So... Use PM and threats to secrecy

So I just updated the replacement r620 from < 1.47 to 2.60 and used HTML5 console.

As was written here earlier the path is
* Do every step up to 1.66.65, in my case, 1.46, 1.57, 1.66.65.
* Update Lifecycle from every step up to 1.4.2.12, in my case 1.3.1.13, 1.4.0.128, 1.4.2.12.
* Update BIOS to the latest every step. I chickened out here and updated every step, there may be better paths.
* Install Lifecycle in the following order
** 2.10.10.10
** 2.15.10.10
** 2.20.20.20
** 2.21.21.21
** 2.30.30.30
** 2.60.60.60

Now, under virtual console you can choose type of console, pick HTML5.

When choosing what to download on dell.com, choose the win32 exe, upload in iDRAC
Except for BIOS, I just downloaded the standalone EXE and used unzip to unpack the image inside.

Now I have to wait until the 12th for my kit to arrive so I can test this out on the broken MB.

bios update should be fine, looking forward to hearing from you on this!

A few flashing and servers later, I believe that stepping through each firmware is paranoid, albeit justified. I have not had issues in bringing the server to 1.66.65 for iDRAC and the latest 1.x BIOS, then upgrading iDRAC to 2.10.10.10 and the BIOS to the earliest 2.x available. Once at that point, it becomes possible again to use the Lifecycle Controller, which updates straight to the latest and greatest available for both.

I'm also struggling with a dead IDRAC but on a T620. I'm struggling to find the UART connections -- any chance you could post a pic or describe where they are on the motherboard?

I don't have a system like that, but you should still be able to find 4 pins aligned around the area where the iDRAC connector and its CPU are.

Hi. I also have an iDRAC initialisation error on my R620 server. Bricked it yesterday on attempt to update FW. Take a look on this picture, I marked the position of UART interface on the MB. They didn't soldered any connector, and I think I'm going to solder there 4 pins to be able to connect.

Hi. Two days ago I have done an experiment. I soldered out the 25DF321 FLASH memory of the iDRAC and tried to read it out using a programmer. To my surprise it was completely unreadable. Moreover, the programmer wasn't able to recognise the type of the IC. At this point I looked at my warehouse for some old motherboards and found that some of them have a 25L64 FLASH memory as a BIOS chip. Which is suppose to be exactly the same as 25L32, but it has a twice as much memory. So I soldered it out and was able to write into this chip the R620 iDRAC dump (Nannerkins provide it to us in some of the previous posts). And soldered it into to the server's board. Now when power was connected to the server, iDRAC's indicator lamp on the back panel started to blink constantly with an amber light. My first thought was that I solved the problem and it just might need to make a reset to the iDRAC because before I changed the chip iDRAC was bricked completely, and there was no light on the back panel at all. To my disappointment it wasn't true - iDRAC still does not respond to long press and the amber light continues to blink.
My next steps would be:
1. Try to use exactly the same FLASH type(25DF321) which I have already ordered from EBAY
2. Try to find another R620 and read the memory dump out of it.
3. Try to connect to UART interface and see if there any activity.

Below you can see some pictures describing desoldering an soldering process

What software did you use to read/write the 25L64 and what is it's full name? If it's W25L64FV (Choose BV and use software 1.30, if BV try FV instead)

Its full name is 25L6406E. I have a USB MiniPro Universal programmer. You can find it on EBAY. It has a pretty large list of supported chips.
As for FLASH itself, I prefer to wait until I'll get 25DF321A and a clips socket adapter for in-system programming that I ordered and give it a try again. Moreover, it might be that I'm going to get the access to another R620 server to read out iDRAC BIOS. I think it will be much better to try this first. PCB don't likes multiple soldering/soldering, and there is no option for SOIC8 adapter that can be installed between PCB and IC to allow the installation without soldering(ike you have it for ICs with DIP body type).

SOIC8 test clip cable is for BIOS chips that are soldered on, not for DIP BIOS. So once you have one you wont need to desolder chips.

This thread has been interesting, but on my r420 what I'm finding is that the idrac spi chip may indicate that the real meat and potatoes isn't in the 25Q03213 chip at all (on my board that's what it was).
I was able to read it with the gold CH341A, and it verified fine (after pressing down firmly a few times and using a magnifying glass to make sure both sides had pin contact).

I copied a unpatched chip, then my new patched chip. Off the bat the 0-FFFFF portion of both of those is identical.

In the next section starting at 100000 all the configuration was identical, except for 4 mac addresses.

These settings in particular make me wonder if the actual firmware isn't stored on /dev/mmcblk0p2 (the sd card inside the server)?
hostname=idrac7
bootnfs=tftp 82000000 $(bootfile); run nfsargs; bootm
runtime_args=mode=vhard reset_cause=ac nmi_buf=0x83000000
fwudev=evb
bootargs=root=/dev/mmcblk0p2 rootwait rw rootfstype=squashfs mem=239616k console=ttyS2,115200 <NULL> mac1=00:00:00:00:00:01 mac2=00:00:00:00:00:02 mode=vhard reset_cause=ac nmi_buf=0x83000000 quiet
fwu_kernel_start=1
fwu_rootfs_start=8001
fwu_platform_start=120006
fwu_uboot_start=3fc01
fwu_kernel_size=135d
fwu_rootfs_size=19330
fwummc=emmc
ver=U-Boot 2009.08 (Jun 03 2013 - 03:47:19) Avocent (0.0.3) EVB

Has anyone actually unbricked the idrac with the bios only flash approach? My next guess is to image the sdcard and go to those offsets and see if I can find the same bytes in the firmimg.d7 file (and if so then I'll try copying the sdcard out of the working dell).

