| Message ID | 1583158895-31342-1-git-send-email-jilin@nvidia.com (mailing list archive) |
|---|---|
| State | Mainlined |
| Commit | 86d92f5465958752481269348d474414dccb1552 |
| Headers | show |
| Series | [1/1] usb: storage: Add quirk for Samsung Fit flash | expand |
On Mon, 2 Mar 2020, Jim Lin wrote: > Current driver has 240 (USB2.0) and 2048 (USB3.0) as max_sectors, > e.g., /sys/bus/scsi/devices/0:0:0:0/max_sectors > > If data access times out, driver error handling will issue a port > reset. > Sometimes Samsung Fit (090C:1000) flash disk will not respond to > later Set Address or Get Descriptor command. > > Adding this quirk to limit max_sectors to 64 sectors to avoid issue > occurring. > > Signed-off-by: Jim Lin <jilin@nvidia.com> > --- > drivers/usb/storage/unusual_devs.h | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h > index 1cd9b6305b06..1880f3e13f57 100644 > --- a/drivers/usb/storage/unusual_devs.h > +++ b/drivers/usb/storage/unusual_devs.h > @@ -1258,6 +1258,12 @@ UNUSUAL_DEV( 0x090a, 0x1200, 0x0000, 0x9999, > USB_SC_RBC, USB_PR_BULK, NULL, > 0 ), > > +UNUSUAL_DEV(0x090c, 0x1000, 0x1100, 0x1100, > + "Samsung", > + "Flash Drive FIT", > + USB_SC_DEVICE, USB_PR_DEVICE, NULL, > + US_FL_MAX_SECTORS_64), > + > /* aeb */ > UNUSUAL_DEV( 0x090c, 0x1132, 0x0000, 0xffff, > "Feiya", Acked-by: Alan Stern <stern@rowland.harvard.edu>
Hello, Jim Lin wrote on 02.03.20 15:21: > Current driver has 240 (USB2.0) and 2048 (USB3.0) as max_sectors, > e.g., /sys/bus/scsi/devices/0:0:0:0/max_sectors > > If data access times out, driver error handling will issue a port > reset. > Sometimes Samsung Fit (090C:1000) flash disk will not respond to > later Set Address or Get Descriptor command. > > Adding this quirk to limit max_sectors to 64 sectors to avoid issue > occurring. > This may need revisiting as it appears to be a performance killer (3-4 times slower seq reads) for otherwise perfectly working sticks. Going down from 2048 to 64 seems to cause a pretty significant speed degradation. Here are a few examples: # lsusb Bus 002 Device 012: ID 090c:1000 Silicon Motion, Inc. - Taiwan (formerly Feiya Technology Corp.) Flash Drive # lsusb -t /: Bus 02.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/6p, 5000M |__ Port 3: Dev 12, If 0, Class=Mass Storage, Driver=usb-storage, 5000M # dmesg [23153.493726] usb 2-3: Product: Flash Drive FIT [23153.493729] usb 2-3: Manufacturer: Samsung [23153.493731] usb 2-3: SerialNumber: 0375119090033353 [23153.575386] usb-storage 2-3:1.0: USB Mass Storage device detected [23153.575514] usb-storage 2-3:1.0: Quirks match for vid 090c pid 1000: 400 [23153.575559] scsi host2: usb-storage 2-3:1.0 [23153.576529] usbcore: registered new interface driver usb-storage [23153.578645] usbcore: registered new interface driver uas # cat /proc/scsi/usb-storage/* Host scsi2: usb-storage Vendor: Samsung Product: Flash Drive FIT Serial Number: 0375119090033353 Protocol: Transparent SCSI Transport: Bulk Quirks: MAX_SECTORS_64 SANE_SENSE # hdparm -t /dev/sdb Timing buffered disk reads: 132 MB in 3.03 seconds = 43.62 MB/sec # dd