Considering giving up on X220: HDD not always found during boot

[markstos]

Hello,

I've got an X220 with i7 motherboard. I bought it used on eBay to run Linux on.

I found two problems with that prevent may from using it and seem like they may be hardware-related, so the laptop has seen little use and I'm considering giving up on it now.

The first issue that every few boots-- more than 1 in 10-- the BIOS fails to find the HDD and then can't boot. The current drive in it is a Samsung SSD 840 Pro. The error is "2100: Detection ereror on HDD0 (Main HDD)". I also tried an Intel SSD in it and HDD (possibly the stock HDD). This seems like it must be hardware related. If my memory is correct that this has happened with multiple drives, it it seems like the issue is likely to be the laptop and not with the drives.

The second problem is that I've had trouble getting the laptop to resume from suspend (with Linux) reliably. Sometimes it resumes, but not always. I've run Linux on many Thinkpads and suspend/resume has generally been solid in recent years with the exception of this machine. I believe I tried Ubuntu 12.04, 14.04 and Arch Linux and they all had problems. This /could/ be a Linux compatibility issue, but it strikes me that it could also be the issue I have with booting: If there's a problem bootstrapping the electrical connection to the drives, that could affect resuming as well as booting.

Does this sound like I got a lemon or is there anything I might to do make this laptop reliable?

I also installed an after-market battery the same time I got the laptop, so perhaps that's contributing some how as well.

Thanks,

Mark

[ajkula66]

Welcome to the forum!

My advice would be to try another SSD, and preferably from a different manufacturer. The chances are that the problem will go away, right then and there. SanDisk Extreme Pro would be my pick, but it's your money to spend. Stay away from any SF-based SSDs, though.

I can't really offer any wisdom when it comes to your second question. Hopefully someone with deeper knowledge of *nix than myself will chime in.

[mpcook]

Two obvious things to check, one, make sure you are booting in AHCI mode (with drivers installed). And make sure drive is carefully and fully seated. I know little about linux, except what I remember from my unix days. One way to eliminate HW as the issue is to run a drive with Win7 and see if the same issues crop up.

[brchan]

I also have an 840 PRO drive. It has worked in all of my thinkpads (T60, 2 T61s, W530) that use the standard sata drives. I also used arch linux on every one of my machines, and standby worked out of the box on all of them. Since it seems that even with your other 2 drives the machine runs into the same problem, I am thinking of a loose motherboard sata connector or some kind of motherboard related issue. You could try restoring the bios defaults, but that probably won't solve it.

[RealBlackStuff]

Seems to be a common error, caused by incompatible driver-combinations.
https://forums.lenovo.com/t5/ThinkPad-S ... 8591#M9759

[markstos]

Thank for all the feedback and ideas! That was a warm welcome.

I'll consider all the options shared here and see if I can make progress.

Thanks.

[jaspen-meyer]

What happens when you boot the harddrive using a sata-to-usb attachment?

What version of bios are you running?

The 'change history' of bios updates https://download.lenovo.com/ibmdl/pub/p ... uj26us.txt
lists fixes to several similar problems:

1.37 "Fixed an issue where specific SATA devices might not be detected after reboot."
1.33 Fixed an issue where specific hard disks might not be detected after
resuming normal operation from standby/sleep state.
1.26 - (Fix) Fixed an issue that might not resume from suspend state with combination of
specific CPU and BIOS setting for Intel Hyper-Threading Technology.
1.25 - (Fix) Fixed an issue where SATA BIOS might return timeout error with
SATA compatibility mode.

Try running memtest, download from their site and then
dd if=Memtest86-6.3.0.iso of=/dev/sdX

[bit_twiddler]

You could just pick up a spinning hard drive for about $30 or so.
There are a lot of them available on ebay which are either new
or pulled from PlayStations.

That would help isolate any suspend/resume/detection problems in the
BIOS because a spinner is a lot dumber.

[markstos]

jaspen-meyer,

Good idea to check the BIOS changelog. I'm running BIOS 1.39, so I already the latest fixes.

bit_twiddler,

You are right that trying an HDD is easy and cheap. I found my old notes about trying that. I had tried a 500GB Hitachi drive. I had a major pain dealing with UEFI vs BIOS issues and Master Boot Record vs GPT issues, although those were just a detour. I got Ubuntu 14.04 installed and bootable on it. I suspended and resumed it 4 times successfully. On the 5th time, It spewed errors about not being able to access the disk, which was the same kind of thing that was happening with the SSDs.

