[PATCH] [Outreachy kernel] staging: speakup: prefer usleep_range over udelay

[PATCH] [Outreachy kernel] staging: speakup: prefer usleep_range over udelay

[Payal Kshirsagar]

Since usleep_range is built on top of hrtimers, the
wakeup will be very precise (ish), thus a simple
usleep function would likely introduce a large number
of undesired interrupts.
Thus udelay replaced with usleep_range.

Signed-off-by: Payal Kshirsagar <payal.s.kshirsagar.98@gmail.com>
---
 drivers/staging/speakup/speakup_decpc.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/drivers/staging/speakup/speakup_decpc.c b/drivers/staging/speakup/speakup_decpc.c
index 798c42d..10ca19b 100644
--- a/drivers/staging/speakup/speakup_decpc.c
+++ b/drivers/staging/speakup/speakup_decpc.c
@@ -251,7 +251,7 @@ static int dt_waitbit(int bit)
 	while (--timeout > 0) {
 		if ((dt_getstatus() & bit) == bit)
 			return 1;
-		udelay(50);
+		usleep_range(50, 50 + 10);
 	}
 	return 0;
 }
@@ -265,7 +265,7 @@ static int dt_wait_dma(void)
 	while (--timeout > 0) {
 		if ((dt_getstatus() & STAT_dma_state) == state)
 			return 1;
-		udelay(50);
+		usleep_range(50, 50 + 10);
 	}
 	dma_state = dt_getstatus() & STAT_dma_state;
 	return 1;
@@ -283,7 +283,7 @@ static int dt_ctrl(u_int cmd)
 	dt_sendcmd(CMD_control | cmd);
 	outb_p(0, speakup_info.port_tts + 6);
 	while (dt_getstatus() & STAT_cmd_ready) {
-		udelay(20);
+		usleep_range(20, 20 + 10);
 		if (--timeout == 0)
 			break;
 	}
@@ -302,20 +302,20 @@ static void synth_flush(struct spk_synth *synth)
 	while (dt_ctrl(CTRL_flush)) {
 		if (--timeout == 0)
 			break;
-		udelay(50);
+		usleep_range(50, 50 + 10);
 	}
 	for (timeout = 0; timeout < 10; timeout++) {
 		if (dt_waitbit(STAT_dma_ready))
 			break;
-		udelay(50);
+		usleep_range(50, 50 + 10);
 	}
 	outb_p(DMA_sync, speakup_info.port_tts + 4);
 	outb_p(0, speakup_info.port_tts + 4);
-	udelay(100);
+	usleep_range(100, 100 + 10);
 	for (timeout = 0; timeout < 10; timeout++) {
 		if (!(dt_getstatus() & STAT_flushing))
 			break;
-		udelay(50);
+		usleep_range(50, 50 + 10);
 	}
 	dma_state = dt_getstatus() & STAT_dma_state;
 	dma_state ^= STAT_dma_state;
-- 
2.7.4

Re: [PATCH] [Outreachy kernel] staging: speakup: prefer usleep_range over udelay

[Greg KH]

On Mon, Mar 18, 2019 at 11:36:24PM +0530, Payal Kshirsagar wrote:
> Since usleep_range is built on top of hrtimers, the
> wakeup will be very precise (ish), thus a simple
> usleep function would likely introduce a large number
> of undesired interrupts.
> Thus udelay replaced with usleep_range.
> 
> Signed-off-by: Payal Kshirsagar <payal.s.kshirsagar.98@gmail.com>
> ---
>  drivers/staging/speakup/speakup_decpc.c | 14 +++++++-------
>  1 file changed, 7 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/staging/speakup/speakup_decpc.c b/drivers/staging/speakup/speakup_decpc.c
> index 798c42d..10ca19b 100644
> --- a/drivers/staging/speakup/speakup_decpc.c
> +++ b/drivers/staging/speakup/speakup_decpc.c
> @@ -251,7 +251,7 @@ static int dt_waitbit(int bit)
>  	while (--timeout > 0) {
>  		if ((dt_getstatus() & bit) == bit)
>  			return 1;
> -		udelay(50);
> +		usleep_range(50, 50 + 10);

I've said this in the past, only do this type of change if you can
actually test it, and know it will be ok.  For some stuff like this, it
_might_ be ok, but you just expanded the timeout range potentially a lot
longer than before, and do we know that is ok?

I sure do not :)

So just leave these things alone unless you know the hardware will keep
working properly.

thanks,

greg k-h

Re: [PATCH] [Outreachy kernel] staging: speakup: prefer usleep_range over udelay

[Payal Kshirsagar]

[-- Attachment #1: Type: text/plain, Size: 1537 bytes --]

On Tue, Mar 19, 2019, 12:15 AM Greg KH <gregkh@linuxfoundation.org> wrote:

> On Mon, Mar 18, 2019 at 11:36:24PM +0530, Payal Kshirsagar wrote:
> > Since usleep_range is built on top of hrtimers, the
> > wakeup will be very precise (ish), thus a simple
> > usleep function would likely introduce a large number
> > of undesired interrupts.
> > Thus udelay replaced with usleep_range.
> >
> > Signed-off-by: Payal Kshirsagar <payal.s.kshirsagar.98@gmail.com>
> > ---
> >  drivers/staging/speakup/speakup_decpc.c | 14 +++++++-------
> >  1 file changed, 7 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/staging/speakup/speakup_decpc.c
> b/drivers/staging/speakup/speakup_decpc.c
> > index 798c42d..10ca19b 100644
> > --- a/drivers/staging/speakup/speakup_decpc.c
> > +++ b/drivers/staging/speakup/speakup_decpc.c
> > @@ -251,7 +251,7 @@ static int dt_waitbit(int bit)
> >       while (--timeout > 0) {
> >               if ((dt_getstatus() & bit) == bit)
> >                       return 1;
> > -             udelay(50);
> > +             usleep_range(50, 50 + 10);
>
> I've said this in the past, only do this type of change if you can
> actually test it, and know it will be ok.  For some stuff like this, it
> _might_ be ok, but you just expanded the timeout range potentially a lot
> longer than before, and do we know that is ok?
>
> I sure do not :)
>
> So just leave these things alone unless you know the hardware will keep
> working properly.
>
> thanks,
>
> greg k-h
>


Okay I'll drop this patch .
Regards,
Payal

>

[-- Attachment #2: Type: text/html, Size: 2403 bytes --]