4.17.87 patch SVR5

4.17.87 patch SVR5

[Robert Lipe]

Much better.   Missed it by only one line this time. :-)

(robertl) rjlinux:/home/robertl/tmp/gdb/gdb-4.17.87/gdb
$ diff -u configure.tgt_ configure.tgt
--- configure.tgt_      Fri Mar 26 12:43:35 1999
+++ configure.tgt       Fri Mar 26 12:45:41 1999
@@ -100,6 +100,7 @@
 i[3456]86-*-sunos*)    gdb_target=sun386 ;;
 i[3456]86-*-sysv4.2MP) gdb_target=i386v42mp ;;
 i[3456]86-*-sysv4.2uw2*)       gdb_target=i386v42mp ;;
+i[3456]86-*-sysv5*)    gdb_target=i386v42mp ;;
 i[3456]86-*-sysv4*)    gdb_target=i386v4 ;;
 i[3456]86-*-unixware2*)        gdb_target=i386v42mp ;;
 i[3456]86-*-unixware*) gdb_target=i386v4 ;;


Test results to follow.

Re: 4.17.87 patch SVR5

[Stan Shebs]

   Date: Fri, 26 Mar 1999 13:09:14 -0600
   From: Robert Lipe <robertl@sco.com>

   Much better.   Missed it by only one line this time. :-)

Ah ha, you've been holding out on me, you dirty software hoarder. :-) :-)

I'm adding that to the repository now.  In exchange for the added
line, I collapsed the two 4.2 configs:

    i[3456]86-*-sysv4.2MP) gdb_target=i386v42mp ;;
    i[3456]86-*-sysv4.2uw2*)       gdb_target=i386v42mp ;;

to

    i[3456]86-*-sysv4.2*) gdb_target=i386v42mp ;;

I assume this is a plausible thing to do, but I'll go along with people
who know more about the minutiae of SCO/Unixware configs.

							Stan

Re: 4.17.87 patch SVR5

[Robert Lipe]

>    Much better.   Missed it by only one line this time. :-)
> Ah ha, you've been holding out on me, you dirty software hoarder. :-) :-)

I offer exhibit A in which it is clearly shown the defendant is not 
a hoarder and has offered this very patch before:

	http://www.cygnus.com/ml/gdb-patches/1999-Mar/0065.html

> I'm adding that to the repository now.  

Thank you.

> In exchange for the added line

The shame of the bloat was killing me. :-)

> I collapsed the two 4.2 configs:
> 
>     i[3456]86-*-sysv4.2MP) gdb_target=i386v42mp ;;
>     i[3456]86-*-sysv4.2uw2*)       gdb_target=i386v42mp ;;
> 
> to
> 
>     i[3456]86-*-sysv4.2*) gdb_target=i386v42mp ;;
> 
> I assume this is a plausible thing to do, but I'll go along with people
> who know more about the minutiae of SCO/Unixware configs.

Executive summary: Looks fine to me.

I don't know *what* had been going on with this particular config in
various GNU configure mechanisms.  As I remember history (and I could
be wrong) on X86 the only commercially available SVR4.2 product was
UnixWare and it was available only in MP version, but did run on a uni.
UW1* and UW2* were both MP-capable but I think they did differ in how
they represented /proc.  So why these were ever differentiated to GDB
in this way utterly escapes me.  This was also indirectly responsible
for me wandering around inside procfs for completely too long when I
overrode --host and got the "wrong" case on MP and therefor got the
wrong /proc handling.  So I think you're doing the world a favor in
collapsing the above *if* config.guess doesn't outsmart us on that
target.

Perhaps it all predates standardized config.guess schemes and different
packages just picked different representations.  I don't know.  I didn't
closely follow UW during that era.

Rodney, can you confirm this doesn't hose UW2?

Re: 4.17.87 patch SVR5

[Stan Shebs]

   Date: Fri, 26 Mar 1999 14:41:06 -0600
   From: Robert Lipe <robertl@sco.com>

   >    Much better.   Missed it by only one line this time. :-)
   > Ah ha, you've been holding out on me, you dirty software hoarder. :-) :-)

   I offer exhibit A in which it is clearly shown the defendant is not 
   a hoarder and has offered this very patch before:

	   http://www.cygnus.com/ml/gdb-patches/1999-Mar/0065.html

You're right and I'm lame... :-)

   I don't know *what* had been going on with this particular config in
   various GNU configure mechanisms.

Different people with different ideas about which systems are which.
Cygnus has tangled with versions of Unixware a couple times, but the
systems were so difficult to use that they didn't get much dedicated
attention.  This is one area where I'd like to rely on people who
have more hands-on experience with the actual systems...

