| Message ID | 1446670112-29088-2-git-send-email-balbi@ti.com |
|---|---|
| State | New |
| Headers | show |
Hi, Greg KH <greg@kroah.com> writes: > On Wed, Nov 04, 2015 at 02:48:32PM -0600, Felipe Balbi wrote: >> USB3 devices, because they are much newer, have much >> less chance of having issues with larger transfers. >> >> We still keep a limit because anything above 2048 >> sectors really rendered negligible speed >> improvements, so we will simply ignore >> that. Transferring 1MiB should already give us >> pretty good performance. >> >> Signed-off-by: Felipe Balbi <balbi@ti.com> >> --- >> drivers/usb/storage/scsiglue.c | 16 +++++++++++----- >> drivers/usb/storage/usb.c | 4 ++++ >> include/linux/usb_usual.h | 2 ++ >> 3 files changed, 17 insertions(+), 5 deletions(-) >> >> diff --git a/drivers/usb/storage/scsiglue.c b/drivers/usb/storage/scsiglue.c >> index 00aadf3a3857..bda2b3c9ae42 100644 >> --- a/drivers/usb/storage/scsiglue.c >> +++ b/drivers/usb/storage/scsiglue.c >> @@ -115,13 +115,19 @@ static int slave_configure(struct scsi_device *sdev) >> { >> struct us_data *us = host_to_us(sdev->host); >> >> - /* Many devices have trouble transferring more than 32KB at a time, >> - * while others have trouble with more than 64K. At this time we >> - * are limiting both to 32K (64 sectores). >> - */ >> - if (us->fflags & (US_FL_MAX_SECTORS_64 | US_FL_MAX_SECTORS_MIN)) { >> + if (us->fflags & US_FL_USB3) { >> + /* USB3 devices will be limitted at 2048 sectors. This gives us > > "limited"? hehe, that was bad :-p > Other than that, this looks good to me, did you have a chance to test a > Windows machine to see what it does? yeah, it always does 128 sectors no matter if it's usb2 or usb3. Mac OS X will do 256 sectors for USB2 and 2048 for USB3. Tested both by copying formatting and copying large files over. -- balbi
diff --git a/drivers/usb/storage/scsiglue.c b/drivers/usb/storage/scsiglue.c index 00aadf3a3857..bda2b3c9ae42 100644 --- a/drivers/usb/storage/scsiglue.c +++ b/drivers/usb/storage/scsiglue.c @@ -115,13 +115,19 @@ static int slave_configure(struct scsi_device *sdev) { struct us_data *us = host_to_us(sdev->host); - /* Many devices have trouble transferring more than 32KB at a time, - * while others have trouble with more than 64K. At this time we - * are limiting both to 32K (64 sectores). - */ - if (us->fflags & (US_FL_MAX_SECTORS_64 | US_FL_MAX_SECTORS_MIN)) { + if (us->fflags & US_FL_USB3) { + /* USB3 devices will be limitted at 2048 sectors. This gives us + * close to 25% throughput improvement on some devices. + */ + blk_queue_max_hw_sectors(sdev->request_queue, 2048); + } else if (us->fflags & (US_FL_MAX_SECTORS_64 | US_FL_MAX_SECTORS_MIN)) { unsigned int max_sectors = 64; + /* Many devices have trouble transferring more than 32KB at a + * time, while others have trouble with more than 64K. At this + * time we are limiting both to 32K (64 sectores). + */ + if (us->fflags & US_FL_MAX_SECTORS_MIN) max_sectors = PAGE_CACHE_SIZE >> 9; if (queue_max_hw_sectors(sdev->request_queue) > max_sectors) diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c index 43576ed31ccd..b7735c9850f8 100644 --- a/drivers/usb/storage/usb.c +++ b/drivers/usb/storage/usb.c @@ -594,6 +594,10 @@ static int get_device_info(struct us_data *us, const struct usb_device_id *id, return -ENODEV; } + /* Flag USB3 devices so we can increase max_sectors to 2048 sectors. */ + if (dev->speed == USB_SPEED_SUPER) + us->fflags |= US_FL_USB3; + /* * This flag is only needed when we're in high-speed, so let's * disable it if we're in full-speed diff --git a/include/linux/usb_usual.h b/include/linux/usb_usual.h index 7f5f78bd15ad..b76b0faaedc8 100644 --- a/include/linux/usb_usual.h +++ b/include/linux/usb_usual.h @@ -79,6 +79,8 @@ /* Cannot handle MI_REPORT_SUPPORTED_OPERATION_CODES */ \ US_FLAG(MAX_SECTORS_240, 0x08000000) \ /* Sets max_sectors to 240 */ \ + US_FLAG(USB3, 0x10000000) \ + /* This is a USB3 Storage Device */ \ #define US_FLAG(name, value) US_FL_##name = value , enum { US_DO_ALL_FLAGS };
USB3 devices, because they are much newer, have much less chance of having issues with larger transfers. We still keep a limit because anything above 2048 sectors really rendered negligible speed improvements, so we will simply ignore that. Transferring 1MiB should already give us pretty good performance. Signed-off-by: Felipe Balbi <balbi@ti.com> --- drivers/usb/storage/scsiglue.c | 16 +++++++++++----- drivers/usb/storage/usb.c | 4 ++++ include/linux/usb_usual.h | 2 ++ 3 files changed, 17 insertions(+), 5 deletions(-) -- 2.6.2 -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html