    > >
    > > Would anyone be kind enough to download the
    > > source from http://bochs.sourceforge.net and attempt a plain old
    > > ./configure and make/gmake
    > >
    > > I am using gcc.
    > >
    > > The reason I am asking is that one of the maintainers apparently
    > > tested on a Solaris x86 machine and claims there is no problem.
    > >
    > > I sure as hell cant get it to build. And I have been able to build bochs
    > > on a variety of systems in the past (linux/bsd) sucessfully, just a no
    > > go with Solaris.
    > >
    > > Could those who are willing, test for me, and if its broken, append a
    > > comment to the bug report I have open with them? or if a sucess, just
    > > post here.
    > >
    > > Appreciate anyone willing to participate/help
    > >
    > > Regards
    > > Alex

    >
    > Tried under Solaris 10 u0 (sparc).
    > g++ --version is 2.95.3
    >
    > iodev/hdimage.cc did not compile because of void * passing, fixed types
    > and got through compiler.


    i got two errors - one in a mmap call the other in a similar function
    further down.
    My "fix' was not something I care to share with the World but it got me
    through.
    What was your fix? Better than mine Im sure.

    > Link failed because -lsocket and -lnsl were not included. Once I hacked
    > the make file then linked and ran up.


    I set LDFLAGS="-lsocket -lnsl" at configure time...

    > Tried under Solaris 10 u2 (x86)
    > g++ --version is 3.4.3


    I used Studio 11 and set CC CXX accordingly.

    > Get the same error as you
    > gmake[1]: Entering directory `/import/orac/home/jon/bochs-2.3/cpu'
    > g++ -c -I.. -I./.. -I../instrument/stubs -I./../instrument/stubs
    > -D_XOPEN_SOURCE_EXTENDED=1 -D__EXTENSIONS__ -D_FILE_OFFSET_BITS=64
    > -D_LARGE_FILES init.cc -o init.o
    > In file included from cpu.h:38,
    > from init.cc:30:
    > ../disasm/disasm.h:365: error: expected unqualified-id before numeric
    > constant
    > ../disasm/disasm.h:371: error: expected unqualified-id before numeric
    > constant
    > Have not posted this to SF, but you can do that.


    I did not. gmake install failed due to bogus install program.
    Setting INSTALL=/usr/ucb/install at configure time may fix that.
    I just hacked the Makefile.

    Once installed I was able to run bochs but it could find a "boot disk"
    Evidently I need to set up a configure text file...
    Anyway it seems to run even with my ugly mmap hack.

    All this inside VMWare workstation 5.5.2 Sol 64 bit mu2 dual core AMD
    laptop BTW


  5. Re: Bochs X86 Emulator

    gerryt wrote:
    > John Doe wrote:
    >> Alex wrote:
    >>> Hi All,
    >>>
    >>> I am trying to work with the bochs developers on a bug I have
    >>> open with them, in that I cannot get Bochs to compile under
    >>> solaris 10 x86.
    >>>
    >>> I have opened:
    >>>
    >>> https://sourceforge.net/tracker/inde...80&atid=112580
    >>>
    >>>
    >>> Would anyone be kind enough to download the
    >>> source from http://bochs.sourceforge.net and attempt a plain old
    >>> ./configure and make/gmake
    >>>
    >>> I am using gcc.
    >>>
    >>> The reason I am asking is that one of the maintainers apparently
    >>> tested on a Solaris x86 machine and claims there is no problem.
    >>>
    >>> I sure as hell cant get it to build. And I have been able to build bochs
    >>> on a variety of systems in the past (linux/bsd) sucessfully, just a no
    >>> go with Solaris.
    >>>
    >>> Could those who are willing, test for me, and if its broken, append a
    >>> comment to the bug report I have open with them? or if a sucess, just
    >>> post here.
    >>>
    >>> Appreciate anyone willing to participate/help
    >>>
    >>> Regards
    >>> Alex

    >> Tried under Solaris 10 u0 (sparc).
    >> g++ --version is 2.95.3
    >>
    >> iodev/hdimage.cc did not compile because of void * passing, fixed types
    >> and got through compiler.

    >
    > i got two errors - one in a mmap call the other in a similar function
    > further down.


    Yes, same problems.

    > My "fix' was not something I care to share with the World but it got me
    > through.
    > What was your fix? Better than mine Im sure.


    Probably not, just a cast to char *

    >
    >> Link failed because -lsocket and -lnsl were not included. Once I hacked
    >> the make file then linked and ran up.

    >
    > I set LDFLAGS="-lsocket -lnsl" at configure time...


    I just hacked the Makefile! ouch, but yes this is much better.

    >
    >> Tried under Solaris 10 u2 (x86)
    >> g++ --version is 3.4.3

    >
    > I used Studio 11 and set CC CXX accordingly.


    Interesting - I shall give this a go, especially on sparc H/W. Must be a
    GCC 3.4.x problem.

