Skip to contentSkip to conversation

staging: media: Use dev_err() instead of pr_err()

DISCUSSION Mauro Carvalho Chehab 2 participants 4 messages Nov 14, 2013 – Nov 19, 2013 Media (V4L/DVB)

lore
Share & tools
lore

Mauro Carvalho Chehab

author Reviewed-by Josh Triplett
#0 lore raw

Hi,

I'm not sure how this patch got applied upstream:

commit b6ea5ef80aa7fd6f4b18ff2e4174930e8772e812
Author: Dulshani Gunawardhana <[email protected]>
Date: Sun Oct 20 22:58:28 2013 +0530

staging:media: Use dev_dbg() instead of pr_debug()

Use dev_dbg() instead of pr_debug() in go7007-usb.c.

But, from the custody chain, it seems it was not C/C to linux-media ML, doesn't have the driver maintainer's ack[1] and didn't went via my tree.

[1] Dulshani, please next time run the get_maintainer.pl script to get the proper maintainers:

	$ /scripts/get_maintainer.pl -f drivers/staging/media/go7007/go7007-usb.c
	Hans Verkuil <[email protected]> (maintainer:STAGING - GO7007...)
	Mauro Carvalho Chehab <[email protected]> (maintainer:MEDIA INPUT INFRA...)
	Greg Kroah-Hartman <[email protected]> (supporter:STAGING SUBSYSTEM)
	[email protected] (open list:MEDIA INPUT INFRA...)
	[email protected] (open list:STAGING SUBSYSTEM)

Anyway, this patch is clearly wrong, and will cause an OOPS if CONFIG_DEBUG is enabled, during device probing, because of this change:

@@ -1052,21 +1050,21 @@ static int go7007_usb_probe(struct usb_interface *intf,
const struct usb_device_id *id)

 {
        struct go7007 *go;
        struct go7007_usb *usb;
        const struct go7007_usb_board *board;
        struct usb_device *usbdev = interface_to_usbdev(intf);
        unsigned num_i2c_devs;
        char *name;
        int video_pipe, i, v_urb_len;

- pr_debug("probing new GO7007 USB board\n");
+ dev_dbg(go->dev, "probing new GO7007 USB board\n");

        switch (id->driver_info) {
        case GO7007_BOARDID_MATRIX_II:
                name = "WIS Matrix II or compatible";
                board = &board_matrix_ii;
                break;
        case GO7007_BOARDID_MATRIX_RELOAD:
                name = "WIS Matrix Reloaded or compatible";
                board = &board_matrix_reload;
                break;

As it will try to de-reference the uninitialized "go" struct go7007_usb pointer.

The alternative of mixing pr_debug with dev_debug, as Dan is suggesting is, IMHO, worse, as it will lack coherency on the usage of printk macros inside the driver.

So, I think we should just revert this patch.

Comments?

Regards,
Mauro

Forwarded message:

Date: Tue, 5 Nov 2013 23:26:05 +0300
From: Dan Carpenter <[email protected]>
To: [email protected]
Cc: [email protected], [email protected]
Subject: re: staging: media: Use dev_err() instead of pr_err()

Hello Dulshani Gunawardhana,

The patch 44ee8e801137: "staging: media: Use dev_err() instead of pr_err()" from Oct 20, 2013, leads to the following
GCC warning:

drivers/staging/media/go7007/go7007-usb.c: In function ‘go7007_usb_probe’:
drivers/staging/media/go7007/go7007-usb.c:1100:13: warning: ‘go’ may be used uninitialized in this function [-Wuninitialized]

drivers/staging/media/go7007/go7007-usb.c
1049 static int go7007_usb_probe(struct usb_interface *intf,
1050 const struct usb_device_id *id)
1051 {

  1052          struct go7007 *go;
  1053          struct go7007_usb *usb;
  1054          const struct go7007_usb_board *board;
  1055          struct usb_device *usbdev = interface_to_usbdev(intf);
  1056          unsigned num_i2c_devs;
  1057          char *name;
  1058          int video_pipe, i, v_urb_len;
  1059  
  1060          dev_dbg(go->dev, "probing new GO7007 USB board\n");
                        ^^^^^^^
  1061  
  1062          switch (id->driver_info) {
  1063          case GO7007_BOARDID_MATRIX_II:
  1064                  name = "WIS Matrix II or compatible";
  1065                  board = &board_matrix_ii;
  1066                  break;

There are several other uses of "go" before it has been initialized.

