From: horms@verge.net.au (Simon Horman) Date: Fri, 13 Nov 2015 09:51:58 +0900 Subject: [PATCH v2 3/3] ARM: shmobile: gose: Add QSPI device to DT In-Reply-To: References: <1447291763-27527-1-git-send-email-horms+renesas@verge.net.au> <1447291763-27527-4-git-send-email-horms+renesas@verge.net.au> Message-ID: <20151113005157.GF20462@verge.net.au> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Thu, Nov 12, 2015 at 11:44:15AM +0100, Geert Uytterhoeven wrote: > Hi Simon, > > On Thu, Nov 12, 2015 at 2:29 AM, Simon Horman > wrote: > > Enable the QSPI controller in the gose device tree. > > > > Based similar work for the silk board by by Vladimir Barinov and > > Sergei Shtylyov. > > > > Cc: Vladimir Barinov > > Cc: Sergei Shtylyov > > Signed-off-by: Simon Horman > > Acked-by: Geert Uytterhoeven > > > --- > > Tested: Read /dev/mtd0 with 512K blocks at 1.7MB/s > > That's slow: I get 6.7 MB/s on koelsch (69.9 Mbps for Quad Read according > to my in-kernel QSPI debug code). Is this with SYS-DMAC enabled? Yes it does seem very small. I think that the problem is that /dev/mtd0 is very small, only 262kB. And I do not see any interrupts for dma-controllers when reading from hit. /dev/mdt1 is much later, 4.2MB. dd reads that at ~7MB/s using 512kB blocks. In that test I see interrupts for one of the dma-controllers. I plan to go ahead and apply this series.