Jens Gehrlein wrote: > Hi Andy, > > Andy Fleming schrieb: >> The ecm variable in sdram.c was being declared for all 8548, but only >> used by specific 8548 boards, so we make that variable require those >> specific boards, too >> > > [snip] > >> diff --git a/board/tqc/tqm85xx/sdram.c b/board/tqc/tqm85xx/sdram.c >> index 2b3de44..6d73a88 100644 >> --- a/board/tqc/tqm85xx/sdram.c >> +++ b/board/tqc/tqm85xx/sdram.c >> @@ -79,7 +79,9 @@ long int sdram_setup (int casl) >> volatile ccsr_ddr_t *ddr = (void *)(CONFIG_SYS_MPC85xx_DDR_ADDR); >> #ifdef CONFIG_TQM8548 >> volatile ccsr_gur_t *gur = (void *)(CONFIG_SYS_MPC85xx_GUTS_ADDR); >> +#if defined(CONFIG_TQM8548_AG) || defined(CONFIG_TQM8548_BE) >> volatile ccsr_local_ecm_t *ecm = (void *)(CONFIG_SYS_MPC85xx_ECM_ADDR); >> +#endif >> #else /* !CONFIG_TQM8548 */ >> unsigned long cfg_ddr_timing1; >> unsigned long cfg_ddr_mode; > > Currently ecm is only used in DDR20 workaround. > > Strictly speaking, this workaround should apply to all TQM8548, not > only to its variants AG and BE, because it's valid for all CPU > revisions. > > I overlooked this, when reviewing the patch "MPC85xx: TQM8548: > workaround for erratum DDR 19 and 20".
I decided to enable the work-around only for new TQM8548 board variants as I'm not able to test the old ones. > But I'm fine with the current code and your patch for now. Currently, > there are no other TQM8548 variants than AG and BE. Andy, thanks for the quick fix. Wolfgang. _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot