F2A85-M fails to detect 24-32GB of RAM properly
|Status:||Response Needed||Start date:||04/25/2016|
|Assignee:||Kyösti Mälkki||% Done:|
I am running Coreboot 4.3 release on the Asus F2a85-M with 16GB of Crucial Ballistix Sport DDR-1600 DIMMs at 1,5V (2 sticks). When I populate the remaining RAM slots with another 2 sticks of the same type, the system does not boot to the Payload. When I put only 1 of the additional sticks in (so in total I should have 24GB), the system boots fine but I am seeing 78GB (!!) of memory detected on the operating system - see the attached output.
I don't have the coreboot config at hand, but I am just running basically 4.3 release configured for this board together with the vgabios blob.
Is it safe to try the 4.4 prerelease to see if the bug can be reproduced there as well? I am just asking because I have no separate chip as backup nor the proper clip for hardware flashing this device.
#2 Updated by Kyösti Mälkki over 3 years ago
- Assignee set to Kyösti Mälkki
- Status changed from New to Response Needed
You need to collect coreboot log from serial port, but I don't really expect to find the solution there in standard output.
I can enable lots of additional debugging for RAM detection, but I cannot provide you with a binary that is pre-tested to boot f2a85-m to OS. So please get backup SPI parts or programming gear.
#3 Updated by Daniel Kulesz over 3 years ago
I got the gear (+1 spare chip) already, so I just some time for testing it. Can you provide me with a config that has the debug etc. enabled?
And from testing these modules in another machine I found out that they are sort of "grade b" items, since they run stable only at 1333MHz while they are specified to be working at 1600MHz. I assume this is also what is causing the issues in Coreboot.
#4 Updated by Kyösti Mälkki over 3 years ago
I need to modify source tree to get extended debugging about DIMMs. In the meantime, you can post the serial console output of coreboot here.
Attach log with 2, 3 and 4 DIMMs installed.
Even if DIMMs where faulty/unstable at 1600MHz, having total capacity reported incorrectly suggests there could be other issues.
#5 Updated by Kyösti Mälkki over 3 years ago
- File coreboot.rom.bz2 added
You can find the patch with verbose raminit debugging here: https://review.coreboot.org/#/c/15320/
Record of last known-good f2a85-m test was commit ab90f96b and I have attached a binary with my patch applied on top. Note that I did not include video bios in the image.
Author: Kyösti Mälkki firstname.lastname@example.org
Date: Wed Jun 22 22:50:10 2016 +0300
f15tn: enable IDS printing [WIP] Change-Id: Ie5c0de6358b294160f9bf0a202161722f88059c1 Signed-off-by: Kyösti Mälkki <email@example.com>
Author: Furquan Shaikh firstname.lastname@example.org
Date: Tue May 31 23:38:26 2016 -0700