Re: RFA: Testsuite patches...

[Scott Bambrough <scottb@netwinder.org>]

> > 2000-07-26  Scott Bambrough <scottb@netwinder.org>
> >
> >         * gdb.base/environ.exp: Run tests for all targets.
> >         * gdb.base/recurse.exp: Run tests for all targets.
> >
> >         * gdb.base/sect-cmd.exp: Run tests for all targets.  Use
> >         'section .text' clause for Linux in set section command
> >         test.
> >
> >         * gdb.base/so-impl-ld.exp: Added wildcard to
> >         handle the gnu-oldld case on ARM.
> >
> >         * gdb.base/so-indr-cl.exp: Run tests for *-*-linux-gnu*.
> >         * gdb.base/watchpoint.exp (test_stepping): Clear xfail
> >         for ARM targets.
> >
> > I've had these tests in my tree for quite a while now.  Most were HP specific,
> > but I believe they are generic enough for all platforms, or
> > at least all Linux platforms.  I've tested them on Linux on ARM and x86.  I
> > suggest we add them in and deal with failures on specific targets as they
> > happen.  Comments?
> >

Fernando Nasser wrote:
> 
> Will *you* be willing to "deal" with the failures, and on a timely fashion?

I can try, but I don't see why I should have to.  They are new tests for most
targets, previously most of these tests only ran on HP machines.  I looked into
them to find out why they weren't running on my ARM Linux system.  The ones
above I feel should run on all platforms, at least all Linux platforms.  If they
fail the maintainer should look into the problem.  If something like environ.exp
fails on all embedded type platforms, I can look into restricting it to Linux
only, but I think it should run on HPUX, Solaris, DOS, Windows etc.  I don't
have access to all these platforms.

> > BTW, I had to tweak to linux-thread.c to get environ.exp to run
> 
> I belive this touches someone else's mainainership as well.

Mark Kettenis and/or Jim Blandy should probably look at it.

Scott

-- 
Scott Bambrough - Software Engineer
REBEL.COM    http://www.rebel.com
NetWinder    http://www.netwinder.org
From brolley@redhat.com Thu Jul 27 09:59:00 2000
From: Dave Brolley <brolley@redhat.com>
To: gdb-patches@sourceware.cygnus.com
Subject: [PATCH][RFA]: sim/common: Cpu Frequence and Exec Time
Date: Thu, 27 Jul 2000 09:59:00 -0000
Message-id: <39806B05.96F0DA12@redhat.com>
X-SW-Source: 2000-07/msg00336.html
Content-length: 1415

Hi,

The attached patch adds a field to the common cpu profiling
information which represents the clock frequency of the CPU. The
new field is PROFILE_CPU_FREQ and it can be set internally by the
simulator (intended to be used for the default setting) or by the
user via the new simulator command line option:

--profile-cpu-frequency=<frequency>

where <frequency> is an integer optionally followed by the suffix
mhz or khz or any non-ambiguous abbreviation. The suffix is not
case sensitive. Example:

--profile-cpu-frequency=400mhz

Simulators should override a frequency of 0 with the appropriate
default.

If the frequency has been set and profiling is enabled, then the
profile output will contain a calculation of the simulated
execution time based on the number of cycles executed and the
given frequency.

I'm seeking approval to commit this. This was added to on of our
simuators at the request of a customer, and I thought it would be
useful in general.

Dave
2000-07-27  Dave Brolley  <brolley@redhat.com>

	* sim-profile.h (PROFILE_DATA): Add cpu_freq.
	(PROFILE_CPU_FREQ): New macro.
	* sim-profile.c (OPTION_PROFILE_CPU_FREQUENCY): New enumerator.
	(profile-options): Add profile-cpu-frequency.
	(parse_frequency): New function.
	(profile_option_handler): Handle OPTION_PROFILE_CPU_FREQUENCY.
	(profile_print_speed): Print cpu frequency and simulated execution time.
	Re-indent other items to match.
From msnyder@redhat.com Thu Jul 27 11:13:00 2000
From: Michael Snyder <msnyder@redhat.com>
To: Mark Kettenis <kettenis@wins.uva.nl>
Cc: gdb-patches@sources.redhat.com, jimb@cygnus.com
Subject: Re: [PING][RFA] Fall back on dynamic symtab in solib.c:bdf_lookup_symbol
Date: Thu, 27 Jul 2000 11:13:00 -0000
Message-id: <39807BB0.763D@redhat.com>
References: <200007270710.e6R7AjS00711@delius.kettenis.local>
X-SW-Source: 2000-07/msg00337.html
Content-length: 213

