Dear Jason Liu,
In message BANLkTinApAmm6hrcSofdYA6r7KUPMr8L+w@mail.gmail.com you wrote:
Yes, I have already thought about it before I submit patches. The fact is that there is there are two DDR chips connected to CS0 and CS1 of MX53 on loco board, but each DDR chip is 512MB, but CS0 and CS1 can support up to 1GB, which means the memory space is not continuous, when set:
The mapping is fix, and you cannot remap it as a single, contiguous region?
gd->ram_size = get_ram_size((volatile void *)CONFIG_SYS_SDRAM_BASE, PHYS_SDRAM_1_SIZE + PHYS_SDRAM_2_SIZE); it will return the same as:
gd->ram_size = get_ram_size((volatile void *)CONFIG_SYS_SDRAM_BASE, PHYS_SDRAM_1_SIZE);
You have always to test each bank on it's own, i. e.
size1 = get_ram_size(..., PHYS_SDRAM_1_SIZE); size2 = get_ram_size(..., PHYS_SDRAM_2_SIZE);
gd->ram_size = get_ram_size = size1 + size2;
I think the gd->ram_size is just used for uboot, but
Right, but it is supposed to show the full RAM size, and not only half of it (otherwise it should be gd->half_ram_size or similar ;-)
If you still think that I really need fix it, I will do it.
THis needs fixing.
This has been discussed before: this code should be factored out into common code.
Yes, I think I can send out another clean up patch for it to fix all the boards including mx31/35/51/53 later.
No. Wrong order. Please don't add to the mess and clean up later. Instead, clean up now, and base your patch on th already cleaned up code.
Please remove the '1' from all #defines that just enable features, i. e. where no specific numeric value is needed.
Yes, I can send out another clean up patch to fix all the i.mx board config> file
Again: No. Please clean up NOW.
This patch will not be accepted as is.
BTW, this patch set has been send out 2 months ago and Stefano has send out pull request to Albert and Albert has pulled in already, do you agree I send out another patch to address your comments? Thanks for your consideration.
No, this patch has serious bugs (memory size calculation) and other issues (needs cleanup) and is thus rejected.
Best regards,
Wolfgang Denk