[U-Boot] [PATCH 1/2] nand: zynq: Fix driver initialization

[U-Boot] [PATCH 1/2] nand: zynq: Fix driver initialization

[Michal Simek]

From: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>

This driver is currently broken, refusing to initialize properly.

The reason is that get_nand_dev_by_index() was being called before
nand_register(), thus returning a pointer into uninitialized memory.
In other words, the struct mtd_info used by the driver is total junk.

Fix it by getting the correct struct mtd_info, via nand_to_mtd()
on the driver's struct nand_chip.

Tested on a custom board, where the CPU is halted without this patch.

Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Reviewed-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
---

 drivers/mtd/nand/zynq_nand.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/mtd/nand/zynq_nand.c b/drivers/mtd/nand/zynq_nand.c
index 6494196049f1..9f6ff3d045c2 100644
--- a/drivers/mtd/nand/zynq_nand.c
+++ b/drivers/mtd/nand/zynq_nand.c
@@ -1025,7 +1025,7 @@ int zynq_nand_init(struct nand_chip *nand_chip, int devnum)
 	}
 
 	xnand->nand_base = (void __iomem *)ZYNQ_NAND_BASEADDR;
-	mtd = get_nand_dev_by_index(0);
+	mtd = nand_to_mtd(nand_chip);
 
 	nand_chip->priv = xnand;
 	mtd->priv = nand_chip;
-- 
2.17.0

[U-Boot] [PATCH 2/2] nand: zynq: Cleanup initialization

[Michal Simek]

From: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>

CONFIG_NAND_ZYNQ selects CONFIG_SYS_NAND_SELF_INIT, so the
driver doesn't have to play any ifdef game.

Also, we can mark zynq_nand_init() as static and get rid
of the mach-specific nand.h header.

This is really a revert of:
"mtd: zynq: nand: Move board_nand_init() function to board.c"
(sha1: 310995d9f91ae56082b49be06fe8c3d01424f8f6)

Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Reviewed-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
---

 arch/arm/mach-zynq/include/mach/nand.h | 9 ---------
 drivers/mtd/nand/zynq_nand.c           | 6 ++----
 2 files changed, 2 insertions(+), 13 deletions(-)
 delete mode 100644 arch/arm/mach-zynq/include/mach/nand.h

diff --git a/arch/arm/mach-zynq/include/mach/nand.h b/arch/arm/mach-zynq/include/mach/nand.h
deleted file mode 100644
index 61ef45f5828a..000000000000
--- a/arch/arm/mach-zynq/include/mach/nand.h
+++ /dev/null
@@ -1,9 +0,0 @@
-/*
- * Copyright (C) 2017 National Instruments Corp.
- *
- * SPDX-License-Identifier:	GPL-2.0+
- */
-
-#include <nand.h>
-
-void zynq_nand_init(void);
diff --git a/drivers/mtd/nand/zynq_nand.c b/drivers/mtd/nand/zynq_nand.c
index 9f6ff3d045c2..2d4e8b4736ba 100644
--- a/drivers/mtd/nand/zynq_nand.c
+++ b/drivers/mtd/nand/zynq_nand.c
@@ -1006,7 +1006,7 @@ static int zynq_nand_device_ready(struct mtd_info *mtd)
 	return 0;
 }
 
-int zynq_nand_init(struct nand_chip *nand_chip, int devnum)
+static int zynq_nand_init(struct nand_chip *nand_chip, int devnum)
 {
 	struct zynq_nand_info *xnand;
 	struct mtd_info *mtd;
@@ -1192,14 +1192,12 @@ fail:
 	return err;
 }
 
-#ifdef CONFIG_SYS_NAND_SELF_INIT
 static struct nand_chip nand_chip[CONFIG_SYS_MAX_NAND_DEVICE];
 
-void __weak board_nand_init(void)
+void board_nand_init(void)
 {
 	struct nand_chip *nand = &nand_chip[0];
 
 	if (zynq_nand_init(nand, 0))
 		puts("ZYNQ NAND init failed\n");
 }
-#endif
-- 
2.17.0

[U-Boot] [PATCH 1/2] nand: zynq: Fix driver initialization

[Ezequiel Garcia]

Hi Michal,

On 19 April 2018 at 08:10, Michal Simek <michal.simek@xilinx.com> wrote:
> From: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
>
> This driver is currently broken, refusing to initialize properly.
>
> The reason is that get_nand_dev_by_index() was being called before
> nand_register(), thus returning a pointer into uninitialized memory.
> In other words, the struct mtd_info used by the driver is total junk.
>
> Fix it by getting the correct struct mtd_info, via nand_to_mtd()
> on the driver's struct nand_chip.
>
> Tested on a custom board, where the CPU is halted without this patch.
>
> Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
> Reviewed-by: Michal Simek <michal.simek@xilinx.com>
> Signed-off-by: Michal Simek <michal.simek@xilinx.com>

Thanks for collecting these and submitting them!

BTW, I thought there was no need for a reviewed-by on top
of a signed-off-by. It's kind of assumed that if you are signing
something, you have reviewed it in the first place.

Or at least that's how I've been doing it.
-- 
Ezequiel García, VanguardiaSur
www.vanguardiasur.com.ar

[U-Boot] [PATCH 1/2] nand: zynq: Fix driver initialization

[Michal Simek]

Hi Ezequiel,

On 19.4.2018 17:16, Ezequiel Garcia wrote:
> Hi Michal,
> 
> On 19 April 2018 at 08:10, Michal Simek <michal.simek@xilinx.com> wrote:
>> From: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
>>
>> This driver is currently broken, refusing to initialize properly.
>>
>> The reason is that get_nand_dev_by_index() was being called before
>> nand_register(), thus returning a pointer into uninitialized memory.
>> In other words, the struct mtd_info used by the driver is total junk.
>>
>> Fix it by getting the correct struct mtd_info, via nand_to_mtd()
>> on the driver's struct nand_chip.
>>
>> Tested on a custom board, where the CPU is halted without this patch.
>>
>> Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
>> Reviewed-by: Michal Simek <michal.simek@xilinx.com>
>> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
> 
> Thanks for collecting these and submitting them!
> 
> BTW, I thought there was no need for a reviewed-by on top
> of a signed-off-by. It's kind of assumed that if you are signing
> something, you have reviewed it in the first place.
> 
> Or at least that's how I've been doing it.

My initial goal was to send it without my SoB line but my good git setup
is adding it automatically. I didn't touch your patches that's why I
wanted to say that I have reviewed it and at the end my SoB line will be
there too because it will go via my tree.

Thanks,
Michal