Hi Fernando, Thank you for reviewing. From: "ext Guzman Lugo, Fernando" <x0095...@ti.com> Subject: RE: [PATCH 05/13] DSPBRIDGE: enable smart/autoidle for mailbox sysconfig Date: Wed, 15 Jul 2009 23:59:22 +0200
> > Hi, > > > -----Original Message----- > > From: Ameya Palande [mailto:ameya.pala...@nokia.com] > > Sent: Wednesday, July 15, 2009 9:56 AM > > To: linux-omap@vger.kernel.org > > Cc: Ramirez Luna, Omar; Guzman Lugo, Fernando; Menon, Nishanth; > > hiroshi.d...@nokia.com > > Subject: [PATCH 05/13] DSPBRIDGE: enable smart/autoidle for mailbox > > sysconfig > > > > From: Doyu Hiroshi (Nokia-D/Helsinki) <hiroshi.d...@nokia.com> > > > > Reported-by: Tero Kristo <tero.kri...@nokia.com> > > Signed-off-by: Hiroshi DOYU <hiroshi.d...@nokia.com> > > --- > > drivers/dsp/bridge/hw/hw_mbox.c | 4 +++- > > drivers/dsp/bridge/hw/hw_mbox.h | 5 +++++ > > drivers/dsp/bridge/wmd/io_sm.c | 1 + > > 3 files changed, 9 insertions(+), 1 deletions(-) > > > > diff --git a/drivers/dsp/bridge/hw/hw_mbox.c > > b/drivers/dsp/bridge/hw/hw_mbox.c > > index ee79032..5a87597 100644 > > --- a/drivers/dsp/bridge/hw/hw_mbox.c > > +++ b/drivers/dsp/bridge/hw/hw_mbox.c > > @@ -33,7 +33,9 @@ > > /* width in bits of MBOX Id */ > > #define HW_MBOX_ID_WIDTH 2 > > > > -struct MAILBOX_CONTEXT mboxsetting = {0x4, 0x1, 0x1}; > > +static struct MAILBOX_CONTEXT mboxsetting = { > > + .sysconfig = 2 << 3 | 1, /* SMART/AUTO-IDLE */ > > +}; > > > > /* Saves the mailbox context */ > > HW_STATUS HW_MBOX_saveSettings(void __iomem *baseAddress) > > diff --git a/drivers/dsp/bridge/hw/hw_mbox.h > > b/drivers/dsp/bridge/hw/hw_mbox.h > > index ad1a89c..8a5f6bd 100644 > > --- a/drivers/dsp/bridge/hw/hw_mbox.h > > +++ b/drivers/dsp/bridge/hw/hw_mbox.h > > @@ -320,4 +320,9 @@ extern HW_STATUS HW_MBOX_saveSettings(void __iomem > > *baseAddres); > > */ > > extern HW_STATUS HW_MBOX_restoreSettings(void __iomem *baseAddres); > > > > +static inline void HW_MBOX_initSettings(void __iomem *baseAddres) > > +{ > > + HW_MBOX_restoreSettings(baseAddres); > > +} > > + > > #endif /* __MBOX_H */ > > diff --git a/drivers/dsp/bridge/wmd/io_sm.c > > b/drivers/dsp/bridge/wmd/io_sm.c > > index 39c34f7..d8ae1f1 100644 > > --- a/drivers/dsp/bridge/wmd/io_sm.c > > +++ b/drivers/dsp/bridge/wmd/io_sm.c > > @@ -282,6 +282,7 @@ DSP_STATUS WMD_IO_Create(OUT struct IO_MGR **phIOMgr, > > pIOMgr->fSharedIRQ = pMgrAttrs->fShared; > > IO_DisableInterrupt(hWmdContext); > > if (devType == DSP_UNIT) { > > + HW_MBOX_initSettings(hostRes.dwMboxBase); > > What do think about doing a call to __raw_writel to avoid a new function > wrapper as Omar suggested? Something like that: > > /* Enabling mailbox SMART/AUTO-IDLE */ > __raw_writel(2 << 3 | 1, hostRes.dwMboxBase + 10); I think that the basic concept of "hw_mbox.[ch]" is to abstruct h/w mailbox feature, to provide lowlevel mailbox APIs(h/w accessors), and to conceal the direct h/w manipulation from other modules. Considering this concept, manipulating mailbox registers with __raw_writel() looks to conflict this original design purpose, IMHO. > > > > /* Plug the channel ISR:. */ > > if ((request_irq(INT_MAIL_MPU_IRQ, IO_ISR, 0, > > "DspBridge\tmailbox", (void *)pIOMgr)) == 0) > > -- > > 1.6.2.4 > > > > Regards, > Fernando > Guzman > Lugo. > -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
