Hi, a few questions for Mike and Richard: On Mon, 15 Jun 2009, Kevin Hilman wrote: > From: Richard Woodruff <r-woodruff2@xxxxxx> > > I'm thinking DLL state is an exceptional condition which happens based > on some wrong early programming when initially setting DLL up. Some > kind of race between idle features and lock happens early on. This > patch recognizes the issue and moves state machine into locked state. > > Its my guess the kick case won't be executed that often. As long as > DLL is not on you can mess with idle state. When its on to mess with > DDR control you need to be in self-refresh and not making > accesses... like in dvfs code. > > Tested-by: Mike Chan <mikechan@xxxxxxxxxx> The last time I torture-tested CORE DVFS, which admittedly was some time ago, the boards tested were stable across the entire overnight run, doing nothing but CORE DVFS switching. This was on a Beagle and a 3430SDP. So, what is different about your setup? The usual suite of questions: - What chip revisions/boards does this affect? - Is this specific to certain bootloaders? - Has anyone else out there seen this problem? Rather than working around an occasional DLL failure to lock, is it possible to reset the DLL earlier in the boot process, as Richard's commit message suggests? That would be preferable to this approach. A back-of-the-envelope assessment suggests that this patch could cause unpredictable, additional 1.5 millisecond latency spikes whenever the workaround triggers (since OCM RAM is marked uncacheable). - Paul > --- > arch/arm/mach-omap2/sleep34xx.S | 22 ++++++++++++++++++++-- > 1 files changed, 20 insertions(+), 2 deletions(-) > > diff --git a/arch/arm/mach-omap2/sleep34xx.S b/arch/arm/mach-omap2/sleep34xx.S > index aedcf94..c469bbe 100644 > --- a/arch/arm/mach-omap2/sleep34xx.S > +++ b/arch/arm/mach-omap2/sleep34xx.S > @@ -595,20 +595,38 @@ wait_sdrc_ok: > ldr r5, [r4] > bic r5, r5, #0x40 > str r5, [r4] > -wait_dll_lock: > - /* Is dll in lock mode? */ > +is_dll_in_lock_mode: > ldr r4, sdrc_dlla_ctrl > ldr r5, [r4] > tst r5, #0x4 > bxne lr > /* wait till dll locks */ > +wait_dll_lock_timed: > ldr r4, sdrc_dlla_status > + mov r6, #0x2000 > +wait_dll_lock: > + subs r6, r6, #0x1 > + beq kick_dll > ldr r5, [r4] > and r5, r5, #0x4 > cmp r5, #0x4 > bne wait_dll_lock > bx lr > > + /* Kick DLL state machine if lock not started */ > +kick_dll: > + ldr r4, sdrc_dlla_ctrl /* get dlla addr */ > + ldr r5, [r4] /* grab value */ > + mov r6, r5 /* save value */ > + orr r6, r6, #0x10 /* dllidle on */ > + str r6, [r4] > + dsb > + bic r6, #0x10 /* dllidle off */ > + str r6, [r4] > + dsb > + str r6, [r4] /* restore old value */ > + b wait_dll_lock_timed > + > cm_idlest1_core: > .word CM_IDLEST1_CORE_V > sdrc_dlla_status: > -- > 1.6.3.2 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-omap" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html > - Paul -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