Mark Kettenis wrote:
> 
> I'm still looking for approval of the following patch:
> 
>   http://sources.redhat.com/ml/gdb-patches/2000-05/msg00318.html

This looks perfectly safe to me, but I'm not the maintainer.
From msnyder@redhat.com Thu Jul 27 11:34:00 2000
From: Michael Snyder <msnyder@redhat.com>
To: Scott Bambrough <scottb@netwinder.org>
Cc: GDB Patches Mail List <gdb-patches@sourceware.cygnus.com>
Subject: Re: RFA: Testsuite patches...
Date: Thu, 27 Jul 2000 11:34:00 -0000
Message-id: <398080CA.13A5@redhat.com>
References: <39803B5C.9C31BB15@netwinder.org>
X-SW-Source: 2000-07/msg00338.html
Content-length: 2090

Scott Bambrough wrote:
> 
> 2000-07-26  Scott Bambrough <scottb@netwinder.org>
> 
>         * linux-thread.c (linux_threads_pid_to_str):  Change "Pid" to
>         "process" to allow gdb.base/environ.exp to pass all tests.

I don't see any reference to the string "Pid" in environ.exp, 
and the references to "process" are not testing the output
that you changed.  Why do you want this change?

>         Change capitalization on "Thread" for consistency.

Consistency with what?  I'm not strongly opposed, but since
SOMEONE may depend on this output, I'd like to know the
motivation for changing it.

> 
> 2000-07-26  Scott Bambrough <scottb@netwinder.org>
> 
>         * gdb.base/environ.exp: Run tests for all targets.
>         * gdb.base/recurse.exp: Run tests for all targets.
> 
>         * gdb.base/sect-cmd.exp: Run tests for all targets.  Use
>         'section .text' clause for Linux in set section command
>         test.
> 
>         * gdb.base/so-impl-ld.exp: Added wildcard to
>         handle the gnu-oldld case on ARM.
> 
>         * gdb.base/so-indr-cl.exp: Run tests for *-*-linux-gnu*.
>         * gdb.base/watchpoint.exp (test_stepping): Clear xfail
>         for ARM targets.
> 
> I've had these tests in my tree for quite a while now.  Most were HP specific,
> but I believe they are generic enough for all platforms, or
> at least all Linux platforms.  I've tested them on Linux on ARM and x86.  I
> suggest we add them in and deal with failures on specific targets as they
> happen.  Comments?
> 
> BTW, I had to make the change to tweak to linux-thread.c to get environ.exp to
> run correctly.  I changed this rather than the pattern to match in the test for
> fear of breaking the test on HP platforms.
> 
> Scott
> 
> --
> Scott Bambrough - Software Engineer
> REBEL.COM    http://www.rebel.com
> NetWinder    http://www.netwinder.org
> 
>     ---------------------------------------------------------------
> 
>                   Name: gdbtest.patch
>    Part 1.2       Type: application/x-unknown-content-type-patch_auto_file
>               Encoding: base64
From msnyder@redhat.com Thu Jul 27 11:38:00 2000
From: Michael Snyder <msnyder@redhat.com>
To: Scott Bambrough <scottb@netwinder.org>
Cc: Fernando Nasser <fnasser@cygnus.com>, GDB Patches Mail List <gdb-patches@sourceware.cygnus.com>
Subject: Re: RFA: Testsuite patches...
Date: Thu, 27 Jul 2000 11:38:00 -0000
Message-id: <39808189.579F@redhat.com>
References: <39803B5C.9C31BB15@netwinder.org> <3980469B.E0BD7ED2@cygnus.com> <39804FC2.764BCDCC@netwinder.org>
X-SW-Source: 2000-07/msg00339.html
Content-length: 1951

