[U-Boot] cmd: usb: ignore block devices under mass storage device

Message ID 1502430811-21475-1-git-send-email-suneelglinux@gmail.com
State Superseded
Delegated to: Marek Vasut
Headers show

Commit Message

Suneel Garapati Aug. 11, 2017, 5:53 a.m.
usb tree and info commands may cause crash otherwise

Signed-off-by: Suneel Garapati <suneelglinux@gmail.com>
---
 cmd/usb.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

Comments

Simon Glass Aug. 13, 2017, 9:37 p.m. | #1
Hi Suneel,

On 10 August 2017 at 23:53, Suneel Garapati <suneelglinux@gmail.com> wrote:
> usb tree and info commands may cause crash otherwise
>
> Signed-off-by: Suneel Garapati <suneelglinux@gmail.com>
> ---
>  cmd/usb.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>

Thank you for the patch - it certainly looks like a bug. Can you
please expand the commit message a little? E.g. you have
UCLASS_USB_EMUL below.

> diff --git a/cmd/usb.c b/cmd/usb.c
> index 992d414..81e1a7b 100644
> --- a/cmd/usb.c
> +++ b/cmd/usb.c
> @@ -415,7 +415,8 @@ static void usb_show_tree_graph(struct usb_device *dev, char *pre)
>                 udev = dev_get_parent_priv(child);
>
>                 /* Ignore emulators, we only want real devices */
> -               if (device_get_uclass_id(child) != UCLASS_USB_EMUL) {
> +               if (device_get_uclass_id(child) !=
> +                   (UCLASS_USB_EMUL | UCLASS_BLK)) {

This seems odd to me. Do you mean to check that the child uclass is
neither USB_EMUL nor BLK?

Would it be possible to check that the parent is UCLASS_USB? That
seems like a better condition to determine whether the child has USB
parent data.

>                         usb_show_tree_graph(udev, pre);
>                         pre[index] = 0;
>                 }
> @@ -605,7 +606,8 @@ static void usb_show_info(struct usb_device *udev)
>         for (device_find_first_child(udev->dev, &child);
>              child;
>              device_find_next_child(&child)) {
> -               if (device_active(child)) {
> +               if (device_active(child) &&
> +                   (device_get_uclass_id(child) != UCLASS_BLK)) {
>                         udev = dev_get_parent_priv(child);
>                         usb_show_info(udev);
>                 }
> --
> 2.7.4
>

Regards,
Simon
Suneel Garapati Aug. 15, 2017, 3:06 a.m. | #2
Hi Simon,


On Sun, Aug 13, 2017 at 2:37 PM, Simon Glass <sjg@chromium.org> wrote:
> Hi Suneel,
>
> On 10 August 2017 at 23:53, Suneel Garapati <suneelglinux@gmail.com> wrote:
>> usb tree and info commands may cause crash otherwise
>>
>> Signed-off-by: Suneel Garapati <suneelglinux@gmail.com>
>> ---
>>  cmd/usb.c | 6 ++++--
>>  1 file changed, 4 insertions(+), 2 deletions(-)
>>
>
> Thank you for the patch - it certainly looks like a bug. Can you
> please expand the commit message a little? E.g. you have
> UCLASS_USB_EMUL below.

I will change the description

>
>> diff --git a/cmd/usb.c b/cmd/usb.c
>> index 992d414..81e1a7b 100644
>> --- a/cmd/usb.c
>> +++ b/cmd/usb.c
>> @@ -415,7 +415,8 @@ static void usb_show_tree_graph(struct usb_device *dev, char *pre)
>>                 udev = dev_get_parent_priv(child);
>>
>>                 /* Ignore emulators, we only want real devices */
>> -               if (device_get_uclass_id(child) != UCLASS_USB_EMUL) {
>> +               if (device_get_uclass_id(child) !=
>> +                   (UCLASS_USB_EMUL | UCLASS_BLK)) {
>
> This seems odd to me. Do you mean to check that the child uclass is
> neither USB_EMUL nor BLK?
>
> Would it be possible to check that the parent is UCLASS_USB? That
> seems like a better condition to determine whether the child has USB
> parent data.

It is possible to check parent uclass but would that ever fail?
I assume, block device under usb storage device will always have
parent as usb class.
Also, tree is called on only usb class devices. Maybe I am missing something.

