Re: Infinite looping in omap2430.c USB driver
From: Felipe Balbi <hidden>
Date: 2012-08-14 08:01:55
Also in:
lkml
On Tue, Aug 14, 2012 at 07:46:59AM +1000, NeilBrown wrote:
On Mon, 13 Aug 2012 17:32:34 +0300 Felipe Balbi [off-list ref] wrote:quoted
Hi, On Mon, Aug 13, 2012 at 12:34:53PM +1000, NeilBrown wrote:quoted
On Thu, 9 Aug 2012 14:15:51 +0300 Felipe Balbi [off-list ref] wrote:quoted
hehe, that's nasty. Please send a patch converting to a try count and a udelay_range(), or something.how's this? Thanks, NeilBrown From: NeilBrown <redacted> Date: Mon, 13 Aug 2012 12:32:58 +1000 Subject: [PATCH] omap2430: don't loop indefinitely in interrupt. When called during resume_irqs, omap2430_musb_set_vbus() is run with interrupts disabled, In that case 'jiffies' never changes so the loop can loop forever. So impose a maximum loop count and add an 'mdelay' to ensure we wait a reasonable amount of time for bit to be cleared. This fixes a hang on resume. Signed-of-by: NeilBrown [off-list ref]diff --git a/drivers/usb/musb/omap2430.c b/drivers/usb/musb/omap2430.c index c7785e8..8a93381 100644 --- a/drivers/usb/musb/omap2430.c +++ b/drivers/usb/musb/omap2430.c@@ -34,6 +34,7 @@ #include <linux/dma-mapping.h> #include <linux/pm_runtime.h> #include <linux/err.h> +#include <linux/delay.h> #include "musb_core.h" #include "omap2430.h"@@ -145,6 +146,7 @@ static void omap2430_musb_set_vbus(struct musb *musb, int is_on) if (is_on) { if (musb->xceiv->state == OTG_STATE_A_IDLE) { + int loops = 100; /* start the session */ devctl |= MUSB_DEVCTL_SESSION; musb_writeb(musb->mregs, MUSB_DEVCTL, devctl);@@ -154,9 +156,11 @@ static void omap2430_musb_set_vbus(struct musb *musb, int is_on) */ while (musb_readb(musb->mregs, MUSB_DEVCTL) & 0x80) { + mdelay(5);I would prefer udelay_range() as it will let scheduler group timers. Something like: udelay_range(3000, 5000); should do, I gues...Except that there is no udelay_range :-( There is a usleep_range, but that can only be used from non-atomic context and in the problem case interrupts are disabled and a spinlock is held so we
my bad. Got confused. I will apply the original patch then. Thanks. -- balbi
Attachments
- signature.asc [application/pgp-signature] 836 bytes