[RFC 0/4] Fix recent sim/*/config.in fallout: frv, m32r, sh64, iq2000

[RFC 0/4] Fix recent sim/*/config.in fallout: frv, m32r, sh64, iq2000

[Hans-Peter Nilsson]

Recently, a change requiring -ldl due to bfd plugin default changing
from off to on, spread throughout the generated configury files, for
simulators as well as gdb.  In that process, sim/*/config.in for the
sims below was regenerated for the first time since 2012.  This had
build-breaking fallouts for some simulator showing bloopers in changes
from around 2013-03-23, attempts enabling sim-hardware and
dv-sockser.o.  Here's my take on fixing that: let's return to the
state before those attempts.  This is a request-for-comments, I'll
commit this in a few days unless someone has a better idea, one that
lets these simulators build *without* --{enable,disable} options and
where committing this would still not be an incremental improvement.

All this seems to have started with building dv-sockser.o for *all*
targets and that failing for windows hosts.  I tried to find the
reason dv-sockser.o was enabled in the first place, but failed finding
anything earlier than the hint in
<http://sourceware.org/ml/gdb-patches/2013-03/msg00745.html>.  It
seems the introduction of (building) dv-sockser.o and enabling
sim-hardware for *all* sims from that time was just misguided.  After
these patches, all simulators now build, in particular those fixed
below without and with --enable-sim-hardware and
--disable-sim-hardware, but that's mainly because
SIM_AC_OPTION_HARDWARE has a bug; it forces the value of its first
parameter rather than making it the default for
--{enable,disable}-sim-hardware.  See sim/common/acinclude.m4 and the
generated configure files.

brgds, H-P

Re: [RFC 0/4] Fix recent sim/*/config.in fallout: frv, m32r, sh64, iq2000

[Hans-Peter Nilsson]

> From: Hans-Peter Nilsson <hp@axis.com>
> Date: Wed, 20 Aug 2014 06:28:23 +0200

> Recently, a change requiring -ldl due to bfd plugin default changing
> from off to on, spread throughout the generated configury files, for
> simulators as well as gdb.  In that process, sim/*/config.in for the
> sims below was regenerated for the first time since 2012.  This had
> build-breaking fallouts for some simulator showing bloopers in changes
> from around 2013-03-23, attempts enabling sim-hardware and
> dv-sockser.o.  Here's my take on fixing that: let's return to the
> state before those attempts.  This is a request-for-comments, I'll
> commit this in a few days unless someone has a better idea, one that
> lets these simulators build *without* --{enable,disable} options and
> where committing this would still not be an incremental improvement.

Now committed.  I'll stop here, noting that possible further cleanups
would be to fix the buggy SIM_AC_OPTION_HARDWARE and decouple the
prerequisites for dv-sockser.o from the act of actually enabling it
(e.g. the now combined effect of HAVE_DV_SOCKSER).  Please also, if
you consider "re-enabling" simulated hardware for these sims, first
investigate whether all the wires and plumbing is in place for it to
actually make a difference in function for the particular sim.  Also,
test-cases for the enabled simulated hardware would be helpful, even a
simple "hello, world" to a simulated serial port.

brgds, H-P

Re: [RFC 0/4] Fix recent sim/*/config.in fallout: frv, m32r, sh64, iq2000

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 2293 bytes --]

On 23 Aug 2014 05:13, Hans-Peter Nilsson wrote:
> From: Hans-Peter Nilsson <hp@axis.com>
> > Recently, a change requiring -ldl due to bfd plugin default changing
> > from off to on, spread throughout the generated configury files, for
> > simulators as well as gdb.  In that process, sim/*/config.in for the
> > sims below was regenerated for the first time since 2012.  This had
> > build-breaking fallouts for some simulator showing bloopers in changes
> > from around 2013-03-23, attempts enabling sim-hardware and
> > dv-sockser.o.  Here's my take on fixing that: let's return to the
> > state before those attempts.  This is a request-for-comments, I'll
> > commit this in a few days unless someone has a better idea, one that
> > lets these simulators build *without* --{enable,disable} options and
> > where committing this would still not be an incremental improvement.
> 
> Now committed.  I'll stop here, noting that possible further cleanups
> would be to fix the buggy SIM_AC_OPTION_HARDWARE and decouple the
> prerequisites for dv-sockser.o from the act of actually enabling it
> (e.g. the now combined effect of HAVE_DV_SOCKSER).  Please also, if
> you consider "re-enabling" simulated hardware for these sims, first
> investigate whether all the wires and plumbing is in place for it to
> actually make a difference in function for the particular sim.  Also,
> test-cases for the enabled simulated hardware would be helpful, even a
> simple "hello, world" to a simulated serial port.

