| 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
| ||
|
Message-ID: <CADYTM3bJWUQm_Zrxvb+UY5CzTDYODDfQ=MWdqL7tJaetbL1Qzg@mail.gmail.com>
Date: Thu, 15 May 2014 10:19:07 -0500
From: Bin Liu <binmlist@...il.com>
To: George Cherian <george.cherian@...com>
Cc: Daniel Mack <zonque@...il.com>, linux-kernel@...r.kernel.org,
linux-usb@...r.kernel.org, linux-omap@...r.kernel.org,
balbi@...com, gregkh@...uxfoundation.org, "Liu, Bin" <b-liu@...com>
Subject: Re: [PATCH v3 0/5] Add support for SW babble Control
George,
On Thu, May 15, 2014 at 1:28 AM, George Cherian <george.cherian@...com> wrote:
> Hi Bin,
>
>
> On 5/14/2014 10:13 PM, Bin Liu wrote:
>>
>> George,
>>
>> On Wed, May 14, 2014 at 9:34 AM, Bin Liu <binmlist@...il.com> wrote:
>>>
>>> George,
>>>
>>> On Wed, May 14, 2014 at 12:37 AM, George Cherian <george.cherian@...com>
>>> wrote:
>>>>
>>>> On 5/14/2014 12:07 AM, Bin Liu wrote:
>>>>>
>>>>> Hi,
>>>>>
>>>>> On Tue, May 13, 2014 at 8:24 AM, George Cherian <george.cherian@...com>
>>>>> wrote:
>>>>>>
>>>>>> Hi Daniel,
>>>>>>
>>>>>>
>>>>>> On 5/13/2014 6:44 PM, Daniel Mack wrote:
>>>>>>>
>>>>>>> Hi George,
>>>>>>>
>>>>>>> On 05/13/2014 02:57 PM, George Cherian wrote:
>>>>>>>>
>>>>>>>> I never enabled the MUSB_BABBLE_SW_SESSION_CTRL in the
>>>>>>>> MUSB_BABBLE_CTL
>>>>>>>> reg.
>>>>>>>> can you try with the following patch.
>>>>>>>>
>>>>>>>> diff --git a/drivers/usb/musb/musb_dsps.c
>>>>>>>> b/drivers/usb/musb/musb_dsps.c
>>>>>>>> index 1ae6681..1160cd1 100644
>>>>>>>> --- a/drivers/usb/musb/musb_dsps.c
>>>>>>>> +++ b/drivers/usb/musb/musb_dsps.c
>>>>>>>> @@ -477,8 +477,11 @@ static int dsps_musb_init(struct musb *musb)
>>>>>>>> * logic enabled.
>>>>>>>> */
>>>>>>>> val = dsps_readb(musb->mregs, MUSB_BABBLE_CTL);
>>>>>>>> - if (val == MUSB_BABBLE_RCV_DISABLE)
>>>>>>>> + if (val == MUSB_BABBLE_RCV_DISABLE) {
>>>>>>>> glue->sw_babble_enabled = true;
>>>>>>>> + val |= MUSB_BABBLE_SW_SESSION_CTRL;
>>>>>>>> + dsps_writeb(musb->mregs, MUSB_BABBLE_CTL, val);
>>>>>>>> + }
>>>>>>>> ret = dsps_musb_dbg_init(musb, glue);
>>>>>>>> if (ret)
>>>>>>>
>>>>>>> MUSB_BABBLE_STUCK_J still remains unset, so I get the same result as
>>>>>>> without the patch: a full glue reset is conducted. Do I get you right
>>>>>>> that you expect MUSB_BABBLE_STUCK_J to be set in babble conditions
>>>>>>> when
>>>>>>> MUSB_BABBLE_SW_SESSION_CTRL is set?
>>>>>>>
>>>>>> Basically, there are 2 types of babble conditions.
>>>>>> 1) Transient babble condition - which could be recovered from without
>>>>>> an
>>>>>> IP
>>>>>> reset .
>>>>>> 2) Babble condition - which could be recovered from only by doing an
>>>>>> IP
>>>>>> reset.
>>>>>>
>>>>>> Looks like you are always hitting case 2 (Most times am also hitting
>>>>>> the
>>>>>> same).