							Stan

Re: 4.17.87 patch SVR5

[Stan Shebs]

   Date: Fri, 26 Mar 1999 13:09:14 -0600
   From: Robert Lipe <robertl@sco.com>

   Much better.   Missed it by only one line this time. :-)

Ah ha, you've been holding out on me, you dirty software hoarder. :-) :-)

I'm adding that to the repository now.  In exchange for the added
line, I collapsed the two 4.2 configs:

    i[3456]86-*-sysv4.2MP) gdb_target=i386v42mp ;;
    i[3456]86-*-sysv4.2uw2*)       gdb_target=i386v42mp ;;

to

    i[3456]86-*-sysv4.2*) gdb_target=i386v42mp ;;

I assume this is a plausible thing to do, but I'll go along with people
who know more about the minutiae of SCO/Unixware configs.

							Stan

4.17.87 patch SVR5

[Robert Lipe]

Much better.   Missed it by only one line this time. :-)

(robertl) rjlinux:/home/robertl/tmp/gdb/gdb-4.17.87/gdb
$ diff -u configure.tgt_ configure.tgt
--- configure.tgt_      Fri Mar 26 12:43:35 1999
+++ configure.tgt       Fri Mar 26 12:45:41 1999
@@ -100,6 +100,7 @@
 i[3456]86-*-sunos*)    gdb_target=sun386 ;;
 i[3456]86-*-sysv4.2MP) gdb_target=i386v42mp ;;
 i[3456]86-*-sysv4.2uw2*)       gdb_target=i386v42mp ;;
+i[3456]86-*-sysv5*)    gdb_target=i386v42mp ;;
 i[3456]86-*-sysv4*)    gdb_target=i386v4 ;;
 i[3456]86-*-unixware2*)        gdb_target=i386v42mp ;;
 i[3456]86-*-unixware*) gdb_target=i386v4 ;;


Test results to follow.

Re: 4.17.87 patch SVR5

[Stan Shebs]

   Date: Fri, 26 Mar 1999 14:41:06 -0600
   From: Robert Lipe <robertl@sco.com>

   >    Much better.   Missed it by only one line this time. :-)
   > Ah ha, you've been holding out on me, you dirty software hoarder. :-) :-)

   I offer exhibit A in which it is clearly shown the defendant is not 
   a hoarder and has offered this very patch before:

	   http://www.cygnus.com/ml/gdb-patches/1999-Mar/0065.html

You're right and I'm lame... :-)

   I don't know *what* had been going on with this particular config in
   various GNU configure mechanisms.

Different people with different ideas about which systems are which.
Cygnus has tangled with versions of Unixware a couple times, but the
systems were so difficult to use that they didn't get much dedicated
attention.  This is one area where I'd like to rely on people who
have more hands-on experience with the actual systems...

							Stan

Re: 4.17.87 patch SVR5

[Robert Lipe]

>    Much better.   Missed it by only one line this time. :-)
> Ah ha, you've been holding out on me, you dirty software hoarder. :-) :-)

I offer exhibit A in which it is clearly shown the defendant is not 
a hoarder and has offered this very patch before:

	http://www.cygnus.com/ml/gdb-patches/1999-Mar/0065.html

> I'm adding that to the repository now.  

Thank you.

> In exchange for the added line

The shame of the bloat was killing me. :-)

> I collapsed the two 4.2 configs:
> 
>     i[3456]86-*-sysv4.2MP) gdb_target=i386v42mp ;;
>     i[3456]86-*-sysv4.2uw2*)       gdb_target=i386v42mp ;;
> 
> to
> 
>     i[3456]86-*-sysv4.2*) gdb_target=i386v42mp ;;
> 
> I assume this is a plausible thing to do, but I'll go along with people
> who know more about the minutiae of SCO/Unixware configs.

Executive summary: Looks fine to me.

I don't know *what* had been going on with this particular config in
various GNU configure mechanisms.  As I remember history (and I could
be wrong) on X86 the only commercially available SVR4.2 product was
UnixWare and it was available only in MP version, but did run on a uni.
UW1* and UW2* were both MP-capable but I think they did differ in how
they represented /proc.  So why these were ever differentiated to GDB
in this way utterly escapes me.  This was also indirectly responsible
for me wandering around inside procfs for completely too long when I
overrode --host and got the "wrong" case on MP and therefor got the
wrong /proc handling.  So I think you're doing the world a favor in
collapsing the above *if* config.guess doesn't outsmart us on that
target.

Perhaps it all predates standardized config.guess schemes and different
packages just picked different representations.  I don't know.  I didn't
closely follow UW during that era.

Rodney, can you confirm this doesn't hose UW2?