Probably you will just want to change these back to pr_info(). Some of the messages are not very useful like:

	dev_info(go->dev, "Sensoray 2250 found\n");

You can delete that one.

regards,
dan carpenter

--
To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#1 lore raw
Quoted 10 lines · Mauro Carvalho Chehab

Hi,

I'm not sure how this patch got applied upstream:

commit b6ea5ef80aa7fd6f4b18ff2e4174930e8772e812
Author: Dulshani Gunawardhana <[email protected]>
Date: Sun Oct 20 22:58:28 2013 +0530

staging:media: Use dev_dbg() instead of pr_debug()

Quoted 8 lines

Use dev_dbg() instead of pr_debug() in go7007-usb.c.

But, from the custody chain, it seems it was not C/C to linux-media ML, doesn't have the driver maintainer's ack[1] and didn't went via my tree.

It came in through my tree as part of the OPW intern application process.

And yes, sorry, it's broken, I have some follow-on patches to fix this, but you are right, it should just be reverted for now, very sorry about that.

Do you want to do that, or should I?

thanks,

greg k-h

#2 lore raw

Em Fri, 15 Nov 2013 15:29:39 +0900
Greg Kroah-Hartman <[email protected]> escreveu:

Quoted 17 lines · Mauro Carvalho Chehab
Quoted 16 lines

Hi,

I'm not sure how this patch got applied upstream:

commit b6ea5ef80aa7fd6f4b18ff2e4174930e8772e812
Author: Dulshani Gunawardhana <[email protected]>
Date: Sun Oct 20 22:58:28 2013 +0530

staging:media: Use dev_dbg() instead of pr_debug()

Use dev_dbg() instead of pr_debug() in go7007-usb.c.

Quoted 5 lines
Quoted 2 lines

But, from the custody chain, it seems it was not C/C to linux-media ML, doesn't have the driver maintainer's ack[1] and didn't went via my tree.

It came in through my tree as part of the OPW intern application process.

Ah, OK.

I don't mind if you apply those directly, but what makes me a little worried is that at least the final version of the patchset should be c/c to driver/subsystem maintainers for their review and for them to know that the patch will be merged via some other tree, as it might be causing conflicts with their trees.

Quoted 3 lines

And yes, sorry, it's broken, I have some follow-on patches to fix this, but you are right, it should just be reverted for now, very sorry about that.

No problem.

Quoted 1 line

Do you want to do that, or should I?

I prefer if you could do it, as I'm still waiting the merge from my tree, and I don't want to cascade another pull request before the original pull requests get handled. In any case, they won't conflict with this, as I don't have any patch for this driver on my tree for 3.13.

Thanks!
Mauro

#3 lore raw
Quoted 38 lines · Mauro Carvalho Chehab

Em Fri, 15 Nov 2013 15:29:39 +0900
Greg Kroah-Hartman <[email protected]> escreveu:

Quoted 17 lines · Mauro Carvalho Chehab
Quoted 16 lines

Hi,

I'm not sure how this patch got applied upstream:

commit b6ea5ef80aa7fd6f4b18ff2e4174930e8772e812
Author: Dulshani Gunawardhana <[email protected]>
Date: Sun Oct 20 22:58:28 2013 +0530

staging:media: Use dev_dbg() instead of pr_debug()

Use dev_dbg() instead of pr_debug() in go7007-usb.c.

Quoted 5 lines
Quoted 2 lines

But, from the custody chain, it seems it was not C/C to linux-media ML, doesn't have the driver maintainer's ack[1] and didn't went via my tree.

It came in through my tree as part of the OPW intern application process.

Ah, OK.

I don't mind if you apply those directly, but what makes me a little worried is that at least the final version of the patchset should be c/c to driver/subsystem maintainers for their review and for them to know that the patch will be merged via some other tree, as it might be causing conflicts with their trees.

Quoted 3 lines

And yes, sorry, it's broken, I have some follow-on patches to fix this, but you are right, it should just be reverted for now, very sorry about that.

Quoted 8 lines

No problem.

Quoted 1 line

Do you want to do that, or should I?

I prefer if you could do it, as I'm still waiting the merge from my tree, and I don't want to cascade another pull request before the original pull requests get handled. In any case, they won't conflict with this, as I don't have any patch for this driver on my tree for 3.13.

Ok, I'll do this after 3.13-rc1 is out, sorry for the problems.

greg k-h