On 08/06/21, Jorge Ramirez-Ortiz, Foundries wrote: > On 08/06/21, Michal Simek wrote: > > Hi, > > > > On 6/7/21 8:41 PM, Jorge Ramirez-Ortiz, Foundries wrote: > > > On 07/06/21, Jorge Ramirez-Ortiz, Foundries wrote: > > >> hi Michal > > >> > > >> um, when we exchanged emails about enabling ECC support for MPSoC, I > > >> left with the understanding that there already was a DMA driver > > >> available in u-boot that I could use to initialize the memory. > > >> > > >> do you have something in the works or will I have to write such a > > >> driver? compatible would be 'xlnx,zynqmp-dma-1.0' > > > > > > alternatively, would it be acceptable - upstream - If I just init ECC > > > as done by FSBL (ie, without a proper DMA driver, just accessing the > > > controller)? > > > > There are 2 things here. > > The first thing that would be the best to do it proper framework which > > is likely dma where dma_memset should be likely added. > > The second that this code should land in SPL which can't be big that's > > why this feature should be very small. > > the way I am doing it ATM is just a verbatim of XFsbl_EccInit (grabs a > channel and inits all memory to DEADBEEF). About 100 LOC total.
1:1 when ported to SPL (except the coding standards) https://github.com/Xilinx/embeddedsw/blob/master/lib/sw_apps/zynqmp_fsbl/src/xfsbl_initialization.c#L1574 > > I belive the FSBL copyright notice allows for that (below excerpt). > > [..] Permission is hereby granted, free of charge, to any person obtaining a > copy > of this software and associated documentation files (the "Software"), to deal > in the Software without restriction, including without limitation the rights > to use, copy, modify, merge, publish, distribute, sublicense, and/or sell > copies of the Software, and to permit persons to whom the Software is > furnished to do so, [..] > > but yes this just works as expected. > > a bit trickier is where to init from - for the particular product > I am working on I can hardcode the addresses and sizes and init right > after board_init_f in spl.c...havent had the time to think about > something portable yet - if you have a suggestion I could send both > patches (one for the spl_ecc_init(addr, len) function and another one > calling it) > > thanks for your inputs > > > > > Based on above I think that would be good to have long term both of them > > in the tree. It means feel free to send short version and let's look at > > it and dma one should be hopefully pretty much using the same code via > > dma framework. > > > > Thanks, > > Michal > >