Regards,
Suneel

>
>>                         usb_show_tree_graph(udev, pre);
>>                         pre[index] = 0;
>>                 }
>> @@ -605,7 +606,8 @@ static void usb_show_info(struct usb_device *udev)
>>         for (device_find_first_child(udev->dev, &child);
>>              child;
>>              device_find_next_child(&child)) {
>> -               if (device_active(child)) {
>> +               if (device_active(child) &&
>> +                   (device_get_uclass_id(child) != UCLASS_BLK)) {
>>                         udev = dev_get_parent_priv(child);
>>                         usb_show_info(udev);
>>                 }
>> --
>> 2.7.4
>>
>
> Regards,
> Simon
Suneel Garapati Aug. 28, 2017, 6:37 p.m. | #3
Hi,

Request for review on comments below.

Regards,
Suneel

On Mon, Aug 14, 2017 at 8:06 PM, Suneel Garapati <suneelglinux@gmail.com> wrote:
> Hi Simon,
>
>
> On Sun, Aug 13, 2017 at 2:37 PM, Simon Glass <sjg@chromium.org> wrote:
>> Hi Suneel,
>>
>> On 10 August 2017 at 23:53, Suneel Garapati <suneelglinux@gmail.com> wrote:
>>> usb tree and info commands may cause crash otherwise
>>>
>>> Signed-off-by: Suneel Garapati <suneelglinux@gmail.com>
>>> ---
>>>  cmd/usb.c | 6 ++++--
>>>  1 file changed, 4 insertions(+), 2 deletions(-)
>>>
>>
>> Thank you for the patch - it certainly looks like a bug. Can you
>> please expand the commit message a little? E.g. you have
>> UCLASS_USB_EMUL below.
>
> I will change the description
>
>>
>>> diff --git a/cmd/usb.c b/cmd/usb.c
>>> index 992d414..81e1a7b 100644
>>> --- a/cmd/usb.c
>>> +++ b/cmd/usb.c
>>> @@ -415,7 +415,8 @@ static void usb_show_tree_graph(struct usb_device *dev, char *pre)
>>>                 udev = dev_get_parent_priv(child);
>>>
>>>                 /* Ignore emulators, we only want real devices */
>>> -               if (device_get_uclass_id(child) != UCLASS_USB_EMUL) {
>>> +               if (device_get_uclass_id(child) !=
>>> +                   (UCLASS_USB_EMUL | UCLASS_BLK)) {
>>
>> This seems odd to me. Do you mean to check that the child uclass is
>> neither USB_EMUL nor BLK?
>>
>> Would it be possible to check that the parent is UCLASS_USB? That
>> seems like a better condition to determine whether the child has USB
>> parent data.
>
> It is possible to check parent uclass but would that ever fail?
> I assume, block device under usb storage device will always have
> parent as usb class.
> Also, tree is called on only usb class devices. Maybe I am missing something.
>
> Regards,
> Suneel
>
>>
>>>                         usb_show_tree_graph(udev, pre);
>>>                         pre[index] = 0;
>>>                 }
>>> @@ -605,7 +606,8 @@ static void usb_show_info(struct usb_device *udev)
>>>         for (device_find_first_child(udev->dev, &child);
>>>              child;
>>>              device_find_next_child(&child)) {
>>> -               if (device_active(child)) {
>>> +               if (device_active(child) &&
>>> +                   (device_get_uclass_id(child) != UCLASS_BLK)) {
>>>                         udev = dev_get_parent_priv(child);
>>>                         usb_show_info(udev);
>>>                 }
>>> --
>>> 2.7.4
>>>
>>
>> Regards,
>> Simon
Simon Glass Aug. 31, 2017, 12:52 p.m. | #4
Hi Suneel,

On 15 August 2017 at 11:06, Suneel Garapati <suneelglinux@gmail.com> wrote:
> Hi Simon,
>
>
> On Sun, Aug 13, 2017 at 2:37 PM, Simon Glass <sjg@chromium.org> wrote:
>> Hi Suneel,
>>
>> On 10 August 2017 at 23:53, Suneel Garapati <suneelglinux@gmail.com> wrote:
>>> usb tree and info commands may cause crash otherwise
>>>
>>> Signed-off-by: Suneel Garapati <suneelglinux@gmail.com>
>>> ---
>>>  cmd/usb.c | 6 ++++--
>>>  1 file changed, 4 insertions(+), 2 deletions(-)
>>>
>>
>> Thank you for the patch - it certainly looks like a bug. Can you
>> please expand the commit message a little? E.g. you have
>> UCLASS_USB_EMUL below.
>
> I will change the description
>
>>
>>> diff --git a/cmd/usb.c b/cmd/usb.c
>>> index 992d414..81e1a7b 100644
>>> --- a/cmd/usb.c
>>> +++ b/cmd/usb.c
>>> @@ -415,7 +415,8 @@ static void usb_show_tree_graph(struct usb_device *dev, char *pre)
>>>                 udev = dev_get_parent_priv(child);
>>>
>>>                 /* Ignore emulators, we only want real devices */
>>> -               if (device_get_uclass_id(child) != UCLASS_USB_EMUL) {
>>> +               if (device_get_uclass_id(child) !=
>>> +                   (UCLASS_USB_EMUL | UCLASS_BLK)) {
>>
>> This seems odd to me. Do you mean to check that the child uclass is
>> neither USB_EMUL nor BLK?
>>
>> Would it be possible to check that the parent is UCLASS_USB? That
>> seems like a better condition to determine whether the child has USB
>> parent data.
>
> It is possible to check parent uclass but would that ever fail?