if=/dev/sdb of=/dev/null bs=1M count=1000 1048576000 bytes (1,0 GB, 1000 MiB) copied, 22,3564 s, 46,9 MB/s # rmmod uas usb_storage # modprobe usb_storage quirks=090c:1000: # hdparm -t /dev/sdb Timing buffered disk reads: 452 MB in 3.01 seconds = 150.33 MB/sec # dd if=/dev/sdb of=/dev/null bs=1M count=1000 1048576000 bytes (1,0 GB, 1000 MiB) copied, 6,51492 s, 161 MB/s [23612.690798] usb 2-3: Product: Intenso High Speed Line [23612.690799] usb 2-3: Manufacturer: SMI [23612.690801] usb 2-3: SerialNumber: 19112500000332 [23612.780771] usb-storage 2-3:1.0: USB Mass Storage device detected [23612.780895] usb-storage 2-3:1.0: Quirks match for vid 090c pid 1000: 400 [23612.780940] scsi host2: usb-storage 2-3:1.0 [23612.781093] usbcore: registered new interface driver usb-storage [23612.783226] usbcore: registered new interface driver uas # cat /proc/scsi/usb-storage/* Host scsi2: usb-storage Vendor: SMI Product: Intenso High Speed Line Serial Number: 19112500000332 Protocol: Transparent SCSI Transport: Bulk Quirks: MAX_SECTORS_64 SANE_SENSE # hdparm -t /dev/sdb Timing buffered disk reads: 220 MB in 3.00 seconds = 73.22 MB/sec # dd if=/dev/sdb of=/dev/null bs=1M count=1000 1048576000 bytes (1,0 GB, 1000 MiB) copied, 11,5469 s, 90,8 MB/s # rmmod uas usb_storage # modprobe usb_storage quirks=090c:1000: # hdparm -t /dev/sdb Timing buffered disk reads: 1016 MB in 3.00 seconds = 338.51 MB/sec # dd if=/dev/sdb of=/dev/null bs=1M count=1000 1048576000 bytes (1,0 GB, 1000 MiB) copied, 3,31022 s, 317 MB/s I'm using both sticks as a bootable/emergency media (Debian stable, kernel 4.19/no-quirks with X, XFCE, web browser, etc) and haven't had any issues with timeouts, unresponsiveness or whatsoever. When tested with recent kernels (e.g. Debian testing/5.5, Ubuntu 20.04 LTS/5.4) it's slow. Setting "options usb_storage quirks=090c:1000:" in /etc/modprobe.d as a workaround for now. > Signed-off-by: Jim Lin <jilin@nvidia.com> > --- > drivers/usb/storage/unusual_devs.h | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h > index 1cd9b6305b06..1880f3e13f57 100644 > --- a/drivers/usb/storage/unusual_devs.h > +++ b/drivers/usb/storage/unusual_devs.h > @@ -1258,6 +1258,12 @@ UNUSUAL_DEV( 0x090a, 0x1200, 0x0000, 0x9999, > USB_SC_RBC, USB_PR_BULK, NULL, > 0 ), > > +UNUSUAL_DEV(0x090c, 0x1000, 0x1100, 0x1100, > + "Samsung", > + "Flash Drive FIT", > + USB_SC_DEVICE, USB_PR_DEVICE, NULL, > + US_FL_MAX_SECTORS_64), > + > /* aeb */ > UNUSUAL_DEV( 0x090c, 0x1132, 0x0000, 0xffff, > "Feiya", > Thanks, Atanas
In my case device will be in hung state even issuing port reset if without this patch. jim On 2020/5/2 下午9:56, Atanas Dinev wrote: > External email: Use caution opening links or attachments > > > Hello, > > Jim Lin wrote on 02.03.20 15:21: >> Current driver has 240 (USB2.0) and 2048 (USB3.0) as max_sectors, >> e.g., /sys/bus/scsi/devices/0:0:0:0/max_sectors >> >> If data access times out, driver error handling will issue a port >> reset. >> Sometimes Samsung Fit (090C:1000) flash disk will not respond to >> later Set Address or Get Descriptor command. >> >> Adding this quirk