i've finished landing a series of changes into master that can be summarized:
 - clean up dv-sockser code to centralize its enabled/disabled logic
 - let the common code handle building of dv-sockser all the time (sans mingw)
 - fix bugs around SIM_AC_OPTION_HARDWARE default enable
 - delete the broken uart code from frv devices

that leads me back to this patch series.  there's no longer a need to disable
hw in these sims by default, so i've somewhat reverted these changes by setting 
the default to "yes" for them.

testcases would certainly be nice, but i have no idea how to write ones for 
these ports.  i'll be pestering people in the future when they make updates to 
specific arches, and i'll be rejecting new sims that lack any tests.
-mike

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

[PATCH] sim: Revert "Fix frv/iq2000/m32r/sh64 sim, default hardware to off."

[Mike Frysinger]

This partially reverts commits:
	105dd264de9a2fa7eee45eff897aa1d6171c1c4b
	3df3af7c3f9095f9dc951fe680ce76f6e497914b
	c4892a6b37647a0b33a2113b59762f678aabe4b2
	9e3042ec073e05a0a5aa56398fd2662c5dcd5002

Now that dv-sockser is handled entirely by the common build logic, the
failure these targets were hitting isn't really possible anymore.  Lets
reset their hardware status back to defaulting to on.  Some of these
were set to "always" previously, but we don't support that anymore.

Committed.
---
 sim/frv/ChangeLog       | 4 ++++
 sim/frv/configure       | 2 +-
 sim/frv/configure.ac    | 2 +-
 sim/iq2000/ChangeLog    | 4 ++++
 sim/iq2000/configure    | 2 +-
 sim/iq2000/configure.ac | 2 +-
 sim/m32r/ChangeLog      | 4 ++++
 sim/m32r/configure      | 2 +-
 sim/m32r/configure.ac   | 2 +-
 sim/sh64/ChangeLog      | 4 ++++
 sim/sh64/configure      | 3 +--
 sim/sh64/configure.ac   | 3 +--
 12 files changed, 24 insertions(+), 10 deletions(-)

diff --git a/sim/frv/ChangeLog b/sim/frv/ChangeLog
index 2e4f8b6..3a745e3 100644
--- a/sim/frv/ChangeLog
+++ b/sim/frv/ChangeLog
@@ -2,6 +2,10 @@
 
 	* configure: Regenerate.
 
+2015-03-24  Mike Frysinger  <vapier@gentoo.org>
+
+	* configure: Regenerate.
+
 2015-03-23  Mike Frysinger  <vapier@gentoo.org>
 
 	* configure: Regenerate.
diff --git a/sim/frv/configure b/sim/frv/configure
index 39a1567..35e307d 100755
diff --git a/sim/frv/configure.ac b/sim/frv/configure.ac
index 05faf03..c56fbe7 100644
--- a/sim/frv/configure.ac
+++ b/sim/frv/configure.ac
@@ -28,6 +28,6 @@ if test x"$silent" != x"yes" && test x"$sim_trapdump" != x""; then
 fi],[sim_trapdump=""])dnl
 AC_SUBST(sim_trapdump)
 
-SIM_AC_OPTION_HARDWARE(no,"","")
+SIM_AC_OPTION_HARDWARE(yes,"","")
 
 SIM_AC_OUTPUT
diff --git a/sim/iq2000/ChangeLog b/sim/iq2000/ChangeLog
index 1f95a08..b2b91ff 100644
--- a/sim/iq2000/ChangeLog
+++ b/sim/iq2000/ChangeLog
@@ -2,6 +2,10 @@
 
 	* configure: Regenerate.
 
