On 02/20/2018 03:19 PM, Hugo Landau wrote: >> I also gave it a test on an OpenBMC romulus image. Looks fine, but that's >> an old custom U-Boot. Which defconfig did you use for U-Boot HEAD ? > evb-ast2500_defconfig.
ok > FYI, these changes are necessary, but not sufficient to get u-boot HEAD > (or for that matter u-boot 2017.11, another version tested) running. > > The other issues were > - the tests > while (!(readl(®s->ecc_test_ctrl) & SDRAM_TEST_DONE)); > and > while (!(readl(&info->regs->config) & SDRAM_CONF_CACHE_INIT_DONE)); > which appear in various places in the u-boot source and which spin > forever. I made u-boot work by commenting these out in u-boot rather > than patching qemu, not familiar enough with qemu to implement this. This patch : https://github.com/openbmc/qemu/commit/4fb98fffd3115d8d3d0a16a1033f5335b5c0fd9b fakes some more SDMC registers to let the SDRAM initialization run. you might want to take a look at it. Thanks, C. > - the call to reset_assert in ast2500_sdrammc_probe seems to actually > reset the machine rather than just initialize SDRAM as it is > apparently supposed to, leading to an infinite cycle of resets. > Couldn't quite figure out how it was supposed to work, so I > commented this out, since obviously qemu doesn't actually have SDRAM > initialization requirements. > > The above changes plus this patch allowed u-boot to get to the u-boot > CLI. Haven't tried booting anything with it yet though. >