How would it fail? The only device that does not have a parent is the
root device, and we know it cannot be that since it will be
UCLASS_ROOT.

> I assume, block device under usb storage device will always have
> parent as usb class.

Yes that sounds right.

> Also, tree is called on only usb class devices. Maybe I am missing something.

Maybe, but I think it is more robust to check for the thing you want
than the things you don't. If someone adds a new thing that can be a
child then this code will need updating.

>
> Regards,
> Suneel
>
>>
>>>                         usb_show_tree_graph(udev, pre);
>>>                         pre[index] = 0;
>>>                 }
>>> @@ -605,7 +606,8 @@ static void usb_show_info(struct usb_device *udev)
>>>         for (device_find_first_child(udev->dev, &child);
>>>              child;
>>>              device_find_next_child(&child)) {
>>> -               if (device_active(child)) {
>>> +               if (device_active(child) &&
>>> +                   (device_get_uclass_id(child) != UCLASS_BLK)) {
>>>                         udev = dev_get_parent_priv(child);
>>>                         usb_show_info(udev);
>>>                 }
>>> --
>>> 2.7.4
>>>

Regards,
Simon
Lothar Waßmann Sept. 1, 2017, 6:30 a.m. | #5
Hi,

On Thu, 10 Aug 2017 22:53:31 -0700 Suneel Garapati wrote:
> usb tree and info commands may cause crash otherwise
> 
> Signed-off-by: Suneel Garapati <suneelglinux@gmail.com>
> ---
>  cmd/usb.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/cmd/usb.c b/cmd/usb.c
> index 992d414..81e1a7b 100644
> --- a/cmd/usb.c
> +++ b/cmd/usb.c
> @@ -415,7 +415,8 @@ static void usb_show_tree_graph(struct usb_device *dev, char *pre)
>  		udev = dev_get_parent_priv(child);
>  
>  		/* Ignore emulators, we only want real devices */
> -		if (device_get_uclass_id(child) != UCLASS_USB_EMUL) {
> +		if (device_get_uclass_id(child) !=
> +		    (UCLASS_USB_EMUL | UCLASS_BLK)) {
>
This should most probably be:
> +		if ((device_get_uclass_id(child) != UCLASS_USB_EMUL) &&
			(device_get_uclass_id(child) != UCLASS_BLK)) {


Lothar Waßmann
Suneel Garapati Sept. 1, 2017, 7:34 a.m. | #6
Hi Simon,

On Thu, Aug 31, 2017 at 5:52 AM, Simon Glass <sjg@chromium.org> wrote:
> Hi Suneel,
>
> On 15 August 2017 at 11:06, Suneel Garapati <suneelglinux@gmail.com> wrote:
>> Hi Simon,
>>
>>
>> On Sun, Aug 13, 2017 at 2:37 PM, Simon Glass <sjg@chromium.org> wrote:
>>> Hi Suneel,
>>>
>>> On 10 August 2017 at 23:53, Suneel Garapati <suneelglinux@gmail.com> wrote:
>>>> usb tree and info commands may cause crash otherwise
>>>>
>>>> Signed-off-by: Suneel Garapati <suneelglinux@gmail.com>
>>>> ---
>>>>  cmd/usb.c | 6 ++++--
>>>>  1 file changed, 4 insertions(+), 2 deletions(-)
>>>>
>>>
>>> Thank you for the patch - it certainly looks like a bug. Can you
>>> please expand the commit message a little? E.g. you have
>>> UCLASS_USB_EMUL below.
>>
>> I will change the description
>>
>>>
>>>> diff --git a/cmd/usb.c b/cmd/usb.c
>>>> index 992d414..81e1a7b 100644
>>>> --- a/cmd/usb.c
>>>> +++ b/cmd/usb.c
>>>> @@ -415,7 +415,8 @@ static void usb_show_tree_graph(struct usb_device *dev, char *pre)
>>>>                 udev = dev_get_parent_priv(child);
>>>>
>>>>                 /* Ignore emulators, we only want real devices */
>>>> -               if (device_get_uclass_id(child) != UCLASS_USB_EMUL) {
>>>> +               if (device_get_uclass_id(child) !=
>>>> +                   (UCLASS_USB_EMUL | UCLASS_BLK)) {
>>>
>>> This seems odd to me. Do you mean to check that the child uclass is
>>> neither USB_EMUL nor BLK?
>>>
>>> Would it be possible to check that the parent is UCLASS_USB? That
>>> seems like a better condition to determine whether the child has USB
>>> parent data.
>>
>> It is possible to check parent uclass but would that ever fail?
>
> How would it fail? The only device that does not have a parent is the
> root device, and we know it cannot be that since it will be
> UCLASS_ROOT.
>
>> I assume, block device under usb storage device will always have
>> parent as usb class.
>
> Yes that sounds right.
>
>> Also, tree is called on only usb class devices. Maybe I am missing something.
>
> Maybe, but I think it is more robust to check for the thing you want
> than the things you don't. If someone adds a new thing that can be a
> child then this code will need updating.
I will add a check on parent uclass for USB in v1.

