From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Thu, 21 Nov 2013 22:51:04 +1100 From: Michael Ellerman To: Anton Blanchard Subject: Re: [PATCH 2/9] pseries: Add H_SET_MODE to change exception endianness Message-ID: <20131121115104.GC15913@concordia> References: <1384946106-18200-1-git-send-email-anton@samba.org> <1384946106-18200-3-git-send-email-anton@samba.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <1384946106-18200-3-git-send-email-anton@samba.org> Cc: rusty@rustcorp.com.au, Ulrich.Weigand@de.ibm.com, paulus@samba.org, alistair@popple.id.au, linuxppc-dev@lists.ozlabs.org List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Wed, Nov 20, 2013 at 10:14:59PM +1100, Anton Blanchard wrote: > On little endian builds call H_SET_MODE so exceptions have the > correct endianness. We need to reset the endian during kexec > so do that in the MMU hashtable clear callback. > > Signed-off-by: Anton Blanchard > --- > arch/powerpc/include/asm/hvcall.h | 2 ++ > arch/powerpc/include/asm/plpar_wrappers.h | 26 +++++++++++++++++++ > arch/powerpc/platforms/pseries/lpar.c | 17 +++++++++++++ > arch/powerpc/platforms/pseries/setup.c | 42 +++++++++++++++++++++++++++++++ > 4 files changed, 87 insertions(+) > > diff --git a/arch/powerpc/include/asm/hvcall.h b/arch/powerpc/include/asm/hvcall.h > index 0c7f2bf..d8b600b 100644 > --- a/arch/powerpc/include/asm/hvcall.h > +++ b/arch/powerpc/include/asm/hvcall.h > @@ -403,6 +403,8 @@ static inline unsigned long cmo_get_page_size(void) > extern long pSeries_enable_reloc_on_exc(void); > extern long pSeries_disable_reloc_on_exc(void); > > +extern long pseries_big_endian_exceptions(void); > + > #else > > #define pSeries_enable_reloc_on_exc() do {} while (0) > diff --git a/arch/powerpc/include/asm/plpar_wrappers.h b/arch/powerpc/include/asm/plpar_wrappers.h > index a63b045..12c32c5 100644 > --- a/arch/powerpc/include/asm/plpar_wrappers.h > +++ b/arch/powerpc/include/asm/plpar_wrappers.h > @@ -287,6 +287,32 @@ static inline long disable_reloc_on_exceptions(void) { > return plpar_set_mode(0, 3, 0, 0); > } > > +/* > + * Take exceptions in big endian mode on this partition > + * > + * Note: this call has a partition wide scope and can take a while to complete. > + * If it returns H_LONG_BUSY_* it should be retried periodically until it > + * returns H_SUCCESS. > + */ > +static inline long enable_big_endian_exceptions(void) > +{ > + /* mflags = 0: big endian exceptions */ > + return plpar_set_mode(0, 4, 0, 0); > +} > + > +/* > + * Take exceptions in little endian mode on this partition > + * > + * Note: this call has a partition wide scope and can take a while to complete. > + * If it returns H_LONG_BUSY_* it should be retried periodically until it > + * returns H_SUCCESS. > + */ > +static inline long enable_little_endian_exceptions(void) > +{ > + /* mflags = 1: little endian exceptions */ > + return plpar_set_mode(1, 4, 0, 0); > +} > + ... > diff --git a/arch/powerpc/platforms/pseries/setup.c b/arch/powerpc/platforms/pseries/setup.c > index 1f97e2b..c1f1908 100644 > --- a/arch/powerpc/platforms/pseries/setup.c > +++ b/arch/powerpc/platforms/pseries/setup.c > @@ -442,6 +442,32 @@ static void pSeries_machine_kexec(struct kimage *image) > } > #endif Nit picking, but I'm not sure the enable_big/little_endian_exceptions() wrappers buy us much. And they mean you end up with essentially the same routine twice below: > +#ifdef __LITTLE_ENDIAN__ > +long pseries_big_endian_exceptions(void) > +{ > + long rc; > + > + while (1) { > + rc = enable_big_endian_exceptions(); > + if (!H_IS_LONG_BUSY(rc)) > + return rc; > + mdelay(get_longbusy_msecs(rc)); > + } > +} > + > +static long pseries_little_endian_exceptions(void) > +{ > + long rc; > + > + while (1) { > + rc = enable_little_endian_exceptions(); > + if (!H_IS_LONG_BUSY(rc)) > + return rc; > + mdelay(get_longbusy_msecs(rc)); > + } > +} > +#endif Whereas you could do: static long pseries_big_endian_exceptions(bool big_endian) { int mflags; long rc; mflags = big_endian ? 0 : 1; while (1) { rc = plpar_set_mode(mflags, 4, 0, 0); if (!H_IS_LONG_BUSY(rc)) return rc; mdelay(get_longbusy_msecs(rc)); } } Perhaps pseries_big_endian_exceptions(false) looks a bit odd though .. cheers