Re: [PATCH 2/2] arcmsr: adds code for support areca new adapter ARC1203

[Joe Perches <joe@perches.com>]

On Tue, 2015-11-24 at 17:53 +0800, Ching Huang wrote:
> On Tue, 2015-11-24 at 01:33 -0800, Joe Perches wrote:
> > On Tue, 2015-11-24 at 16:17 +0800, Ching Huang wrote:
> > > From: Ching Huang <ching2048@areca.com.tw>
> > > 
> > > Support areca new PCIe to SATA RAID adapter ARC1203
> > 
> > Why add the dma_free_coherent to an old data path?
> > Is that a general bug fix that should be backported?
> 
> That's right. It's need to release the allocated resource for failed
> condition.

Then the dma_free_coherent addition should be a separate patch.

Style trivia:

The goto to another error path like that is odd and
the label is unintelligible.

Ideally error condition handling would use a goto and
a separate and obviously named label.  Use multiple
labels for cases with more complicated unwinding.

Dan Carpenter has written about this several times.

For this use, something like:

	writel(ARCMSR_MESSAGE_START_DRIVER_MODE, reg->drv2iop_doorbell);
	if (!arcmsr_hbaB_wait_msgint_ready(acb)) {
		logging_message(...);
		goto
err_free_resource;
	}
	writel(ARCMSR_MESSAGE_GET_CONFIG, reg->drv2iop_doorbell);
	if (!arcmsr_hbaB_wait_msgint_ready(acb)) {
		logging_message(...);
		goto err_free_resource;
	}

	[success path...]

	return true;

err_free_resource:
	dma_free_coherent(...);

	return false;
}