to limit max_sectors to 64 sectors to avoid issue >> occurring. >> > This may need revisiting as it appears to be a performance killer (3-4 > times slower seq reads) for otherwise perfectly working sticks. > Going down from 2048 to 64 seems to cause a pretty significant speed > degradation. > Here are a few examples: > > # lsusb > Bus 002 Device 012: ID 090c:1000 Silicon Motion, Inc. - Taiwan > (formerly Feiya Technology Corp.) Flash Drive > # lsusb -t > /: Bus 02.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/6p, 5000M > |__ Port 3: Dev 12, If 0, Class=Mass Storage, Driver=usb-storage, > 5000M > > # dmesg > [23153.493726] usb 2-3: Product: Flash Drive FIT > [23153.493729] usb 2-3: Manufacturer: Samsung > [23153.493731] usb 2-3: SerialNumber: 0375119090033353 > [23153.575386] usb-storage 2-3:1.0: USB Mass Storage device detected > [23153.575514] usb-storage 2-3:1.0: Quirks match for vid 090c pid > 1000: 400 > [23153.575559] scsi host2: usb-storage 2-3:1.0 > [23153.576529] usbcore: registered new interface driver usb-storage > [23153.578645] usbcore: registered new interface driver uas > > # cat /proc/scsi/usb-storage/* > Host scsi2: usb-storage > Vendor: Samsung > Product: Flash Drive FIT > Serial Number: 0375119090033353 > Protocol: Transparent SCSI > Transport: Bulk > Quirks: MAX_SECTORS_64 SANE_SENSE > > # hdparm -t /dev/sdb > Timing buffered disk reads: 132 MB in 3.03 seconds = 43.62 MB/sec > # dd if=/dev/sdb of=/dev/null bs=1M count=1000 > 1048576000 bytes (1,0 GB, 1000 MiB) copied, 22,3564 s, 46,9 MB/s > > # rmmod uas usb_storage > # modprobe usb_storage quirks=090c:1000: > > # hdparm -t /dev/sdb > Timing buffered disk reads: 452 MB in 3.01 seconds = 150.33 MB/sec > # dd if=/dev/sdb of=/dev/null bs=1M count=1000 > 1048576000 bytes (1,0 GB, 1000 MiB) copied, 6,51492 s, 161 MB/s > > > [23612.690798] usb 2-3: Product: Intenso High Speed Line > [23612.690799] usb 2-3: Manufacturer: SMI > [23612.690801] usb 2-3: SerialNumber: 19112500000332 > [23612.780771] usb-storage 2-3:1.0: USB Mass Storage device detected > [23612.780895] usb-storage 2-3:1.0: Quirks match for vid 090c pid > 1000: 400 > [23612.780940] scsi host2: usb-storage 2-3:1.0 > [23612.781093] usbcore: registered new interface driver usb-storage > [23612.783226] usbcore: registered new interface driver uas > > # cat /proc/scsi/usb-storage/* > Host scsi2: usb-storage > Vendor: SMI > Product: Intenso High Speed Line > Serial Number: 19112500000332 > Protocol: Transparent SCSI > Transport: Bulk > Quirks: MAX_SECTORS_64 SANE_SENSE > > # hdparm -t /dev/sdb > Timing buffered disk reads: 220 MB in 3.00 seconds = 73.22 MB/sec > # dd if=/dev/sdb of=/dev/null bs=1M count=1000 > 1048576000 bytes (1,0 GB, 1000 MiB) copied, 11,5469 s, 90,8 MB/s > > # rmmod uas usb_storage > # modprobe usb_storage quirks=090c:1000: > # hdparm -t /dev/sdb > Timing buffered disk reads: 1016 MB in 3.00 seconds = 338.51 MB/sec > # dd if=/dev/sdb of=/dev/null bs=1M count=1000 > 1048576000 bytes (1,0 GB, 1000 MiB) copied, 3,31022 s, 317 MB/s > > > I'm using both sticks as a bootable/emergency media (Debian