Regards,
Suneel

>
>>
>> Regards,
>> Suneel
>>
>>>
>>>>                         usb_show_tree_graph(udev, pre);
>>>>                         pre[index] = 0;
>>>>                 }
>>>> @@ -605,7 +606,8 @@ static void usb_show_info(struct usb_device *udev)
>>>>         for (device_find_first_child(udev->dev, &child);
>>>>              child;
>>>>              device_find_next_child(&child)) {
>>>> -               if (device_active(child)) {
>>>> +               if (device_active(child) &&
>>>> +                   (device_get_uclass_id(child) != UCLASS_BLK)) {
>>>>                         udev = dev_get_parent_priv(child);
>>>>                         usb_show_info(udev);
>>>>                 }
>>>> --
>>>> 2.7.4
>>>>
>
> Regards,
> Simon
Suneel Garapati Sept. 1, 2017, 7:35 a.m. | #7
Hi,

On Thu, Aug 31, 2017 at 11:30 PM, Lothar Waßmann <LW@karo-electronics.de> wrote:
> Hi,
>
> On Thu, 10 Aug 2017 22:53:31 -0700 Suneel Garapati wrote:
>> usb tree and info commands may cause crash otherwise
>>
>> Signed-off-by: Suneel Garapati <suneelglinux@gmail.com>
>> ---
>>  cmd/usb.c | 6 ++++--
>>  1 file changed, 4 insertions(+), 2 deletions(-)
>>
>> diff --git a/cmd/usb.c b/cmd/usb.c
>> index 992d414..81e1a7b 100644
>> --- a/cmd/usb.c
>> +++ b/cmd/usb.c
>> @@ -415,7 +415,8 @@ static void usb_show_tree_graph(struct usb_device *dev, char *pre)
>>               udev = dev_get_parent_priv(child);
>>
>>               /* Ignore emulators, we only want real devices */
>> -             if (device_get_uclass_id(child) != UCLASS_USB_EMUL) {
>> +             if (device_get_uclass_id(child) !=
>> +                 (UCLASS_USB_EMUL | UCLASS_BLK)) {
>>
> This should most probably be:
>> +             if ((device_get_uclass_id(child) != UCLASS_USB_EMUL) &&
>                         (device_get_uclass_id(child) != UCLASS_BLK)) {
Agree. Will make change in v1.

Regards,
Suneel
>
>
> Lothar Waßmann

Patch

diff --git a/cmd/usb.c b/cmd/usb.c
index 992d414..81e1a7b 100644
--- a/cmd/usb.c
+++ b/cmd/usb.c
@@ -415,7 +415,8 @@  static void usb_show_tree_graph(struct usb_device *dev, char *pre)
 		udev = dev_get_parent_priv(child);
 
 		/* Ignore emulators, we only want real devices */
-		if (device_get_uclass_id(child) != UCLASS_USB_EMUL) {
+		if (device_get_uclass_id(child) !=
+		    (UCLASS_USB_EMUL | UCLASS_BLK)) {
 			usb_show_tree_graph(udev, pre);
 			pre[index] = 0;
 		}
@@ -605,7 +606,8 @@  static void usb_show_info(struct usb_device *udev)
 	for (device_find_first_child(udev->dev, &child);
 	     child;
 	     device_find_next_child(&child)) {
-		if (device_active(child)) {
+		if (device_active(child) &&
+		    (device_get_uclass_id(child) != UCLASS_BLK)) {
 			udev = dev_get_parent_priv(child);
 			usb_show_info(udev);
 		}