Re: [patch 0/4] Instrumentation menu removal, against2.6.24-rc4-mm1 (mmotm)

[Mathieu Desnoyers]

* Ananth N Mavinakayanahalli (ananth@xxxxxxxxxx) wrote:
> On Mon, Dec 10, 2007 at 02:27:41PM +0530, Ananth N Mavinakayanahalli wrote:
> > On Sat, Dec 08, 2007 at 10:32:55AM -0500, Mathieu Desnoyers wrote:
> > > Hi Andrew,
> > > 
> > > This time I am taking no chance :
> > > 
> > > The instrumentation menu removal patchset here applies against 2.6.24-rc4-mm1
> > > _and_ against mmotm (dated : stamp-2007-12-05-15-24) without problem.
> > > 
> > > We should hopefully be able to stop racing against other architecture specific
> > > fixes done underneath. Please be aware that the following fix :
> > > 
> > > - fix-oprofile-configuration-breakage.patch
> > > 
> > > from MIPS did not show up in your mmotm tree. I guess you just sent it upstream
> > > without keeping it in your own tree. I have applied the content of this fix in
> > > my patchset (meaning : select HAVE_OPROFILE if !MIPS_MT_SMTC in
> > > add-have-oprofile.patch), but I think you might have a reject if you still have
> > > this fix-oprofile-configuration-breakage.patch in your local tree but not in
> > > mmotm.
> > 
> > Mathieu,
> > 
> > With this patchset, a `make defconfig' results in:
> > 
> > CONFIG_OPROFILE=y
> > CONFIG_HAVE_OPROFILE=y
> > CONFIG_KPROBES=y
> > CONFIG_HAVE_KPROBES=y
> > 
> > You probably also need to change the defconfigs...
> 
> Nevermind. If the intention is to key off the build based on CONFIG_<FOO>
> and have the CONFIG_HAVE_<FOO> settings just to make the config simpler,
> this change is fine. The side effect is just that the CONFIG_HAVE_<FOO>
> will still have default settings even when CONFIG_<FOO> is not set.
> 

On x86_32, a diff between defconfig before and after the patchset :

97a98,103
> CONFIG_PROFILING=y
> # CONFIG_MARKERS is not set
> CONFIG_OPROFILE=y
> CONFIG_HAVE_OPROFILE=y
> CONFIG_KPROBES=y
> CONFIG_HAVE_KPROBES=y
1401,1405d1406
< CONFIG_INSTRUMENTATION=y
< CONFIG_PROFILING=y
< CONFIG_OPROFILE=y
< CONFIG_KPROBES=y
< # CONFIG_MARKERS is not set

Shows the same result. As you explain in your message, the HAVE_* is
just a way to know if the architecture provides a feature. The menu
entry is CONFIG_<FOO>, which depends on CONFIG_HAVE_<FOO>, will allow
y/n/m selection.

Mathieu

-- 
Mathieu Desnoyers
Computer Engineering Ph.D. Student, Ecole Polytechnique de Montreal
OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F  BA06 3F25 A8FE 3BAE 9A68
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/