[U-Boot] [PATCH 1/2] mxc nand: Add support for i.MX5

[Marek Vasut <marex@denx.de>]

Dear Beno?t Th?baudeau,

> Dear Marek Vasut,
> 
> On Tuesday, January 8, 2013 10:33:29 AM, Marek Vasut wrote:
> > Subject: [PATCH 1/2] mxc nand: Add support for i.MX5
> 
> Where is the 2/2?

Nowhere, it's a typo.

> > From: Beno?t Th?baudeau <benoit.thebaudeau@advansee.com>
> > 
> > Signed-off-by: Beno?t Th?baudeau <benoit.thebaudeau@advansee.com>
> > Cc: Scott Wood <scottwood@freescale.com>
> > Cc: Stefano Babic <sbabic@denx.de>
> > ---
> > 
> >  arch/arm/include/asm/arch-mx5/imx-regs.h |    9 ++
> >  drivers/mtd/nand/mxc_nand.c              |  219
> >  ++++++++++++++++++++++--------
> >  include/fsl_nfc.h                        |  149 +++++++++++++-------
> >  nand_spl/nand_boot_fsl_nfc.c             |  114 ++++++++++++----
> >  4 files changed, 364 insertions(+), 127 deletions(-)
> > 
> > NOTE: I'm seeing issues when this is compiled into U-Boot. U-Boot
> > won't boot on
> > my MX53 board and will hang right after printing "DRAM:" and before
> > printing the
> > CPU info. I suspect it's some kind of unaligned access.
> 
> OK. I'm waiting for your test results. It works for me on i.MX51 with
> 2012.07.

It works with 2013.01-rc1, but I have issues. When I boot the board from SD, it 
hangs. When I load via JTAG, it doesn't hang. I suspect it's something related 
to the MXC NAND driver, since if the driver is not compiled in, it works like 
charm both ways.

Otherwise, when booted, the NAND works fine.

I wonder if there's some unaligned access or something happening in the NAND 
driver. Or maybe some other issue where the code is miscompiled.

> > NOTE2: I fixed a few complaints from GCC
> 
> Which ones? I don't see any such change in this new version.

This one I think:

354 @@ -1167,8 +1226,8 @@ static struct nand_bbt_descr bbt_mirror_descr = {
355  int board_nand_init(struct nand_chip *this)
356  {
357         struct mtd_info *mtd;
358 -#ifdef MXC_NFC_V2_1
359 -       uint16_t tmp;
360 +#if defined(MXC_NFC_V2_1) || defined(MXC_NFC_V3_2)
361 +       uint32_t tmp;
362  #endif
363 
364  #ifdef CONFIG_SYS_NAND_USE_FLASH_BBT

> > and rebased on top of
> > master.
> 
> OK.
> 
> [--snip--]
> 
> > @@ -698,7 +757,7 @@ static int mxc_nand_correct_data(struct mtd_info
> > *mtd, u_char *dat,
> > 
> >  	 * additional correction.  2-Bit errors cannot be corrected by
> >  	 * HW ECC, so we need to return failure
> >  	 */
> > 
> > -	uint16_t ecc_status = readw(&host->regs->ecc_status_result);
> > +	uint32_t ecc_status = readnfc(&host->regs->ecc_status_result);
> 
>         ^
> Here it was uint16_t in my last version, and this was correct since this is
> code for MXC_NFC_V1. This change should not hurt, but it was not
> necessary.

But in case readnfc() results in readl(), u16 is too small.

> Is this change related to the GCC complaints that you mentioned?

No, there's one more, see above.

> >  	if (((ecc_status & 0x3) == 2) || ((ecc_status >> 2) == 2)) {
> >  	
> >  		MTDDEBUG(MTD_DEBUG_LEVEL0,
> 
> [--snip--]
> 
> The other changes (i.e. the rebase) are fine with me.

I hope I didn't screw up the attribution. I used "From:" which I hope is 
correct.

> Best regards,
> Beno?t

Best regards,
Marek Vasut