How about the jtag port, have any of you had luck getting into that or the SPI_DBG port instead of the regular rs232 redirect?

This appears to be the emmc, as it doesn't change even if I pull out all the sd cards.

That said, I broke down and bought a new mb from ebay, copied it's chip, then updated it completely, and recopied it's chip.

I then took that and flashed it back to my broken server that wouldn't enter recovery mode and at least now I have the flashing yellow light.

I soldered the tty pins and connected with putty, and I was happy to see this:
######################################################################
** No bootable iDRAC image is found **
** (System Health & ID LED is flashing amber at ~1/2 second rate).
To Recover iDRAC via an SD card.
1) Format SD using FAT on a Windows box or EXT3 using Linux.
2) Copy 'firmimg.d7' to root path.
3) Insert SD card.
4) System Health & ID button solid amber durning recover.
5) Both boot paths are flashed.
######################################################################

Polling for SD card state change

However I can't find any firmimg.d7 that will work.
I've tried 1.31.30, 1.35.35, 1.40.40, 1.65.65, 1.66.65, and 2.10.10.10 and they all pretty much do something like this:
Found '/firmimg.d7' in 'FAT' file system


UTIL RECOVER:Transport:sd TargetMMC:EMMC File:/firmimg.d7
reading /firmimg.d7

95344358 bytes read
UTIL RECOVER:SD load passed from FAT fs.
UTIL RECOVER:Transport time [sec:mil]: 16:525

Clear OS images in partition/s.
Clear kernelN, rootfsN, ubootN
'EMMC' blk size=[0x200][512] Erase to 0xffffffff
Mem buf size=[0x06000000][100663296] Total bytes=[0x08000000][134217728]
blocks[0x1:0x40001][1:262145]
mmc write failed000] Buf[0x88000000]..fill buffer..erase/write
util flash:
Failure writing blocks [1:196609] RC=0xffffffac
/nRecover returned failure, RC=0xffffffff
Firmimg.d7 might be bad.
Polling for SD card state change

The upgrade path for the ebay server indicated the original idrac upgrade was from 1.40.40 to 1.65.65, so I expected 1.40.40 to work. When it finally got to this state I was able to get into F2 (but not do anything), at least I could see the settings were 1.65.65, but alas that doesn't work any better.

Now I'm back to being stumped. I suppose I'll try flashing the post updated chip and then see if it takes 1.65.65...

Hi,
Last time when I wrote here I said that I was able to program iDRAC BIOS. After that it's indicator started to blink with an amber light. By that time I didn't know that this indicates "pooling for MMC card" status. Only after I had a UART adapter connected I found out what's going on behind the scene. First, I was glad, and thought that I'm one-step from the solution. But finally all my experiments are ended up exactely as yours. No matter what firmimg.d7 file version tried.
What looks confuzing for me, is that string, where it says "Firmimg.d7 might be bad", which comes right after "Failure writing blocks [1:196609] RC=0xffffffac
/nRecover returned failure, RC=0xffffffff"
Here a qustion arise in my head: Either a "firmimg.d7" has to be some sort of "recovery version" which is not a part of a regular firmware package and should be downloaded hell knows where from. Or this is an eMMC chip failure.

Polling for SD card state change \
SD card state change, previous=0x700 new=0x780


######################################################################
** No bootable iDRAC image is found **
** (System Health & ID LED is flashing amber at ~1/2 second rate).
To Recover iDRAC via an SD card.
1) Format SD using FAT on a Windows box or EXT3 using Linux.
2) Copy 'firmimg.d7' to root path.
3) Insert SD card.
4) System Health & ID button solid amber durning recover.
5) Both boot paths are flashed.
######################################################################


Found '/firmimg.d7' in 'FAT' file system


UTIL RECOVER:Transport:sd TargetMMC:EMMC File:/firmimg.d7
reading /firmimg.d7

62953181 bytes read
UTIL RECOVER:SD load passed from FAT fs.
UTIL RECOVER:Transport time [sec:mil]: 10:987

Clear OS images in partition/s.
Clear kernelN, rootfsN, ubootN
'EMMC' blk size=[0x200][512] Erase to 0xffffffff
Mem buf size=[0x06000000][100663296] Total bytes=[0x08000000][134217728]
blocks[0x1:0x40001][1:262145]
mmc write failed000] Buf[0x88000000]..fill buffer..erase/write
util flash:
Failure writing blocks [1:196609] RC=0xffffffac
/nRecover returned failure, RC=0xffffffff
Firmimg.d7 might be bad.
Polling for SD card state change \

I would have thought so too, but what are the odds both our healthy systems would have failed in exactly the same spot. Plus, googling that I found another person recently posted that same error to pastebin. I confirmed the crc on the download from Dell, so I know my firmware isn't corrupt. it does seem the emmc fails to write though.

I reflashed the post-updated chip to my other server and it's still doing the same thing, I did see something in the bios that's probably something I need to consider.

iDRAC Settings version....1.65.65.04
iDRAC Firmware Version...2.61.60 (Build 8)

Maybe I should try giving it 2.61.60.

This particular server killed itself when I connected the lifecycle controller to the internet and told it to just download all the updates from ftp.dell.com

Which bugs me to no end, but that aside, it clearly downloaded a 2.61.60 build that I haven't come across yet...I'm not sure why I didn't see that when I was at work but I guess I'll try that tomorrow. But maybe the problem is that the settings don't match the bios.

In that case I'll make sure to update the other server to 2.61.60 and then dump the idrac and write that...they sure make it hard to back out of this situation.