    >
    >> Get the same error as you
    >> gmake[1]: Entering directory `/import/orac/home/jon/bochs-2.3/cpu'
    >> g++ -c -I.. -I./.. -I../instrument/stubs -I./../instrument/stubs
    >> -D_XOPEN_SOURCE_EXTENDED=1 -D__EXTENSIONS__ -D_FILE_OFFSET_BITS=64
    >> -D_LARGE_FILES init.cc -o init.o
    >> In file included from cpu.h:38,
    >> from init.cc:30:
    >> ../disasm/disasm.h:365: error: expected unqualified-id before numeric
    >> constant
    >> ../disasm/disasm.h:371: error: expected unqualified-id before numeric
    >> constant
    >> Have not posted this to SF, but you can do that.

    >
    > I did not. gmake install failed due to bogus install program.
    > Setting INSTALL=/usr/ucb/install at configure time may fix that.
    > I just hacked the Makefile.


    With u2 companion there is GNU install in /opt/sfw/bin (if I remember
    rightly) which is called "install", previous releases this has been
    "ginstall".

    >
    > Once installed I was able to run bochs but it could find a "boot disk"
    > Evidently I need to set up a configure text file...
    > Anyway it seems to run even with my ugly mmap hack.
    >
    > All this inside VMWare workstation 5.5.2 Sol 64 bit mu2 dual core AMD
    > laptop BTW
    >


  6. Re: Bochs X86 Emulator


    John Doe wrote:
    > gerryt wrote:
    > > John Doe wrote:
    > >> Alex wrote:
    > >>> Hi All,
    > >>>
    > >>> I am trying to work with the bochs developers on a bug I have
    > >>> open with them, in that I cannot get Bochs to compile under
    > >>> solaris 10 x86.
    > >>>
    > >>> I have opened:
    > >>>
    > >>> https://sourceforge.net/tracker/inde...80&atid=112580
    > >>>
    > >>>
    > >>> Would anyone be kind enough to download the
    > >>> source from http://bochs.sourceforge.net and attempt a plain old
    > >>> ./configure and make/gmake
    > >>>
    > >>> I am using gcc.
    > >>>
    > >>> The reason I am asking is that one of the maintainers apparently
    > >>> tested on a Solaris x86 machine and claims there is no problem.
    > >>>
    > >>> I sure as hell cant get it to build. And I have been able to build bochs
    > >>> on a variety of systems in the past (linux/bsd) sucessfully, just a no
    > >>> go with Solaris.
    > >>>
    > >>> Could those who are willing, test for me, and if its broken, append a
    > >>> comment to the bug report I have open with them? or if a sucess, just
    > >>> post here.
    > >>>
    > >>> Appreciate anyone willing to participate/help
    > >>>
    > >>> Regards
    > >>> Alex
    > >> Tried under Solaris 10 u0 (sparc).
    > >> g++ --version is 2.95.3
    > >> iodev/hdimage.cc did not compile because of void * passing, fixed types
    > >> and got through compiler.

    > > i got two errors - one in a mmap call the other in a similar function
    > > further down.

    > Yes, same problems.
    > > My "fix' was not something I care to share with the World but it got me
    > > through.
    > > What was your fix? Better than mine Im sure.

    > Probably not, just a cast to char *


    Tried that but CC didnt like it...
    So I "copied" it
    char *s = (char *) var (type void *)
    passed s instead of var and CC didnt complain.

    The real fix is to set "var" correctly in the first place but
    I can barely READ c++ nowadays : <

    > >> Link failed because -lsocket and -lnsl were not included. Once I hacked
    > >> the make file then linked and ran up.

    > > I set LDFLAGS="-lsocket -lnsl" at configure time...

    > I just hacked the Makefile! ouch, but yes this is much better.
    > >> Tried under Solaris 10 u2 (x86)
    > >> g++ --version is 3.4.3

    > > I used Studio 11 and set CC CXX accordingly.

    > Interesting - I shall give this a go, especially on sparc H/W. Must be a
    > GCC 3.4.x problem.


    A C++ problem more like : >

    > >> Get the same error as you
    > >> gmake[1]: Entering directory `/import/orac/home/jon/bochs-2.3/cpu'
    > >> g++ -c -I.. -I./.. -I../instrument/stubs -I./../instrument/stubs
    > >> -D_XOPEN_SOURCE_EXTENDED=1 -D__EXTENSIONS__ -D_FILE_OFFSET_BITS=64
    > >> -D_LARGE_FILES init.cc -o init.o
    > >> In file included from cpu.h:38,
    > >> from init.cc:30:
    > >> ../disasm/disasm.h:365: error: expected unqualified-id before numeric
    > >> constant
    > >> ../disasm/disasm.h:371: error: expected unqualified-id before numeric
    > >> constant
    > >> Have not posted this to SF, but you can do that.

    > > I did not. gmake install failed due to bogus install program.
    > > Setting INSTALL=/usr/ucb/install at configure time may fix that.
    > > I just hacked the Makefile.

    > With u2 companion there is GNU install in /opt/sfw/bin (if I remember
    > rightly) which is called "install", previous releases this has been
    > "ginstall".


    Yes but Im not sure how "install" is set during configure - its
    probably
    looked for in the PATH but thats about it. Some 'install's looked to be
    hard coded when they should be $(INSTALL) or whatever the Make
    syntax is - I always forget : >


+ Reply to Thread