On 4 December 2015 at 21:34, Daniel Lenski <dlenski@xxxxxxxxx> wrote: > On Fri, Dec 4, 2015 at 11:21 AM, Daniel Lenski <dlenski@xxxxxxxxx> wrote: >> On 21 October 2015 at 12:02, Ulf Hansson <ulf.hansson <at> linaro.org> wrote: >>> As I don't know the HW, I can just provide some guesses of how to narrow down the problem. >>> >>> If you re-build the kernel and make changes around which MMC_CAPS the host supports, you can try to narrow down the problem if it's related to speed modes. >>> >>> For example start by using *only* the following MMC_CAPS (updates to be made in rtsx_usb_init_host() - drivers/mmc/host/rtsx_usb.c): >>> >>> MMC_CAP_4_BIT_DATA | MMC_CAP_SD_HIGHSPEED | MMC_CAP_NEEDS_POLL >> >>> ..and if no error, try add cap by cap to see what happens. > > Hi Ulf, > Thanks! I had success with this approach :-) > > I started with the stock Ubuntu 4.2.0-21 kernel, and rebuilt ONLY the > rtsx_usb and rtsx_usb_sdmmc modules: > > $ uname -a > Linux dlenski-ultra 4.2.0-21-generic #25-Ubuntu SMP Wed Dec 2 18:42:25 > UTC 2015 x86_64 x86_64 x86_64 GNU/Linux > $ make -C /lib/modules/`uname -r`/build M=$PWD drivers/mfd/rtsx_usb.ko > $ make -C /lib/modules/`uname -r`/build M=$PWD > drivers/mmc/host/rtsx_usb_sdmmc.ko > > First I tried disabling SDR50 only. Looks like SDR25 is a no-go for > this card. I get the same errors as with SDR50 (-110): > > [ 4441.001959] mmc0: new ultra high speed SDR25 SDHC card at address aaaa > [ 4441.002292] mmcblk0: mmc0:aaaa SU32G 29.7 GiB > [ 4441.006009] mmcblk0: > [ 4441.678714] mmcblk0: error -110 sending stop command, original cmd > response 0x900, card status 0x900 > > Next, I disabled SDR25 as well. SDR12 *seems* to work fine at first, > although I get a different error in dmesg immediately when the card is > detected (-32), and then start to get a blizzard of error (-110) when > I try to copy lots of files to the card. > > [ 4583.052995] mmc0: new ultra high speed SDR12 SDHC card at address aaaa > [ 4583.053304] mmcblk0: mmc0:aaaa SU32G 29.7 GiB > [ 4583.055852] mmcblk0: > [ 4583.328276] FAT-fs (mmcblk0): Volume was not properly unmounted. > Some data may be corrupt. Please run fsck. > [ 4691.924909] mmcblk0: error -32 sending stop command, original cmd > response 0x900, card status 0xc00 > [ 4695.869462] mmcblk0: error -32 sending stop command, original cmd > response 0x900, card status 0xc00 > [ 4709.745164] FAT-fs (mmcblk0): Volume was not properly unmounted. > Some data may be corrupt. Please run fsck. > [ 5082.557400] mmcblk0: error -110 sending status command, retrying > [ 5083.158663] mmcblk0: error -110 sending status command, aborting > > Finally, I disabled SDR12 as well. And now the card works just fine: > > mmc->caps = MMC_CAP_4_BIT_DATA | MMC_CAP_SD_HIGHSPEED | > MMC_CAP_MMC_HIGHSPEED | MMC_CAP_BUS_WIDTH_TEST | > /* MMC_CAP_UHS_SDR12 | MMC_CAP_UHS_SDR25 | MMC_CAP_UHS_SDR50 |*/ > MMC_CAP_NEEDS_POLL; > > [ 5196.200933] mmc0: new high speed SDHC card at address aaaa > [ 5196.201239] mmcblk0: mmc0:aaaa SU32G 29.7 GiB > [ 5196.203600] mmcblk0: > [ 5196.462521] FAT-fs (mmcblk0): Volume was not properly unmounted. > Some data may be corrupt. Please run fsck. > > In conclusion... it seems that the kernel is attempting to drive this > card faster than it can actually work. What I'm trying to figure out > now: > - How does the kernel attempt to figure out the capability of a given > SD card? Where is that in the kernel source tree? drivers/mmc/core/* It's kind of complicated, but please go ahead and have a look. > - Is it possible to make the list of supported caps a module > parameter, to more easily throttle high-speed modes that don't work > with a particular card? Maybe a sysfs/debugfs interface would be better, but I like the idea! Currently there is no way for userspace to disable some of these "speed mode" caps, which would be very useful in cases like yours. I don't know if I ever will get time to hack on such code, so feel free to have a try yourself! Regarding how to fix this problem right now, I guess we should remove the caps which the host driver wrongly claims to support. Do you want to send a patch for that? Kind regards Uffe -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
