linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Trent Piepho <xyzzy@speakeasy.org>
To: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: v4l-dvb maintainer list <v4l-dvb-maintainer@linuxtv.org>,
	Linux and Kernel Video <video4linux-list@redhat.com>,
	linux-kernel@vger.kernel.org
Subject: Re: [v4l-dvb-maintainer] Re: [PATCH] V4L: struct video_device corruption
Date: Sun, 23 Jul 2006 19:16:00 -0700 (PDT)	[thread overview]
Message-ID: <Pine.LNX.4.58.0607231828430.16282@shell3.speakeasy.net> (raw)
In-Reply-To: <1153647324.22089.32.camel@praia>

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: TEXT/PLAIN; charset=X-UNKNOWN, Size: 2501 bytes --]

On Sun, 23 Jul 2006, Mauro Carvalho Chehab wrote:
> Ok, I had some time to fix the broken dependencies. Please look at the
> two commits I did today.

Two things I noticed.  bttv didn't use to _require_ V4L1.  I use bttv and have
V4L1 turned off.  I guess this is just because bttv ignored the V4L1 setting?

Second, the fix for class_device_create_file() doesn't roll-back properly
on failure.  I already posted a patch which does this correctly.  Attached
is the patch against the current Hg, go ahead and import it.

Also attached is patch to bttv-driver.c that has it use
class_device_create_file(), with an error message and handling of failure.


> One patch removes HAVE_V4L1 check at drivers. It also include some checks for
> V4L1_COMPAT on some core files that should implement both calls to provide
> support for V4L1 drivers.
>
> The other fixes the broken dependencies for drivers that still need V4L1
> to work properly.
>
> Cheers,
> Mauro.
>
> Em Sex, 2006-07-21 às 15:55 -0700, Trent Piepho escreveu:
> > On Fri, 21 Jul 2006, Mauro Carvalho Chehab wrote:
> > > config VIDEO_BT848
> > >         tristate "BT848 Video For Linux"
> > >         depends on VIDEO_DEV && PCI && I2C && VIDEO_V4L2
> > >
> > > Argh! it should be V4L1 instead!
> >
> > You can compile and use bt848 without V4L1 turned on.  It still has some
> > V4L1 functions defined.
> >
> > > > All these files include v4l2-dev.h and have HAVE_V4L1 defined when V4L1 is
> > > > not turned on in Kconfig.  There files are all buildable when V4L1 is off;
> > > > they don't depend on it in Kconfig.
> > > Some of the above drivers are V4L2, like tda9887, tuner-core,
> > > tuner-simple, msp3400, cs53l32a, tveeprom, wm87xx. Maybe they are just
> > > including the wrong headers. We should try to change to videodev2.h and
> > > see what happens with all those drivers. The ones that break should me
> > > marked with the proper requirement on Kconfig.
> > >
> > > Some of they need some #ifdef inside. For example, compat_ioctl32 should
> > > handle both APIs, since it is a generic code to fix 32 bit calls to 64
> > > bit kernel.
> >
> > I think this is pretty much what I've been saying.  Drivers should:
> > A. Not include videodev.h, but use videodev2.h
> > B. Include videodev.h, but be marked V4L1 in Kconfig
> > C. #ifdef around videodev.h (and code that needs videodev.h), so it
> >    is not included or needed when V4L1 is turned off.
> Cheers,
> Mauro.
>
>

[-- Attachment #2: Type: TEXT/PLAIN, Size: 2823 bytes --]

# HG changeset patch
# User Trent Piepho <xyzzy@speakeasy.org>
# Node ID b900796fdfb4bc87642c470515f88ed7ae92ced3
# Parent  e33331856212e707150dc85fda486240eb7feae0
Handle class_device_create_file errors

From: Trent Piepho <xyzzy@speakeasy.org>

Add proper error checking and roll-back for failure of
class_device_create_file() in videodev.c.  Print error messages and
unroll partially created sysfs entries.

Also, failure of class_device_register() in video_register_device() is
handled correctly.  It was failing to de-allocate the minor number.  This
must be done in video_register_device(), since the caller has no way of
knowing if failure occurred before or after the class device was
registered.

Also added an error message if video_register_device() is called with
an unknown type, which should never happen.

Signed-off-by: Trent Piepho <xyzzy@speakeasy.org>

diff -r e33331856212 -r b900796fdfb4 linux/drivers/media/video/videodev.c
--- a/linux/drivers/media/video/videodev.c	Sun Jul 23 18:36:01 2006 -0700
+++ b/linux/drivers/media/video/videodev.c	Sun Jul 23 19:08:21 2006 -0700
@@ -1557,6 +1557,8 @@ int video_register_device(struct video_d
 			name_base = "radio";
 			break;
 		default:
+			printk(KERN_ERR "%s called with unknown type: %d\n",
+			       __FUNCTION__, type);
 			return -1;
 	}
 
