From mboxrd@z Thu Jan 1 00:00:00 1970 From: will.deacon@arm.com (Will Deacon) Date: Wed, 15 Jun 2011 10:02:57 +0100 Subject: [PATCH] ARM: mm: cache-l2x0: Add support for re-enabling l2x0 In-Reply-To: <20110614173013.GA9234@e102568-lin.cambridge.arm.com> References: <1307926019-28958-1-git-send-email-ccross@android.com> <20110613101942.GA7924@e102568-lin.cambridge.arm.com> <20110614173013.GA9234@e102568-lin.cambridge.arm.com> Message-ID: <20110615090257.GC31110@e102144-lin.cambridge.arm.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Hi guys, On Tue, Jun 14, 2011 at 06:30:19PM +0100, Lorenzo Pieralisi wrote: > On Mon, Jun 13, 2011 at 07:20:34PM +0100, Colin Cross wrote: > > According to the PL310 TRM (r3p2 3-9) on disable the PL310 syncs, which means > it flushes internal buffers, but it is not supposed to clean dirty lines. > It seems unsafe to me to use the stack and other cacheable data right before > disabling. > > > Otherwise, wouldn't a flush_cache_all() and l2x0_flush_all() in > > l2x0_disable() ensure the stack lines were flushed out? Probably > > needs a readl_releaxed as well, to ensure the disable has taken > > effect. > > > > I reckon the safest option is to clean and disable in one C > function. Moreover at that point in time spinlocks might not be an option > anymore (CPU out of coherency), which makes those functions unusable. I was planning to add this functionality as part of my kexec patches since otherwise the l2x0_lock gets blown away on disable. I initially attributed this to loss of coherency, but that's clearly not the case as I still had L1 running. Colin - I'll add you to CC for v3 of that series. Will