On Tue, Apr 9, 2019 at 12:47 PM Heiko Stübner heiko@sntech.de wrote:
Hi Kever,
Am Dienstag, 2. April 2019, 10:46:54 CEST schrieb Kever Yang:
Rockchip SOC's mmc controller does not support read data from mmc to sram, we need a bounce buffer(in sdram), and then copy to sram.
Signed-off-by: Kever Yang kever.yang@rock-chips.com
common/bouncebuf.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)
diff --git a/common/bouncebuf.c b/common/bouncebuf.c index a7098e2caf..364fb17c96 100644 --- a/common/bouncebuf.c +++ b/common/bouncebuf.c @@ -26,6 +26,18 @@ static int addr_aligned(struct bounce_buffer *state) return 0; }
+#ifdef CONFIG_ARCH_ROCKCHIP
/** Rockchip SOC's mmc controller does not support read data* from mmc to sram, we need a bounce buffer(in sdram), and then* copy to sram.*/if (((ulong)state->user_buffer & 0xfff80000) ==CONFIG_ROCKCHIP_IRAM_BASE) {debug("Unsupport IRAM space %p\n", state->user_buffer);return 0;}wouldn't it be easier to just check for "in-ddr-region"? Rockchip SoCs may have multiple sram areas one might want to use.
Also I get the feeling this should not live in an ARCH_ROCKCHIP ifdef. Instead maybe define some sort of kconfig settings to describe the directly usable memory areas, that the bounce_buffer then could use?
Indeed this still looks like a hack and should not be applied. My comments to [1] still apply to this patch: arch-specific defines should not be located in generic files if we can avoid it.
[1] https://patchwork.ozlabs.org/patch/1069730/
Regards, Simon
Heiko
+#endif /* Aligned */ return 1; }
U-Boot mailing list U-Boot@lists.denx.de https://lists.denx.de/listinfo/u-boot