RE: TPM driver changes to support multiple locality

["Agarwal, Lomesh" <lomesh.agarwal@intel.com>]

-----Original Message-----
From: Randy Dunlap [mailto:randy.dunlap@oracle.com] 
Sent: Thursday, October 11, 2007 4:02 PM
To: Agarwal, Lomesh
Cc: Valdis.Kletnieks@vt.edu; linux-kernel@vger.kernel.org
Subject: Re: TPM driver changes to support multiple locality

On Thu, 11 Oct 2007 15:46:03 -0700 Agarwal, Lomesh wrote:

> Attached is the patch which resolves all the comments.

Inline patches are preferred so that reviewers can comment on them
more easily.

What mail client are you using?
[Agarwal, Lomesh] I am using MS Outlook. Earlier you said inline patches
have problem because of mail client. That's why I sent it as attachment.

The patch has trailing CRs on each line ("DOS mode").

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Just verifying:  this TPM device has interrupts per locality?

+	/* check if interrupt is meant for this locality */
+	if (check_locality(chip, locality) < 0)
+		return IRQ_NONE;

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[Agarwal, Lomesh] TPM device has only one interrupt. so on receiving
interrupt driver has to make sure that its meant for its locality.

init_tis() still seems to have some problems.

 static int __init init_tis(void)
 {
+#define DEVNAME_SIZE 10
+
 	int rc;
 
+	if ((locality < 0) || (locality > 4))
+		return PTR_ERR(pdev);

pdev hasn't been set (so it's NULL).

+		pdev = platform_device_register_simple(devname, -1,
NULL, 0);
+		if (IS_ERR(pdev))
 			return PTR_ERR(pdev);

Error path above needs to call driver_unregister().
[Agarwal, Lomesh] attached is the new patch.


---
~Randy

Attachment: tpm_tis.c.patch
Description: tpm_tis.c.patch