Hi,
Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> writes:
> On Fri, 15 Apr 2016, Felipe Balbi wrote:
>
>> Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> writes:
>> > On Thu, 14 Apr 2016, Felipe Balbi wrote:
>> >
>> >> >> --- a/drivers/usb/storage/scsiglue.c
>> >> >> +++ b/drivers/usb/storage/scsiglue.c
>> >> >> @@ -127,6 +127,11 @@ static int slave_configure(struct scsi_device *sdev)
>> >> >> if (queue_max_hw_sectors(sdev->request_queue) > max_sectors)
>> >> >> blk_queue_max_hw_sectors(sdev->request_queue,
>> >> >> max_sectors);
>> >> >> + } else if (us->pusb_dev->speed >= USB_SPEED_SUPER) {
>> >> >> + /* USB3 devices will be limited to 2048 sectors. This gives us
>> >> >> + * better throughput on most devices.
>> >> >> + */
>> >> >> + blk_queue_max_hw_sectors(sdev->request_queue, 2048);
>> >> >> } else if (sdev->type == TYPE_TAPE) {
>> >> >> /* Tapes need much higher max_sector limits, so just
>> >> >> * raise it to the maximum possible (4 GB / 512) and
>> >> >
>> >> > Argh! This has the same kind of problem as before. What will happen
>> >> > when somebody has a USB-3 tape drive?
>> >>
>> >> I didn't know that was even plausible :-) Anyway, I'll update, but while
>> >> at that, so I use for bcdUSB instead of speed as Oliver suggested ? I
>> >> mean, a USB3 stick running on high-speed can also support 2048 max
>> >> sectors, right ?
>> >>
>> >> let me know
>> >
>> > To tell the truth, I have no idea. There probably aren't enough USB-3
>> > products in existence yet to tell -- not to mention that with the
>> > existing code, we wouldn't detect any exceptions.
>> >
>> > It sounds reasonable... But won't a USB-3 device running at high speed
>> > provide a device descriptor that has bcdUSB set to 0x0210? (See the
>> > second paragraph of section 9.6.1 in the USB-3.1 spec.)
>>
>> right, but an HS USB 2.1 device is also recent enough that it's likely
>> to work similarly, no ?
>
> Again, I don't know. But in my experience, relying on USB device
> designers to do things correctly tends not to work out. :-(
>
> Besides, whether something works right has less to do with how recent
> it is and more to do with whether or not it was ever tested. If
> Windows doesn't transfer more than 240 sectors at a time at high speed
> then we probably shouldn't either.
fair enough, I'll send v5 shortly.
--
balbi
Attachment:
signature.asc
Description: PGP signature
