RE: Comments on 2.6.10 schedule_timeout please

[Mukund JB.]

Dear Nish,
Sorry that I cannot share the entire code as I have to face some licensing issues.
Please find the required code snapshots below. 
please ask if I have missed something ...

Please see the source code explained briefly below:-

/* my_msleep() sleep milli(msec) seconds */
void my_msleep(int msec)
{
	current->state = TASK_INTERRUPTIBLE;
	schedule_timeout((msec * HZ)/1000);
}
{
MMCSD_RESPONSE2
{
	..............
	.............
	/* gather the hardware interrupt reg status that is updated in the ISR context*/
	..............
	............
}

StandbyMMCSD(PCMMCSD pSD)
{
	do
	{
		tifm_msleep(300);
	}while(!MMCSD_RESPONSE2(pSD, 31, 31, false));
}

Code Description:
This is a part SD card Driver for 2.6.10 kernel & it works fine.
This part of the code is called in the ISR context from the bottomhalf when the SD Card is inserted to initialize the SD slot & card.

How did I diagnose the delay:

1) After inserting the card, I tried pressing the character '1' expecting the shedule_tiemout works by giving the keyboard  process its time to execute and print 1's onto the screen. Instead, I found the a BIG delay of 20 characters. i.e. 20  characters printed at once after the SD card initialization is done.
2) With debug messages immediately before & after the schedule_timeout call.
3) Commenting the schedule_timeout call.
 
Regards,
Mukund Jampala



-----Original Message-----
From: Nish Aravamudan [mailto:nish.aravamudan@xxxxxxxxx]
Sent: Monday, November 07, 2005 8:24 PM
To: Mukund JB.
Cc: Adrian Bunk; linux-kernel@xxxxxxxxxxxxxxx
Subject: Re: Comments on 2.6.10 schedule_timeout please


On 11/7/05, Mukund JB. <mukundjb@xxxxxxxxxxxxxxxxxxxxx> wrote:
>
> Dear Kernel Developers,
>
> I have noticed the schedule_timeout behaving somewhat different as penned
> from the Linux 2.6 Oreelly books.
> I have developed a SD card Driver for 2.6.10 kernel & it works fine.
> I needed a hardware reg to update that take a time of 300ms. I have issued a
> call like..
>
> set_current_state(TASK_INTERRUPTIBLE);
> schedule_timeout (300*HZ/1000);

Full code or function snippet, please.

> But, when I finally use it I get a sufficient delay which looks like a looped delay
> not allowing the keyboard to print messages on the screen.

This would be easier to diagnose if you shared all of the code you are
using *and* verified this occurred with a current kernel (2.6.10 is
old.).

Thanks,
Nish
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/