Scott Bambrough wrote:
> 
> > > 2000-07-26  Scott Bambrough <scottb@netwinder.org>
> > >
> > >         * gdb.base/environ.exp: Run tests for all targets.
> > >         * gdb.base/recurse.exp: Run tests for all targets.
> > >
> > >         * gdb.base/sect-cmd.exp: Run tests for all targets.  Use
> > >         'section .text' clause for Linux in set section command
> > >         test.
> > >
> > >         * gdb.base/so-impl-ld.exp: Added wildcard to
> > >         handle the gnu-oldld case on ARM.
> > >
> > >         * gdb.base/so-indr-cl.exp: Run tests for *-*-linux-gnu*.
> > >         * gdb.base/watchpoint.exp (test_stepping): Clear xfail
> > >         for ARM targets.
> > >
> > > I've had these tests in my tree for quite a while now.  Most were HP specific,
> > > but I believe they are generic enough for all platforms, or
> > > at least all Linux platforms.  I've tested them on Linux on ARM and x86.  I
> > > suggest we add them in and deal with failures on specific targets as they
> > > happen.  Comments?
> > >
> 
> Fernando Nasser wrote:
> >
> > Will *you* be willing to "deal" with the failures, and on a timely fashion?
> 
> I can try, but I don't see why I should have to.  They are new tests for most
> targets, previously most of these tests only ran on HP machines.  I looked into
> them to find out why they weren't running on my ARM Linux system.  The ones
> above I feel should run on all platforms, at least all Linux platforms.  If they
> fail the maintainer should look into the problem.  If something like environ.exp
> fails on all embedded type platforms, I can look into restricting it to Linux
> only, but I think it should run on HPUX, Solaris, DOS, Windows etc.  I don't
> have access to all these platforms.

Then can you tell us more about the tests that you 
think should run on them?  And who you think should
do the work?  We like to get problem reports about
problems, not tests that are not expected to pass.
From guo@cup.hp.com Thu Jul 27 17:26:00 2000
From: Jimmy Guo <guo@cup.hp.com>
To: gdb-patches@sourceware.cygnus.com
Subject: [PATCH] gdb.java/configure.in
Date: Thu, 27 Jul 2000 17:26:00 -0000
Message-id: <Pine.LNX.4.10.10007271725050.2248-100000@hpcll168.cup.hp.com>
X-SW-Source: 2000-07/msg00340.html
Content-length: 1581

gdb.java/configure.in erroneously specified a nonexistent file javamisc.exp
as unique file.

- Jimmy

Thu Jul 27 17:23:22	Jimmy Guo	<guo@cup.hp.com>
	* gdb.java/configure.in: Use jmisc.exp as unique file.
	* gdb.java/configure: Rebuilt.

Index: gdb.java/configure.in
/usr/local/bin/diff -c -w -L gdb.java/configure.in gdb.java/configure.in@@/main/cygnus/0 gdb.java/configure.in
*** gdb.java/configure.in
--- gdb.java/configure.in	Thu Jul 27 17:21:23 2000
***************
*** 1,7 ****
  dnl Process this file file with autoconf to produce a configure script.
  
  AC_PREREQ(2.5)
! AC_INIT(javamisc.exp)
  
  CC=${CC-cc}
  AC_SUBST(CC)
--- 1,7 ----
  dnl Process this file file with autoconf to produce a configure script.
  
  AC_PREREQ(2.5)
! AC_INIT(jmisc.exp)
  
  CC=${CC-cc}
  AC_SUBST(CC)
Index: gdb.java/configure
/usr/local/bin/diff -c -w -L gdb.java/configure gdb.java/configure@@/main/cygnus/0 gdb.java/configure
*** gdb.java/configure
--- gdb.java/configure	Thu Jul 27 17:22:21 2000
***************
*** 458,464 ****
  
  # A filename unique to this package, relative to the directory that
  # configure is in, which we can look for to find out if srcdir is correct.
! ac_unique_file=javamisc.exp
  
  # Find the source files, if location was not specified.
  if test -z "$srcdir"; then
--- 458,464 ----
  
  # A filename unique to this package, relative to the directory that
  # configure is in, which we can look for to find out if srcdir is correct.
! ac_unique_file=jmisc.exp
  
  # Find the source files, if location was not specified.
  if test -z "$srcdir"; then