>>>>>> Case 1 is really hard to reproduce. I don't have a reliable method as
>>>>>> of
>>>>>> now
>>>>>> to
>>>>>> reproduce this case consistently.
>>>>>>
>>>>>>> [ 19.672373] CAUTION: musb: Babble Interrupt Occurred
>>>>>>> [ 19.677776] musb_stage0_irq 789: unhandled DISCONNECT transition
>>>>>>> (a_wait_bcon)
>>>>>>> [ 19.685815] usb 1-1: USB disconnect, device number 3
>>>>>>> [ 19.769720] musb-hdrc musb-hdrc.0.auto: babble: MUSB_BABBLE_CTL
>>>>>>> value
>>>>>>> 44
>>>>>>> [ 19.776765] musb-hdrc musb-hdrc.0.auto: STUCK_J is reset
>>>>>>>
>>>>>>>
>>>>>>> I don't quite follow, especially as I lack documentation of the IP
>>>>>>> core.
>>>>>>> How do you test babble errors, is there any way to force them to
>>>>>>> happen
>>>>>>> reliably?
>>>>>>
>>>>>>
>>>>>> There is no 100% reliable method to force it to happen. Following is
>>>>>
>>>>> I have a way to force babble happen reliably - shorting DP or DM to
>>>>> VBUS. I opened the far-end plug of the USB cable, so I can easily
>>>>> short DP or DM to VBUS.
>>>>
>>>> Good to know that you have a reliable way to test babble condition.
>>>> Can you please do a quick test on 3.15.0-rc4 with the series applied?
>>>> In case of any assistance please do let me know.
>>>
>>> Sure, but could you please re-send those patches to my corporate email
>>> so that I can apply them from Thunderbird?
>>
>> You don't have to resend the patches. Nishanth Menon showed me a way
>> to extract the patch from Gmail - Thanks Nishanth.
>>
>> But which repo do you want to me test with? The first patch ([PATCH v2
>> 1/5] usb: musb: core: Convert babble recover work to delayed work)
>> does not apply to v3.15-rc4 tag. the current musb_core.c does not have
>> the recovery work for musb. Please let me know what I missed.
>
> Oops I missed to mention the same.
> Please try on
> git://git.kernel.org/pub/scm/linux/kernel/git/balbi/usb.git master
The test is done. The babble always causes STUCK_J is reset,
MUSB_BABBLE_CTL is 0x44 or 0x4. MUSB reset happens.
Do you think when re-start happens the driver should print a message
on console saying re-start due to babble? It would help debug the
babble problems while the dynamic debug is off.
Thanks,
-Bin.
>
>> Thanks,
>> -Bin.
>>
>>> I read these linux-usb emails in Gmail, and am not aware of any easy
>>> way to extract patches from Gmail.
>>>
>>> BTY, I tested with TI 3.12.10 kernel, in which I guess the babble
>>> handling is similar to this patch set. With TI3.12.10, MISC is always
>>> 0x64, so MUSB never restarts.
>>>
>>> Thanks,
>>> -Bin.
>>>
>>>>> But the interesting thing is that with TI 3.2 kernel, shorting DP or
>>>>> DM to VBUS causes MISC register to be 0x4, but the result is
>>>>> completely opposite in TI 3.12.10 kernel, which cause MISC to be 0x64.
>>>>>
>>>>> So in the 3.2 kernel, the babble handing resets the controller, but
>>>>> the 3.12.10 does not.
>>>>>
>>>>> Regards,
>>>>> -Bin.
>>>>>
>>>>>> my setup ,
>>>>>> I have a HUB with 4 devices connected , which gives me a Babble
>>>>>> interrupt
>>>>>> on both connects and disconnects ( Not always though).
>>>>>>
>>>>>>> Anyway, the full glue layer solves this rare condition quite well for
>>>>>>> me. Is there any downside of this?
>>>>>>>
>>>>>>>
>>>>>>> Daniel
>>>>>>>
>>>>>> --
>>>>>> -George
>>>>>>
>>>>>> --
>>>>>> To unsubscribe from this list: send the line "unsubscribe linux-omap"
>>>>>> in
>>>>>>
>>>>>> the body of a message to majordomo@...r.kernel.org
>>>>>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>>>>
>>>>
>>>>
>>>> --
>>>> -George
>>>>
>
>
> --
> -George
>
--
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/
Powered by blists - more mailing lists