On Fri, 6 Sep 2019 at 14:26, Belisko Marek <marek.belisko@xxxxxxxxx> wrote: > > On Fri, Sep 6, 2019 at 2:19 PM Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote: > > > > On Fri, 6 Sep 2019 at 11:14, Belisko Marek <marek.belisko@xxxxxxxxx> wrote: > > > > > > Hi, > > > > > > I'm trying to get some insights about how to debug emmc speed issue. > > > We're using board with am437x CPU using mainline 4.12 kernel (added > > > only devicetree for our custom board). eMMC is 8bit connect (using > > > sandisk SDINBDG4-8G-I1 chip). > > > > > > Issue is that when copying data from SD card to eMMC we get throughput > > > only ~300kB/s (it should be bit higher IIRC). I'm just looking for > > > some advice what to check and how to find out where issue can be > > > (still unsure if SW or HW). Thanks a lot. > > > > Mount debugfs and look at the ios node for the mmc host in question. > > There is information about clocks, busdwidth, etc. > Thanks for info. From below it looks quite ok: > clock: 52000000 Hz > vdd: 21 (3.3 ~ 3.4 V) > bus mode: 2 (push-pull) > chip select: 0 (don't care) > power mode: 2 (on) > bus width: 3 (8 bits) > timing spec: 1 (mmc high-speed) > signal voltage: 0 (3.30 V) > driver type: 0 (driver type B) > > compared to gumstix poblano (same cpu): > cat /sys/kernel/debug/mmc1/ios > clock: 50000000 Hz > vdd: 21 (3.3 ~ 3.4 V) > bus mode: 2 (push-pull) > chip select: 0 (don't care) > power mode: 2 (on) > bus width: 2 (4 bits) > timing spec: 2 (sd high-speed) > signal voltage: 0 (3.30 V) > driver type: 0 (driver type B) > > difference is in bus width (I'm confused as poblano according dts + > schematics should also have 8 bit interface) and timing spec are > different for my case. Anything else I can check? Thanks. This looks fine to me. I think you need to do some deeper debug to find out what the issue is, sorry for not being able to help any more than this. Of course you already tried reading/writing to the eMMC and SD independently, so you have a reference of the throughput values? Kind regards Uffe