stable, > kernel 4.19/no-quirks with X, XFCE, web browser, etc) and haven't had > any issues with timeouts, unresponsiveness or whatsoever. > > When tested with recent kernels (e.g. Debian testing/5.5, Ubuntu 20.04 > LTS/5.4) it's slow. > > Setting "options usb_storage quirks=090c:1000:" in /etc/modprobe.d as > a workaround for now. > >> Signed-off-by: Jim Lin <jilin@nvidia.com> >> --- >> drivers/usb/storage/unusual_devs.h | 6 ++++++ >> 1 file changed, 6 insertions(+) >> >> diff --git a/drivers/usb/storage/unusual_devs.h >> b/drivers/usb/storage/unusual_devs.h >> index 1cd9b6305b06..1880f3e13f57 100644 >> --- a/drivers/usb/storage/unusual_devs.h >> +++ b/drivers/usb/storage/unusual_devs.h >> @@ -1258,6 +1258,12 @@ UNUSUAL_DEV( 0x090a, 0x1200, 0x0000, 0x9999, >> USB_SC_RBC, USB_PR_BULK, NULL, >> 0 ), >> >> +UNUSUAL_DEV(0x090c, 0x1000, 0x1100, 0x1100, >> + "Samsung", >> + "Flash Drive FIT", >> + USB_SC_DEVICE, USB_PR_DEVICE, NULL, >> + US_FL_MAX_SECTORS_64), >> + >> /* aeb */ >> UNUSUAL_DEV( 0x090c, 0x1132, 0x0000, 0xffff, >> "Feiya", >> --nvpublic
On Mon, Sep 12, 2022 at 7:56 PM Jim Lin <jilin@nvidia.com> wrote: >On Sat, 2022-09-10 at 20:52 +0900, SungHwan Jung wrote: >External email: Use caution opening links or attachments >>I have found that the read rate of "samsung Bar plus" is slower than on windows 11 (210mb/s -> 70mb/s) and recovered by disabling quirks (using /etc/modprobe.d) >>This patch affects not only "Samsung Flash Driver FIT", but also other usb flash storages. (They may use the same controller?) >>But I can't reproduce the timeout problem without quirks. >>Could you provide information to reproduce the timeout problem or logs? >>It may help us find other solutions to fix it. >> >>Thanks, >>SungHwan. >Issue was reproduced after device has bad block. > >--nvpublic If bad blocks cause the issue, in my opinion, it's better to apply this quirk by users (additional kernel parameters with a bootloader) if their flash drive has bad blocks, not the kernel patch, because the performance degradation is severe for several USB storage that didn't have bad blocks.
diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h index 1cd9b6305b06..1880f3e13f57 100644 --- a/drivers/usb/storage/unusual_devs.h +++ b/drivers/usb/storage/unusual_devs.h @@ -1258,6 +1258,12 @@ UNUSUAL_DEV( 0x090a, 0x1200, 0x0000, 0x9999, USB_SC_RBC, USB_PR_BULK, NULL, 0 ), +UNUSUAL_DEV(0x090c, 0x1000, 0x1100, 0x1100, + "Samsung", + "Flash Drive FIT", + USB_SC_DEVICE, USB_PR_DEVICE, NULL, + US_FL_MAX_SECTORS_64), + /* aeb */ UNUSUAL_DEV( 0x090c, 0x1132, 0x0000, 0xffff, "Feiya",
Current driver has 240 (USB2.0) and 2048 (USB3.0) as max_sectors, e.g., /sys/bus/scsi/devices/0:0:0:0/max_sectors If data access times out, driver error handling will issue a port reset. Sometimes Samsung Fit (090C:1000) flash disk will not respond to later Set Address or Get Descriptor command. Adding this quirk to limit max_sectors to 64 sectors to avoid issue occurring. Signed-off-by: Jim Lin <jilin@nvidia.com> --- drivers/usb/storage/unusual_devs.h | 6 ++++++ 1 file changed, 6 insertions(+)