Bug 4852

Summary:	system boot stops after installing efi memory
Product:	[Yocto Project Subprojects] Kernel	Reporter:	David Anders <danders>
Component:	linux-yocto	Assignee:	Darren Hart <dvhart>
Status:	VERIFIED MOVED	QA Contact:
Severity:	major
Priority:	Medium+	CC:	alexandrux.palalau, scott.a.garman
Version:	unspecified
Target Milestone:	1.5 M4
Hardware:	x86
OS:	Multiple
Whiteboard:
OS type for building Yocto:	---	Type of Regression:	---
Verified:		Documentation change:	Don't know

Description David Anders 2013-07-15 20:22:10 UTC

Install PPI: 057A449A-1FDC-4C06-BFC9-F53F6A99BB92
PROGRESS CODE: V3020003 I0
Evaluate PEI DEPEX for FFS(9618C0DC-50A4-496C-994F-7241F282ED01)
  PUSH GUID(2AB86EF5-ECB5-4134-B556-3854CA1FE1B4) = TRUE
  PUSH GUID(3ACF33EE-D892-40F4-A2FC-3854D2E1323D) = TRUE
  PUSH GUID(A55D6970-1306-440C-8C72-8F51FAFB2926) = TRUE
  PUSH GUID(01F34D25-4DE2-23AD-3FF3-36353FF323F1) = TRUE
  PUSH GUID(490E9D85-8AEF-4193-8E56-F734A9FFAC8B) = TRUE
  AND
  AND
  AND
  AND
  END
  RESULT = TRUE
Loading PEIM at 0x000FFFDA3BC EntryPoint=0x000FFFDA494 PlatformPeim.efi
PROGRESS CODE: V3020002 I0
PeiTcPreMemInit()
UpdateBootMode()
IsFirstBoot = 0 , EnableFastBoot= 1.
Boot with Minimum configuration
Install PPI: 7408D748-FC8C-4EE6-9288-C4BEC092A410
Notify: PPI Guid: 7408D748-FC8C-4EE6-9288-C4BEC092A410, Peim notify entry point: FFFD1DFE
Reinstall PPI: 6F8C2B35-FEF4-448D-8256-E11B19D61077
TCClearSmiAndWake()
CallSecPlatformInformationPpi()
Detect graphic Memory Before MRC
IGD enabled.
MRC Entry
InstallEfiMemory.

Comment 1 Lee Leahy 2013-09-12 17:59:35 UTC

I have assigned this to Darren Hart since "hardware" is not an option under Product or Component.  This issue should be assigned back to Dave Anders at Circuitco.

Sent: Thursday, August 01, 2013 2:45 PM
To: David Anders
Cc: Garman, Scott A; Bob Smith; Hart, Darren; Dave Albert; Stewart, David C; Mitchell, Mark L; Krau, Michael P; Rehn, Peter; Peterson, Charlie; Chin, Kenyon B
Subject: RE: MinnowBoard FirstBoot Issue

Hi Dave,

The auto-power-on issue is a hardware issue with the board.  The pull-up resistors (R51, R182, R171 and R411) for (E6XX_GPIO_SUS0, E6XX_GPIO_SUS5, E6XX_GPIO_SUS6 and E6XX_GPIO_SUS8) need to be pulled up to 3V3, not 3V3_S.

During the failing boot the memory configuration is incorrectly communicated to the processor and the resulting configuration is:

    Ranks: 1 - 2 sets of memory chips
    Width: 1 - 16 bits
    Density: 0 - 256M

This corresponds to the strapping inputs being at ground (0).  The correct input requires all of these strapping inputs to be at 3.3V which is a logical 1.  This is the case when the board is powered-off using the on-board  power switch, delaying for 5 seconds and then powering the board back on using the on-board power button.  However when you increase the delay to 2 minutes instead of 5 seconds the 3V3_S supply has a chance to discharge fully and the board fails to boot again.

During a successful boot the output is:
    Ranks: 0 - 1 sets of memory chips
    Width: 0 - 8 bits
    Density: 3 - 2G

The attached debug firmware displays the memory configuration prior to running the memory test that has been failing.  You may use this to verify the issue at Circuitco as well as verify the solution.  This image also includes the other debug output that you requested and the setting of the LEDs.

Comment 2 Darren Hart 2013-09-17 17:02:13 UTC

Hardware issues are not tracked in this bugzilla, so this will be closed and needs to be tracked in Circuit Co's issue tracker.

Marking as MOVED for lack of a better term.

David, can you please take care of that?

Comment 3 Alexandru Palalau 2013-10-14 14:00:14 UTC

Switching to verified in order to get it out of QA "To verify" bug list.