@@ -1597,18 +1599,20 @@ int video_register_device(struct video_d
 	if (ret) {
 		printk(KERN_ERR "%s: class_device_register failed\n",
 		       __FUNCTION__);
-		return ret;
-	}
-        ret = class_device_create_file(&vfd->class_dev, &class_device_attr_name);
-        if (ret < 0) {
-                printk(KERN_WARNING "%s error: %d\n", __FUNCTION__, ret);
-		return ret;
+		goto fail_minor;
+	}
+	ret = class_device_create_file(&vfd->class_dev, &class_device_attr_name);
+	if (ret) {
+		printk(KERN_ERR "%s: class_device_create_file 'name' failed\n",
+		       __FUNCTION__);
+		goto fail_classdev;
 	}
 #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,12)
-        ret = class_device_create_file(&vfd->class_dev, &class_device_attr_dev);
-        if (ret < 0) {
-                printk(KERN_WARNING "%s error: %d\n", __FUNCTION__, ret);
-		return ret;
+	ret = class_device_create_file(&vfd->class_dev, &class_device_attr_dev);
+	if (ret) {
+		printk(KERN_ERR "%s: class_device_create_file 'dev' failed\n",
+		       __FUNCTION__);
+		goto fail_classdev;
 	}
 #endif
 
@@ -1620,6 +1624,15 @@ int video_register_device(struct video_d
 		       "http://lwn.net/Articles/36850/\n", vfd->name);
 #endif
 	return 0;
+
+fail_classdev:
+	class_device_unregister(&vfd->class_dev);
+fail_minor:
+	mutex_lock(&videodev_lock);
+	video_device[vfd->minor] = NULL;
+	vfd->minor = -1;
+	mutex_unlock(&videodev_lock);
+	return ret;
 }
 
 /**

[-- Attachment #3: Type: TEXT/PLAIN, Size: 1424 bytes --]

# HG changeset patch
# User Trent Piepho <xyzzy@speakeasy.org>
# Node ID 092b04007c0f1aa2d3f9b4467ea8ab3837723196
# Parent  b900796fdfb4bc87642c470515f88ed7ae92ced3
bttv: use class_device_create_file and handle errors

From: Trent Piepho <xyzzy@speakeasy.org>

Revery bttv-driver.c from video_device_create_file() to use
class_device_create_file() again.  video_device_create_file() is only
available when V4L1 is on.

Proper error checking is added for failure of class_device_create_file(). 
Will print error message and unroll partially created sysfs entries.

Signed-off-by: Trent Piepho <xyzzy@speakeasy.org>

diff -r b900796fdfb4 -r 092b04007c0f linux/drivers/media/video/bt8xx/bttv-driver.c
--- a/linux/drivers/media/video/bt8xx/bttv-driver.c	Sun Jul 23 19:08:21 2006 -0700
+++ b/linux/drivers/media/video/bt8xx/bttv-driver.c	Sun Jul 23 19:10:52 2006 -0700
@@ -3957,8 +3957,12 @@ static int __devinit bttv_register_video
 	printk(KERN_INFO "bttv%d: registered device video%d\n",
 	       btv->c.nr,btv->video_dev->minor & 0x1f);
 #if LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)
-
-	video_device_create_file(btv->video_dev, &class_device_attr_card);
+	if (class_device_create_file(&btv->video_dev->class_dev,
+				     &class_device_attr_card)<0) {
+		printk(KERN_ERR "bttv%d: class_device_create_file 'card' "
+		       "failed\n", btv->c.nr);
+		goto err;
+	}
 #endif
 
 	/* vbi */

  reply	other threads:[~2006-07-24  2:16 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-07-13  4:43 oops in bttv Chuck Ebbert
2006-07-13  5:05 ` Greg KH
2006-07-13  5:24   ` Randy.Dunlap
2006-07-13  5:44     ` Andrew Morton
2006-07-15 23:08       ` [PATCH] V4L: struct video_device corruption Robert Fitzsimons
2006-07-16  1:31         ` Mauro Carvalho Chehab
2006-07-18  0:25           ` [v4l-dvb-maintainer] " Trent Piepho
2006-07-19 11:54             ` Mauro Carvalho Chehab
2006-07-19 15:36               ` Michael Krufky
2006-07-20 21:57               ` Trent Piepho
2006-07-21 12:26                 ` Mauro Carvalho Chehab
2006-07-21 20:06                   ` Trent Piepho
2006-07-21 20:30                     ` Mauro Carvalho Chehab
2006-07-21 22:55                       ` Trent Piepho
2006-07-23  9:35                         ` Mauro Carvalho Chehab
2006-07-24  2:16                           ` Trent Piepho [this message]
2006-07-24 12:34                             ` Mauro Carvalho Chehab
2006-07-24 22:06                               ` Trent Piepho
2006-07-25 17:59                                 ` Mauro Carvalho Chehab
2006-07-25  3:08           ` Andrew Morton
2006-07-25  7:57             ` [v4l-dvb-maintainer] " Manu Abraham
2006-07-25  8:42             ` Trent Piepho
2006-07-25  9:06               ` Andrew Morton
2006-07-25 12:28                 ` Mauro Carvalho Chehab

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=Pine.LNX.4.58.0607231828430.16282@shell3.speakeasy.net \
    --to=xyzzy@speakeasy.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mchehab@infradead.org \
    --cc=v4l-dvb-maintainer@linuxtv.org \
    --cc=video4linux-list@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).