linux-kernel - Re: [PATCH 3/7] ahci: Use new interfaces for MSI/MSI-X enablement

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Tue, 14 Jan 2014 09:50:40 +0100
From:	Alexander Gordeev <agordeev@...hat.com>
To:	Bjorn Helgaas <bhelgaas@...gle.com>
Cc:	linux-kernel@...r.kernel.org, Tejun Heo <tj@...nel.org>,
	linux-ide@...r.kernel.org, linux-pci@...r.kernel.org
Subject: Re: [PATCH 3/7] ahci: Use new interfaces for MSI/MSI-X enablement

On Mon, Jan 13, 2014 at 12:12:20PM -0700, Bjorn Helgaas wrote:
> > -	nvec = rc;
> > -	rc = pci_enable_msi_block(pdev, nvec);
> > -	if (rc)
> > +	if (pci_enable_msi_range(pdev, nvec, nvec) < 0)
> >  		goto intx;
> >  
> >  	return nvec;
> >  
> >  single_msi:
> > -	rc = pci_enable_msi(pdev);
> > -	if (rc)
> > +	if (pci_enable_msi_range(pdev, 1, 1) < 0)
> 
> This part doesn't seem like an improvement.  There are a hundred or so
> callers of pci_enable_msi() that only want a single MSI.  Is there any
> benefit in changing them to use pci_enable_msi_range()?

In this particular case it reads better to me as one sees on the screen
pci_enable_msi_range(pdev, nvec, nvec) and pci_enable_msi_range(pdev, 1, 1)
calls. That allows to avoid switching in mind between negative-or-positive
return in the former call and negative-or-zero return from pci_enable_msi()
if we had it.

But in most cases when single MSI is enabled we do cause complication
with the patterns below (which I expect I am going be hated for ;) ):


-	rc = pci_enable_msi(pdev);
-	if (rc)
+	rc = pci_enable_msi_range(pdev, 1, 1);
+	if (rc < 0)
		...


-	rc = pci_enable_msi(pdev);
-	if (!rc)
+	rc = pci_enable_msi_range(pdev, 1, 1);
+	if (rc > 0)
		...

 
I think we have a tradeoff between the interface simplicity and code clarity.
What if we try to address both goals by making pci_enable_msi() a helper over
pci_enable_msi_range(pdev, 1, 1)? In this case the return value will match
negative-or-positive binary semantics while reads almost as good as it used to:


-	rc = pci_enable_msi(pdev);
-	if (rc)
+	rc = pci_enable_msi(pdev);
+	if (rc < 0)
		...


-	rc = pci_enable_msi(pdev);
-	if (!rc)
+	rc = pci_enable_msi(pdev);
+	if (rc > 0)
		...


The whole interface would not be inflated as well, with just:

diff --git a/Documentation/PCI/MSI-HOWTO.txt b/Documentation/PCI/MSI-HOWTO.txt
index a8d0100..fa0b27d 100644
--- a/Documentation/PCI/MSI-HOWTO.txt
+++ b/Documentation/PCI/MSI-HOWTO.txt
@@ -158,6 +158,9 @@ static int foo_driver_enable_single_msi(struct pci_dev *pdev)
 	return pci_enable_msi_range(pdev, 1, 1);
 }
 
+A helper function pci_enable_msi() could be used instead. Note, as just
+one MSI is requested it could return either a negative errno or 1.
+
 4.2.2 pci_disable_msi
 
 void pci_disable_msi(struct pci_dev *dev)


-- 
Regards,
Alexander Gordeev
agordeev@...hat.com
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/