From mboxrd@z Thu Jan 1 00:00:00 1970 From: Peter Tyser Date: Fri, 10 Jul 2009 15:51:45 -0500 Subject: [U-Boot] [PATCH 3/7] 86xx: Unlock l1 cache unconditionally In-Reply-To: <20090710203723.34C41832E416@gemini.denx.de> References: <05de848db92da7ff85fa3c1bb285247aba99ec05.1243005212.git.ptyser@xes-inc.com> <4B5A7126-FCDF-4EB0-9181-FB1BE571C715@kernel.crashing.org> <20090710203723.34C41832E416@gemini.denx.de> Message-ID: <1247259105.32367.22.camel@localhost.localdomain> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de On Fri, 2009-07-10 at 22:37 +0200, Wolfgang Denk wrote: > Dear Kumar Gala, > > In message <4B5A7126-FCDF-4EB0-9181-FB1BE571C715@kernel.crashing.org> you wrote: > > > > On May 22, 2009, at 10:26 AM, Peter Tyser wrote: > > > > > Previously, it was only unlocked when Linux was executed using the > > > "bootm" command. Unlocking it unconditionally improves U-Boot > > > performance and provides a common cache state when booting OSes. > > > > > > Signed-off-by: Peter Tyser > > > --- > > > lib_ppc/board.c | 8 +++++--- > > > lib_ppc/bootm.c | 3 ++- > > > 2 files changed, 7 insertions(+), 4 deletions(-) > > > > Wolfgang, do you know any reason we can't do the unlock for ALL ppc's > > in board_init_r()? > > Sorry, I don't remember if there was any speciofic reason, or what > that might have been. Maybe the CHANGELOGs contain some hints? The only platforms the define unlock_ram_in_cache() are 83xx, 85xx, 86xx, and 74xx_7xx. It looks like the most of them had "issues" in their implementation that could explain why there weren't enabled: 83xx - ade50c7fa1b16ef98be17e9c3ae286aecf4f5605, 6eb2a44e27919fdc601e0c05404b298a7602c0e3 86xx - 392438406041415fe64ab8748ec5ab5ad01d1cf7 74xx - d685b74c64a38849f1a129b3ab846fbf67dd937e 85xx - this one works My best guess was all the non-85xx implementations had bugs or other issues that caused U-Boot to become unstable after unlocking the cache. Perhaps those bugs are fixed now, but I only have 86xx hardware to test on. I've been running a few 86xx boards with the cache unlocked with no noticeable problems. Peter