[PATCH] [MTD] NAND: Prevents NAND subsystem from hogging the CPU.

[PATCH] [MTD] NAND: Prevents NAND subsystem from hogging the CPU.

[Rick Bronson]

  This is my first patch for mtd so bear with me...  sorry if I missed
something.

[PATCH] [MTD] NAND: Prevents NAND subsystem from hogging the
CPU waiting for ready line.

This patch prevents the nand_wait_ready chip from using too much
CPU time.  This was causing very long thread latencies.  The
solution, which is used elsewhere in this same file, is to call
cond_resched() in the wait loop.

[From: Rick Bronson <rick@efn.org>]
[Acked-by: Realname <realmail>]
Signed-off-by: Realname <realmail>


diff -ruN mtd-2.6.orig/drivers/mtd/nand/nand_base.c mtd-2.6/drivers/mtd/nand/nand_base.c
--- mtd-2.6.orig/drivers/mtd/nand/nand_base.c   2007-09-06 07:46:18.000000000 -0700
+++ mtd-2.6/drivers/mtd/nand/nand_base.c        2007-09-06 08:31:20.000000000 -0700
@@ -429,6 +429,7 @@
                if (chip->dev_ready(mtd))
                        break;
                touch_softlockup_watchdog();
+               cond_resched();  /* let others run */
        } while (time_before(jiffies, timeo));
        led_trigger_event(nand_led_trigger, LED_OFF);
 }

Re: [PATCH] [MTD] NAND: Prevents NAND subsystem from hogging the CPU.

[Jörn Engel]

On Thu, 6 September 2007 11:06:22 -0700, Rick Bronson wrote:
> 
> [From: Rick Bronson <rick@efn.org>]
> [Acked-by: Realname <realmail>]
> Signed-off-by: Realname <realmail>

Looks like whatever tool you used needs a bit of configuration.

Jörn

-- 
To announce that there must be no criticism of the President, or that we
are to stand by the President, right or wrong, is not only unpatriotic
and servile, but is morally treasonable to the American public.
-- Theodore Roosevelt, Kansas City Star, 1918

Re: [PATCH] [MTD] NAND: Prevents NAND subsystem from hogging the CPU.

[Artem Bityutskiy]

On Thu, 2007-09-06 at 11:06 -0700, Rick Bronson wrote:
>                 if (chip->dev_ready(mtd))
>                         break;
>                 touch_softlockup_watchdog();
> +               cond_resched();  /* let others run */
>         } while (time_before(jiffies, timeo));
>         led_trigger_event(nand_led_trigger, LED_OFF);
>  }

And I think this touch_softlockup_watchdog() is bogus.

-- 
Best regards,
Artem Bityutskiy (Битюцкий Артём)