[PATCH RESEND v4 1/4] omap3: nand: prefetch in irq mode support

Tony,

-----Original Message-----
From: Ghorai, Sukumar
Sent: Saturday, September 18, 2010 11:55 PM
To: 'Tony Lindgren'
Cc: 'linux-omap at vger.kernel.org'; 'linux-mtd at lists.infradead.org'; 'linux-
arm-kernel at lists.infradead.org'
Subject: RE: [PATCH RESEND v4 1/4] omap3: nand: prefetch in irq mode
support

Tony,

quoted

-----Original Message-----
From: Tony Lindgren [mailto:tony at atomide.com]
Sent: Friday, September 17, 2010 11:25 PM
To: Ghorai, Sukumar
Cc: linux-omap at vger.kernel.org; linux-mtd at lists.infradead.org; linux-

arm-

quoted

kernel at lists.infradead.org; Vimal Singh
Subject: Re: [PATCH RESEND v4 1/4] omap3: nand: prefetch in irq mode
support

* Sukumar Ghorai [off-list ref] [100916 00:53]:

quoted

This patch enable prefetch-irq mode for NAND.

--- a/drivers/mtd/nand/omap2.c
+++ b/drivers/mtd/nand/omap2.c

@@ -467,6 +485,152 @@ static void omap_write_buf_dma_pref(struct

mtd_info *mtd,

quoted

 		omap_nand_dma_transfer(mtd, (u_char *) buf, len, 0x1);
 }

+/*
+ * omap_nand_irq - GMPC irq handler
+ * @this_irq: gpmc irq number
+ * @dev: omap_nand_info structure pointer is passed here
+ */
+static irqreturn_t omap_nand_irq(int this_irq, void *dev)
+{
+	struct omap_nand_info *info = (struct omap_nand_info *) dev;
+	u32 bytes;
+	u32 irq_stat;
+
+	irq_stat = gpmc_read_status(GPMC_GET_IRQ_STATUS);
+	bytes = gpmc_read_status(GPMC_PREFETCH_FIFO_CNT);
+	bytes = bytes  & 0xFFFC; /* io in multiple of 4 bytes */
+	if (info->iomode == OMAP_NAND_IO_WRITE) { /* checks for write io */
+		if (irq_stat & 0x2)
+			goto done;
+
+		if (info->buf_len & (info->buf_len < bytes))
+			bytes = info->buf_len;
+		else if (!info->buf_len)
+			bytes = 0;
+		iowrite32_rep(info->nand.IO_ADDR_W,
+						(u32 *)info->buf, bytes >> 2);
+		info->buf = info->buf + bytes;
+		info->buf_len -= bytes;
+
+	} else {
+		ioread32_rep(info->nand.IO_ADDR_R,
+						(u32 *)info->buf, bytes >> 2);
+		info->buf = info->buf + bytes;
+
+		if (irq_stat & 0x2)
+			goto done;
+	}
+	gpmc_cs_configure(info->gpmc_cs, GPMC_SET_IRQ_STATUS, irq_stat);
+	irq_stat = gpmc_read_status(GPMC_GET_IRQ_STATUS);
+
+	return IRQ_HANDLED;
+
+done:
+	complete(&info->comp);
+	/* disable irq */
+	gpmc_cs_configure(info->gpmc_cs, GPMC_ENABLE_IRQ, 0);
+
+	/* clear status */
+	gpmc_cs_configure(info->gpmc_cs, GPMC_SET_IRQ_STATUS, irq_stat);
+	irq_stat = gpmc_read_status(GPMC_GET_IRQ_STATUS);
+
+	return IRQ_HANDLED;
+}

This handler should be in gpmc.c as it may be needed for other GPMC
connected devices on the same system. You can use chained irq handlers
to allow all the drivers to use the interrupt then.

[Ghorai]
You mean as this function used the gpmc-irq number in nand file, so
handler should move to gpmc.c file?
	1. For that we need to add one io-struct (to keep io buffer status)
in gpmc.c;

	2. Also need help how to sync between gpmc.c/omap_nand_irq() and
omap2.c/omap_write_buf_irq_pref(), men how read/write function know that
work done in interrupt-context? Or you prefer to move the complete IO
function (omap_read/write_buf_irq_pref) to gpmc.c?

	3. gpmc does not now about the read and write address that's
applicable for NAND. So how to pass the IO address from omap2.c to gpmc.c,
interrupt handler?


So, please let me know your suggestion again such that I can post this
time itself. Otherwise again it will miss from coming release, this was
posted/reviewed for last release too. And suggest to void repeating of
missing release window again.

[Ghorai] Please reply with your 2nd suggestion.

quoted

Regards,

Tony