2014-01-07 23:20:07

by EUNBONG SONG

Subject: [PATCH] mtd: nand: make more readable panic_nand_wait_ready() and nand_wait_ready()

panic_nand_wait_ready() and nand_wait_ready() calls dev_ready() without checking if it exists.
This patch add check routine dev_ready() before run dev_ready()
and this makes the code more readable

Signed-off-by: Eunbong Song <[email protected]>
---
drivers/mtd/nand/nand_base.c | 12 ++++++++----
1 files changed, 8 insertions(+), 4 deletions(-)

diff --git a/drivers/mtd/nand/nand_base.c b/drivers/mtd/nand/nand_base.c
index bd39f7b..110db78 100644
--- a/drivers/mtd/nand/nand_base.c
+++ b/drivers/mtd/nand/nand_base.c
@@ -473,8 +473,10 @@ static void panic_nand_wait_ready(struct mtd_info *mtd, unsigned long timeo)

/* Wait for the device to get ready */
for (i = 0; i < timeo; i++) {
- if (chip->dev_ready(mtd))
- break;
+ if(chip->dev_ready){
+ if (chip->dev_ready(mtd))
+ break;
+ }
touch_softlockup_watchdog();
mdelay(1);
}
@@ -493,8 +495,10 @@ void nand_wait_ready(struct mtd_info *mtd)
led_trigger_event(nand_led_trigger, LED_FULL);
/* Wait until command is processed or timeout occurs */
do {
- if (chip->dev_ready(mtd))
- break;
+ if(chip->dev_ready){
+ if (chip->dev_ready(mtd))
+ break;
+ }
touch_softlockup_watchdog();
} while (time_before(jiffies, timeo));
led_trigger_event(nand_led_trigger, LED_OFF);
--
1.7.0.4
????{.n?+???????+%?????ݶ??w??{.n?+????{??G?????{ay?ʇڙ?,j??f???h?????????z_??(?階?ݢj"???m??????G????????????&???~???iO???z??v?^?m????????????I?

2014-01-20 19:19:15

by Brian Norris

Subject: Re: [PATCH] mtd: nand: make more readable panic_nand_wait_ready() and nand_wait_ready()

On Tue, Jan 07, 2014 at 11:19:56PM +0000, Eunbong Song wrote:
>
> panic_nand_wait_ready() and nand_wait_ready() calls dev_ready() without checking if it exists.

There are many occasions where we call a function pointer without
(locally) checking that it is non-NULL. That is by design.

If you see a problem with this behavior, please fix this properly rather
than covering it up. This patch will just make faulty drivers spin for
20ms rather than hitting a bug. Instead, you should aim to fix buggy
drivers that call dev_ready() without assigning it; or even better,
defensively improve nand_base to detect those drivers which are
utilizing nand_command_lp() or nand_command() without assigning
dev_ready().

> This patch add check routine dev_ready() before run dev_ready()
> and this makes the code more readable

This patch does not make the code more readable, nor does it improve the
state of the original code much. Please solve and document a real
problem.

Thanks,
Brian