2018-01-25 10:30:07

by Jia-Ju Bai

Subject: [PATCH] ata: pata_it821x: Replace mdelay with usleep_range in it821x_firmware_command

After checking all possible call chains to it821x_firmware_command here,
my tool finds that it821x_firmware_command
is never called in atomic context,
namely never in an interrupt handler or holding a spinlock.
And it821x_firmware_command calls kmalloc(GFP_KERNEL),
so it proves again that it821x_firmware_command
can call functions which can sleep.
Thus mdelay can be replaced with usleep_range to avoid busy wait.

This is found by a static analysis tool named DCNS written by myself.

Signed-off-by: Jia-Ju Bai <[email protected]>
---
drivers/ata/pata_it821x.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/ata/pata_it821x.c b/drivers/ata/pata_it821x.c
index 7a21edf..8468b30 100644
--- a/drivers/ata/pata_it821x.c
+++ b/drivers/ata/pata_it821x.c
@@ -683,7 +683,7 @@ static u8 *it821x_firmware_command(struct ata_port *ap, u8 cmd, int len)
ioread16_rep(ap->ioaddr.data_addr, buf, len/2);
return (u8 *)buf;
}
- mdelay(1);
+ usleep_range(500, 1000);
}
kfree(buf);
printk(KERN_ERR "it821x_firmware_command: timeout\n");
--
1.7.9.5

2018-01-25 15:32:42

by Tejun Heo

Subject: Re: [PATCH] ata: pata_it821x: Replace mdelay with usleep_range in it821x_firmware_command

On Thu, Jan 25, 2018 at 06:32:59PM +0800, Jia-Ju Bai wrote:
> After checking all possible call chains to it821x_firmware_command here,
> my tool finds that it821x_firmware_command
> is never called in atomic context,
> namely never in an interrupt handler or holding a spinlock.
> And it821x_firmware_command calls kmalloc(GFP_KERNEL),
> so it proves again that it821x_firmware_command
> can call functions which can sleep.
> Thus mdelay can be replaced with usleep_range to avoid busy wait.
>
> This is found by a static analysis tool named DCNS written by myself.
>
> Signed-off-by: Jia-Ju Bai <[email protected]>

Applied to libata/for-4.16.

Thanks.

--
tejun