Hi,

Bin Liu <[email protected]> writes:
> Hi,
>
> On 11/06/2015 11:56 AM, Felipe Balbi wrote:
>>
>> Hi,
>>
>> Bin Liu <[email protected]> writes:
>>> This reverts commit 2035772010db634ec8566b658fb1cd87ec47ac77.
>>>
>>> Commit 20357720 claims throughput improvement for MSC/UVC, but I
>>> don't see much improvement. Following are the MSC measurement using
>>> dd on AM335x GP EVM.
>>>
>>> with BCD_BH:    read: 14.9MB/s, write: 20.9MB/s
>>> without BCD_BH: read: 15.2MB/s, write: 21.2MB/s
>>>
>>> However with this commit the following regressions have been observed.
>>>
>>> 1. ASIX usb-ethernet dongle is completely broken on UDP RX.
>>>
>>> 2. Unpluging a 3G modem, which uses option driver, behind a hub causes
>>>     console log flooding with the following message.
>>>
>>>     option_instat_callback: error -71
>>>
>>> Signed-off-by: Bin Liu <[email protected]>
>>> ---
>>>   drivers/usb/musb/musb_host.c | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/usb/musb/musb_host.c b/drivers/usb/musb/musb_host.c
>>> index 0e29184..1c7f858 100644
>>> --- a/drivers/usb/musb/musb_host.c
>>> +++ b/drivers/usb/musb/musb_host.c
>>> @@ -2613,7 +2613,7 @@ static const struct hc_driver musb_hc_driver = {
>>>     .description            = "musb-hcd",
>>>     .product_desc           = "MUSB HDRC host driver",
>>>     .hcd_priv_size          = sizeof(struct musb *),
>>> -   .flags                  = HCD_USB2 | HCD_MEMORY | HCD_BH,
>>> +   .flags                  = HCD_USB2 | HCD_MEMORY,
>>
>> reverting will revert isochronous devices, IIRC. Can you check that
>> cameras still work ? That commit actually mentions that with this commit
>> we got 640x480@30fps working. Can you make sure it works without HCD_BH
>> as well ?
>>
> I don't see Isoch improvement with HCD_BH either.
> All UVC cameras that I have do not work at YUV 640x480@30fps with or 
> without HCD_BH, mainly due the 50% bandwidth utilization in Isoch 
> transfer, which I talked to you months ago. The best I can get is YUV 
> 640x480@20fps, which does not change with or without HCD_BH.

fair enough. Unless someone else has any complaints, I'll take this patch.

-- 
balbi

Attachment: signature.asc
Description: PGP signature

Reply via email to