The first SSD I had tried with it was an Intel 520. With that, I got errors upon resume like: ""ata1: COMRESET failed (errno=-32)".

Here's a recent "resume" error when using the Samsung 840 Pro SSD with Ubuntu 14.04:

ata1: SError: { DevExch }
ata1: SError: { RecovComm PhYRdyChg CommWake DevExch }
ata1: exception Emask 0x10 Sact 0x0 SErr 0X400000 action 0xe frozen
ata1: irq_stat 0x00000040, connection status changed

All my issues seem to have in common fundamental connection problems with the hard drive, across three brands, and SSD vs HDD. It's difficult to think it's a "seating" issue after so many cycles of pulling drives in and out. I believe there were even different drive rails in use between SSD vs HDD. (plastic vs rubber, I think).

I'm going to give all the ideas above a second pass and see what's left to try.

Thanks for all the suggestions and feedback!

[bit_twiddler]

I don't know much about the disk driver, but you might want to look at:

https://lime-technology.com/wiki/index. ... ive_Issues

especially driver interface issue #4.

According to the author, the lack of a CRC error usually indicates a bad cable.
There aren't any cables in your laptop, but you might have a bad trace or solder
joint on your motherboard.

[jaspen-meyer]

For what it's worth, I just suspended 10x without fail with my x220 and xubuntu 14.04 and a
samsung evo 840 drive. Also, there are a bunch of people on this forum running linux on the x220 and I haven't read many complaints - the machine is very compatible with modern versions of linux.

You mention GPT trouble. gdisk is a useful program.

Code: Select all

sudo gdisk /dev/sdX

and then 'v' will tell you if there are problems with the disk or gpt table.

Could pick up a motherboard on ebay.

[markstos]

jasper,

I've also had many great experiences with Ubuntu on a number of Thinkpad models, with few problems. That's another reason I thought it might be a hardware problem.

Thanks for letting me know that Ubuntu and a Samsung drive on working your model!

When I was researching my issue, it seems there were some issues now that were specific to the i7, like this one: https://bugs.launchpad.net/ubuntu/+sour ... ug/1233794

Do you have an i7 by chance?

I see what motherboard prices are like on eBay.

Mark

[rjwilmsi]

As you've replicated the issue with multiple hard drives (SSD or HDD) then it's not a disk issue. You're on the latest BIOS so not a known firmware issue. Obviously Linux generally works on X220 etc. So yes, sounds like a connection/mobo/soldering issue (on my X200T I have intermittent issues with Wi-Fi card not being detected on boot).

I would try two things
* memory test as suggested by others (probably not the issue but always worth checking)
* physically cleaning/dusting the SATA port carefully using a vacuum cleaner/dry brush.

If that doesn't solve it, I would take advantage of the fact that the X220 has an mSATA slot, I would buy an mSATA drive and test that. If it doesn't solve the problem you should be able to resell the drive for a net cost of only a few $ plus shipping.

I think buying/testing an mSATA drive would be a lot faster and easier than swapping the motherboard.

[axur-delmeria]

I think buying/testing an mSATA drive would be a lot faster and easier than swapping the motherboard.

As someone who has already installed mSATA SSDs on three X220's, and removed the system board on at least one of them, I agree completely.

[markstos]

This forum is so helpful!

I took out the SSD and shined a flashlight down on the connectors. I saw a tiny spec of something near the soldering points for the SATA connector and removed that. There was also a small bit of hair near the SATA connection, and I removed that.

I put the laptop back together and started to run boot tests again. On the 9th boot test the BIOS failed to recognize the SSD. The problem persisted.

I ordered a new i5 motherboard for $40 on eBay and am going to try replacing that. Hopefully the combination of the new mobo and switching from i7 to i5 will solve the issue with the SSD being recognized.

If this works, I think will be a better deal for me in the end then selling the laptop in a broken state.

I've got some travel coming up, but when I get this project done I'll share the results of trying the new motherboard here!

Thanks again.

