On Mon, Jun 8, 2015 at 6:29 PM, Gary Thomas <g...@mlbassoc.com> wrote: > On 2015-06-08 10:19, Eric Nelson wrote: >> >> Hi Andreas, >> >> On 06/08/2015 05:48 AM, Andreas Müller wrote: >>> >>> On Variscite Varsom we have three devices connected to MMC >>> >>> * MMC1: eMMC >>> * MMC2: SDCard >>> * MMC3: Wifi >>> >>> Without this patch only the first device was detected properly. >>> I found two workarounds which can be removed by this patch: >>> >>> 1. Variscite: booting from SDCard made possible by renumbering MMC >>> devices [1]. >>> 2. BoundaryDevices: Add workqueues for each mmc_host [2]. This had been >>> tested >>> here too but causes crashing sdcard for huge storage traffic. > > > Can you (Andreas) elaborate on this? Is there a procedure to show this > failure? > I'd like to test it on my board(s) that have multiple [active] MMC devices. > As this patch fixes some race there are likely many configurations around working properly and it it not easy to say which configuration causes reproducibility. Interesting note regarding race: Before I found this patch I tried to gather more information and enabled CONFIG_MMC_DEBUG and all MMC devices were properly detected! So thought 'OK not the best solution but a workaround fixing my problem'. Fortunately my colleague complained next morning for massive spam in kernel log so I had to look for a proper solution...
Andreas -- _______________________________________________ meta-freescale mailing list meta-freescale@yoctoproject.org https://lists.yoctoproject.org/listinfo/meta-freescale