Re: [PATCH v12 0/3] add the GPMI controller driver for IMX23/IMX28

From: "Lothar Waßmann" <LW@KARO-electronics.de>
To: Huang Shijie <b32955@freescale.com>
Cc: dedekind1@gmail.com, koen.beel.barco@gmail.com,
	w.sang@pengutronix.de, marek.vasut@gmail.com,
	linux-mtd@lists.infradead.org, shijie8@gmail.com,
	s.hauer@pengutronix.de, linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH v12 0/3] add the GPMI controller driver for IMX23/IMX28
Date: Fri, 16 Sep 2011 15:43:30 +0200	[thread overview]
Message-ID: <20083.21122.830407.146310@ipc1.ka-ro> (raw)
In-Reply-To: <4E71CDB2.4000205@freescale.com>

Hi,

Huang Shijie writes:
> Hi,
> > Hi,
> >
> > Huang Shijie writes:
> >> The patch set is based on Artem's tree:
> >> 	http://git.infradead.org/users/dedekind/l2-mtd-2.6.git
> >>
> >> The general-purpose media interface(GPMI) controller is a flexible interface
> >> to up to several NAND flashs.
> >>
> >> The Bose Ray-Choudhury Hocquenghem(BCH) module is a hardware ECC accelerator.
> >>
> >> With the help of BCH, the GPMI controller can choose to do the hardware ECC or
> >> not.
> >>
> >> This driver is a _pure_ MTD NAND controller driver now.
> >>
> >>
> >> The driver depends on another GPMI-NAND device patch set, you can find them at :
> >> 	[1] http://marc.info/?l=linux-arm-kernel&m=131416901319573&w=2
> >> 	[2] http://marc.info/?l=linux-arm-kernel&m=131416912319668&w=2
> >> 	[3] http://marc.info/?l=linux-arm-kernel&m=131416891119504&w=2
> >> 	[4] http://marc.info/?l=linux-arm-kernel&m=131416896219539&w=2
> >>
> >> Test environment:
> >> 	Using imx23 and imx28 boards for test.
> >>
> >> 	imx23 :
> >> 	console=ttyAMA0,115200 mtdparts=gpmi-nfc:20m(boot),-(user)
> >> 	Tested by USB boot and NAND boot.
> >>
> >>          imx28 :
> >> 	#console=ttyAMA0,115200 root=/dev/mmcblk0p3 rw rootwait
> >> 	Tested by SD card boot mode.
> >>
> > How did you perform your tests?
> >
> > I tried this driver on our TX28 board with the result that with
> > concurrent access of the SD card and the NAND flash, I still get the
> > dreaded DMA timeouts within seconds:
> > DMA timeout, last DMA :1
> > GPMI regs:
> >              HW_GPMI_CTRL0[000]=21800001
> >            HW_GPMI_COMPARE[010]=00000000
> >            HW_GPMI_ECCCTRL[020]=00000000
> >           HW_GPMI_ECCCOUNT[030]=00000840
> >            HW_GPMI_PAYLOAD[040]=46578000
> >          HW_GPMI_AUXILIARY[050]=46578800
> >              HW_GPMI_CTRL1[060]=0004000c
> >            HW_GPMI_TIMING0[070]=00010203
> >            HW_GPMI_TIMING1[080]=05000000
> >            HW_GPMI_TIMING2[090]=09020101
> >               HW_GPMI_STAT[0b0]=0d000003
> >              HW_GPMI_DEBUG[0c0]=01000000
> > BCH regs:
> >                HW_BCH_CTRL[000]=00000100
> >             HW_BCH_STATUS0[010]=00000010
> >                HW_BCH_MODE[020]=00000000
> >           HW_BCH_ENCODEPTR[030]=00000000
> >             HW_BCH_DATAPTR[040]=00000000
> >             HW_BCH_METAPTR[050]=00000000
> >        HW_BCH_LAYOUTSELECT[070]=00000000
> >       HW_BCH_FLASH0LAYOUT0[080]=030a4200
> >       HW_BCH_FLASH0LAYOUT1[090]=08404200
> >              HW_BCH_DEBUG0[100]=00000000
> > DMA regs:
> >             HW_APBHX_CTRL0[000]=30000000
> >             HW_APBHX_CTRL1[010]=ffff0000
> >             HW_APBHX_CTRL2[020]=00000000
> >      HW_APBHX_CHANNEL_CTRL[030]=00000000
> >      HW_APBHX_CHn_CURCMDAR(0)[100]=46e1c000
> >      HW_APBHX_CHn_NXTCMDAR(0)[110]=46e1c04c
> >           HW_APBHX_CHn_CMD(0)[120]=000001c8
> >           HW_APBHX_CHn_BAR(0)[130]=00000000
> >          HW_APBHX_CHn_SEMA(0)[140]=00000000
> >        HW_APBHX_CHn_DEBUG1(0)[150]=00a0001e
> >        HW_APBHX_CHn_DEBUG2(0)[160]=00000000
> >      HW_APBHX_CHn_CURCMDAR(4)[2c0]=4652c098
> >      HW_APBHX_CHn_NXTCMDAR(4)[2d0]=4652c0e4
> >           HW_APBHX_CHn_CMD(4)[2e0]=000001c9
> >           HW_APBHX_CHn_BAR(4)[2f0]=46553241
> >          HW_APBHX_CHn_SEMA(4)[300]=00010000
> >        HW_APBHX_CHn_DEBUG1(4)[310]=27a00015
> >        HW_APBHX_CHn_DEBUG2(4)[320]=00000000
> > BCH Geometry :
> > GF length              : 13
> > ECC Strength           : 8
> > Page Size in Bytes     : 2112
> > Metadata Size in Bytes : 10
> > ECC Chunk Size in Bytes: 512
> > ECC Chunk Count        : 4
> > Payload Size in Bytes  : 2048
> > Auxiliary Size in Bytes: 16
> > Auxiliary Status Offset: 12
> > Block Mark Byte Offset : 1999
> > Block Mark Bit Offset  : 0
> >
> > I'm doing a:
> > dd if=/dev/mtd0>  /dev/null&  dd if=/dev/mmcblk0>  /dev/null
> > Sometimes the 'dd' accessing the SD card will stall indefinitely.
> > Also copying data from SD card to flash will produce the error, though
> > less likely.
> >
> it reproduces in my side too.
> > This looks like the problem arises in the DMA driver when multiple
> anyway, I will debug it.
> 
> but i will on vacation next week.
> 
> I am not sure I can fix it tomorrow.
> 
I now found, that selecting 'CONFIG_PREEMPT_NONE' in the kernel config
dramatically decreases the chance of hitting this problem.

I've checked the mxs-dma and gpmi-nand drivers, but could not find
anything fishy wrt preemption.

Lothar Waßmann
-- 
___________________________________________________________

Ka-Ro electronics GmbH | Pascalstraße 22 | D - 52076 Aachen
Phone: +49 2408 1402-0 | Fax: +49 2408 1402-10
Geschäftsführer: Matthias Kaussen
Handelsregistereintrag: Amtsgericht Aachen, HRB 4996

www.karo-electronics.de | info@karo-electronics.de
___________________________________________________________