[RealBlackStuff]

FYI: in case you need it, X220 i5 mobos do NOT have a USB3.0 port

[jaspen-meyer]

Do you have an i7 by chance?

model name : Intel(R) Core(TM) i5-2410M CPU @ 2.30GHz

[markstos]

FYI: in case you need it, X220 i5 mobos do NOT have a USB3.0 port

Thanks for the tip. I noticed that in the specs. USB 3.0 is a nice-to-have for me, but not a must-have.

[markstos]

I got the new i5 motherboard and installed it. Installation went fairly smoothly.

However, when I try to boot it. nothing comes on the screen, I just get a 1-3-3-1 beep code. Some references refer to it as: "One short beep, pause, three short beeps, pause, three more short beeps, and one short beep."

Online documentation say to check the DIMMs and System Board if this beep code happens. The DIMM cheaps were the same as before, so I doubt they are the problem. I re-seated them for good measure, which didn't help. It seems my options are then:

1. The new motherboard is bad.
2. I didn't hook up the new board correctly somehow.
3. The BIOS has detected that the motherboard or motherboard model has changed and is refusing to boot with a different motherboard.

I'm hoping it's not option 3. Anyone have experience with this?

[bit_twiddler]

Do you have another machine that you can test the RAM in?

AFAIK, the BIOS is on the motherboard, so I don't see how that could be the problem.

[RealBlackStuff]

1-3-3-1 beeping is always a sign of bad RAM.

[markstos]

bit_twiddler and RealBlackStuff,

You were both right. I removed one of the DIMM sticks and the system booted and seems to work fine now. I had to run a "Boot Repair Disk" for Ubuntu, but that was easy enough.

I'll play around to see if I can get the troublesome DIMM to work in the other slot. But even with 4 GB of RAM instead of 8 GB, the laptop is quite usable.

Now I have to wonder if the troublesome DIMM played a part in the original problem (as some suggested a memory test, but I ignored that advice), or if something in the mobo switch to make it quit working.

The Bottom Line though is that the people here are forums.thinkpads.com SAVED MY X220!

THANKS!

[jaspen-meyer]

bit_twiddler and RealBlackStuff,

You were both right. I removed one of the DIMM sticks and the system booted and seems to work fine now. I had to run a "Boot Repair Disk" for Ubuntu, but that was easy enough.

I'll play around to see if I can get the troublesome DIMM to work in the other slot. But even with 4 GB of RAM instead of 8 GB, the laptop is quite usable.

Now I have to wonder if the troublesome DIMM played a part in the original problem (as some suggested a memory test, but I ignored that advice), or if something in the mobo switch to make it quit working.

The Bottom Line though is that the people here are forums.thinkpads.com SAVED MY X220!

THANKS!

Glad it's working.

Yeah, run memtest and be sure it's THE MOST CURRENT VERSION!!
Pay special attention to how forcefully I'm trying to tell you to use the current version (i.e not the default version which comes with debian et al).
Instructions for copying to usb are above and here's some inspiration for you, a snap shot of what I was doing last week


It's a free and easy first step when debugging hardware. Second step is to disconnect all peripherals.

btw, you can (likely) use the machine with the bad ram.
Someone here may have a better how to on the topic but what I do is:
1. ground myself [loop]periodically touch a piece of metal[/loop]
2. figure out which dimm is failing
3. remove and clean contacts of failing dimm using cotton dipped in Isopropyl alcohol
4. inspect dimm slots using magnifying glass and lots of light
4b. [as needed] fix dimm slot
5. reinsert dimm and hope thinkpad screen shows up

Code: Select all

if [no boot screen]
then try other slot && double check everything
fi

6. run mem test
7. set mode to report with badRAM format
8. plug the badRAM values into /etc/default/grub
9. `sudo update-grub`
10. done

edit: basic overview of 'badRAM' is that the command line arguement tells the boot kernel to tell the actual kernel to have root reserve the bad block(s) of ram and NEVER let them go! Pretty cool.

[rjwilmsi]

I have an X200s installed with 2 x 4 GB RAM. One of the DIMMs has an intermittent error at a single address, so I have configured grub with memmap command so the kernel ignores 10 KB of memory from that address upwards.

