Hi Matthijs, I have heard you have 4GB working! Congrats!
What is the status? Still a hack or already something we should integrate into our git repo?
And what were the missing pieces?
BR, Nikolaus
On 24 February 2017 at 18:59, H. Nikolaus Schaller hns@goldelico.com wrote:
I have heard you have 4GB working! Congrats!
Indeed! In retrospect it basically just worked with trivial changes, I was just getting distracted by various unrelated problems.
In particular, I discovered that one SD card had a tendency to return different data than I had written to it. That probably didn't help with the crash probability :P
What is the status? Still a hack or already something we should integrate into our git repo?
Hack :-)
https://github.com/mvduin/u-boot/commit/dd32990a8f3c8e1ee3849c380d01372b6af0...
Right now it also still tells the kernel it has 2GB via DT. For testing I've overridden that using kernel parameters. I still need to check how the memory info ends up in DT exactly.
And preferably I'd do auto-detection of the memory size, which shouldn't be hard though.
Matthijs
Am 24.02.2017 um 23:34 schrieb Matthijs van Duin matthijsvanduin@gmail.com:
On 24 February 2017 at 18:59, H. Nikolaus Schaller hns@goldelico.com wrote: I have heard you have 4GB working! Congrats!
Indeed! In retrospect it basically just worked with trivial changes, I was just getting distracted by various unrelated problems.
In particular, I discovered that one SD card had a tendency to return different data than I had written to it. That probably didn't help with the crash probability :P
I already had that once... I wrote files onto the SD card, checked md5 sums but reading on the target device showed different md5.
What is the status? Still a hack or already something we should integrate into our git repo?
Hack :-)
https://github.com/mvduin/u-boot/commit/dd32990a8f3c8e1ee3849c380d01372b6af0...
Oh, that's all?
Maybe lisa_map_4G_x_2_x_2 could be set in the board file hence not modifying other boards.
This seems to be possible by emif_get_dmm_regs
http://git.goldelico.com/?p=gta04-uboot.git;a=blob;f=board/ti/dra7xx/evm.c;h...
or
http://git.goldelico.com/?p=gta04-uboot.git;a=blob;f=board/ti/panda/panda.c;...
I think we could copy the size detection of the dra7. Except that it appears to read RAM size from a config EEPROM by board_ti_get_emif1_size().
Right now it also still tells the kernel it has 2GB via DT. For testing I've overridden that using kernel parameters. I still need to check how the memory info ends up in DT exactly.
And preferably I'd do auto-detection of the memory size, which shouldn't be hard though.
Yes, that would be the best since we have ca. 30 boards with only 2GB and it would not be easy to ask to install different U-Boot and kernels...
BR and applause for the breakthrough, Nikolaus
-
H. Nikolaus Schaller -
Matthijs van Duin