| Message ID | 20221026064021.16683-1-zev@bewilderbeest.net |
|---|---|
| State | New |
| Headers | show |
| Series | [u-boot,v2019.04-aspeed-openbmc] Revert "config/ast2500: Enable RAM devices" | expand |
On Tue, Oct 25, 2022 at 11:40:21PM PDT, Zev Weiss wrote: >This reverts commit ba91e9df1e16db0d209177148e864c65e58eb00f. > >CONFIG_RAM=y currently breaks DRAM initialization on ast2500; Aspeed >recommends against using it [0]. > >[0] https://lore.kernel.org/openbmc/HK0PR06MB2834AE1581020A5B7CE191839C5B9@HK0PR06MB2834.apcprd06.prod.outlook.com/ > >Signed-off-by: Zev Weiss <zev@bewilderbeest.net> >--- > >Note that I'm not at all tied to this particular patch as the fix if >the interested parties can agree on better course of action; I'm >mostly just hoping to spur some further conversation given that >currently a vanilla OpenBMC build bricks my ast2500 systems in a way >that's not real easy to recover without a flash programmer. > Ping...I'm hoping to get e3c246d4i migrated to u-boot-aspeed-sdk soon (bearing in mind the end-of-year deadline suggested in Joel's email a few months ago [0]), but with this issue outstanding doing so will make any such systems self-bricking unless I hack around it with a bandaid kconfig fragment in a bbappend, which doesn't seem like a great fix since it's not really an e3c246d4i-specific problem. Thanks, Zev [0] https://lore.kernel.org/openbmc/CACPK8Xe4ijKWnURT4T9em2pUqifNdkZUfg0dd5osATYnqqutSw@mail.gmail.com/
On Wed, 9 Nov 2022 at 11:48, Zev Weiss <zweiss@equinix.com> wrote: > > On Tue, Oct 25, 2022 at 11:40:21PM PDT, Zev Weiss wrote: > >This reverts commit ba91e9df1e16db0d209177148e864c65e58eb00f. > > > >CONFIG_RAM=y currently breaks DRAM initialization on ast2500; Aspeed > >recommends against using it [0]. > > > >[0] https://lore.kernel.org/openbmc/HK0PR06MB2834AE1581020A5B7CE191839C5B9@HK0PR06MB2834.apcprd06.prod.outlook.com/ > > > >Signed-off-by: Zev Weiss <zev@bewilderbeest.net> > >--- > > > >Note that I'm not at all tied to this particular patch as the fix if > >the interested parties can agree on better course of action; I'm > >mostly just hoping to spur some further conversation given that > >currently a vanilla OpenBMC build bricks my ast2500 systems in a way > >that's not real easy to recover without a flash programmer. > > > > Ping...I'm hoping to get e3c246d4i migrated to u-boot-aspeed-sdk soon > (bearing in mind the end-of-year deadline suggested in Joel's email a > few months ago [0]), but with this issue outstanding doing so will make > any such systems self-bricking unless I hack around it with a bandaid > kconfig fragment in a bbappend, which doesn't seem like a great fix > since it's not really an e3c246d4i-specific problem. I've merged this for now. We should consider removing the C DRAM training, or re-working it to match what the assembly does (IIRC it is different in structure compared to the assembly. I don't know why). > > > Thanks, > Zev > > [0] https://lore.kernel.org/openbmc/CACPK8Xe4ijKWnURT4T9em2pUqifNdkZUfg0dd5osATYnqqutSw@mail.gmail.com/
diff --git a/configs/evb-ast2500_defconfig b/configs/evb-ast2500_defconfig index d2a4e835dd12..b83c315e5057 100644 --- a/configs/evb-ast2500_defconfig +++ b/configs/evb-ast2500_defconfig @@ -64,7 +64,6 @@ CONFIG_PHY_GIGE=y CONFIG_FTGMAC100=y CONFIG_PHY=y CONFIG_PINCTRL=y -CONFIG_RAM=y CONFIG_DM_SERIAL=y CONFIG_SYS_NS16550=y CONFIG_SPI=y
This reverts commit ba91e9df1e16db0d209177148e864c65e58eb00f. CONFIG_RAM=y currently breaks DRAM initialization on ast2500; Aspeed recommends against using it [0]. [0] https://lore.kernel.org/openbmc/HK0PR06MB2834AE1581020A5B7CE191839C5B9@HK0PR06MB2834.apcprd06.prod.outlook.com/ Signed-off-by: Zev Weiss <zev@bewilderbeest.net> --- Note that I'm not at all tied to this particular patch as the fix if the interested parties can agree on better course of action; I'm mostly just hoping to spur some further conversation given that currently a vanilla OpenBMC build bricks my ast2500 systems in a way that's not real easy to recover without a flash programmer. configs/evb-ast2500_defconfig | 1 - 1 file changed, 1 deletion(-)