+2015-03-24  Mike Frysinger  <vapier@gentoo.org>
+
+	* configure: Regenerate.
+
 2015-03-23  Mike Frysinger  <vapier@gentoo.org>
 
 	* Makefile.in (CONFIG_DEVICES): Delete.
diff --git a/sim/iq2000/configure b/sim/iq2000/configure
index 49c0f3a..37b6bac 100755
diff --git a/sim/iq2000/configure.ac b/sim/iq2000/configure.ac
index dd26a18..78dcbb6 100644
--- a/sim/iq2000/configure.ac
+++ b/sim/iq2000/configure.ac
@@ -14,6 +14,6 @@ SIM_AC_OPTION_ENVIRONMENT
 SIM_AC_OPTION_INLINE()
 SIM_AC_OPTION_CGEN_MAINT
 
-SIM_AC_OPTION_HARDWARE(no,"","")
+SIM_AC_OPTION_HARDWARE(yes,"","")
 
 SIM_AC_OUTPUT
diff --git a/sim/m32r/ChangeLog b/sim/m32r/ChangeLog
index 48f2e47..530e768 100644
--- a/sim/m32r/ChangeLog
+++ b/sim/m32r/ChangeLog
@@ -2,6 +2,10 @@
 
 	* configure: Regenerate.
 
+2015-03-24  Mike Frysinger  <vapier@gentoo.org>
+
+	* configure: Regenerate.
+
 2015-03-23  Mike Frysinger  <vapier@gentoo.org>
 
 	* configure: Regenerate.
diff --git a/sim/m32r/configure b/sim/m32r/configure
index bd17e87..cfce044 100755
diff --git a/sim/m32r/configure.ac b/sim/m32r/configure.ac
index d5c728e..56a7caa 100644
--- a/sim/m32r/configure.ac
+++ b/sim/m32r/configure.ac
@@ -27,6 +27,6 @@ SIM_AC_OPTION_CGEN_MAINT
 AC_SUBST(traps_obj)
 AC_SUBST(sim_extra_cflags)
 
-SIM_AC_OPTION_HARDWARE(no,"","")
+SIM_AC_OPTION_HARDWARE(yes,"","")
 
 SIM_AC_OUTPUT
diff --git a/sim/sh64/ChangeLog b/sim/sh64/ChangeLog
index 8c3c872..8692f0f 100644
--- a/sim/sh64/ChangeLog
+++ b/sim/sh64/ChangeLog
@@ -2,6 +2,10 @@
 
 	* configure: Regenerate.
 
+2015-03-24  Mike Frysinger  <vapier@gentoo.org>
+
+	* configure: Regenerate.
+
 2015-03-23  Mike Frysinger  <vapier@gentoo.org>
 
 	* configure: Regenerate.
diff --git a/sim/sh64/configure b/sim/sh64/configure
index 6851755..87e270e 100755
diff --git a/sim/sh64/configure.ac b/sim/sh64/configure.ac
index b0a875d..c979458 100644
--- a/sim/sh64/configure.ac
+++ b/sim/sh64/configure.ac
@@ -14,7 +14,6 @@ SIM_AC_OPTION_ENVIRONMENT
 SIM_AC_OPTION_INLINE()
 SIM_AC_OPTION_CGEN_MAINT
 
-SIM_AC_OPTION_HARDWARE(no,"","")
+SIM_AC_OPTION_HARDWARE(yes,"","")
 
 SIM_AC_OUTPUT
-
-- 
2.3.3

Re: [RFC 0/4] Fix recent sim/*/config.in fallout: frv, m32r, sh64, iq2000

[Hans-Peter Nilsson]

> From: Mike Frysinger <vapier@gentoo.org>
> Date: Tue, 24 Mar 2015 05:18:56 +0100

> i've finished landing a series of changes into master that can be summarized:
>  - clean up dv-sockser code to centralize its enabled/disabled logic
>  - let the common code handle building of dv-sockser all the time (sans mingw)
>  - fix bugs around SIM_AC_OPTION_HARDWARE default enable
>  - delete the broken uart code from frv devices

Sounds good.

> that leads me back to this patch series.  there's no longer a need to disable
> hw in these sims by default, so i've somewhat reverted these changes by setting 
> the default to "yes" for them.

'k.  FWIW, nothing broke for my autotester setup (targets likely
mentioned in the quoted post).

> testcases would certainly be nice, but i have no idea how to write ones for 
> these ports.  i'll be pestering people in the future when they make updates to 
> specific arches, and i'll be rejecting new sims that lack any tests.

Good! :)

brgds, H-P