However, firstly to use that you would need to have a faulty DIMM with a consistent, relatively small set of bad addresses (probably need to run memtest several times to be confident of that). We don't yet know if in this case the DIMM has a small number of errors or thousands. A friend's machine had a bad memory DIMM that produced hundreds of errors per second in memtest, so no hope of using memmap for that. Secondly, I believe not all Linux distributions have memmap/badram enabled/compiled in their kernels, so have to confirm compatibility.

Frankly, with the cost of a used 4 GB DDR3 Laptop DIMM under GBP 15 on eBay etc., it's probably a better use of time/effort/money to replace the DIMM (and take faulty one to electronics recycling).

[jaspen-meyer]

However, firstly to use that you would need to have a faulty DIMM with a consistent, relatively small set of bad addresses.

That's false. The kernel can, and does, assign as many segments of ram as it wants.
Further, the newer versions of MemTest are able to export results to disk. The number of bad segments is not relevant.

[rjwilmsi]

Sure, I didn't mean that was a limitation of memtest/badram/memmap, but rather in my view if a DIMM has a large volume of bad addresses I would not be confident that more bad addresses wouldn't appear soon and would not be confident that one pass of memtest would be certain to identify all the current bad ones, so I would not use the DIMM at all. Though yes, as you say, you could continue to use a DIMM with thousands of addresses blacklisted by badram/memmap.

[bit_twiddler]

Advanced features like bad ram mapping are really most useful in a server/workstation environment
in conjunction with ECC, chipkill, and other error detection/correction features.
You're better off just chucking any non-ECC ram that gives you any errors at all.

I seem to recall that the author of memtest recommended 7-8 passes to test all possible
combinations. Anyway, a week of burn-in of non-premium RAM is a good idea to get through
the first part of the "bathtub curve."

[markstos]

The replacement i5 motherboard had a brief few days of great stability, but now it too has serious problems.

It started with a sudden loss of power. Not like a freeze or OS crash, but the power cutting out.

I tried removing the battery and only running off the power cord. I tried running it off the battery but not the power cord. I tried running it on both.

Unprovoked crashes reminded me of experiences with bad memory, so I ran 4 passes of memtest overnight, and got a 100% pass rate.

The next day the crashes were increasingly frequent, to the point where were the machine was only running or a minute or two before cutting out. A couple times it even gave a 1-3-3-1 beep code (indicating a mobo or memory problem) and it didn't boot at all.

The increasing frequency of the crashes made me think the crashes might be temperature related, so I let it rest overnight.

The next morning, I quickly installed "psensor" (for Ubuntu), but before I could even launch it, the power cut out again. I rebooted and launched 'psensor' just long enough to see temperature readings in a normal range (about 50C). Nothing near the 100 degree limit the mobo is rated for.

SOO, after my problems with the last motherboard, this is pretty disheartening. I've bought a string of used Thinkpads in the past and had fairly consistently great results. This experience just about has me ready to head to Best Buy...

I'm fairly certain the issue is not with Linux. I think sometimes it cuts out even before the OS gets a chance to boot. It definitely dies no matter what's happening. The cut-outs don't seem to be triggered by any software action.

I see a couple possible problems that could exist:

* It really is a memory problem, but is a rare case that memtest didn't catch. If it will stay running, perhaps I should try running mem pass for another 10 or passes overnight.
* I hamfisted the motherboard swap and there's some bad connection. I try re-seating all the connections.

I guess I know what options to try, but at this point I feel like running for the hills (or Best Buy, as it were).

Exorcism advice welcome.

[brchan]

Are you using a genuine Lenovo power supply or an aftermarket one? Aftermarket or fake thinkpad power supplies (See: https://ithinkpad.wordpress.com/2012/10 ... c-adapter/) are notorious for causing problems and even damage to laptops.

I would also make sure the mobo is grounded and that there are no loose screws or objects that would cause shorts, but since the unit was working fine at first, this probably isn't the issue.

[jaspen-meyer]

Are
you using the current version of memtest (6.3) or the dated version
which appears in the grub menu of Ubuntu installs?

How many watts (V*A) is the adapter rated for?

You've got a lemon. Disappointing but heading to Best Buy won't get you a
decent thinkpad. Head back to Ebay or place an ad in the Marketplace
in this forum for an x220 with ram and an adapter.