Skip to contentSkip to conversation

mtd: nand: fsl-ifc: fix support of multiple NAND devices

PATCH New Kurt Kanzenbach 3 participants 4 messages Oct 20, 2017 – Nov 17, 2017

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 mtd: nand: fsl-ifc: fix support of multiple NAND devices

Currently the chipselect used to identify the corresponding NAND chip is stored at the controller and only set during fsl_ifc_chip_init(). This way, only the last NAND chip is working, as the previous value of cs_nand gets overwritten.

In order to solve this issue the chipselect is computed on demand by evaluating the bank variable. Thus, the correct chipselect for each NAND chip operation is used.

Tested on hardware with two NAND chips connected to the IFC controller.

---
Changes for v2:
- get rid of cs_nand and compute chipselect on demand
- pass priv instead of mtd to fsl_ifc_sram_init()

drivers/mtd/nand/fsl_ifc_nand.c | 15 ++++++---------

 1 file changed, 6 insertions(+), 9 deletions(-)
drivers/mtd/nand/fsl_ifc_nand.c +6 -9
index bc6bdc9b2c..a757a3be44 100644
@@ -48,7 +48,6 @@ struct fsl_ifc_ctrl {
 	/* device info */
 	struct fsl_ifc regs;
 	void __iomem *addr;      /* Address of assigned IFC buffer        */
-	unsigned int cs_nand;    /* On which chipsel NAND is connected	  */
 	unsigned int page;       /* Last page written to / read from      */
 	unsigned int read_bytes; /* Number of bytes read during command   */
 	unsigned int column;     /* Saved column from SEQIN               */
@@ -296,7 +295,7 @@ static int fsl_ifc_run_command(struct mtd_info *mtd)
 	int i;
 
 	/* set the chip select for NAND Transaction */
-	ifc_out32(&ifc->ifc_nand.nand_csel, ifc_ctrl->cs_nand);
+	ifc_out32(&ifc->ifc_nand.nand_csel, priv->bank << IFC_NAND_CSEL_SHIFT);
 
 	/* start read/write seq */
 	ifc_out32(&ifc->ifc_nand.nandseq_strt,
@@ -798,7 +797,7 @@ static void fsl_ifc_select_chip(struct mtd_info *mtd, int chip)
 {
 }
 
-static int fsl_ifc_sram_init(uint32_t ver)
+static int fsl_ifc_sram_init(struct fsl_ifc_mtd *priv, uint32_t ver)
 {
 	struct fsl_ifc_runtime *ifc = ifc_ctrl->regs.rregs;
 	uint32_t cs = 0, csor = 0, csor_8k = 0, csor_ext = 0;
@@ -823,7 +822,7 @@ static int fsl_ifc_sram_init(uint32_t ver)
 		return 1;
 	}
 
-	cs = ifc_ctrl->cs_nand >> IFC_NAND_CSEL_SHIFT;
+	cs = priv->bank;
 
 	/* Save CSOR and CSOR_ext */
 	csor = ifc_in32(&ifc_ctrl->regs.gregs->csor_cs[cs].csor);
@@ -850,7 +849,7 @@ static int fsl_ifc_sram_init(uint32_t ver)
 	ifc_out32(&ifc->ifc_nand.col0, 0x0);
 
 	/* set the chip select for NAND Transaction */
-	ifc_out32(&ifc->ifc_nand.nand_csel, ifc_ctrl->cs_nand);
+	ifc_out32(&ifc->ifc_nand.nand_csel, priv->bank << IFC_NAND_CSEL_SHIFT);
 
 	/* start read seq */
 	ifc_out32(&ifc->ifc_nand.nandseq_strt, IFC_NAND_SEQ_STRT_FIR_STRT);
@@ -911,10 +910,8 @@ static int fsl_ifc_chip_init(int devnum, u8 *addr)
 		csor = ifc_in32(&gregs->csor_cs[priv->bank].csor);
 
 		if ((cspr & CSPR_V) && (cspr & CSPR_MSEL) == CSPR_MSEL_NAND &&
-		    (cspr & CSPR_BA) == CSPR_PHYS_ADDR(phys_addr)) {
-			ifc_ctrl->cs_nand = priv->bank << IFC_NAND_CSEL_SHIFT;
+		    (cspr & CSPR_BA) == CSPR_PHYS_ADDR(phys_addr))
 			break;
-		}
 	}
 
 	if (priv->bank >= MAX_BANKS) {
@@ -1029,7 +1026,7 @@ static int fsl_ifc_chip_init(int devnum, u8 *addr)
 
 	ver = ifc_in32(&gregs->ifc_rev);
 	if (ver >= FSL_IFC_V1_1_0)
-		ret = fsl_ifc_sram_init(ver);
+		ret = fsl_ifc_sram_init(priv, ver);
 	if (ret)
 		return ret;
 
Signature · 1 line
2.11.0
#1 lore raw

PATCH v2 1/1 mtd: nand: fsl-ifc: fix support of multiple NAND devices

Quoted 8 lines · Kurt Kanzenbach

Currently the chipselect used to identify the corresponding NAND chip is stored at the controller and only set during fsl_ifc_chip_init(). This way, only the last NAND chip is working, as the previous value of cs_nand gets overwritten.

In order to solve this issue the chipselect is computed on demand by evaluating the bank variable. Thus, the correct chipselect for each NAND chip operation is used.

Quoted 7 lines

Tested on hardware with two NAND chips connected to the IFC controller.

---
Changes for v2:
- get rid of cs_nand and compute chipselect on demand
- pass priv instead of mtd to fsl_ifc_sram_init()

Scott,

Are you OK with this version? Somehow this patch was assigned me. I can bring it in if you ack it.

York

#2 lore raw

PATCH v2 1/1 mtd: nand: fsl-ifc: fix support of multiple NAND devices

Quoted 22 lines · York Sun
Quoted 13 lines · Kurt Kanzenbach

Currently the chipselect used to identify the corresponding NAND chip is stored
at the controller and only set during fsl_ifc_chip_init(). This way, only the
last NAND chip is working, as the previous value of cs_nand gets overwritten.

In order to solve this issue the chipselect is computed on demand by evaluating
the bank variable. Thus, the correct chipselect for each NAND chip operation is
used.

Quoted 7 lines

Tested on hardware with two NAND chips connected to the IFC controller.

---
Changes for v2:
   - get rid of cs_nand and compute chipselect on demand
   - pass priv instead of mtd to fsl_ifc_sram_init()

Quoted 4 lines

Scott,

Are you OK with this version? Somehow this patch was assigned me. I can bring it in if you ack it.

-Scott

Applied
#3 lore raw

PATCH v2 1/1 mtd: nand: fsl-ifc: fix support of multiple NAND devices

Quoted 23 lines · Scott Wood
Quoted 22 lines · York Sun
Quoted 20 lines · Kurt Kanzenbach

Currently the chipselect used to identify the corresponding NAND chip is stored
at the controller and only set during fsl_ifc_chip_init(). This way, only the
last NAND chip is working, as the previous value of cs_nand gets overwritten.

In order to solve this issue the chipselect is computed on demand by evaluating
the bank variable. Thus, the correct chipselect for each NAND chip operation is
used.

Tested on hardware with two NAND chips connected to the IFC controller.

---
Changes for v2:
   - get rid of cs_nand and compute chipselect on demand
   - pass priv instead of mtd to fsl_ifc_sram_init()

Quoted 7 lines
Quoted 4 lines

Scott,

Are you OK with this version? Somehow this patch was assigned me. I can bring it in if you ack it.

Applied to fsl-qoriq master. Thanks.

York