Re: [PATCH] synchronize_irq needs a barrier

On Saturday 20 October 2007 07:46:24 Benjamin Herrenschmidt wrote:

quoted

I probably need to add this synchronize_irq() logic in dmfe.c too, but I probably do it later,
I think I am overestimating this race, since most drivers don't do dev->insuspend checks in IRQ handler.
Maybe even just use free_irq() after all....

Most drivers are probably underestimating the race :-)

free_irq() would work provided that you did the masking on chip before
(and unmask only after request_irq on the way back in). But it's a bit
like using a 10 tons truck to crush an ant... 

Agreed.

So, I will add synchronize_irq() to both saa7134, and dmfe, the two drivers that their .suspend/.resume 
routines were written by me.

I already added a synchronize_irq()  plus few more fixes to the driver , but those patches are still in v4l tree.

I now has this:

	saa_writel(SAA7134_IRQ1, 0);
	saa_writel(SAA7134_IRQ2, 0);
	saa_writel(SAA7134_MAIN_CTRL, 0);

	synchronize_irq(pci_dev->irq);
	dev->insuspend = 1;

and I will probably need (with the synchronize_irq patch applied)


	/* Disable interrupts, DMA, and rest of the chip*/
	saa_writel(SAA7134_IRQ1, 0);
	saa_writel(SAA7134_IRQ2, 0);
	saa_writel(SAA7134_MAIN_CTRL, 0);
	
	dev->insuspend = 1;
	synchronize_irq(pci_dev->irq);

	/* ACK pending interrupts just in case*/
	saa_writel(SAA7134_IRQ_REPORT,saa_readl(SAA7134_IRQ_REPORT));

	......
This should be bullet-proof.

Ben.

Best regards,
	Maxim Levitsky