linux-kernel - Re: [PATCH 148/196] Infiniband: make ipath driver use default driver groups.

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: Windows password security audit tool. GUI, reports in PDF.

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

Date:	Fri, 25 Jan 2008 14:23:41 -0800
From:	Greg KH <gregkh@...e.de>
To:	Roland Dreier <rdreier@...co.com>
Cc:	linux-kernel@...r.kernel.org,
	Bryan O'Sullivan <bryan.osullivan@...gic.com>,
	Arthur Jones <arthur.jones@...gic.com>,
	Cornelia Huck <cornelia.huck@...ibm.com>
Subject: Re: [PATCH 148/196] Infiniband: make ipath driver use default
	driver groups.

On Fri, Jan 25, 2008 at 02:11:14PM -0800, Roland Dreier wrote:
> So I think it is coming from the following code in ipath_sysfs.c:
> 
> 	int ipath_device_create_group(struct device *dev, struct ipath_devdata *dd)
> 	{
> 		int ret;
> 	
> 		ret = sysfs_create_group(&dev->kobj, &dev_attr_group);
> 		if (ret)
> 			goto bail;
> 	
> 		ret = sysfs_create_group(&dev->kobj, &dev_counter_attr_group);
> 		if (ret)
> 			goto bail_attrs;
> 	
> 		sysfs_remove_group(&dev->kobj, &dev_counter_attr_group);
> 	bail_attrs:
> 		sysfs_remove_group(&dev->kobj, &dev_attr_group);
> 	bail:
> 		return ret;
> 	}
> 
> note that the success path falls through into the error path and
> removes the groups it just created, which means that on initialization
> the groups are already removed, so we hit the BUG on the second try to
> remove the groups on module exit.
> 
> And I think this chunk of your patch:
> 
> 	-	snprintf(unit, sizeof(unit), "%02d", dd->ipath_unit);
> 	-	ret = sysfs_create_link(&dev->driver->kobj, &dev->kobj, unit);
> 	-	if (ret == 0)
> 	-		goto bail;
> 	-
> 
> was what broke it.  (Note the "if (ret == 0) then skip error
> unwinding" code that got deleted by mistake -- not surprising given
> how hard to read that construction is)
> 
> My bad for not testing -mm better.  I'll queue this up to fix it:

Thanks for the fix, and sorry for causing that, it was a complex error
path :)

Although a BUG() for when we try to remove a file that is no longer
there seems pretty harsh, I think I'll change that to a WARN_ON() so
that people just know to fix up their code, not kill the kernel
entirely.

thanks,

greg k-h
--
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/