Bug #22
closedX230 native raminit failed
0%
Description
Native raminit failed with: "edge write discovery failed".
It's failing on channel 1 with two DIMMs installed, while it has no problems booting with a single DIMM.
The test fails on lane 1.
These are the Cvals of channel 1:
Cval: 1, 1, 0, 26
Cval: 1, 1, 1, 21
Cval: 1, 1, 2, 41
Cval: 1, 1, 3, 4d
Cval: 1, 1, 4, 4e
Cval: 1, 1, 5, 52
Cval: 1, 1, 6, 2c
Cval: 1, 1, 7, 5e
Cval: 1, 0, 0, 22
Cval: 1, 0, 1, 5e
Cval: 1, 0, 2, 3b
Cval: 1, 0, 3, 4a
Cval: 1, 0, 4, 4a
Cval: 1, 0, 5, 4c
Cval: 1, 0, 6, 29
Cval: 1, 0, 7, 5b
Values for the same lane are about the same (+- 6units), but only lane 1
shows a huge offset of 61.
Files
Updated by Patrick Rudolph almost 9 years ago
By adding a simple check to timB discovery:
FOR_ALL_LANES {
if (((ctrl->timings[channel][slotrank].lanes[lane].timB & 0x3f) == 0x3e) ||
((ctrl->timings[channel][slotrank].lanes[lane].timB & 0x3f) == 0x3f))
ctrl->timings[channel][slotrank].lanes[lane].timB += 2;
}
The issue is fixed for me.
This patch just covers the problem instead of fixing it.
Updated by Patrick Rudolph over 8 years ago
- Status changed from New to Closed
Should be fixed in latest git.
Needs tests on x230.
Updated by Alexander Couzens over 8 years ago
- Status changed from Closed to Needs Testing
Updated by Martin Kepplinger over 7 years ago
- File debugboot1.txt debugboot1.txt added
Native ram init should really work, as I'm using 2 DIMMs and at least with ram init I don't have a problem.
I append a cdbmem log, with ram debug enabled, in case that helps.
Updated by Patrick Rudolph over 5 years ago
More raminit fixes landed in the meantime. Does it still happen?
Updated by Martin Kepplinger over 5 years ago
Haven't had issues with raminit in a long time on the X230.
Updated by Felix Singer almost 3 years ago
- Status changed from Needs Testing to Resolved